Glutathionylation of Trypanosomal Thiol Redox Proteins*

Trypanosomatids, the causative agents of several tropical diseases, lack glutathione reductase and thioredoxin reductase but have a trypanothione reductase instead. The main low molecular weight thiols are trypanothione (N1,N8-bis-(glutathionyl)spermidine) and glutathionyl-spermidine, but the parasites also contain free glutathione. To elucidate whether trypanosomes employ S-thiolation for regulatory or protection purposes, six recombinant parasite thiol redox proteins were studied by ESI-MS and MALDI-TOF-MS for their ability to form mixed disulfides with glutathione or glutathionylspermidine. Trypanosoma brucei mono-Cys-glutaredoxin 1 is specifically thiolated at Cys181. Thiolation of this residue induced formation of an intramolecular disulfide bridge with the putative active site Cys104. This contrasts with mono-Cys-glutaredoxins from other sources that have been reported to be glutathionylated at the active site cysteine. Both disulfide forms of the T. brucei protein were reduced by tryparedoxin and trypanothione, whereas glutathione cleaved only the protein disulfide. In the glutathione peroxidase-type tryparedoxin peroxidase III of T. brucei, either Cys47 or Cys95 became glutathionylated but not both residues in the same protein molecule. T. brucei thioredoxin contains a third cysteine (Cys68) in addition to the redox active dithiol/disulfide. Treatment of the reduced protein with GSSG caused glutathionylation of Cys68, which did not affect its capacity to catalyze reduction of insulin disulfide. Reduced T. brucei tryparedoxin possesses only the redox active Cys32-Cys35 couple, which upon reaction with GSSG formed a disulfide. Also glyoxalase II and Trypanosoma cruzi trypanothione reductase were not sensitive to thiolation at physiological GSSG concentrations.

The reversible S-glutathionylation of specific cysteine residues is a recently discovered mechanism for the regulation of redox-sensitive thiol proteins and is the preponderant mode of redox signal transduction (1). Other functions of glutathionylation are the protection of proteins from irre-versible overoxidation of cysteine residues and the formation of one GSH from one GSSG molecule without NADPH consumption. Accumulation of glutathionylated proteins under oxidative stress conditions has been reported for different cell types (1)(2)(3). In addition, glutathionylation of abundant proteins under basal conditions may serve as a glutathione store (4). The specific mechanisms causing formation of protein-SSG intermediates are largely unknown. In contrast, it is well established that glutaredoxins (thioltransferases) are specific and efficient catalysts of protein-SSG deglutathionylation (3,(5)(6)(7).
Trypanosomes and Leishmania are the causative agents of severe tropical diseases such as African sleeping sickness (Trypanosoma brucei gambiense and Trypanosoma brucei rhodesiense), Chagas disease (Trypanosoma cruzi), and the different forms of leishmaniasis (Leishmania donovani, Leishmania major, Leishmania mexicana). All of these parasites have in common that the ubiquitous glutathione reductase is replaced by a trypanothione reductase. The flavoenzyme maintains trypanothione (N 1 ,N 8 -bis(glutathionyl)spermidine) and glutathionylspermidine (Gsp) 2 in the reduced state (8,9). The parasite-specific trypanothione is synthesized from glutathione and spermidine in two consecutive steps. In the first reaction, Gsp is formed, which reacts with a second glutathione molecule to trypanothione (10,11). The cellular concentration of Gsp in T. brucei is about 50 M (12). In the insect parasite Crithidia fasciculata, the Gsp concentration is higher and can rise up to 2.5 mM in the stationary phase. In addition, the ratio between trypanothione and Gsp strongly depends on the growth stage of the parasites (13). Gsp, but not trypanothione, is also formed in Escherichia coli when the bacteria enter the stationary phase (14). Although trypanothione forms the basis of the parasite thiol metabolism, trypanosomatids contain also significant levels of free glutathione. In logarithmic and nearly stationary culture forms of bloodstream T. brucei, the concentration of free glutathione is 1.2 mM and 170 M, respectively (15), which is even higher than the respective trypanothione concentrations (340 and 100 M). In T. brucei isolated from infected mice, as well as in cultured procyclic parasites, the GSH concentration is 200 -300 M (12). Promastigote L. donovani contain up to 1.8 mM GSH (16). Thus, the question arises whether the parasites employ glutathione only for the synthesis of glutathionylspermidine conjugates or also for the glutathionylation of thiol redox proteins.
T. brucei mono-Cys-glutaredoxin 1 is a glutaredoxin-like protein with a CAYS motif replacing the CPYC active site of classical glutaredoxins. The biological function of the protein is not yet known. Related proteins from other organisms form mixed disulfides with glutathione (7,17,18), which may indicate a function in redox signaling. Yeast cells that are deficient in mono-Cys-glutaredoxin 5 (Grx5) are more sensitive toward oxidative stress and the protein has been shown to be involved in the biosynthesis of iron sulfur clusters (19). T. brucei peroxidase III is a cysteine homologue of the classical selenocysteinecontaining glutathione peroxidases (20). The enzyme catalyzes the trypanothione/tryparedoxin-dependent reduction of a variety of hydroperoxides and is essential for both the mammalian and the insect stages of T. brucei (21,22). T. brucei thioredoxin catalyzes reactions typical for thioredoxins such as the reduction of ribonucleotide reductase and insulin disulfide (23). The parasite thioredoxin is probably kept reduced by the spontaneous reaction with trypanothione, because the completely sequenced genomes of T. brucei as well as T. cruzi and L. major do not reveal any gene for a thioredoxin reductase (24). Tryparedoxins are small parasite-specific dithiol proteins with a CPPC active site motif. They belong to the super family of proteins with a CXXC motif, which comprises thioredoxins and glutaredoxins (25). T. brucei tryparedoxin transfers electrons from trypanothione onto different peroxidases (see above), as well as ribonucleotide reductase (26), and is involved in the parasite replication of mitochondrial (kinetoplastid) DNA (27,28). T. brucei glyoxalase II catalyzes the hydrolysis of lactoyltrypanothione, which is formed from methylglyoxal and trypanothione (29). Trypanothione reductase, the key enzyme of the trypanothione metabolism, is essential for the parasites (30). It is responsible for maintaining a reducing intracellular milieu by catalyzing the NADPH-dependent reduction of trypanothione disulfide. Here we report on the thiolation of specific cysteine residues as revealed by ESI-and MALDI-TOF-MS analyses of the parasite proteins after treatment with GSSG or Gsp disulfide.
Purification of Recombinant T. brucei Mono-Cys-Glutaredoxin 1-5 ml Overnight culture of E. coli Nova Blue cells with the pQE30 plasmid containing the grx1 gene (without the mitochondrial targeting sequence and with N-terminal His 6 tag) 3 were diluted in 750 ml of 2ϫ YT (yeast tryptone) medium containing 100 g/ml carbenicillin. The cells were grown at 37°C to an optical density of 0.6. Expression was induced by adding 1 mM isopropyl-␤-D-thiogalactopyranoside overnight at 30°C. After centrifugation, the cells were suspended in 10 ml of buffer A (50 mM sodium phosphate, 300 mM NaCl, pH 7.2) containing 150 nM pepstatin, 4 nM cystatin, 20 M phenylmethylsulfonyl fluoride, 5 mg lysozyme, and 0.5 mg DNase I and disintegrated by sonication. Following centrifugation, the supernatant was applied onto a 15-ml TALON Superflow metal affinity resin (Clontech) column at 4°C. The column was washed with 75 ml of buffer A followed by 150 ml each of 10 mM and 20 mM imidazole in buffer A. The His-tagged protein was eluted with 250 mM imidazole in buffer A.
Determination of Free Thiol Groups-The number of free thiol groups in the reduced proteins was determined photometrically with DTNB (⑀ 412 ϭ 13.6 mM Ϫ1 cm Ϫ1 (34,35)). The reaction mixture contained 200 M DTNB and 10 -100 M thiol in a 1:1 mixture of 50 mM potassium phosphate, pH 8.0, and 50 mM ammonium bicarbonate/HCl, pH 7.5.
Reaction of Reduced Thiol Proteins with the Disulfide Forms of Glutathione, Gsp, and Trypanothione-10 -50 M reduced protein in 50 mM ammonium bicarbonate/HCl, pH 7.5, was incubated at room temperature for 4 h with 0.25-5 mM GSSG, 0.5 mM Gsp disulfide, and 0.5 mM TS 2 . For assigning the binding site of Gsp to Grx1 by MALDI-TOF-MS, the protein was treated with 0.5 mM Gsp disulfide in 50 mM Tris/HCl, pH 9.0. Completely modified peptides as the standard for quantification were generated by incubating the reduced proteins with 20 mM GSSG or Gsp disulfide for 48 h at 4°C.
Proteolytic Digestion of Proteins-Fully reduced proteins in 50 mM ammonium bicarbonate/HCl, pH 7.5, were treated with different concentrations of low molecular weight disulfides as described above. Iodoacetamide was added to a final concentration of 9 mM to alkylate residual free cysteine residues. The modified and unmodified proteins were digested with trypsin (T. brucei Grx1, T. brucei glyoxalase II, T. brucei tryparedoxin, T. cruzi trypanothione reductase), GluC (T. brucei Grx1, T. brucei PxIII, T. brucei glyoxalase II), chymotrypsin (T. brucei Trx), and LysC (T. brucei Trx) at a protease:protein ratio of 1/10 (w/w). The reaction was allowed to proceed for 2 h at 37°C, and the resulting peptides were analyzed by MALDI-TOF-MS as described in the next section.
Mass Spectrometric Analysis-The intact masses of modified and unmodified proteins were determined by ESI-MS on an API-QSTAR TM Pulsar instrument (Applied Biosystems) with a high pressure liquid chromatography system (HPLC; Agilent) on-line-coupled to the ESI-QTOF instrument. 100 l of a 1 to 10 M protein solution diluted with water was loaded onto a 50 Poros R1 trapping column. After 1.5 min of washing (0.1% trifluoroacetic acid, 0.4 ml/min), the protein was eluted into the electrospray ion source with 80% acetonitrile, 0.1% trifluoroacetic acid at 20 l/min essentially as described in Ref. 36. The QTOF instrument was calibrated with apomyoglobin (Sigma). The mass accuracy was better than 100 ppm. For relative quantification of thiolation, mixtures of known composition were evaluated, changing the amount of thiolated protein against the unmodified protein and vice versa (see supplemental Figs. 1 and 2). Peptides were analyzed by MALDI-TOF-MS on a Bruker Ultraflex mass spectrometer. The peptide solution in 50 mM ammonium bicarbonate/HCl, pH 7.5, was desalted on a ZipTip C-18 column (Millipore). The MALDI-matrix was prepared as a saturated solution of ␣-cyano-4-hydroxycinnamic acid in acetonitrile, 0.1% trifluoroacetic acid (1:1 v/v). A nitrogen laser (337 nm) was employed for desorption/ionization, and the ion acceleration voltage was 20 kV. The spectra from 400 laser shots were averaged. The instrument was calibrated externally with peptide standard II from Bruker, resulting in a mass accuracy of 100 ppm in the range up to 4000 Da. For higher masses a slightly decreased accuracy was observed (150 ppm). Data were compared with the theoretical masses of the proteins or peptides taking into consideration the possible modification of cysteine residues. Relative peak intensities of the different forms of a peptide were calculated by comparing the peak areas given by the SNAP (sophisticated numerical annotation procedure) algorithm of the program flex analysis (version 2.2) from Bruker, which searches for known patterns in the measured spectrum and performs its own internal base-line correction and noise determination. The SNAP algorithm calculates the monoisotopic peaks based on a fitting of the isotopic distribution of the profile spectrum placing isotope peaks within millidaltons of their true centroids. This procedure, which uses Fourier transform methods, allows the separation of overlapping and low-resolved isotopic patterns and a precise determination of the monoisotopic masses as well as the areas of the separated isotopic patterns. Suppression or enhancement of a certain peptide by thiolation was analyzed by measuring the unmodified and completely modified peptide alone and in mixture. Then the peak areas of modified or unmodified peptides were compared with the peak areas of peptides containing no cys-teines as internal standards. For peptide sequencing by MALDI-TOF-TOF-MS, peptide fragments were generated by post-source decay, and fragment spectra were acquired on a Bruker Ultraflex mass spectrometer using the LIFT method. The precursor mass window was 1% of the mass of the parent ion.
pH Dependence of the Reaction of T. brucei Mono-Cys-Glutaredoxin 1 with GSSG or Gsp Disulfide-Reduced Grx1 was prepared as described above. The protein was eluted from the PD10 column in 50 mM Tris/HCl, pH 9.5. The 50 M Grx1 solution was titrated with 50 mM MES to pH values between 9 and 5 (at lower pH values the protein precipitated). 0.5 mM GSSG or Gsp disulfide was added, and the reaction mixture was incubated at room temperature for 4 h. The reaction was stopped by freezing the sample, and the intact mass of the protein was determined by ESI-MS. The degree of thiolation was estimated by comparing the peak intensities of modified and unmodified protein species.
Formation of an Intramolecular Disulfide between Cys 104 and Cys 181 in Grx1-50 M Reduced Grx1 was treated with 2 mM GSSG, Gsp disulfide, TS 2 , or hydroxyethyl disulfide at room temperature for 24 h in 50 mM ammonium bicarbonate/HCl, pH 7.5. Excess low molecular weight reagents were removed by size exclusion chromatography on a PD10 column. In addition, reduced Grx1 was incubated with 100 or 500 M hydrogen peroxide at room temperature for 20 min, and the reaction was stopped by the addition of 50 units of bovine catalase. The thiol content of the protein samples was determined with DTNB as described above. Remaining free thiols were alkylated by adding 15 mM iodoacetamide, and the intact mass of the protein was determined by ESI-MS. The incorporation of carboxamidomethyl groups showed cysteine residues that are not involved in an intramolecular or a mixed disulfide. To further analyze formation of an intramolecular disulfide, the tryptic digests of the treated Grx1 samples were subjected to MALDI-TOF-MS and examined for a fragment containing the linked Cys 104 and Cys 181 peptides. The identity of the disulfide-forming fragments was confirmed by MALDI-TOF-TOF-MS peptide sequencing and by the fact that the respective peak disappeared after the addition of 15 mM DTT.
Assaying Potential Catalytic Activities of T. brucei Grx1-To analyze whether Grx1 can catalyze the glutathionylation of other proteins, 50 M reduced Grx1 in 50 mM ammonium bicarbonate/HCl, pH 7.5, was incubated with an equimolar concentration of reduced peroxidase III or thioredoxin and 0.5 mM GSSG. After 2, 5, 10, 20, 30 min, and 4 h, aliquots were subjected to ESI-MS. To measure a possible deglutathionylation activity, T. brucei thioredoxin was glutathionylated as described above, and excess GSSG was removed on a PD10 column. 10 M glutathionylated thioredoxin in 50 mM ammonium bicarbonate/HCl, pH 7.5 and 8.0, was treated with 10 M reduced Grx1. The ability of Grx1 to reduce protein disulfides was studied by incubating 50 M reduced Grx1 with an equal concentration of oxidized tryparedoxin in 50 mM ammonium bicarbonate/HCl, pH 7.5, followed by ESI-MS.
Deglutathionylation and Reduction of Grx1-To elucidate whether other thiol proteins catalyze the deglutathionylation and/or reduction of the intramolecular disulfide bridge of Grx1, 50 M reduced Grx1 in 50 mM ammonium bicarbonate/HCl, pH 7.5, was treated with 2 mM GSSG, which led to a mixture of glutathionylated protein and the protein with an intramolecular disulfide. 50 M reduced T. brucei tryparedoxin or 5 M D. melanogaster thioredoxin reductase and 200 M NADPH in 50 mM ammonium bicarbonate/HCl, pH 7.5, were added. After 10 min, 15 mM iodoacetamide was added, and the samples were subjected to ESI-MS. To reveal whether GSH and trypanothione are able to reduce the intramolecular or the mixed disulfide of the protein, Grx1 treated with GSSG or Gsp disulfide was incubated with either 3 mM GSH or 0.8 mM TS 2 , 2 M T. cruzi trypanothione reductase, and 4 mM NADPH in 50 mM ammonium bicarbonate/HCl, pH 7.5, and processed as described above.
Circular Dichroism Spectroscopy-50 M Reduced Grx1 was incubated with 2 mM GSSG for 24 h at room temperature and desalted on a PD10 column. CD spectra of reduced (14 M) and oxidized (10 M) Grx1 were recorded at 20°C from 195 to 300 nm on a JASCO J-810 spectropolarimeter in 5 mM potassium phosphate, 100 mM KCl, pH 7.5, with a 1-mm path length cuvette. Data were collected every nanometer with an averaging time of 2 s and a bandwidth of 1.5 nm averaging over four repeated scans. Thermal denaturations of reduced (112 M) and oxidized (20 M) Grx1 were followed at 222 nm by measuring the change in ellipticity at increasing temperatures from 20 to 95°C at a speed of 0.5°C/min. Data were recorded for each degree with an 8-s averaging time and a 1.5-nm bandwidth.
Reduction of Insulin Disulfide by Glutathionylated T. brucei Thioredoxin-T. brucei Trx was reduced as described above and incubated with 5 mM GSSG for 72 h at 4°C. Excess glutathione was removed by centrifugation through a 10-kDa cutoff Amicon filter. Reduction of insulin disulfide was measured essentially as described by Casagrande et al. (37). In a total volume of 800 l of 100 mM potassium phosphate, pH 7.0, 2 mM EDTA, the assays contained 160 M insulin, 200 M NADPH, and 1 or 2 M of glutathionylated and untreated Trx, respectively. The reactions were started by adding 535 nM human thioredoxin reductase, and the absorption decrease was followed at 340 nm.

Intact Mass Determination of Thiolated T. brucei Mono-
Cys-Glutaredoxin 1-Recombinant Grx1 was treated with GSSG and subjected to ESI-MS as described under "Experimental Procedures." The untreated protein showed a molecular mass of 17,330.1 Da (Fig. 1A). The reduced and oxidized forms of Grx1 have theoretical masses of 17,330.9 and 17,328.9 Da. The observed mass of the protein with and without intramolecular disulfide was 17,328.6 Ϯ 0.5 and 17,330.7 Ϯ 0.3 Da, respectively (mean Ϯ S.D. of seven intact mass determinations). Incubation of reduced Grx1 with 0.25 or 0.5 mM GSSG resulted in a peak of 17,636 Da (Fig. 1, B and C), which is compatible with the addition of one glutathione molecule (305 Da). With millimolar GSSG concentrations, two peaks appeared, in agreement with the incorporation of one and two glutathione molecules (Fig. 1, D and E).
Peak intensities can be used for the relative quantification of protein species if ion suppression is excluded. To elucidate whether the glutathionylation of Grx1 suppresses the ionization/detection of one protein species by the other in the mix-  (17,941 Da). The small peaks with slightly higher masses that accompany the main peaks are probably oxidation products of the protein. F, reduced Grx1 treated with 0.5 mM TS 2 . No mixed disulfide is formed. G, reduced protein after incubation with 0.5 mM Gsp disulfide. The small peak at 17,763.3 Da corresponds exactly to one Gsp molecule bound to Grx1 (17,330 ϩ 433 Da). The most prominent new peak, at 17,795.5 Da, has a 32-Da higher mass, which is probably caused by the additional incorporation of two oxygen atoms. The peak at 18,196 Da represents Grx1 modified by two Gsp molecules. The ESI-MS spectrum of the reduced unmodified protein shows 15 charge states from ϩ11 to ϩ25 with a maximum at ϩ21. Interestingly, in the samples treated with the low molecular weight disulfides, the charge distribution for the unmodified protein showed a maximum of ϩ15. This indicates that formation of the intramolecular disulfide bridge lowers the number of charge states. The mono-and bis-glutathionylated protein species occur in 13 charge states, from ϩ13 to ϩ25. Grx1 modified by one or two Gsp (G) molecules yielded 12 and 8 charge states from ϩ14 to ϩ25 and from ϩ18 to ϩ25; the maximum of all modified species of Grx1 was at ϩ21. The reduced number of charge states observed for the protein with two bound Gsp molecules is only due to the low amount of this species in the mixture. When Grx1 was treated with 20 mM Gsp disulfide, which resulted in the protein with two bound Gsp molecules becoming the main product, the same charge states as in the case of Grx1 modified by a single Gsp molecule were observed. ture, Grx1 was modified at both cysteine residues by treating the protein with 20 mM GSSG for 48 h at 4°C. Different ratios of unmodified and bis-glutathionylated Grx1 were mixed and subjected to ESI-MS. In the first series of experiments, the unmodified protein was kept constant, and bis-glutathionylated Grx1 was varied. The percentage of bis-glutathionylated Grx1 detected in the spectrum was depicted against the theoretical percentage of this species in the mixture (see supplemental Fig. 1A). In the second series, bis-glutathionylated Grx1 was kept constant, and the free protein was varied (see supplemental Fig. 1B). The slopes of the lines obtained were 0.96 and 1.02, respectively, which shows that glutathionylation of both cysteine residues in Grx1 does not significantly affect the ionization efficiency. Because Grx1 with two bound glutathione molecules is formed only at millimolar concentrations of GSSG, even under extreme oxidative stress conditions (with a cellular GSH/GSSG ratio of 2 (38)), glutathionylation should be specific for one cysteine residue. Treatment of reduced Grx1 with 0.5 mM TS 2 did not result in any protein modification (Fig.  1F). As expected, the intramolecular disulfide did not form stable mixed disulfides with protein thiols.
The ESI-MS spectrum of Grx1 treated with 0.5 mM Gsp disulfide is shown in Fig. 1G. The peak at 17,763.3 Da corresponds to the protein with one bound Gsp molecule (433 Da). The most prominent peak with a mass of 17,795.5 Da is consistent with the incorporation of one Gsp molecule plus an additional mass of 32 Da. Another minor peak at 18,196 Da corresponds to the protein with two bound Gsp molecules. As described above for the quantification of the glutathionylated protein, Grx1 was thiolated by two Gsp molecules, and different mixtures of free and modified protein were analyzed by ESI-MS. The slopes of the calibration lines were 0.74 and 1.25 when modified and free Grx1, respectively, were varied (see supplemental Fig. 1, C and D). This indicates that binding of two Gsp molecules to Grx1 lowers the ionization efficiency by about 25% and that the protein species with a mass of 17,795.5 Da is indeed the most prominent one in the reaction mixture (Fig. 1G). The intact mass peaks of all untreated and thiolated forms of Grx1 are accompanied by minor peaks with higher masses that probably reflect oxidation products ( Fig. 1, A-G). Interestingly, only in the case of Grx1 with one bound Gsp, the 32-Da higher mass (17,795.5 Da peak in Fig. 1G) is the most prominent peak. Several lines of evidence support the interpretation that this species represents the protein with Gsp bound at Cys 181 and Cys 104 being oxidized to a sulfinic acid. 1) The reactivity of both cysteine residues was lost when Cys 181 was modified by Gsp. Treatment of the protein sample with iodoacetamide caused a peak shift of 57 Da for the mono-modified protein species but not in the case of monomodified Grx1 with the additional 32-Da mass. 2) The peak at 18,196.4 Da matches the theoretical mass of Grx1 modified by two Gsp molecules, and a peak with an additional 32 Da was not observed for this species. 3) A protein species that lacks Gsp but carries the additional 32 Da did also not occur (see Table 4). The latter two species could occur if methionine residues were oxidized. Taken together, these data strongly suggest that the binding of one Gsp molecule to Grx1 leads to the oxidation of the second cysteine to a sulfinic acid. Treatment of Grx1 with 0.5 mM hydroxyethyl disulfide also generated monothiolated protein (not shown). Incubation of reduced Grx1 with 2 mM Gsp disulfide at room temperature for 24 h led mainly to the thiolation of both cysteines, which indicates that high concentrations of disulfide and long incubation times support this unspecific reaction (see Table 4).
pH Dependence of the Thiolation of T. brucei Mono-Cys-Glutaredoxin 1-Grx1 was allowed to react with GSSG and Gsp disulfide at pH values between 5.0 and 9.0. As described in the previous section and shown in supplemental Fig. 1, the binding of two glutathione or Gsp molecules affects the ionization and detection of Grx1 by less than 5% and by about 25%, respectively. This allowed us to estimate the relative amounts of modified and unmodified protein species from the ESI-MS spectra (Fig. 2). A minor thiolation of both cysteines was observed at pH 5.0, which probably reflects reaction of the unfolded protein, because recombinant Grx1 starts to denature at pH values Յ5.0. At pH 9.0, where cysteine residues should be present in the reactive thiolate state, almost exclusively monothiolated protein was obtained. In the case of Gsp disulfide, modification of the protein increased hyperbolically from pH 5.0 to 9.0. In comparison, incorporation of a glutathione moiety showed a maximum around pH 5.5 to 6.0, which then slightly decreased and remained constant up to pH 9.0. This was not due to the redox potential of the reagents, because glutathione and trypanothione have practically identical values (8). Probably the pK values of the leaving thiol groups play a major role. The pK of GSH is 8.2, and that of trypanothione is 7.4 because the secondary amine in the spermidine bridge favors deprotonation of the thiol groups (39). The pK of Gsp is not known. Because the spermidine moiety in Gsp has two positive charges, its pK value may be even lower than that of trypanothione. Gsp is probably a better leaving group than glutathione (38), which would favor its reaction with the protein kinetically. The higher degree of modification by Gsp may also be caused by favorable electrostatic interactions. Deprotonation of protein residues at FIGURE 2. pH dependence of the S-thiolation of Grx1. As described under "Experimental Procedures," reduced Grx1 was incubated with 0.5 mM GSSG or Gsp disulfide, and the extent of thiolation was estimated from the ratio of the peak intensities for the modified and unmodified protein species in the ESI-MS spectra. f, Grx1 with one bound Gsp; OE, Grx1 with two bound Gsp molecules; Ⅺ, monoglutathionylated protein; ‚, bis-glutathionylated protein.
high pH values may still allow an interaction with Gsp but weaken an interaction with the overall negatively charged glutathione. The fraction of protein with two bound glutathione or Gsp molecules was low at pH 5.0 and negligible at pH 9.0 ( Fig.  2). Therefore, over the whole pH range studied, thiolation of Grx1 occurs at a single site.
Mono-Cys-Glutaredoxin 1 Is Thiolated at Cys 181 -As outlined in the previous section, incubation of Grx1 with 0.25 or 0.5 mM GSSG leads to the modification of the protein by one GSH molecule (Fig. 1, B and C). T. brucei Grx1 possesses two cysteinyl residues at positions 104 and 181. To evaluate whether the Cys-containing proteolytic peptides are in principle detectable in modified form, the protein was treated with iodoacetamide, GSSG, or Gsp disulfide under conditions in which both cysteine residues were modified. The full-range spectra of the GluC-derived peptides of fully alkylated (see supplemental Fig.  3) and fully glutathionylated Grx1 (Figs. 3 and 4B) as well as the protein with two bound Gsp molecules (Fig. 4A) revealed that the peaks of the modified Cys 104 -containing peptides have much higher intensities than the Cys 181 -containing peptides. The peak intensity of the 90 -115 peptide with bound Gsp (m/z 3358.2) is more than an order of magnitude higher than that of the 167-184 peptide with bound Gsp (m/z 2481.6) in the same GluC digest (Fig. 4A). The same is true for the 90 -115 (m/z 3231.1) and 167-184 peptides (m/z 2354.3) with bound glutathione (Fig. 4B). To disclose whether inefficient cleavage of Grx1 by GluC is responsible for the low peak intensities of the Cys 181 -containing peptides, the completely modified Grx1 species were also digested with trypsin. Again, the peaks for the glutathionylated or alkylated Cys 104 -containing peptides had much higher intensities than the glutathionylated Cys 181 -containing peptides (not shown).
To elucidate which of the two cysteines is glutathionylated in the mono-modified protein, Grx1 was treated with 0.5 and 1 mM GSSG, the remaining free thiols were alkylated by iodoacetamide, and the protein was digested with trypsin or GluC (The sequences of the Cys-containing peptides are depicted in Table 1). Treatment of Grx1 with 0.5 mM GSSG caused glutathionylation of Cys 181 . The ions at 2049 and 2354 m/z correspond to the free and glutathionylated 167-184 peptide ( Fig. 5B and Table 2). These ions did not appear in the GluC digest of the protein that was treated only with iodoacetamide (Fig. 5A). Under these conditions, a glutathionylated Cys 104 -containing peptide was not observed (Fig. 6B; for the full-range spectrum In the case of the two Cys 104 -containing peptides, the relative peak intensities of the glutathionylated forms are higher than that of the free peptide, whereas the intensity of the peak for the glutathionylated Cys 181 -containing peptide is much lower than that of the free peptide. see supplemental Fig. 4). Upon treatment of the protein with 1 mM GSSG before alkylation, both glutathionylated and alkylated Cys 104 -containing peptides were observed in the GluC as well as in the tryptic digests (Fig. 6C, for the full-range spectrum, see supplemental Fig. 5 and Table 3). The intensities of the peaks in the MALDI-TOF-MS spectra do not reflect the ratio of the peptide species in the digest. This is evident from the analysis of a 1:1 mixture of fully carboxamidomethylated and fully glutathionylated GluC peptides. In the case of Cys 104 , the peak intensities for both peptides (90 -115 and 101-141) with bound glutathione were much higher than those for the carboxamidomethylated peptides, whereas for the Cys 181 -containing peptides the peak intensities of both modified peptides were very low in comparison with the free peptide (not shown). This finding again underlines the specific glutathionylation of Cys 181 .
When Grx1 was digested with trypsin, cleavage at Lys 177 was not complete, and two Cys 181 -containing peptides were obtained ( Table 3). The unmodified singly charged 178 -182 and 176 -182 peptide ions occur at 549 and 792 m/z. The smaller ion was shown by MALDI-TOF-MS in an experiment with decreased ion suppression. With the general settings of the mass spectrometer, the peptide appeared only in glutathionylated form (854 m/z). Reaction of Grx1 with 0.5 or 1 mM GSSG yielded for the two tryptic Cys 181 -containing peptides new peaks at 854 and 1097 m/z that are consistent with glutathionylation of the smaller (549 ϩ 305 m/z) and the larger (792 ϩ 305 m/z) fragment. Again, the peptide mixture of Grx1 treated with 1 mM but not with 0.5 mM GSSG showed glutathionylation of Cys 104 in addition to Cys 181 . Also, when the protein treated with 0.5 mM GSSG was digested with trypsin without prior alkylation of the remaining free thiol groups, Cys 104 became modified because of the reaction of the free peptide in solution. The addition of DTT to Grx1 that had been treated with 0.5 mM GSSG and iodoacetamide before digestion abolished the glutathione-containing peaks for the Cys 181 -containing peptide in accordance with GSSG forming a mixed disulfide with the protein (Table 3).
T. brucei Mono-Cys-Grx1 was treated with 0.5 mM Gsp disulfide, remaining free thiols were alkylated with iodoacetamide, and the protein was cleaved by trypsin or GluC. At pH 7.5, the MALDI-TOF-MS anal-

TABLE 1 Possible Cys-containing peptides of T. brucei mono-Cys-glutaredoxin 1
Grx1 was cleaved with trypsin or GluC as described under "Experimental Procedures." In the case of trypsin, all Cys-containing peptides with one miscleavage site were selected. Because in the GluC digestion both detected Cys 104 peptides contained two miscleavage sites, all peptides with two miscleavage sites were considered. The cysteine residues are given in bold letters.   18,195 Da for the protein with two bound Gsp molecules, the spectrum showed a peak at 18,068 Da that reflects the protein with one Gsp and one glutathione molecule bound. Long range storage of the protein sample leads to the slow hydrolysis of the spermidine also from Gsp bound to Cys 181 (see supplemental Fig. 9).

Position
Cys 104 and Cys 181 of Grx1 Can Form an Intramolecular Disulfide Bridge-The ESI-MS spectrum of a Grx1 sample stored for 2 weeks at 4°C showed a mass of 17,330 Da (Fig. 1A, Table 4). This species mainly represents the reduced protein, because addition of iodoacetamide caused a mass shift of 114 Da in accordance with the alkylation of two cysteine residues (Table 4). When 50 M Grx1 was incubated with 2 mM GSSG for 24 h, desalted, and then alkylated, the ESI-MS spectrum revealed three masses. The lowest one (17,328 Da) represents the protein with an intramolecular disulfide bridge (Table 4). This oxidized protein form can easily be distinguished from the reduced protein because, as described above, the mass of the reduced but not the oxidized form would have been shifted to 17,444 Da by the alkylation of the two cysteine residues. The third peak (17,693 Da) represents monoglutathionylated and

TABLE 2 Predicted and detected ions of Cys-containing GluC-peptides of Grx1
The protein was thiolated and alkylated as described under "Experimental Procedures." The observed mass-to-charge values are the average from at least three measurements. Binding of a carboxamidomethyl group (CM), glutathione (GSϪ), and glutathionylspermidine (GspϪ) increases the mass of the peptides by 57, 305, and 433 Da, respectively. a Monoisotopic mass-to-charge ratios of the singly charged peptide ions are given for peptides up to 4000 m/z. Because peptides with mass-to-charge ratios higher than 4000 do not show single isotope peaks, the average m/z is given. monocarboxamidomethylated Grx1. Incubation of Grx1 with 2 mM Gsp disulfide (Table 4), trypanothione disulfide or hydroxyethyl disulfide for 24 h (not shown) also caused formation of the intramolecular disulfide. ESI-MS analysis as well as thiol determinations with DTNB indicated that the intramolecular TS 2 is less efficient in inducing formation of the protein disulfide bond than the intermolecular glutathione, Gsp, and hydroxyethyl disulfides. Vice versa this shows that Grx1 is able to reduce these low molecular weight disulfides. As described above, recombinant Grx1 does not contain a significant amount of protein with intramolecular disulfide bridge; only long range storage generates some intramolecular disulfide together with unspecific protein dimers and polymers. 3 The presence of a mixed disulfide at Cys 181 obviously triggers formation of the intramolecular disulfide. Treatment of Grx1 with 100 M hydrogen peroxide for 20 min at room temperature only slightly affected the thiol content of the protein. Incubation with 500 M H 2 O 2 lowered the free thiol groups to about 0.6 per protein molecule (data not shown). ESI-MS analysis showed that treatment with hydrogen peroxide does not lead to the incorporation of oxygen into Grx1, but it induces also formation of the intramolecular disulfide. Incubation of the GSSG-treated Grx1 with DTT regenerated the fully reduced protein.
To further analyze the formation of the intramolecular disulfide, the GSSG-treated and subsequently alkylated Grx1 sample was cleaved by trypsin, and the peptides were analyzed by MALDI-TOF-MS. Two ions (1813 and 2056 m/z, data not shown) were obtained that correspond to the Cys 181 -containing 178 -182 and 176 -182 peptides linked to the Cys 104 -containing 97-108 peptide by a disulfide bridge. Both peaks disappeared upon addition of DTT. The identity of the peptides forming the disulfide bond was confirmed by sequencing the larger fragment by MALDI-TOF-TOF-MS. Cys 104 -Cys 104 -or Cys 181 -Cys 181 -containing peptides were not observed, which corroborates the specific formation of the intramolecular disulfide bridge. Reaction of Grx1 with 0.5 mM Gsp disulfide also generated the protein with intramolecular disulfide bridge (17,328 Da), but the most prominent peak was the protein with one bound Gsp and an additional mass of 32 Da (Fig. 1G, Table 4). As outlined above, this species most probably represents Grx thiolated at Cys 181 carrying a sulfinic acid residue at position 104.

TABLE 3 MALDI-TOF-MS analysis of the tryptic Cys 181 -and Cys 104 -containing peptides of Grx1 treated with GSSG
Grx1 was glutathionylated, alkylated by iodoacetamide (IAM), and digested with trypsin as described under "Experimental Procedures." Binding of a carboxamidomethyl group (CM) and glutathione (GSϪ) leads to a mass increase of 57 and 305 Da, respectively. Monoisotopic mass-to-charge ratio of the singly charged peptide ions. b The relative intensity (Rel. int.) of a peptide form was calculated from the peak areas of all species of this peptide given by the SNAP algorithm of the program Flex Analysis from Bruker. As outlined in the text, the values do not necessarily reflect the real content of this species in the mixture. c The free singly charged peptide ion (792 m/z) is occasionally generated by breakage of the disulfide bond in the laser beam (see also Fig. 5, B and C). d Because of incomplete cleavage at Lys 177 , two Cys 181 -containing peptides were obtained. The smaller peptide (178 -182) appears at 549.3 and 606.3 m/z in the free and carboxamidomethylated form, respectively. Because these ions were detected only in a separate experiment with decreased ion suppression, relative quantification of the peptide was not possible.
Circular Dichroism Spectroscopy-The circular dichroism spectra of reduced and GSSG treated Grx1 are identical (Fig.  8A). Formation of the intramolecular disulfide or glutathionylation does not significantly change the secondary structure of the protein. The pronounced minimum at 207 nm indicates relatively high loop content. Unfolding of Grx1 was studied by thermal denaturation. The melting point of the reduced protein was 59°C, whereas the oxidized protein did not show an inflection point up to 95°C (Fig. 8B). The shift by more than 25°C to higher temperatures indicates that the oxidized protein has a more stable conformation than the reduced protein.

Reduction of the Intramolecular Disulfide and Deglutathionylation of T. brucei Mono-Cys-Grx1 by Different Thiol
Systems-50 M Grx1 was treated with 2 mM GSSG, which, after removal of excess GSSG on a PD10 column, results in a mixture of protein with intramolecular disulfide bridge and the protein with one and (to a minor degree) two bound glutathione molecules. This sample was incubated at pH 7.5 with an equimolar concentration of reduced T. brucei tryparedoxin for 20 min at room temperature followed by alkylation with iodoacetamide. The intact mass determination by ESI-MS revealed a single protein species with two bound carboxamidomethyl groups in accordance with Grx1 being completely reduced and deglutathionylated ( Table 5). The trypanothione/ trypanothione reductase system also reduced both the intramolecular disulfide and the protein disulfide with glutathione (Table 5) or Gsp (not shown). In contrast, D. melanogaster thioredoxin reductase catalyzed the reduction of the intramolecular disulfide but did not deglutathionylate the protein.
Trypanothione reductase and NADPH alone had no effect. GSH reduced the protein disulfide, and the small amount of mixed disulfide between Cys 104 and glutathione or Gsp but did not cleave the mixed disulfides at Cys 181 (Table 5). Interestingly, treatment of the protein with bound Gsp with GSH caused a thiol/disulfide exchange at Cys 181 resulting in the glutathionylated protein with an identical degree of modification (not shown).

TABLE 4 Disulfide bond formation between Cys 104 and Cys 181 in Grx1 analyzed by ESI-MS
The protein was treated with GSSG or Gsp disulfide for 24 h and then desalted, and the remaining free thiols were alkylated with iodoacetamide (IAM). This treatment generated the thiolated protein as well as the protein with intramolecular disulfide. Binding of a carboxamidomethyl group (CM), glutathione (GSϪ), and glutathionylspermidine (GspϪ) increases the mass of the protein by 57, 305, and 433 Da, respectively. S-S, intramolecular disulfide.

Treatment
Mass Cys modifications Observed Calculated [Da] None (control) 17

T. brucei Mono-Cys-Glutaredoxin 1 Lacks (De)
Glutathionylation Activity-As described under "Experimental Procedures," 50 M reduced thioredoxin or peroxidase III was incubated for different times with 0.5 mM GSSG in the presence and absence of 50 M reduced Grx1. The degree of glutathionylation of the proteins was the same with and without Grx1 (not shown). Thus, Grx1 became glutathionylated but did not catalyze the glutathionylation of these proteins. Glutathionylated thioredoxin (see below) was then treated with an equimolar concentration of reduced Grx1 and subjected to ESI-MS. No deglutathionylation activity of Grx1 was observed. ESI-MS analysis of reduced Grx1 incubated with oxidized tryparedoxin revealed that Grx1 is also unable to reduce the intramolecular disulfide of tryparedoxin. In addition, no intermolecular disulfide between Grx1 and tryparedoxin was observed.
T. brucei Peroxidase III Is Monothiolated by Glutathione and Gsp-50 M glutathione peroxidase-type tryparedoxin peroxidase III (20,22) was reduced, excess thiols were removed, and the protein was treated with 0.5 mM GSSG or Gsp disulfide.  Table 6). Incubation of reduced PxIII with TS 2 did not generate any new peak. The intramolecular disulfide did not form a stable mixed disulfide with the protein. To confirm the specific thiolation of a single residue, PxIII was treated with 5 mM GSSG (not shown). Even at this high GSSG concentration, prominent peaks were observed for the monoglutathionylated and unmodified protein, whereas the peak intensity for a protein species with two glutathione molecules was very low. This indicates that glutathionylation of a cysteine residue interferes with the formation of the intramolecular disulfide shown to be an intermediate during catalysis. 4 47 or Cys 95 -Peroxidase III possesses three cysteine residues at positions 47, 76, and 95 (20). To identify the site of thiolation, the reduced protein was treated with GSSG or Gsp disulfide, the remaining free cysteine residues were alkylated, and the protein was digested by GluC. The resulting peptides were analyzed by MALDI-TOF-MS. Table 7 gives the sequences of the possible Cys-containing GluC-peptides. Treatment of the reduced peroxidase with GSSG caused a mass increase of 305 Da for both the Cys 47and the Cys 95 -containing peptides (Table 8). In the folded protein, Cys 76 is not modified by glutathione or Gsp. Glutathionylation of Cys 76 was observed only when the protein sample containing excess GSSG was digested without prior alkylation of the remaining free thiols. Under these conditions, the free Cys 76 peptide reacted with GSSG, which proves that the modified peptide is detectable if it is formed. The intact mass determination of the GSSG-treated peroxidase revealed the binding of a single glutathione molecule (Table 6). Thus, glutathione binds either to Cys 47 or to Cys 95 but not to both cysteine residues in the same protein molecule. To elucidate whether there is a preference for one of the two cysteines, the incubation time was decreased from 4 to 1 h. Again, both residues became modified, indicating that the two cysteines have similar reactivity toward GSSG. Reaction of the reduced PxIII with Gsp disulfide resulted in the selective modification of Cys 95 ( Table 8).

T. brucei Peroxidase III Is Thiolated at Cys
Thiolation of T. brucei Thioredoxin-T. brucei thioredoxin possesses three cysteinyl residues, namely the redox active Cys 31 -Cys 34 couple and Cys 68 in the C-terminal moiety of the protein (23,31). ESI-MS analysis of the reduced protein treated with 0.5 mM GSSG or Gsp disulfide revealed monothiolated protein species (Fig. 9). To elucidate whether the peak intensities of modified and unmodified thioredoxin are comparable and thus reflect the relative amount of the respective species in

Reduction and/or deglutathionylation of GSSG-treated Grx1 by different thiol systems
Incubation of 50 M reduced T. brucei Grx1 with 2 mM GSSG for 24 h resulted in a mixture of Grx1 with an intramolecular disulfide bridge (17,328 Da) and the protein glutathionylated at Cys 181 (17,636 Da) and, to a minor degree, also at Cys 104 (17,941 Da). This protein sample was treated with different reducing systems, the free thiol groups generated were carboxamidomethylated by iodoacetamide, and the proteins were subjected to ESI-MS. Binding of a carboxamidomethyl group (CM), glutathione (GSϪ), and glutathionylspermidine (GspϪ) increases the mass of the protein by 57, 305, and 433 Da, respectively. S-S, intramolecular disulfide.    (14,677 Da in the case of glutathione and 14,804 Da in the case of Gsp). Different ratios of free and monothiolated thioredoxin were subjected to ESI-MS. The percentage of the varied protein species detected in the spectrum was depicted against the theoretical value. As shown in supplemental Fig. 2, A and B, the slopes of the calibration lines were 1.03 and 1.04 when modified or free thioredoxin was varied. This shows that the binding of glutathione does not affect the ionization efficiency of thioredoxin, as is the case with Grx1 (see supplemental Fig. 1). The reactivity of thioredoxin toward glutathione is higher than that of Grx1. Incubation of 50 M reduced thioredoxin with 0.5 mM GSSG for 4 h resulted in about 60% glutathionylated protein. Incubation of thioredoxin with 20 mM Gsp disulfide also resulted in the modification of one cysteine residue (14,804 Da). When varying this protein species against free thioredoxin and vice versa, calibration lines with a slope of 1.1 and 0.8, respectively, were obtained. Thus, binding of Gsp increases the ionization efficiency of thioredoxin by 10 -20%. Taking this into account, the ESI-MS spectra (Fig. 9, B and C) indicate that thioredoxin has a slight preference for glutathione over Gsp binding. This contrasts with Grx1, where Gsp disulfide causes a more pronounced modification than does GSSG under identical conditions (Fig. 1, C and G; supplemental Fig. 1).

Reducing system
T. brucei Thioredoxin Is Thiolated at Cys 68 -Reduced thioredoxin was incubated with 0.5 and 5 mM GSSG or Gsp disulfide, alkylated by iodoacetamide, and digested with chymotrypsin or LysC. The MALDI-TOF-MS analysis of the chymotryptic digest revealed an ion at 1837 m/z that corresponds to the 31-46 peptide with Cys 31 -Cys 34 in oxidized form. The reduced alkylated protein yielded an ion at 1953 m/z for the 31-46 peptide in accordance with the binding of two carboxamidomethyl groups to Cys 31 and Cys 34 . As expected, the active site cysteines did not form mixed disulfides with glutathione or Gsp but were oxidized to the intramolecular disulfide (not shown). Because a peptide containing Cys 68 was not detectable, the thiolated proteins were subsequently digested with LysC. Table 9 gives the sequences of the possible Cys-containing peptides. The MALDI-TOF-MS spectra of the LysC digests showed ions at 2107 and 2234 m/z that correspond to the 68 -83 peptide with bound glutathione and Gsp, respectively (Table 10).
Glutathionylated Thioredoxin Catalyzes Reduction of Insulin Disulfide-Reduced T. brucei thioredoxin was incubated with 5 mM GSSG for 72 h, which resulted in the complete glutathionylation of Cys 68 as revealed by MALDI-TOF-MS analysis of the modified protein. Reduction of insulin disulfide was followed in an assay system containing NADPH, human thioredoxin reductase, and either glutathionylated or unmodified T. brucei thioredoxin. Both protein species showed the same activity, which means that glutathionylation of Cys 68 does not affect the ability of thioredoxin to reduce protein disulfides. MALDI-TOF-MS analysis of the modified protein confirmed that thioredoxin reductase and NADPH do not reduce the mixed disulfide with glutathione.
T. brucei Tryparedoxin Does Not Form Mixed Disulfides with Glutathione-T. brucei tryparedoxin (25, 32) possesses only two cysteinyl residues that form the redox active dithiol/disulfide couple (Cys 41 and Cys 44 ). ESI-MS analysis of reduced tryparedoxin treated with GSSG showed a single peak with a mass (15,757.8 Ϯ 0.4 Da, mean Ϯ S.D. of six intact mass determinations) that was 2 Da lower than the theoretical mass of the reduced protein (15,759.8 Da, not shown). This was a first indication that, as observed for thioredoxin (see above), GSSG causes the active site cysteine residues to form an intramolecular disulfide bond. Disulfide bond formation was subsequently confirmed by peptide analysis with MALDI-TOF-MS (not shown). Reduced untreated and GSSG-treated tryparedoxin were alkylated by iodoacetamide and digested with trypsin. The MALDI-TOF-MS analysis of the GSSG-treated sample revealed an ion at 1775 m/z that represents the 31-45 peptide (TVFLYFSASWCPPCR) in oxidized form. In the reduced protein (but not in the GSSG-treated protein) the active site cys- The relative intensity (Rel. int.) of a peptide form was calculated from the peak areas of all species of this peptide given by the SNAP algorithm of the program Flex Analysis from Bruker. As outlined in the text, the values do not necessarily reflect the real content of this species in the mixture. b Mass-to-charge ratios of the monoisotopic singly charged peptide ions are given except for the ion at 4447.2 m/z, which is the average mass-to-charge ratio. c Because of incomplete cleavage at Glu 89 , two Cys 76 -containing peptides were obtained. teines became alkylated by iodoacetamide, which shifted the singly charged peptide ion by 116 Da to 1891 m/z. In addition, the oxidized peptide appeared at 2 Da lower than the ion of the reduced one, and sequencing by MALDI-TOF-TOF-MS did not yield fragments of the two proline residues between the active site cysteines. Because the reduced peptide shows very strong fragment signals in this region, the sequencing supported the formation of an intramolecular disulfide upon treatment with glutathione disulfide.
Glutathionylation of T. brucei Glyoxalase II-T. brucei Glyoxalase II contains six cysteinyl residues (29). 50 M reduced protein was treated with 0.5 mM GSSG at room temperature for 24 h and then subjected to ESI-MS. Even after this extended incubation time, glutathionylated protein species were not observed. The spectrum showed a peak with a mass of 34,083.6 Da, which corresponds to the calculated mass of glyoxalase II with all cysteines in disulfide form. The most prominent peak at 34,098.9 Ϯ 0.3 Da (mean Ϯ S.D.) had a 16-Da higher mass, which may have been caused by oxidation of a methionine residue. Exposure of glyoxalase II to 5 mM GSSG for 24 h resulted in an additional peak (34,710 Da) with a 611-Da higher mass, which is consistent with the incorporation of two glutathione moieties. The reactivity of two of the six cysteinyl residues of glyoxalase II toward glutathione must be similar because no mono-modified protein species was observed. Because glyoxalase II reacted with GSSG only under drastic conditions, glutathionylation of the protein under cellular conditions is unlikely and therefore was not further analyzed.
T. cruzi Trypanothione Reductase Does Not Form Mixed Disulfides with Glutathione-Trypanothione reductase has a total of seven cysteinyl residues (33). ESI-MS analysis of the GSSG (0.5 and 5 mM)-treated reduced enzyme did not show any peak compatible with the binding of glutathione. MALDI-TOF-MS analysis of the tryptic peptides revealed that GSSG induced formation of the intramolecular disulfide between the redox active Cys 52 and Cys 57 couple (not shown).

DISCUSSION
Glutathionylation of proteins is a protection mechanism against oxidative damage as well as a regulation mechanism of   Monoisotopic mass-to-charge ratio of the singly charged peptide ions. b The rather low relative intensities of the Cys 68 peptide with bound glutathione or Gsp may be caused by decreased ionization efficiency or by miscleavage of the thiolated peptide and does not reflect the real quantity of modification. Analysis of the unmodified peptide, the completely modified peptide and a 1:1 mixture of both showed that the peak for the glutathionylated peptide is decreased by a factor of 5 to 7. The peak intensity of the peptide modified by Gsp is also decreased but only 2-fold. Fig. 9 and the calibration experiment (see supplemental Fig. 2) show that about 60% of the protein is glutathionylated under identical conditions. Even with this low relative intensity, the peak that represents the glutathionylated Cys 68 -containing peptide is detected with a signal-to-noise ratio of 22.
enzyme activities (1)(2)(3)(4). Here we show that T. brucei mono-Cys-glutaredoxin 1, the glutathione peroxidase-type tryparedoxin peroxidase III, and thioredoxin can be glutathionylated and S-glutathionylspermidinylated. In the case of T. brucei glyoxalase II, glutathionylation occurs only at unphysiologically high GSSG concentrations. T. brucei tryparedoxin and T. cruzi trypanothione reductase do not show any binding of glutathione. As expected, trypanothione disulfide does not form a stable mixed disulfide with any of the proteins. Treatment of T. brucei mono-Cys-glutaredoxin 1 with 0.25 or 0.5 mM GSSG resulted in the specific attachment of one glutathione molecule at Cys 181 . At higher GSSG concentrations, the reaction became less specific, and both cysteine residues were modified. 0.5 mM Gsp disulfide also reacted preferably with Cys 181 . The thiolation of this nonconserved residue in the C-terminal part of Grx1 contrasted with data from mono-Cys-glutaredoxins of other organisms, which are reported to be glutathionylated at the active site cysteine corresponding to Cys 104 of the parasite protein (Fig. 10). Saccharomyces cerevisiae glutaredoxin 5 shares 29% of all residues with the T. brucei Grx1. The yeast protein contains the putative active site Cys 60 as well as a cysteine residue at position 117 that is conserved in Grx4 from E. coli and in some other 1-and 2-Cys-glutaredoxins (Ref. 40; Fig. 10). Treatment of reduced S. cerevisiae Grx5 with GSSG yielded a mixture of protein species with glutathione bound to one or both cysteine residues as well as the protein with an intramolecular disulfide bridge (7). It was suggested that glutathione binds first to Cys 60 and is then transferred to Cys 117 , and the glutathionylated Cys 117 acts as an intermediate in the formation of the protein disulfide. Recently E. coli Grx4 has been structurally (40) and functionally (17) investigated. The protein possesses three cysteine residues. Cys 30 represents the putative active site, Cys 43 is conserved in some bacterial proteins, and Cys 84 corresponds to Cys 117 of yeast Grx5 (7) and is also present in the 2-Cys-glutaredoxins E. coli Grx3 and human Grx1 (Fig. 10). Reaction of reduced E. coli Grx4 with 2 mM GSSG has been reported to cause glutathionylation of one cysteine residue and the simultaneous formation of an intramolecular disulfide (17). It was suggested that glutathione binds to Cys 30 and that this reaction is supported by interactions with conserved residues that may form a glutathione binding surface (40 -42). So far a three-dimensional structure of a mono-Cys-glutaredoxin is available only for unmodified reduced E. coli Grx4 (40). For the classical human Grx1 and E. coli Grx3, mutants were generated that retained only the first active site cysteine, and the NMR structures of the mixed disulfides with glutathione have been solved (41,42). Several residues reversibly interacted with the glutathione covalently bound at the active site cysteine and/or provided a putative docking site for the second glutathione molecule that attacks the mixed disulfide during catalysis. Because these studies used mutants in which all but the first active site cysteine had been replaced, a possible glutathionylation of another cysteine residue would not have been detected. Most of the residues supposed to be involved in the binding of glutathione to E. coli Grx3 or human Grx1 are conserved in the parasite protein (Fig. 10); but, as outlined above, neither glutathione nor Gsp binds to the active site cysteine. In contrast to the classical 2-Cys-glutaredoxins, all mono-Cys-glutaredoxins investigated so far do not catalyze the glutathione-dependent reduction of hydroxyethyl disulfide. Thus, a glutathione mixed disulfide at the active site cysteine probably does not occur as a catalytic intermediate. In T. brucei Grx1, Cys 181 is sensitive to thiolation, and a mixed disulfide at this residue triggers formation of an intramolecular disulfide bridge. One may speculate that the glutathione bound to Cys 181 interacts with the conserved residues, which could facilitate Cys 181 to approach Cys 104 . However, an interaction based mainly on an opposite charge distribution is unlikely, because the positively charged Gsp and the uncharged S-hydroxyethyl also drive formation of the intramolecular disulfide when bound to Cys 181 .
Incubation of the reduced Plasmodium falciparum glutaredoxin-like protein 2 (Glp2) with 2 mM GSSG leads to the binding of one glutathione molecule (18), but which of the two cysteinyl residues of the protein is modified has not been determined. When reduced Glp2 was treated with 2 mM GSH under aerobic conditions for 24 h, a mixture of mono-and bis-glutathionylated protein was obtained, which indicates that both cysteine residues are accessible. In P. falciparum Glp2 the second cysteine (Cys 216 ) is the fourth residue from the C terminus as is the case for Cys 181 in T. brucei Grx1 (Fig. 10). Formation of an intramolecular disulfide has not been investigated in P. falciparum Glp2, and from model studies it has been suggested that this formation would be unlikely (18).
As outlined above, thiolation of Cys 181 induces an intramolecular disulfide bridge between Cys 104 and Cys 181 . TS 2 , which does not form stable mixed disulfides, is less efficient in triggering the formation of the protein disulfide, although the redox potentials of glutathione and trypanothione are identical (8). The intramolecular disulfide is virtually not formed when the protein upon oxidation. Alternatively, the two cysteines are also in relative proximity in the reduced enzyme, and glutathionylation of one residue interferes sterically with glutathionylation of the second one. In contrast to glutathione, Gsp reacts exclusively with Cys 95 . The larger Gsp molecule or the charge differences may prevent binding to Cys 47 . The peak intensity of the Gsp-modified protein was much higher than that of the glutathionylated protein. This could be because of the different thiol pK values and/or the specific interactions between the spermidine moiety of Gsp and the protein, as outlined above. Cys 76 , the third cysteine of PxIII, did not bind glutathione or Gsp under all conditions tested.
In T. brucei thioredoxin, besides the redox active dithiol/disulfide, position 68 is occupied by a cysteine residue. In the presence of 0.25 or 0.5 mM GSSG about 40 and 60%, respectively, of the total protein occurs in glutathionylated form. In contrast to Grx1 and PxIII, which reacted more readily with Gsp disulfide, T. brucei thioredoxin showed a slight preference for glutathione binding. Cys 68 is embedded between Arg 67 and Lys 69 . The positive charges may favor the interaction with the negatively charged glutathione and cause an electrostatic repulsion of the positively charged Gsp. Such charge interactions have also been discussed in the case of human thioredoxin. Mammalian thioredoxins possess three additional conserved cysteine residues (Cys 61 , Cys 68 , and Cys 72 in the human protein). Treatment of human thioredoxin with 5 mM GSSG leads to glutathionylation of Cys 72 but not of Cys 68 (37), which corresponds to the residue thiolated in T. brucei thioredoxin. In the human protein, Cys 68 is flanked by two glutamate residues. In contrast, a lysine residue precedes Cys 72 , which may trigger the adduct formation by interacting electrostatically with the ␥-glutamyl moiety of glutathione. Cys 72 is also involved in the formation of covalent dimers (52), and glutathionylation of the residue prevents dimerization of human thioredoxin (37). Prolonged storage of T. brucei thioredoxin generates covalent dimers (31). Because Cys 68 is the only additional cysteine in the protein it may be involved in dimer formation (53). Treatment of reduced T. brucei thioredoxin with GSSG does not cause dimer formation but glutathionylation of Cys 68 . The chloroplast thioredoxin f has been shown sensitive to glutathionylation. Modification of Cys 60 (which is neither conserved in the trypanosomal nor the mammalian proteins) results in an impaired activation of target enzymes (54). Thus, nonconserved cysteine residues in the C-terminal moiety of thioredoxins are susceptible to glutathionylation affecting the protein functions. Glutathionylation of Cys 68 of T. brucei thioredoxin did not impair its ability to reduce insulin disulfide. In contrast, glutathionylation of Cys 72 lowered the ability of human thioredoxin to catalyze the reduction of insulin disulfide to 66% (37). Although the physiological function of Cys 68 in the T. brucei thioredoxin is not known, glutathionylation of the residue can prevent dimerization of the protein and/or the irreversible overoxidation of this residue.
To reveal whether glutathionylation of an individual cysteine residue is mainly determined by the primary or tertiary structure of the protein, Grx1 and PxIII were treated with GSSG and digested without prior blocking of the remaining free thiols. This resulted in the unspecific glutathionylation of all cysteine residues, which shows that the three-dimensional structure of the protein is much more decisive for the selective glutathionylation than the physico-chemical properties of neighboring residues. As expected and shown previously for human thioredoxin (37), the redox active cysteines (Cys 31 and Cys 34 in T. brucei thioredoxin) do not form a stable mixed disulfide when the reduced protein is treated with GSSG, but they are oxidized to the intramolecular disulfide. The same is true for the active site cysteines in T. brucei tryparedoxin and T. cruzi trypanothione reductase. In the crystal structure of trypanothione reductase (Protein Data Bank accession number 1NDA), none of the five cysteine residues (in addition to the active site dithiol/disulfide couple) is exposed to the solvent. This agrees with the observed failure of trypanothione reductase to be glutathionylated and renders regulation of the enzyme by thiolation very unlikely.
As shown here, the small thiol proteins mono-Cys-glutaredoxin 1 and thioredoxin, as well as the glutathione peroxidasetype tryparedoxin peroxidase III of T. brucei, are accessible to specific and reversible glutathionylation. In addition, Gsp can form mixed disulfides with these proteins. Under cellular conditions, reaction of protein cysteinyl residues with the disulfide form of glutathione and/or Gsp is one mechanism; another one is the reaction of protein sulfenic acids with GSH. All in vitro studies on the glutathionylation of proteins employed reduced proteins and GSSG. In this work we used 250 or 500 M GSSG. Such concentrations can occur in the cell under oxidative stress conditions and definitely reflect the physiological conditions better than the 2 or 5 mM GSSG (17,18,37,54) or even 40 mM GSSG (43) used in related in vitro studies. Future studies will focus on the physiological conditions that lead to thiolation of the proteins in living parasites.