Kinetics of formation and reactivity of the persulfide in the one-cysteine peroxiredoxin from Mycobacterium tuberculosis

Hydrogen sulfide (H2S) participates in prokaryotic metabolism and is associated with several physiological functions in mammals. H2S reacts with oxidized thiol derivatives (i.e. disulfides and sulfenic acids) and thereby forms persulfides, which are plausible transducers of the H2S-mediated signaling effects. The one-cysteine peroxiredoxin alkyl hydroperoxide reductase E from Mycobacterium tuberculosis (MtAhpE–SH) reacts fast with hydroperoxides, forming a stable sulfenic acid (MtAhpE–SOH), which we chose here as a model to study the interactions between H2S and peroxiredoxins (Prx). MtAhpE–SOH reacted with H2S, forming a persulfide (MtAhpE–SSH) detectable by mass spectrometry. The rate constant for this reaction was (1.4 ± 0.2) × 103 m−1 s−1 (pH 7.4, 25 °C), six times higher than that reported for the reaction with the main low-molecular-weight thiol in M. tuberculosis, mycothiol. H2S was able to complete the catalytic cycle of MtAhpE and, according to kinetic considerations, it could represent an alternative substrate in M. tuberculosis. MtAhpE–SSH reacted 43 times faster than did MtAhpE–SH with the unspecific electrophile 4,4′-dithiodipyridine, a disulfide that exhibits no preferential reactivity with peroxidatic cysteines, but MtAhpE–SSH was less reactive toward specific Prx substrates such as hydrogen peroxide and peroxynitrite. According to molecular dynamics simulations, this loss of specific reactivity could be explained by alterations in the MtAhpE active site. MtAhpE–SSH could transfer its sulfane sulfur to a low-molecular-weight thiol, a process likely facilitated by the low pKa of the leaving thiol MtAhpE–SH, highlighting the possibility that Prx participates in transpersulfidation. The findings of our study contribute to the understanding of persulfide formation and reactivity.

Persulfides are important intermediates in sulfur metabolism in bacteria, where they are produced in enzymatic catalytic cycles (16,17). Several enzymes, some also present in mammals, produce or transfer these functional groups; these include cystathionine ␥-lyase, cystathionine ␤-synthase (18), mercaptopyruvate sulfurtransferase (19), sulfide:quinone oxidoreductase (20,21) and thiosulfate sulfurtransferases (22,23). Other enzymes are able to react with persulfides; a dioxygenase can use GSH persulfide as substrate and is encoded by the ethe1 gene, which is mutated in ethylmalonic encephalopathy, a severe infantile metabolic disorder (24). Recently proposed as intermediates in the transduction of the signaling effects observed after the administration of H 2 S (25,26), persulfides have been generating increased interest. According to the hypothesis of persulfide-mediated signaling, the formation of a persulfide in certain cysteines could unleash changes in the activity of effector proteins, like the inhibition of papain, a cysteine-dependent protease (27), PTEN, a lipid phosphatase (28), and aquaporin-8, a membrane channel (29), among others. Possible roles in regulation and catalysis are still being explored, and the reactivity and physicochemical features of these species are poorly understood. Lately, some molecular models have been proposed and analytical methods have been developed to study persulfides both in vivo and in vitro (15, 30 -38). When thiols are modified to persulfides, nucleophilicity is maintained and probably increased due to two factors: (a) increased acidity with respect to thiols (39), which results in increased availability of the deprotonated, more nucleophilic form at neutral pH; and (b) the ␣ effect, i.e. the enhanced reactivity of a nucleophilic atom when it is adjacent to an atom containing one or more unshared pairs of electrons (40). In addition, a new property is acquired: electrophilicity. The reduction and the oxidation of persulfides are also possible; either H 2 S and thiols or perthiosulfenic acids (RSSOH) and polysulfides are produced, respectively. The high reactivity of persulfides determines the instability of these compounds in aqueous solutions (35), limiting their study and highlighting the importance of developing suitable models.
Peroxiredoxins (Prxs) are a family of antioxidant enzymes that play crucial roles in redox signaling (41)(42)(43). These enzymes are thiol-dependent peroxidases with ping-pong kinetic mechanisms. The oxidizing substrate (H 2 O 2 , organic hydroperoxide or peroxynitrite) reacts with the thiolate at the peroxidatic cysteine in the reduced enzyme to form a sulfenic acid (44). The reactivities of the thiolates in peroxidatic cysteines of Prx with hydroperoxides are several orders of magnitude faster than those of typical low-or high-molecular-weight thiols. This can be explained by the decrease in the energy of activation of the reaction by an exquisite network of electrostatic and hydrogen-bonding interactions involving the functional groups of an arginine and a threonine among others (45)(46)(47). Besides, the environment of the peroxidatic cysteine lowers the pK a of the thiol by several units relative to free cysteine (48). Once oxidized, the sulfenic acid is then reduced back to thiol by the reducing substrate(s), either directly or after a resolution step that involves the formation of a disulfide bond with a second cysteine residue (resolving cysteine), depending on the Prx subfamily (44,49). Often, a thioredoxin/thioredoxin reductase system reduces the disulfide bond to complete the catalytic cycle (44). The direct reduction of the sulfenic acid occurs in the so-called one-cysteine Prxs, such as alkyl hydroperoxide reductase E of M. tuberculosis (MtAhpE). This Prx catalyzes the reduction of several hydroperoxides, being most active with peroxynitrite and fatty acid hydroperoxides (50,51). The sulfenic acid of MtAhpE (MtAhpE-SOH) is reduced by the glutaredoxin-like protein mycoredoxin-1, either directly or after formation of a mixed disulfide with mycothiol (52)(53)(54). H 2 S is another possible reducing substrate for MtAhpE-SOH; however, it is not clear how effective its contribution could be. Moreover, both the ability of the resulting persulfide (MtAhpE-SSH) to react with typical Prx substrates or, alternatively, the capacity to be transferred to acceptor thiols remain unexplored.
In this work, MtAhpE was chosen as a model for persulfidation studies because this one-cysteine Prx presents the advantage that its sulfenic acid is relatively stable (52). We focused on the kinetic characterization of the reaction between MtAhpE-SOH and H 2 S to form a persulfide. Kinetic methods were employed to assess the possibility that H 2 S could act as a reducing substrate of the sulfenic acid and the relative contribution with respect to the better characterized mycobacterial reducing systems (mycothiol and mycoredoxin-1) is discussed. To compare the reactivity of the persulfide in the peroxidatic cysteine to that of the thiol, we evaluated the kinetics with specific substrates and unspecific reactants of Prxs. Additionally, we performed computational simulations to analyze the structural basis of the effects observed. Furthermore, the possibility of Prx assistance in persulfidation reactions (transpersulfidation) was explored.

Detection of the persulfide in MtAhpE
The formation of the persulfide from the reaction of H 2 S with MtAhpE-SOH was revealed by the detection of its alkylation product after treatment with iodoacetamide and by the reduction of this product to MtAhpE-SH with DTT (Fig. 1A). Cysteine modifications to MtAhpE-SSH and sulfinic acid (MtAhpE-SO 2 H) involve similar mass shifts. The use of an alkylating agent allows us to distinguish unequivocally the nature of the product, because only persulfides will be alkylated due to their high nucleophilicity in opposition to the poor one of sulfinic acids. Furthermore, alkylated persulfides are characteristically reduced by thiol-containing compounds like DTT (55). A species with a molecular mass of 19,408 Da, consistent with the S-carbamidomethyl derivative of the peroxidatic cysteine persulfide (MtAhpE-SS-CAM), was detected in equimolar mixtures of MtAhpE-SH (19,319 Da), H 2 S and H 2 O 2 at different incubation times. The species was already present when iodoacetamide was added 30 s after mixing, and maxi-Formation and reactivity of persulfide in MtAhpE mum yields were obtained after 2-5 min (Fig. 1B). MtAhpE-SSH was relatively stable, because it could still be detected after incubation times before alkylation of 15 and 30 min (Fig. S1), although the signal of the MtAhpE-SS-CAM derivative decreased while that corresponding to reduced MtAhpE (MtAhpE-S-CAM, 19,376 Da) increased. No alkylated MtAhpE polysulfide derivatives (i.e. MtAhpE-SS n S-CAM, with n Ն1) were detected under these experimental conditions. MtAhpE-SS-CAM, an unsymmetrical disulfide, was reduced by treatment with DTT to form the original thiol (19,319 Da) (Fig. 1C). Additionally, a peak corresponding to 19,352 Da was detected in all samples and was particularly evident in those where MtAhpE-SH was incubated with H 2 S and H 2 O 2 . Because this species remained after DTT addition, it most probably reflects the presence of protein over-oxidized to MtAhpE-SO 2 H.

Kinetics of the reaction of H 2 S with MtAhpE-SOH
As shown above, H 2 S is able to react with MtAhpE-SOH forming a persulfide. With the aim of evaluating the viability of this reaction among alternative reducing systems, the kinetics of the reaction was studied. Because direct measurements of concentration changes are not straightforward in this time scale, determinations were performed by two competition assays. The first one was a competition between H 2 S and 5-thio-2-nitrobenzoic acid (TNB) for MtAhpE-SOH ( Fig. 2A). The incubation of MtAhpE-SOH with the colored thiol TNB under pseudo-first-order conditions in the absence of H 2 S (Fig. 2B, blue trace) reproduced previous observations and confirmed the second-order rate constant of the direct reaction between MtAhpE-SOH and TNB to be (2.2 Ϯ 0.1) ϫ 10 3 M Ϫ1 s Ϫ1 (50). The reaction yielded a mixed disulfide (MtAhpE-S-TNB) that slowly reacted with a second molecule of TNB regenerating the reduced enzyme, as reported previously (50). When MtAhpE-SOH was mixed with TNB in the presence of H 2 S, the observed exponential rate constants of TNB decay (k obs ) increased linearly with the concentration of H 2 S while the amplitudes decreased (Fig. 2, B and C), as expected for competition kinetics (56). A second-order rate constant of (1.4 Ϯ 0.2) ϫ 10 3 M Ϫ1 s Ϫ1 for the reaction of H 2 S with MtAhpE-SOH (pH 7.4, 25°C) was obtained from this competition assay. The MtAhpE-S-TNB product reacted with excess H 2 S and produced TNB, explaining the linear increases in absorbance after the exponential decays of TNB in Fig. 2B.
A second approach for determining the kinetics of the reaction of H 2 S with MtAhpE-SOH consisted of a competition assay with the over-oxidation reaction of MtAhpE-SOH to MtAhpE-SO 2 H following the changes in the protein intrinsic fluorescence emission that occur when the enzyme is exposed to excess H 2 O 2 (50,52). The fast oxidation of MtAhpE-SH produced a rapid decrease in its fluorescence emission due to the formation of the MtAhpE-SOH during the first s, which is consistent with the reported rate constant of 8.2 ϫ 10 4 M Ϫ1 s Ϫ1 (50). A second phase of the reaction showed a slow recovery of the emission due to enzyme over-oxidation to MtAhpE-SO 2 H, as shown previously (50 -52). The presence of H 2 S produced changes in the amplitude and in the observed rate constants of this second phase (Fig. 3). The global fitting of the fluorescence changes according to a simple competition model C, mass spectra of MtAhpE-SH alone (black) or incubated with H 2 O 2 and Na 2 S for 5 min and treated with iodoacetamide and gel-filtered as in A, before (red) or after treatment with DTT (2 mM) (blue). (Fig. 3A) yielded a rate constant of (1.0 Ϯ 0.4) ϫ 10 3 M Ϫ1 s Ϫ1 for the reaction of H 2 S with MtAhpE-SOH, which is similar to the value determined using TNB described above. Also, the rate constant of over-oxidation yielded by the global fitting was 60 M Ϫ1 s Ϫ1 , in agreement with our previous report (50).

Reactivity of MtAhpE-SSH with an unspecific target, the disulfide DTDPy
The relative stability of the persulfide derivative of this Prx facilitated the study of its reactivity. The kinetics of the reaction of the persulfide toward the electrophile DTDPy was evaluated. DTDPy is a synthetic disulfude used as a reporter, which allows the study of the reactivity of both thiol and persulfide and is not expected to establish specific interactions with the active site. A mixture of MtAhpE-SOH and H 2 S was aged for increasing time periods and mixed with DTDPy under pseudo-first-order conditions (Fig. 4). The 4-thiopyridone released by reaction of DTDPy with H 2 S and MtAhpE-SSH was followed at 324 nm, and the kinetics showed a biphasic profile. The fast phase, attributable to the reaction of MtAhpE-SSH with DTDPy, increased its amplitude during the first 5 min while MtAhpE-SSH was being slowly produced, as expected from the concentrations of MtAhpE-SOH and H 2 S, and the rate constant of (1.4 Ϯ 0.2) ϫ 10 3 M Ϫ1 s Ϫ1 reported in the previous section. After ϳ400 s, the amplitude of the fast phase became constant, showing that all persulfide had been formed and remained stable during at least 30 min under these experimental conditions (Fig. 4B). From the amplitude of the fast phase, the amount of MtAhpE-SSH formed was calculated as 1.1 M, representing 81% of total MtAhpE. The slow phase that decreased in amplitude with time was attributed to the reaction of H 2 S with DTDPy and had a rate constant of 545 Ϯ 1 M Ϫ1 s Ϫ1 at pH 7.4 and 25°C. Given that at ϳ900 s the formation of MtAhpE-S-SH had reached a plateau while H 2 S had been depleted, experiments with varying concentrations of DTDPy were performed after this aging period. The observed rate constants for the fast phase increased linearly with DTDPy concentration (Fig. 4C). From the slope, the second-order rate constant for the reaction of MtAhpE-SSH with DTDPy was calculated to be (1.8 Ϯ 0.1) ϫ 10 3 M Ϫ1 s Ϫ1 at pH 7.4 and 25°C. In comparison, the rate constant for the reaction of the thiol MtAhpE-SH with DTDPy was 42 Ϯ 8 M Ϫ1 s Ϫ1 under the same conditions (Fig. 4C). The gray trace represents the best fit to an exponential plus straight-line function (Abs 412 nm ϭ A⅐exp(Ϫk obs ⅐t) ϩ b⅐t ϩ c, where A is the amplitude of the exponential contribution; k obs is the observed exponential rate constant; b is the slope of the linear contribution; and c is the offset). The data used for the fits corresponded to 10 half-lives minus the first second. C, exponential rate constants obtained from the exponential phase of absorbance decay versus Na 2 S concentration. Representative results of an experiment performed three independent times are shown.

Reactivity of MtAhpE-SSH toward the specific Prx substrates, peroxynitrite and H 2 O 2
Despite the increased intrinsic reactivity of the persulfide with respect to the thiol, the ability of MtAhpE-SSH to con-sume specific substrates of the enzyme, which implies additional factors, remained to be evaluated. The decay of peroxynitrite (12 M) was highly accelerated in the presence of 15 M reduced MtAhpE-SH (Fig. 5A, black trace) and occurred mostly in the mixing time of the apparatus (1.1 ms), in agreement with the fast reaction previously reported (1.7 ϫ 10 7 M Ϫ1 s Ϫ1 at pH 7.4 and 25°C (50)). On the contrary, both MtAhpE-SOH and MtAhpE-SSH caused only modest increases in peroxynitrite decay rate (Fig. 5A, blue and red traces, respectively). Considering ϳ80% MtAhpE-SSH formation yield, as determined above (Fig. 4A), the rate constant of the reaction with peroxynitrite was estimated as ϳ10 4 M Ϫ1 s Ϫ1 at pH 7.4 and 25°C.
In the case of H 2 O 2 , its consumption by MtAhpE-SH (45 M) was almost 1:1 during the first 30 s, as already described (51), and in agreement with the reported rate constant of 8.2 ϫ 10 4 M Ϫ1 s Ϫ1 at pH 7.4 (Fig. 5B, black squares) (50). In the presence of MtAhpE-SOH (Fig. 5B, blue circles), the decay of H 2 O 2 was slower, and from the initial rate of H 2 O 2 decay a rate constant of 78 M Ϫ1 s Ϫ1 was estimated, which agrees with the low rate constant of the reaction between MtAhpE-SOH and H 2 O 2 previously determined (40 M Ϫ1 s Ϫ1 ) (50). The rate of H 2 O 2 consumption by MtAhpE-SSH (Fig. 5B, red triangles) was similar to that of MtAhpE-SOH. From the initial rate of H 2 O 2 decay and considering persulfide formation yields of ϳ80%, the rate constant of MtAhpE-SSH oxidation by H 2 O 2 was estimated to be 109 M Ϫ1 s Ϫ1 at pH 7.4 and 25°C.

Catalytic consumption of H 2 S and H 2 O 2 by MtAhpE
To study whether H 2 S is able to complete the catalytic cycle of the enzyme, MtAhpE-SH was added to a solution containing H 2 S and H 2 O 2 , and aliquots were removed to determine H 2 S as well as H 2 O 2 consumption during the incubation. Although H 2 O 2 is able to consume H 2 S in the absence of catalysts, in agreement with the slow uncatalyzed reaction already reported (0.35 M Ϫ1 s Ϫ1 at pH 7.4 and 25°C (57)), increased rates of H 2 S and H 2 O 2 consumption were observed when substoichiometric amounts of the enzyme were included (Fig. 6, A and B). The initial rates of H 2 S decay showed a linear dependence on the concentration of MtAhpE-SH with a slope 2.2 ϫ 10 Ϫ2 s Ϫ1 at

Formation and reactivity of persulfide in MtAhpE
the assayed substrate concentrations, confirming the H 2 S-consuming activity in the presence of H 2 O 2 (Fig. 6C).

Molecular dynamics of MtAhpE-SSH
Molecular dynamics (MD) simulations of the thiolate (MtAhpE-S Ϫ ) and the persulfide forms (MtAhpE-SS Ϫ ) of the peroxidatic cysteine (Cys45) were performed to evaluate the structural and dynamic effect of persulfidation. As shown in Fig. 7, A and C, upon 700 ns of MD simulations, neither global structural rearrangements nor big conformational changes were detected when comparing the thiolate and the persulfide forms on the MtAhpE dimer (Fig. 7C). However, when taking a closer look at the active sites, significant disruptions in the interactions of Cys45 with the active-site residues Thr42 and Arg116 were identified. These interactions were previously highlighted as essential for hydroperoxide recognition and reduction, not only for this particular Prx (45,58) but also for the entire Prx family (46 -48, 59, 60). Specifically, a gain in the Thr42 region's flexibility upon persulfidation was observed (Fig. 7D). Furthermore, neither S ␥ nor S ␦ atoms of the persulfide in Cys45 interacted strongly with Thr42 and Arg116 (Fig. 7,  B and D), whereas strong interactions are established with the sulfur of the corresponding thiolate (45). In summary, MD simulations suggest that persulfidation of MtAhpE leads to a significant disruption in the topology of the active site that alters key interactions involved in catalysis.

Transfer of the sulfane sulfur of MtAhpE-SSH to a thiol probe
With the aim of determining whether MtAhpE-SSH is able to transfer its sulfane sulfur to other thiols, MtAhpE-SSH was incubated with sulfane sulfur probe 4 (SSP4), a nonfluorescent thiol-containing compound used as a probe for sulfane sulfurs. The transfer of a sulfane sulfur to a thiol of the probe leads to the formation of a persulfide, which undergoes spontaneous cyclization to release a fluorophore (61,62). Incubation of SSP4 with MtAhpE-SSH led to an increase in the fluorescence emission (Fig. 8). This increase occurred to a much larger extent than in controls with MtAhpE-SH or MtAhpE-SOH. According to calibration curves using Na 2 S 2 , the yield of the transsulfuration process was 95 Ϯ 5%.

Discussion
A plausible fate of H 2 S in cells is represented by its reactions with cysteine sulfenic acids (15,36). The peroxidatic cysteines in Prxs constitute preferential targets for hydroperoxides due to the high reactivity and cellular abundance (63,64). Their reaction constitutes a source of sulfenic acids, which are then reduced by several pathways. The feasibility of the reaction of H 2 S with a sulfenic acid is determined by kinetic aspects and is favored when the latter is long-lived. We particularly focused on MtAhpE-SH, which produces a relatively stable sulfenic acid due to the absence of thiols in the vicinity of the active site (52).
The sulfenic acid form of MtAhpE, MtAhpE-SOH, reacts with H 2 S to form a persulfide (MtAhpE-SSH). The rate constant for this reaction was determined to be (1.4 Ϯ 0.2) ϫ 10 3 M Ϫ1 s Ϫ1 at pH 7.4 and 25°C. Considering that the reactive species are protonated MtAhpE-SOH and ionized HS Ϫ (65) and because the reported pK a values are 6.6 for MtAhpE-SOH (50) and 6.9 for H 2 S (14), the pH-independent rate constant can be calculated to be 1.4 ϫ 10 4 M Ϫ1 s Ϫ1 . 8 The values obtained for MtAhpE-SOH can be compared with those obtained for the reaction of the sulfenic acid in human serum albumin (HSA-SOH) with H 2 S (15). The rate constant for HSA-SOH was (2.7 Ϯ 0.8) ϫ 10 2 M Ϫ1 s Ϫ1 at pH 7.4

Formation and reactivity of persulfide in MtAhpE
and 25°C, whereas the pH-independent rate constant was ϳ4 ϫ 10 2 M Ϫ1 s Ϫ1 (assuming that most HSA-SOH was protonated at pH 7.4). Thus, it can be concluded that MtAhpE-SOH is 30 times more reactive than HSA-SOH with HS Ϫ . Substoichiometric concentrations of MtAhpE were able to consume H 2 S and H 2 O 2 catalytically, suggesting that MtAhpE-SH can be regenerated and that the enzyme can initiate a new catalytic cycle. As precedent, it was proposed that bovine Prx6, another one-cysteine Prx from a different subfamily, is able to consume H 2 O 2 using H 2 S as a reducing substrate via the formation of a persulfide in its peroxidatic cysteine (66). In contrast, in Prx6 from Arenicola marina, no evidence could be obtained for H 2 S participation in the catalytic cycle (67 Equation 4). Then, MtAhpE-SSH can react with H 2 S to recover the original thiol and produce HSSH and other polysulfide by-products (Equation 5). Alternatively, MtAhpE-SSH can react with H 2 O 2 to form a perthiosulfenic acid (RSSOH) and other higher oxidation states (RSSO 2 H and RSSO 3 H), which can then be reduced to thiol or persulfide by H 2 S. The favored pathway is determined by kinetic aspects that remain to be elucidated. Nevertheless, the slope of the plot of rate versus enzyme concentration indicated a turnover of 2.2 ϫ 10 Ϫ2 s Ϫ1 at the used H 2 S and   (Fig. 6C). Assuming that the rate-limiting step is a second-order reaction, either the reaction of MtAhpE-SSH with H 2 S or with H 2 O 2 , and considering that the concentrations of H 2 S and H 2 O 2 were 125 and 127 M, respectively, it can be calculated that the rate constant for the rate-limiting second-order reaction is ϳ170 M Ϫ1 s Ϫ1 . Clearly, the reaction of MtAhpE-SOH with H 2 S, which has an 8-fold higher rate constant ((1.4 Ϯ 0.2) ϫ 10 3 M Ϫ1 s Ϫ1 , pH 7.4), is not rate-limiting in the catalytic process. Besides, the rate of the reaction of MtAhpE-SSH with H 2 O 2 (ϳ109 M Ϫ1 s Ϫ1 ) is below the expected rate-limiting step, leading to the reaction in Equation  5 as the most likely to participate in the catalytic cycling.

Formation and reactivity of persulfide in MtAhpE
In cellular contexts, the reaction of H 2 S with sulfenic acids could be of relevance in one-cysteine Prxs, where the resolving cysteine is absent and the sulfenic acid could be long-lived. It could also be relevant in eukaryotic two-cysteine Prx, particularly in those cases where the reaction of the resolving cysteine with the sulfenic acid to form a disulfide is relatively slow so that the sulfenic acid would have a significant half-life, and that is the case of eukaryotic typical two-cysteine Prxs in opposition to bacterial counterparts (68).
The second-order rate constant of the reaction of MtAhpE-SOH with H 2 S ((1.4 Ϯ 0.2) ϫ 10 3 M Ϫ1 s Ϫ1 ) is six times higher than that reported for the reaction of MtAhpE-SOH with mycothiol and is similar to that reported for mycoredoxin-1 (237 and 1.6 ϫ 10 3 M Ϫ1 s Ϫ1 , respectively (52)), which are the endogenous substrates in M. tuberculosis known up to date. The main fate of MtAhpE-SOH in cells is dictated not only by kinetic constants but also by the concentration of the targets. Although reports on the steady-state concentrations of mycoredoxin-1 in M. tuberculosis are still lacking, levels of both mycothiol (1-8 mM (69)) and H 2 S (ϳ370 M (69, 70)) have been estimated. Thus, H 2 S could represent an effective substrate in M. tuberculosis, an alternative to the mycothiol and mycoredoxin-1. The mechanisms of regulation of mycothiol and H 2 S synthesis in the bacterium are only starting to be unraveled (71,72); therefore, further work is required to establish their relative contribution for MtAhpE reduction during different metabolic conditions. Furthermore, the roles of H 2 S and mycothiol/mycoredoxin-1 as electron donors for AhpE and AhpE-like proteins expressed in other Actinomycetes (csb.wfu.edu/prex.test/ prxInfo.php?subfamilyϭ6) (94), which differ in mycothiol content (70) and can be exposed, depending on their habitat, to high H 2 S concentrations, deserve further investigation. Indeed, H 2 S supplementation was shown to complement the growth defect of bacterial strains with decreased ability to regenerate the reduced form of mycothiol (8). In addition, it was shown that Rv2238c, the gene encoding MtAhpE, is transcriptionally up-regulated in a cellular model of intraocular tuberculosis (73). Our study gives insights into possible mechanisms of cross-talk between the pathogen and its host at a junction between H 2 S signaling and the antioxidant defense systems.
The reactivity of MtAhpE-SSH toward an unspecific electrophilic target was probed using DTDPy. This synthetic disulfide was chosen because it has high intrinsic reactivity, because the reaction can be followed through the absorbance of 4-thiopyridone, because it can be used in pseudo-first-order excess so that the concentration of persulfide does not need to be exactly known, and because it has been used with HSA before (15). In addition, it constitutes an unspecific "substrate" for MtAhpE, which would allow us to interrogate thiol and persulfide reactivity in the absence of specificity aspects. Apparent second-order rate constants of (1.8 Ϯ 0.1) ϫ 10 3 M Ϫ1 s Ϫ1 and 42 Ϯ 8 M Ϫ1 s Ϫ1 were obtained at pH 7.4 for MtAhpE-SSH and for MtAhpE-SH, respectively. The value obtained for the persulfide was 43 times higher than that for the thiol. Considering that the reactive species are ionized, because the pK a of MtAhpE-SH is 5.2 (50), and the pK a of MtAhpE-SSH is also likely to be much lower than 7.4 (39), the values obtained at pH 7.4 are likely to reflect pH-independent values. Thus, the increased reactivity of MtAhpE-SSH with respect to MtAhpE-SH cannot be ascribed to changes in availability of the ionized species. Rather, they can be ascribed to an increase in intrinsic reactivity. A previous publication (15) reported rate constants for the reaction of DTDPy with the persulfide (HSA-SSH) and the thiol (HSA-SH) in human serum albumin as (1.7 Ϯ 0.1) ϫ 10 4 M Ϫ1 s Ϫ1 and (7.6 Ϯ 0.4) ϫ 10 2 M Ϫ1 s Ϫ1 , respectively, at pH 7.4 and 25°C (Table 1). These values translate into pH-independent rate constants of ϳ2 ϫ 10 4 M Ϫ1 s Ϫ1 for HSA-SSH and 7 ϫ 10 3 M Ϫ1 s Ϫ1 for HSA-SH, which has a pK a of 8.1 (74). Thus, the formation of a persulfide produced a 20-fold increase in the reactivity at pH 7.4 and just a 3-fold increase in pH-independent rate constants. The increase in intrinsic reactivity with DTDPy of the persulfide relative to the thiolate can be due to the ␣-effect, to changes in solvation, to alterations in weak interactions in the environment of the cysteine or to combinations of these effects.
Remarkably, the reactivity of MtAhpE-SH toward DTDPy at pH 7.4 was 1 order of magnitude lower than that of HSA-SH. In contrast, the reactivity toward hydroperoxides, the specific substrates of Prxs, is several orders of magnitude higher for MtAhpE-SH than for HSA-SH (Table 1). This is another example of the low reactivity of Prxs toward nonspecific compounds and contributes to the concept that there is no such thing as a general reactive cysteine (46, 75). The peroxidatic cysteine microenvironment in Prxs specifically accelerates the reaction with their hydroperoxide substrates.
In contrast to the increased reactivity toward the synthetic disulfide DTDPy of MtAhpE-SSH versus MtAhpE-SH, the

Formation and reactivity of persulfide in MtAhpE
reactivity of MtAhpE-SSH toward H 2 O 2 and peroxynitrite was several orders of magnitude lower than that of MtAhpE-SH.
The mild reactivity of MtAhpE-SSH with these specific Prx substrates appears to be an effect of geometrical distortion of the catalytic site, which seems to fit the requirements for the correct interaction of hydroperoxides with the peroxidatic cysteine in the thiolate but not the persulfide state, through hydrogen bonds with Arg116 and Thr42 during both the formation of the substrate complex and the transition state (45). In addition to changing the reactivity of this site due to shifts in distances, the formation of a persulfide could also change the value of the pK a . Although the acidity appears to be lower in low-molecular-weight persulfides with respect to their analogous thiols (39), it is not easy to predict the persulfide pK a in the case of a Prx because of the special environment of the active site. Furthermore, persulfides are expected to improve the reactivity as soft bases (15), which make them more likely to react with disulfides than with hard peroxides. Once formed in a Prx, what fate could the persulfide have? Although further reaction with H 2 S or H 2 O 2 can occur in vitro, in vivo it is likely that reactions with thiols predominate, considering the high cellular concentrations of low-and high-molecular-weight thiols. In fact, proteins of the thioredoxin and glutaredoxin families have been shown to react with persulfidated proteins (36,37). The possibility of direct attack of a protein thiolate in the outer sulfur of a Prx persulfide would be promoted by the relatively high acidity of the leaving group thiol. The result would be the formation of a persulfide in the attacking protein. This would constitute a mechanism for transpersulfidation that could contribute to the relatively high levels of persulfidation that have been detected (36 -38). As proof of concept, MtAhpE-SSH was able to transfer the persulfide to a low-molecular-weight thiol in high yield (Fig. 8). Thus, the reaction of H 2 S with Prx sulfenic acids shown in this study opens up the possibility of Prx participation in the persulfidation of proteins.

Protein thiol reduction and quantitation
MtAhpE was reduced immediately before use by incubation with 2 mM DTT for 30 min at 4°C. Excess reductant was removed by gel filtration using a HiTrap desalting column (Amersham Biosciences) and UV detection at 280 nm. Protein thiol content was measured by reaction with either DTNB (⑀ 412 ϭ 14,150 M Ϫ1 cm Ϫ1 ) (82) or DTDPy (⑀ 324 ϭ 21,400 M Ϫ1 cm Ϫ1 ) (83). As expected, the reduced enzyme contained one thiol per protein subunit.

Preparation of MtAhpE derivatives
MtAhpE-SOH was obtained by treatment of the reduced enzyme with an equivalent amount of H 2 O 2 in 0.1 M phosphate buffer with 0.1 mM DTPA (pH 7.4, 25°C). Incubation times required for completion of the reaction under the experimental conditions employed were determined by computational modeling using Gepasi (84). MtAhpE-SSH was prepared by mixing equimolar concentrations of both H 2 O 2 and Na 2 S with MtAhpE-SH. Because H 2 O 2 reacts several orders of magnitude faster with MtAhpE-SH than with H 2 S (8.2 ϫ 10 4 (50) versus 0.35 M Ϫ1 s Ϫ1 (57) at pH 7.4 and 25°C, respectively), MtAhpE-SOH is formed, which in turn reacts with H 2 S to form MtAhpE-SSH.

Detection of MtAhpE-SSH by ESI-Q MS
The detection of persulfides was carried out in samples of MtAhpE-SH (10 M) and Na 2 S (10 M) treated with H 2 O 2 (10 M) in 0.1 M phosphate buffer with 0.1 mM DTPA (pH 7.4, 25°C). At different incubation times, iodoacetamide (40 mM) was added for 30 min at 25°C. The excess alkylating agent was removed by gel filtration using PD SpinTrap G-25 (GE Healthcare) in 20 mM ammonium bicarbonate buffer (pH 7.4, 25°C). After this gel filtration step, some samples were further treated with DTT (2 mM) before analysis. All samples were loaded into a C4 column (GraceVydac 214MS5115) for HPLC separation. Mobile phase consisted of 0.1% formic acid in nanopure water (solvent A) and 0.1% formic acid in CH 3 CN (solvent B), and elution of the protein was performed with a 10-min linear gradient of solvent B (5-50%) followed by an additional 10 min at 50% solvent B at 100 l/min. An ESI-triple quadrupole mass Formation and reactivity of persulfide in MtAhpE spectrometer (QTrap4500, ABSciex) was employed for detection. The spectrometer was set in Q1 positive mode in the 500 -2000 m/z range with a scan rate of 200 Da/s and Q1 resolution in UNIT. Parameters used were as follows: IS, 5000; TEM, 300; DP, 120; EP, 10; CUR, 20; GS1, 30; GS2, 20. Data acquisition was done using Analyst 1.6.2 (ABSciex) and PeakView 2.2 (ABSciex) software was used for data analysis and deconvolution of all spectra.

Kinetics of H 2 S reaction with MtAhpE-SOH
The kinetics of the reaction of MtAhpE-SOH with H 2 S was determined by two competition approaches, using a SX20 Applied Photophysics stopped-flow spectrophotometer and either absorbance or fluorescence detectors. In the competition of H 2 S and TNB for MtAhpE-SOH, the latter (1.  (82). The exponential rate constants were obtained from the best fits to exponential plus straight line functions. The second-order rate constant of the reaction between H 2 S and MtAhpE-SOH was obtained from the slope of the plot of k obs versus Na 2 S concentration.
The competition of H 2 S and H 2 O 2 for MtAhpE-SOH was studied following total fluorescence emission ( ex 295 nm), taking advantage of the changes in the protein intrinsic fluorescence that occur during MtAhpE-SH oxidation to MtAhpE-SOH and over-oxidation to sulfinic acid (MtAhpE-SO 2 H) as described before (50). MtAhpE-SH (1 M) was mixed with solutions of Na 2 S (0 or 1 M) and H 2 O 2 (10, 100, 250 M) in 0.1 M phosphate buffer with 0.1 mM DTPA (pH 7.4, 25°C). Results were fitted and modeled using DynaFit (85) to determine the rate constant for the reaction of H 2 S toward MtAhpE-SOH. The initial concentration of the reagents was considered as well as the previously reported rate constant for the oxidation of MtAhpE-SH to MtAhpE-SOH and that of MtAhpE-SOH to MtAhpE-SO 2 H (50).

Reactivity of MtAhpE-SSH toward DTDPy
MtAhpE-SOH (1.3 M) was prepared as described above in a syringe of the SX20 Applied Photophysics stopped-flow spectrophotometer. Then, Na 2 S (1.3 M) was added to initiate the formation of MtAhpE-SSH. This solution was mixed at different aging times with DTDPy (100 -700 M), and the absorbance at 324 nm (⑀ 324 ϭ 21,400 M Ϫ1 cm Ϫ1 ) was recorded. Data were fitted to double-exponential functions to obtain the k obs and amplitudes of the phases. The rate constant of the reaction of MtAhpE-SSH with DTDPy was calculated from the slope of the plot of the observed rate constants for the fast phase versus DTDPy concentration obtained for mixtures of MtAhpE-SH, H 2 O 2 and Na 2 S aged for at least 900 s.

Reactivity of MtAhpE-SSH toward peroxynitrite and H 2 O 2
The decay of peroxynitrite (12 M) in the absence or presence of MtAhpE-SH, MtAhpE-SOH or MtAhpE-SSH (15 M) in 0.1 M phosphate buffer with 0.1 mM DTPA (pH 7.4, 25°C) was followed at 310 nm using an SX20 Applied Photophysics stopped-flow spectrophotometer. The reduction of H 2 O 2 by the different forms of MtAhpE was followed using the ferrous oxidation-xylenol orange method (FOX assay) (51,86 . Aliquots (100 l) were taken at different times, mixed with 900 l of the FOX reagent, and further incubated for 30 min at room temperature followed by absorbance measurement at 560 nm. The extinction coefficient for H 2 O 2 using this assay was determined (⑀ 560 ϭ 51,520 M Ϫ1 cm Ϫ1 ) and was in close agreement with previously reported values (51). The second-order rate constants of the reduction of these oxidants by the different forms of MtAhpE were estimated by initial velocity kinetics. ) using sealed vials. Samples were removed using gas-tight syringes and the remaining H 2 S was determined by the methylene blue method (87). Alternatively, the same concentrations of H 2 O 2 and Na 2 S used in the previous experiment were mixed in the absence or presence of MtAhpE-SH (2.9 M) in 0.05 M phosphate buffer (pH 7.4, 25°C), and the remaining H 2 O 2 was determined by the FOX assay as described above.

Molecular dynamics of MtAhpE-SH and MtAhpE-SSH
Classical MD of the MtAhpE dimer were performed for the thiolate form of MtAhpE-SH (MtAhpE-S Ϫ ) and for the persulfide form (MtAhpE-SS Ϫ ). For the MtAhpE-S Ϫ MD, the recently reviewed crystal structure of MtAhpE (PDB code 4X0X) (88) was used as the starting structure. The persulfide initial model was generated by a modification of the oxidized structure of MtAhpE (PDB code 4X1U) (88) in which the sulfenic acid was in silico transformed to the persulfide form (MtAhpE-SS Ϫ ). Classical parameters for simulating cysteine persulfide were developed using standard protocols (89).
Both initial models were submitted to the same MD protocol as we have previously performed for this and other related enzymes (45,58,90). Briefly, the system was solvated with an octahedral box 12 Å in radius with TIP3P water molecules (91). With the exception of cysteine persulfide in the MtAhpE-SS Ϫ model, residue parameters correspond to the parm14SB AMBER force field (92). Simulations were performed using periodic boundary conditions with a 10-Å cutoff and a particle mesh Ewald summation method for treating the electrostatic interactions. The hydrogen bond lengths were kept at their equilibrium distance by using the SHAKE algorithm, whereas temperature and pressure were kept constant with a Langevin thermostat and barostat, respectively, as implemented in the AMBER14 program. The system was minimized in 1000 steps (10 with steep gradient and the rest with the conjugate gradient). Then, it was heated from 0 to 300 K for 20 ps at constant pressure with a Berendsen thermostat, and pressure was equilibrated at 1 bar for 5 ps. After these two steps, a 10-ns MD long simulation at a constant temperature (300 K) and a constant volume was performed. An unrestrained 700-ns-long production MD at the NPT ensemble was performed. All dynamics

Formation and reactivity of persulfide in MtAhpE
visualizations and molecular drawings were performed with VMD 1.9.1 (93).

Transfer of the sulfane sulfur of MtAhpE-SSH to the thiol in SSP4
MtAhpE-SOH was prepared by incubation of MtAhpE-SH (4 M) with H 2 O 2 (4 M) in 0.1 M phosphate buffer with 0.1 mM DTPA (pH 7.4, 25°C) during 2 min. MtAhpE-SSH was formed by adding Na 2 S (4 M) to sulfenic acid preparations and aging during 15 min. Then, samples were diluted 2-fold in phosphate buffer and mixed with SSP4 (20 M and 4% DMSO, final concentrations). Fluorescence emission at 515 nm ( ex ϭ 482 nm) was recorded in a Varioskan Flash plate reader (Thermo Fisher Scientific). Calibration curves were performed with Na 2 S 2 assuming equimolarity of sulfane sulfur and Na 2 S 2 in reference solutions. The yield of sulfane sulfur transfer to the probe was estimated based on the amplitude of the fit of the fluorescence increase to an exponential function.

Data processing
Data were plotted and analyzed using OriginPro 8.0 (Origin-Lab). Unless specified, results are expressed as the mean Ϯ S.D. of independent experiments.