The Mycobacterial Thioredoxin Peroxidase Can Act as a One-cysteine Peroxiredoxin*

Thioredoxin peroxidase (TPx) has been reported to dominate the defense against H2O2, other hydroperoxides, and peroxynitrite at the expense of thioredoxin (Trx) B and C in Mycobacterium tuberculosis (Mt). By homology, the enzyme has been classified as an atypical 2-C-peroxiredoxin (Prx), with Cys60 as the “peroxidatic” cysteine (CP) forming a complex catalytic center with Cys93 as the “resolving” cysteine (CR). Site-directed mutagenesis confirms Cys60 to be CP and Cys80 to be catalytically irrelevant. Replacing Cys93 with serine leads to fast inactivation as seen by conventional activity determination, which is associated with oxidation of Cys60 to a sulfinic acid derivative. However, in comparative stopped-flow analysis, WT-MtTPx and MtTPx C93S reduce peroxynitrite and react with TrxB and -C similarly fast. Reduction of pre-oxidized WT-MtTPx and MtTPx C93S by MtTrxB is demonstrated by monitoring the redox-dependent tryptophan fluorescence of MtTrxB. Furthermore, MtTPx C93S remains stable for 10 min at a morpholinosydnonimine hydrochloride-generated low flux of peroxynitrite and excess MtTrxB in a dihydrorhodamine oxidation model. Liquid chromatography-tandem mass spectrometry analysis revealed disulfide bridges between Cys60 and Cys93 and between Cys60 and Cys80 in oxidized WT-MtTPx. Reaction of pre-oxidized WT-MtTPx and MtTPx C93S with MtTrxB C34S or MtTrxC C40S yielded dead-end intermediates in which the Trx mutants are preferentially linked via disulfide bonds to Cys60 and never to Cys93 of the TPx. It is concluded that neither Cys80 nor Cys93 is required for the catalytic cycle of the peroxidase. Instead, MtTPx can react as a 1-C-Prx with Cys60 being the site of attack for both the oxidizing and the reducing substrate. The role of Cys93 is likely to conserve the oxidation equivalents of the sulfenic acid state of CP as a disulfide bond to prevent overoxidation of Cys60 under a restricted supply of reducing substrate.

fore be considered to be the dominant antioxidant device that guarantees the survival of the pathogen in the oxidant environment of the phagocytes of the host. The enzyme has a limited similarity with human Prxs, the most similar one being PrxII with only 21% identities. This dissimilarity to mammalian sequences as well as the putative functional importance underscores the potential use of MtTPx as a drug target.
By sequence, MtTPx is classified as an atypical 2-cysteine peroxiredoxin (2-C-Prx). It is closely related to a variety of thiol peroxidases, of which the thioredoxin peroxidase of Escherichia coli has been most extensively investigated in terms of structure (9) and functional aspects (10). The catalytic mechanism common to all peroxiredoxins consists of the oxidation of a highly reactive "peroxidatic" cysteine (C P ) to a sulfenic acid derivative, which is followed by a direct or indirect reduction by a thiol substrate (11,12). As was first shown for the tryparedoxin peroxidase of Leishmania donovani (13), the activation of C P is achieved by the positive charge of a neighboring arginine residue and hydrogen bonding with a threonine or, less commonly, a serine residue, which forces the active site cysteine into the thiolate form that readily reacts with any kind of ROOH (14). The reduction of the resulting sulfenic acid differs between different types of Prxs. In the case of the typical 2-C-Prxs, the primary reductant is a cysteine conserved near the C terminus ("C R " for resolving cysteine) that forms a disulfide bridge with C P , which then reacts with the reducing substrate, typically a protein containing a CXXC motif such as thioredoxin, tryparedoxin, or AhpF (14,15). However, because C R is sterically out of reach for C P within the same protein subunit, the disulfide is formed between two inversely oriented subunits. During this process C P is delocalized in a way that it is no longer accessible to the reducing substrate. The catalysis would therefore be arrested in the oxidized form, if the reducing substrate could not "resolve" this block by attacking the sulfur of the distally conserved cysteine, which thus becomes the C R . In 1-C-Prxs the C-terminal cysteine is not conserved, and the reducing thiol substrate has to react directly with the oxidized C P . In the atypical 2-C-Prxs the C R typical of the 2-C-Prxs is also missing, but another cysteine is presumed to substitute for C R in a way that it forms an intra-subunit disulfide bridge that is the prerequisite to terminate the catalytic cycle (9).
The recently resolved x-ray structures of MtTPx (PDB 1XVQ), of the molecular mutant MtTPx C60S (PDB 1Y25 (16)), and of the related peroxiredoxins of Haemophilus influenzae (PDB 1Q98), E. coli (PDB 1QXH (9)), and Salmonella pneumoniae (PDB 1PSQ) reveal that Cys 60 in MtTPx builds the center of the catalytic triad typical of Prxs, which here is composed of Cys 60 , Thr 57 , and Arg 130 . Cys 60 thus should be the peroxidatic cysteine, whereas Cys 93 is homologous to the presumed C R of related atypical 2-C-Prxs, and the third cysteine in MtTPx, Cys 80 , remains of questionable functional relevance. The x-ray structures, however, did not provide unambiguous evidence for any obligatory involvement of Cys 93 in the catalysis. In fact, the C-␣ distance between Ser 60 (corresponding to Cys 60 ) and Cys 93 in MtTPx C60S appeared too large to allow a disulfide formation without major rearrangements (Fig. 1). The expected Cys 60 -Cys 93 disulfide bond is not clearly visible in the structure of oxidized MtTPx (PDB 1XVQ) either, and the sub-stantial residual activity of EcTPx C95S that has been reported recently by Baker and Poole (10) sheds further doubt on the hypothesis that the distal conserved cysteine of the atypical 2-C-Prxs is always functionally equivalent to the resolving cysteine of typical 2-C-Prxs.
Here we present compelling evidence for a novel variation of Prx catalysis. The catalytic mechanism of MtTPx does not necessarily involve the presumed C R but rather corresponds to that FIGURE 1. Three-dimensional structure of MtTPx. A, ribbon representation of MtTPx in its reduced state. The model was generated from the MtTPx C60S variant (PDB 1Y25) where the serine residue was mutated back to cysteine using the Swiss-PDB Viewer program. B, ribbon representation of MtTPx in its oxidized form (PDB 1XVQ). In 1XVQ Cys 80 is oxidized and lacks the side chain probably because of radiation damage. Distance between cysteine-sulfur atoms is displayed as black lines (Ångströ m). Figures were produced using the program 3D-Mol from the Vector NTI-Advance software (Invitrogen). of a 1-C-Prx. The cysteine homologous to C R , however, prevents inactivation of the enzyme by hydroperoxides under limited supply of reductants by saving the oxidation equivalents in a readily reversible oxidation state.

Kinetic Measurements
Conventional activity determinations of MtTPx and derived mutants were performed by means of a coupled test that measures thioredoxin-mediated NADPH consumption by thioredoxin reductase as described by Jaeger et al. (7). Preincubation conditions were 10 M thioredoxin reductase, 10 M thioredoxin B or C, 5 M MtTPx, 450 M NADPH, 1 mM EDTA in 50 mM Hepes buffer, pH 7.4, at 25°C. The reaction was started with 73 M tert-butyl hydroperoxide (final concentration) and monitored continuously.
Direct stopped-flow measurement of peroxynitrite reduction by MtTPx was performed as follows. Peroxynitrite was freshly synthesized in a quenched-flow reactor from sodium nitrite and hydrogen peroxide (H 2 O 2 ) under acidic conditions and quantified as described previously (18). Stock solutions of peroxynitrite were treated with granular manganese dioxide to eliminate H 2 O 2 remaining from the synthesis. Nitrite content in samples of peroxynitrite was typically Ͻ30% of the peroxynitrite concentration. MtTPx (either wild-type or mutants) and MtTrxC or -B were reduced overnight by the addition of a Ͼ10fold excess dithiothreitol (DTT). Excess reductant was removed immediately before use by passing the proteins through a high pressure liquid chromatography-connected Hitrap column (Amersham Biosciences) with UV-visible detection at 280 nm and collected manually in rubber-capped tubes, which were subsequently bubbled with argon at 4°C. Total thiol content was determined by Ellman's reagent to exclude the presence of any relevant residual DTT. The extensively degassed elution buffer was 100 mM sodium phosphate, pH 7, plus 0.1 mM DTPA. The reduction of peroxynitrite by reduced MtTPx (either wild-type or the mutants C60S, C80S, or C93S, respectively) was then studied at pH 7.4 and 25°C in a stoppedflow spectrophotometer (SX-17MV; Applied Photophysics) with a mixing time of less than 2 ms, monitoring peroxynitrite decomposition at 310 nm with or without pre-reduced enzyme at the concentrations indicated (19). For the determinations of initial rates of peroxynitrite decomposition, 200 absorbance points were acquired during the first 20 ms of the reaction, and 200 further points were acquired until more than 90% peroxynitrite had decomposed (0.02-20 s). Computer-assisted simulations were performed using Gepasi software (20).
The reduction of peroxynitrite-oxidized MtTPx or its mutants by its natural reductants, i.e. MtTrxB or MtTrxC, was evaluated by monitoring the peroxynitrite decomposition rate at a low MtTPx concentration and its acceleration by MtTrxC or MtTrxB. Precisely, 1.1 M pre-reduced MtTPx was reacted with 5 M peroxynitrite to yield a fast initial peroxynitrite reduction followed by the slope typical of spontaneous peroxynitrite decomposition. The slope thus obtained provides the base line to be compared with that observed upon addition of pre-reduced MtTrxB or MtTrxC (33 M), which by themselves only marginally affect the peroxynitrite decay. The experimen-tal results were then compared with curves calculated by means of known (7) or estimated rate constants.
Alternatively, rates of reduction of MtTPx (or mutants) by TrxB were measured directly by taking advantage of redox-dependent fluorescence changes of MtTrxB. The reaction of 2.25 M MtTrxB, pre-reduced by DTT, with excess MtTPx (or mutants; 15-40 M) was analyzed by the stopped-flow technique monitoring the tryptophan fluorescence of MtTrxB at Ͼ335 nm and an excitation wavelength of ϭ 280 nm.
Qualitatively, the reaction of MtTrxB with MtTPx (and molecular mutants) was investigated by monitoring the oxidation of dihydrorhodamine (DHR) by a low flux of peroxynitrite (ϳ0.7 M/min) generated by 3-morpholinosydnonimine hydrochloride (SIN-1) as follows. 120 M DHR was exposed to 50 M SIN-1 yielding a constant oxidation rate of 0.28 M/min, as monitored at 500 nm (⑀ ϭ 78,800). The influence of MtTrxB, mutants MtTPx, and combinations thereof on slope and/or lag phase is then taken as a semi-quantitative measure of interference with the oxidation process (19).

LC-ESI-MS/MS Analysis of Intermediates of the Catalytic Cycle
Protein Reduction and Oxidation-If not stated differently, each protein was reduced for 80 min at 37°C with 50 mM DTT in 50 mM phosphate buffer, pH 7.5. Following removal of the reductant by buffer exchange (gel filtration on a Bio-Spin 6 column, twice), TPx 25 M was mixed in different experiments with 100 M hydrogen peroxide or 150 M TrxB. The sequence of additions and the incubation times are reported for specific experiments under "Results." The reaction was stopped by adjusting the sample to pH 2 by HCl.
Enzymatic Digestions-Pepsin was added at an enzyme/substrate ratio of 1:100 (w/w) in 100 mM ammonium acetate, pH 2.5. After 3 h of incubation at room temperature, the reaction was stopped by cooling on ice. Trypsin was added at an enzyme/ substrate ratio of 1:10 (w/w) in 100 mM ammonium bicarbonate, pH 8. After overnight incubation at 37°C, the pH was adjusted to 2 with trifluoroacetic acid to stop the reaction. Ten l of the peptide mixtures were applied to LC-ESI-MS/MS.
LC-ESI-MS/MS Analysis of Peptides-For liquid chromatography/mass spectrometry, a Surveyor MS high pressure liquid chromatography pump equipped with a MicroAS autosampler (loop 20 l) was coupled to an LTQ ion trap mass spectrometer by an electrospray interface (ThermoElectron, Milan, Italy). An ACE C18 column (1 ϫ 150 mm, 5 m; ACT, Aberdeen, Scotland, UK) with an acetonitrile gradient was used (eluent A,0.1% formic acid in water; eluent B, 0.1% formic acid in acetonitrile) at a flow rate of 50 l/min. The gradient profile was 5% eluent B for 4 min followed by 5-50% eluent B within 40 min.
For mass spectrometry, the heated capillary was held at 180°C and voltage at 38 V. Spray voltage was 5.2 kV. Spectra were acquired in positive mode (in the range 400 -2000 m/z) using dynamic exclusion for MS/MS analysis (relative collision energy of 35%, repeat count 3).
Data Handling of Mass Spectra-Computer analysis of peptide MS/MS spectra was performed using Bioworks 3.1, based on SEQUEST algorithm (University of Washington and licensed to ThermoFinnigan Corp.). For the peptic peptide mixture, the "no enzyme" option was used because of the limited specificity of pepsin cleavage. As confidence of peptide identification, the minimum values of Xcorr were greater than 1.5, 2.0, and 2.5 for single, double, and triple charge ions, respectively.
Identification of disulfides was carried out as follows. First, the most frequent cysteine-containing peptides obtained by pepsin digestion were identified in both TPx and Trx. Then these peptides were used for manual calculation of possible peptides linked by disulfide bridges. The minimum consecutive multicharge ions of a possible "hybrid peptide" were 3. When possible, manual evaluation of related MS/MS spectra was performed to obtain partial amino acid sequences of one or both peptides.

Apparent Activity of Molecular Mutants in Conventional
Test-To define the catalytic role of each of the three cysteines in MtTPx, the molecular mutants C60S, C80S, and C93S were first investigated for catalytic efficiency by means of the conventional coupled test system that mimics the physiological context of the enzyme as follows: NADPH consumption for hydroperoxide reduction via thioredoxin reductase, MtTrxC, and MtTPx. As is evident from Fig. 2 MtTPx C60S is absolutely inactive, whereas MtTPx C80S is indistinguishable from the wild-type enzyme. The first finding complies with the expectation that Cys 60 is the C P , and the second observation rules out any substantial role of Cys 80 in the catalysis. At first glance the FIGURE 2. Activity determination of MtTPx and molecular mutants. Substrate turnover by MtTPx and molecular mutants was monitored continuously by NADPH consumption in a coupled test system at 25°C and pH 7.4 with t-butyl hydroperoxide as substrate and MtTrxC as cosubstrate essentially as described previously (7). Upon addition of tert-butyl hydroperoxide (arrows; sudden drop of absorbance due to dilution), the decline in absorbance caused by WT-MtTPx (D) and MtTPx C80S (C ) is identical, proving full activity of the latter. MtTPx C60S (A) is completely inactive, whereas MtTPx C93S (B) adopts an interim position. For the first 20 s after reaction start activity is clearly detected, although thereafter the rate of substrate turnover becomes indistinguishable from that of the inactive variant MtTPx C60S.
relevance of Cys 93 to catalysis is corroborated by a marginal residual activity of MtTPx C93S. A closer inspection of the trace, however, discloses that for a few seconds a discrete residual activity is retained in MtTPx C93S, although thereafter NADPH consumption approaches that of the negative control. This finding points rather to a fast inactivation than to an abrogation of catalytic efficacy itself.
Stopped-flow Analysis of MtTPx Oxidation by Peroxynitrite-In a first set of experiments, the oxidation of WT-MtTPx and MtTPx mutants by peroxynitrite was measured directly in the millisecond range by monitoring the absorption at ϭ 310, which is characteristic of peroxynitrite. For this purpose the enzyme samples were reduced by DTT and freed of reductant by gel filtration. Their total thiol content was determined to range between 2.6 and 3.4 thiol groups per subunit of protein, which is consistent with the absence of relevant amounts of low molecular thiols. Any traces of undetected residual DTT in this range of concentrations would not have affected the measurements, because its second-order rate constant with peroxynitrite is only 1100 M Ϫ1 s Ϫ1 (data not shown). In the presence of 40 M wild-type enzyme, peroxynitrite (18 M) decays very rapidly, in agreement with the fast second-order rate constant (k 1 Ј ϭ 1.5 ϫ 10 7 M Ϫ1 s Ϫ1 , pH 7.4, 25°C) that was reported previously for peroxynitrite-dependent MtTPx oxidation (7). The reaction is so fast that an important fraction of peroxynitrite (ϳ50%) already decayed during the first 2 ms, i.e. the mixing time of the apparatus (Fig. 3). With MtTPx C60S (46 M) the rate of peroxynitrite decay is practically the same as in the absence of enzyme, confirming that Cys 60 is responsible for the fast peroxynitrite reduction, i.e. Cys 60 is C P . MtTPx C80S and MtTPx C93S also react very fast with peroxynitrite. The rate constant for MtTPx C93S (2 ϫ 10 6 M Ϫ1 s Ϫ1 ) is, in fact, close to that established previously for the wild-type enzyme (1.5 ϫ 10 7 M Ϫ1 s Ϫ1 ). In view of the pronounced instability of MtTPx C93S under oxidizing conditions, the difference is likely explained by an overestimation of the native enzyme. In essence, the results are hardly compatible with an essential involvement of Cys 93 (or Cys 80 ) in the oxidative part of the catalytic process.
Reduction of Oxidized MtTPx by Its Natural Substrates-As Cys 93 is, by analogy, suspected to be the resolving cysteine, its replacement by serine might selectively affect the interaction with the reducing substrate. The limiting apparent rate constant k 2 Ј for the reduction of oxidized MtTPx by MtTrxB or -C had to be analyzed to demonstrate if MtTPx C93S can also act as a peroxynitrite reductase when using its natural reductants. This task, however, afforded a control model for which the relevant rate constants had been established previously or were accessible. Therefore, peroxynitrite decomposition in the presence of WT-MtTPx, MtTrxC, and combinations thereof was analyzed to evaluate how the system behaves, if the peroxynitrite-oxidized enzyme is being recycled by its reductant. The addition of pre-reduced MtTPx (1.1 M) to peroxynitrite (5.5 M) causes a fast initial peroxynitrite decomposition. After consumption of the under-stoichiometric amount of enzyme, peroxynitrite decays at the rate of spontaneous decomposition (Fig. 4A). The addition of pre-reduced MtTrxC (33 M) alone, which is also a thiol-containing protein, causes a modest increase in the peroxynitrite decomposition rate (from 0.26 to 0.59 s Ϫ1 ), from which a second-order rate constant for the reaction between peroxynitrite and MtTrxC of 1 ϫ 10 4 M Ϫ1 s Ϫ1 , pH 7.4, and 25°C is estimated. However, in the presence of both MtTPx (1.1 M) and pre-reduced MtTrxC (33 M), peroxynitrite decomposition increases substantially, clearly revealing the regeneration of reduced MtTPx by the thioredoxin. In fact, computer-assisted simulations of the reactions according to Equations 1-4 with the established rate constants perfectly match with the experimental data (Fig. 4B).
The same kind of experiment was then performed with the mutated forms of MtTPx to unravel the role of Cys 93 (and Cys 60 or Cys 80 ) in the reductive part of the catalytic cycle. As is shown in Fig. 4C has the same effect on the peroxynitrite decomposition rate as the wild-type enzyme (Fig. 4A) indicating that it is equally able to sustain the entire catalytic cycle. The experiment was repeated with the second natural reductant of MtTPx, MtTrxB. Again, the C93S mutant accelerated the rate of peroxynitrite reduction by the thioredoxin (Fig. 4D). These observations strongly suggest that Cys 93 is not required to resolve the oxidized state of the enzyme.
For final proof of a reduction of oxidized MtTPx without the aid of Cys 93 , i.e. without the necessity to first form a disulfide bridge between Cys 60 and Cys 93 , the decline of tryptophan fluorescence of MtTrxB upon oxidation with MtTPx was measured by the stopped-flow technique. For this purpose MtTrxB was reduced by DTT, freed of reductant, and then reacted with excess oxidized MtTPx (15-40 M). Despite an unfavorable signal to noise ratio, the decline of fluorescence upon oxidation of MtTrxB could be analyzed reliably (Fig. 5A). When plotting the k 2(obs) against the MtTPx concentrations, a k 2 Ј for the reaction of reduced TrxB with WT-MtTPx was calculated (8.2 ϫ 10 3 M Ϫ1 s Ϫ1 ) that is lower than the one obtained previously by steady-state kinetics by a factor of 5, but which is nevertheless indicative of an efficient and specific reaction taking into account the known tendency of peroxiredoxins to lose activity if kept under oxidizing conditions. More importantly, MtTPx C93S turned out Dispensability of Cys 93 at "Physiological" Fluxes of Peroxynitrite-Taken together, the results reported so far clearly indicate that the presumed C R is dispensable for sustaining the catalytic cycle of MtTPx but evidently prevents oxidative inactivation of the enzyme when analyzed at high peroxide concentrations, as are inevitably used for conventional activity determinations. In order to get an idea of the physiological relevance of this protective role of Cys 93 , we tested the activity of MtTPx C93S in a model system that allows us to monitor the prevention of oxidative damage exerted by a low flux of peroxynitrite, as may be expected under in vivo conditions. When DHR is exposed to SIN-1, it is oxidized by peroxynitrite-derived radicals to rhodamine in a process than can be followed at 500 nm. Under the conditions chosen, a constant DHR oxidation rate of 0.28 M/min (Fig. 6, shown in black) is obtained after a lag phase of 10 -15 min. At an estimated oxidation yield of 40%, peroxynitrite fluxes can be calculated to amount to 0.7 M min Ϫ1 . When 1.3 M MtTPx (wild type, C60S, or C93S), which had not been pre-reduced, are added once the lag phase is terminated (15 min after the addition of SIN-1), DHR oxidation by SIN-1 is not significantly affected. Pre-reduced MtTrxB (6.7 M), which at the concentrations chosen should only slowly reduce peroxynitrite, slows down the rate of DHR oxidation to 0.14 M/min, probably by competing for peroxynitrite-derived radicals (Fig. 6, blue line). In the presence of pre-reduced MtTrxB (6.7 M), however, WT-MtTPx as well as MtTPx C93S (each at 1.3 M, shown in red and green, respectively) completely suppress peroxynitrite-dependent DHR oxidation for almost 10 min, which roughly corresponds to the calculated time required for the total consumption of the Trx at given peroxynitrite fluxes. As expected, DHR oxidation is thereafter resumed at almost the same rate as in the absence of proteins. MtTPx C60S, as an inactive control, did not interfere at all with DHR oxidation (not shown). The results first of all confirm that MtTPx C93S is fully active as a Trx-dependent peroxidase and that it remains active despite exposure to peroxynitrite, as long as it is continuously reduced by excess of its natural substrate. Cys 93 thus proves to be not only dispensable  for activity of MtTPx but does not appear to be absolutely required for its stability either.
Mass Spectrometric Analysis of Observed Phenomena-To unravel which of the three cysteines of MtTPx can react with each other or with substrates, WT-MtTPx or MtTPx C93S were reacted with oxidizing and reducing substrates and analyzed by LC-MS/MS for revealing fragments, as described below and compiled in Table 1. Pepsin treatment of reaction products was preferentially applied, because the low pH during peptic digestion reliably prevents rearrangements of disulfide bonds (22), although data analysis is often tedious because of extensive degradation and poor specificity of pepsin. Complementary analyses had therefore to be performed after tryptic digestion, although the results obtained this way may be compromised by thiol/disulfide exchange because of the required alkaline pH. For identification of the site of preferred attack of the TPx by Trx, the molecular mutants of MtTrxB C34S 4 and MtTrxC C40S were used, which have the surface-exposed, reacting cysteines of their CGPC motifs preserved but the coreacting cysteines mutated to serine. Although much less reactive than WT-Trxs, these mutants can be expected to target their reaction partners identically in principle but to remain covalently bound to them via disulfide bonds (23).
When WT-MtTPx was investigated by LC-ESI-MS/MS immediately after reduction by DTT, as expected no disulfidecontaining fragments were obtained if the protein was cleaved under acidic conditions by pepsin, whereas each of the three cysteines of the protein were represented in more than one peptic fragment (data not shown). Upon re-oxidation, a fragment of 4105.7 m/z was isolated from the peptic digest that  (Fig. 7).
When MtTPx C93S was oxidized with H 2 O 2 and then reacted with excess WT-TrxB (Table 1, ii). a fragment suggestive of a disulfide bond within the TPx mutant could no longer be detected. Instead, oxidized TrxB was clearly identified, underscoring that the mutant TPx is enzymatically active. Furthermore, sequencing of a 2343.6 m/z fragment revealed that Cys 60 was oxidized to a sulfinic acid, which complies with the assumption that overoxidation of C P is the reason for the fast inactivation of this mutant under oxidizing conditions (Fig. 8). When reacted with the TrxB mutant C34S, pre-oxidized MtTPx C93S yielded a hybrid peptide that unexpectedly had Cys 80 linked to TrxB but also a fragment indicative of the expected TPx Cys 60 to TrxB Cys 31 bond. Unfortunately, this fragment resisted sequencing and is isobaric with the possible peptic TrxB fragment 42-69. Overoxidation of Cys 60 was evident as in the analogous experiment with WT-TrxB (Table 1, iii). Changing the sequence of substrate additions, i.e. adding TrxB C34S before H 2 O 2 ( Table 1, iv) prevented the overoxidation of the C P . The fragment indicating the reaction of TrxB with Cys 80 of the TPx was no longer seen, which supports the assumption that the latter is an artifact resulting from a reaction with an oxidatively altered enzyme. The "correct" although ambiguous fragment indicating the TPx Cys 60 -TrxB Cys 31 bond remained detectable. Cys 60 as sulfinic acid (Fig. 8); no disulfide-linked fragment detectable wt-TrxB- (24)(25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35)(36)(37)(38)(39)(40)(41) 2076.3 All oxidized; proves enzymatic activity of TPx C93S wt-TrxB- (27)(28)(29)(30)(31)(32)(33)(34)(35)(36)(37)(38)(39)(40)(41) 1698.9 wt-TrxB- (28- Analogous experiments were then carried out with WT-Mt-TPx (Table 1, v and vi). Irrespective of the sequence of substrate additions, a disulfide pattern identical to that obtained by oxidation of WT-TPx without any reductant (Table 1, i) was surprisingly observed, indicating that the formation of disulfide bonds within the WT-TPx is faster than the reduction of the sulfenic acid form by the pseudo-substrate TrxB C34S. When the pseudo-substrate was, however, incubated with the oxidized enzyme overnight (instead of 3 min only), the 2987.4 m/z fragment suggesting the formation of the TPx Cys 60 -TrxB Cys 31 bond was again detected (Table 1, vii). Because of the ambiguity of this fragment, an identically prepared sample was digested with trypsin (Table 1, viii). From the digest a fragment (m/z 5509) was isolated that was proven to represent the tryptic TPx peptide 46 -65 linked to the TrxB peptide 5-35 by MS/MS sequencing (Fig. 9). Similarly, WT-MtTPx, which had been preoxidized with 1 mM t-butyl hydroperoxide and then exposed to TrxC C40S overnight, yielded a tryptic fragment composed of the same TPx peptide (46 -65) and the TrxC peptide 12-41. As anticipated, a variety of additional disulfide-linked fragments could be identified in the tryptic digests (not shown). The unavoidable thiol/disulfide reshuffling during tryptic digestion precludes deciding between disulfide bonds preformed between the native proteins and those generated during sample work-up. Despite extensive search, however, not a single fragment could be detected that had the exposed cysteines of the Trxs bound to Cys 93 of MtTPx, and it may be rated as highly unlikely that particularly such disulfides are completely transformed into the detected ones. It is therefore concluded that the disulfide bridges between MtTrxB Cys 31 or MtTrxC Cys 37 to Cys 60 of MtTPx, which are preferentially seen, indeed reflect the natural attack of the thioredoxins on the peroxidase.

DISCUSSION
The results presented reveal that the catalytic mechanism of atypical 2-C-peroxiredoxins, as proposed for EcTPx by Choi et al. (9), must not be generalized, although many structural and functional parameters of this subfamily of peroxiredoxins appear to be identical. In fact, MtTPx shares with EcTPx (and other Prxs) a peroxidatic cysteine C P that is activated by an arginine and a threonine; like EcTPx, MtTPx is a homodimer but is functionally monomeric (7,16). The resolving Cys 95 of EcTPx is conserved in homologous position of the C-terminal domain (Cys 93 in MtTPx); like EcTPx (9, 10) MtTPx can form a disulfide bridge between the C P and this conserved cysteine within the same subunit; in both enzymes replacement of the distal cysteine (Cys 93 or Cys 95 , respectively) by serine was shown to affect activity, if measured with excess hydroperoxide at conventional time frames (Fig. 2). These similarities have led to the assumptions that in the atypical 2-C-Prxs the cysteine conserved in the N-terminal domain near position 60 is C P , which we confirm, and that the distal cysteine is generally the functional equivalent of the resolving cysteine C R of the typical 2-C-Prxs (9, 10), which is incompatible with our findings. Our stopped-flow kinetics demonstrate that in MtTPx the presumed C R is dispensable for catalysis; neither the rate of oxidation of MtTPx nor the reduction by its natural Trx substrates is significantly affected by mutating Cys 93 to serine. Therefore, in functional terms MtTPx has to be rated as a 1-C-Prx with Cys 60 being the peroxidatic and the resolving cysteine, and the functional role of its Cys 93 has to be redefined.
An attack by the reducing substrate(s) on the sulfenic acid form of C P could not be directly demonstrated by LC-MS/MS analysis of reaction products of the TPx with the pseudo-substrates MtTrxB C34S and TrxC C40S, because the latter are evidently not reactive enough to compete with disulfide formation or overoxidation of the peroxidase. When MtTPx, however, was exposed long term to the pseudo-substrates under oxidizing conditions, the latter became bound to MtTPx Cys 60 and not to Cys 93 . Although the kinetic experiments convincingly demonstrate that the Trxs directly attack C P , the MS data on pseudo-substrate adducts show that, unlike in 2-C-Prxs, the C P remains the site of preferred attack also after formation of the disulfide form. It should be noted that a specific C R is only exceptionally required to complete the catalytic cycle of a peroxidase working with chalcogen catalysis. In the analogous mechanism of glutathione peroxidases, oxidation and reduction take place at the same (seleno)cysteine (22); the identical situation is postulated for 1-C-Prxs, and even typical 2-C-Prxs with a mutated C R can catalyze the entire cycle with low molecular weight reductants such as DTT. A C R proved to be essential only, because a direct approach of the natural proteinaceous reductants to C P is sterically hindered in the 2-C-Prxs investigated. Instead, the sulfur of C R becomes accessible because of conformational changes during formation of a disulfide bridge between the C P and C R (reviewed in Refs. 14 and 15 and also see Refs. 24 and 25). In MtTPx, the initially oxidized C P , i.e. its sulfenic acid form, is evidently accessible to its natural reductants, and even in the Cys 60 -Cys 93 disulfide form the sulfur of Cys 60 remains similarly accessible, whereas that of Cys 93 is not.
Being dispensable for catalysis, Cys 93 nevertheless appears to have a pivotal role, i.e. protecting MtTPx against oxidative inactivation. With exposure to hydroperoxides, MtTPx C93S loses activity within seconds, and the analogous mutant of EcTPx was also reported to be rapidly inactivated by oxidants (10). Furthermore, the MS data (Table 1), although qualitative only, suggest that MtTPx C93S is more readily oxidized to the sulfinic acid form than the WT enzyme. The substrate-induced inactivation reflects the tendency of enzymes working with sulfur or selenium catalysis to become inactivated if their catalytic thiol or selenol groups become oxidized to an oxidation state Ն ϩ4 (26), which is a persistent hazard if the primary oxidation products, i.e. the sulfenic or selenenic acids, are further exposed to hydroperoxides at limiting concentrations of reducing thiols. This inactivation mechanism is common in the chalcogen catalysis of the Prx (12) and GPx families (27) and here could be directly demonstrated to occur in MtTPx, in particular in MtTPx C93S, by mass spectroscopy. Up to the sulfinic or seleninic acid status, such overoxidation is slowly reversed by thiols (28), but reduction of proteinaceous SO 2 H residues can also be accelerated by an ATP-dependent enzyme, sulfiredoxin (29), which has not yet been detected in bacteria.

Enzymatic Mechanism of Mycobacterial TPx
The more readily reversed oxidation level of Prxs is the disulfide status, which in the typical 2-C-Prxs is a regular catalytic intermediate. In the case of MtTPx, this redox status is not an obligatory intermediate of the catalysis but is observed, instead of the sulfenic acid form, in the absence of any reducing substrate. In this context, not only Cys 93 but also Cys 80 of MtTPx has to be considered as reaction partner of oxidized Cys 60 , but the Cys 60 -Cys 93 bond appears to be functionally more important, because Cys 80 cannot substitute for Cys 93 in the protection of the enzyme against oxidative inactivation. The disulfide form may be viewed as an "alternate catalytic intermediate," which can also be reduced by thioredoxins B and C. Pertinent rate constants are evidently high but should be distinct from those of the "regular" sulfenic acid intermediate, which explains the discrete differences of kinetic results depending on whether the analysis is started with reduced or oxidized enzyme. Unfortunately, the precise molarities of the oxidized enzyme forms remain unknown, because the mass spectroscopic analyses do not provide quantitative data. Accordingly, the kinetic experiments shown in Fig. 5 cannot differentiate between reduction of the Cys 60 -Cys 93 and Cys 60 -Cys 80 disulfide bonds. Also, the apparent rate constants estimated for the Trx-dependent reduction of the pre-oxidized peroxidases are certainly compromised by the presence of sulfinic acid forms, which are not readily reduced in the time frame of the experiments, or other oxidatively denatured enzyme species. Rate constants obtained for oxidized enzyme (Fig. 5) must therefore be rated as minimum estimates. Numerically, they are only lower by a factor of 5 than those obtained by steady-state kinetics (7) or stopped-flow approaches in the presence of sufficient reducing substrate (Fig. 4) and are very similar for oxidized WT-MtTPx and MtTPx C93S (Fig. 5).
In conclusion, MtTPx, when oxidized with peroxynitrite, acts as a 1-C-Prx as long as sufficient reduced Trx is available. When the reducing capacity is transiently exhausted, the peroxidatic Cys 60 can form a disulfide bridge with Cys 93 (and possibly with Cys 80 ), which is also quickly reduced by TrxB and TrxC. Unlike the typical 2-C-Prxs, however, formation of an internal disulfide bridge appears not to be an obligatory but a facultative step of MtTPx catalysis, and it certainly depends on the nature and concentration of substrates as to which alternate route is taken. The more important role of this disulfide formation in MtTPx is seen in the prevention of oxidation of Cys 60 to sulfinic acid at excessive concentrations of oxidants, which would lead to functionally incompetent enzymatic species that are not or are slowly reactivated without the support of a sulfiredoxin.