Inactivation of Human Peroxiredoxin I during Catalysis as the Result of the Oxidation of the Catalytic Site Cysteine to Cysteine-sulfinic Acid*

By following peroxiredoxin I (Prx I)-dependent NADPH oxidation spectrophotometrically, we observed that Prx I activity decreased gradually with time. The decay in activity was coincident with the conversion of Prx I to a more acidic species as assessed by two-dimensional gel electrophoresis. Mass spectral analysis and studies with Cys mutants determined that this shift in pI was due to selective oxidation of the catalytic site Cys-SH to Cys-SO2H. Thus, Cys -SOH generated as an intermediate during catalysis appeared to undergo occasional further oxidation to Cys-SO2H, which cannot be reversed by thioredoxin. The presence of H2O2 alone was not sufficient to cause oxidation of Cys to Cys-SO2H. Rather, the presence of complete catalytic components (H2O2, thioredoxin, thioredoxin reductase, and NADPH) was necessary, indicating that such hyperoxidation occurs only when Prx I is engaged in the catalytic cycle. Likewise, hyperoxidation of Cys/Ser mutant Prx I required not only H2O2, but also a catalysis-supporting thiol (dithiothreitol). Kinetic analysis of Prx I inactivation in the presence of a low steady-state level (<1 M) of H2O2 indicated that Prx I was hyperoxidized at a rate of 0.072% per turnover at 30 °C. Hyperoxidation of Prx I was also detected in HeLa cells treated with H2O2.

Peroxiredoxins are a family of peroxidases that reduce hydrogen peroxide and alkyl hydroperoxides to water and alcohol, respectively, with the use of reducing equivalents provided by thiol-containing proteins (1)(2)(3). The first peroxiredoxin (Prx) 1 proteins to be discovered were a 25-kDa yeast protein initially called thiol-specific antioxidant enzyme (4 -6), and a 21-kDa Salmonella typhimurium alkyl-hydroperoxide reductase termed AhpC (7)(8)(9)(10). Subsequently, a mammalian homolog of thiol-specific antioxidant/AhpC was purified; and it, together with thiol-specific antioxidant and AhpC, defined a family of peroxidases that now includes six mammalian isoforms (Prx I-VI) and members identified in organisms from each kingdom (3,10,11).
All Prx proteins contain a conserved Cys residue, which corresponds to Cys 51 in mammalian Prx I, in the N-terminal portion of the molecule (10,11). The majority of Prx proteins, including four (Prx I-IV) of six mammalian peroxiredoxins, contain an additional conserved Cys residue in the C-terminal region that corresponds to Cys 172 in mammalian Prx I (2,3,12,13). The Prx enzymes containing two conserved Cys residues are thus called 2-Cys Prx, in comparison with a small number of Prx proteins termed 1-Cys Prx, which contain only one conserved cysteine residue in the N-terminal domain (2,10). In 2-Cys Prx enzymes, the N-terminal conserved cysteine is oxidized by H 2 O 2 to cysteine-sulfenic acid (Cys 51 -SOH), which then reacts with Cys 172 -SH of the other subunit to produce an intermolecular disulfide (1,14). Reduction of the disulfide intermediate of Prx I-IV is specific in that it can be achieved by thioredoxin (Trx), but not by GSH or glutaredoxin (15,16). Thus, the reducing equivalents for the peroxidase activity of Prx I-IV are ultimately derived from NADPH via thioredoxin reductase (TrxR) and Trx. Not all Prx enzymes containing both conserved Cys residues are reduced by Trx: the bacterial Prx AhpC is reduced by AhpF, which contains both Trx and TrxR domains (10,17). In the absence of a physiological electron donor, the peroxidase activities of 2-Cys Prx enzymes can be supported by small thiol molecules such as dithiothreitol (DTT) and 2-mercaptoethanol, but not by GSH (1,14,15). In 1-Cys Prx enzymes, including mammalian Prx VI, the conserved cysteine is also the site of oxidation, but remains in a sulfenic acid state upon oxidation because there is no nearby partner cysteine to form a disulfide bond (18,19). Trx cannot reduce this sulfenic acid-containing intermediate (18,20). GSH has been proposed to be the electron donor, but these data remain controversial (18, 20 -23).
The crystal structures of 2-Cys and 1-Cys Prx enzymes reveal that the catalytic cysteine is located in a small pocket formed by the N-and C-terminal domains of the two subunits (19, 24 -27). The reactive cysteine is thus protected from larger oxidant molecules that contain disulfide linkages. The structures also show that the N-terminal conserved cysteine is surrounded by positively charged amino acid residues, which stabilize the thiolate (Cys-S Ϫ ) anion. The thiolate anion is more readily oxidized by peroxides than its protonated thiol counterpart (Cys-SH) (28). This provides the mechanistic basis for the observed sensitivity of the active-site cysteine to oxidation by peroxides.
Previously, we reported that Prx purified from yeast is readily inactivated during catalysis (1). We speculated that such inactivation occurs if the sulfenic acid moiety of the reaction intermediate is further oxidized by H 2 O 2 to cysteinesulfinic acid (Cys-SO 2 H) before disulfide formation with Cys 172 can occur (1). Sulfinic acid cannot be reduced by the Trx or DTT included in the assay mixture. Recently, Mitsumoto et al. (29) used two-dimensional PAGE to compare proteins in human umbilical vein endothelial cells before and after exposure of the cells to H 2 O 2 . In H 2 O 2 -treated cells, a number of proteins (including Prx I and Prx II) demonstrated altered migration consistent with decreased pI, suggesting that such oxidative inactivation might also occur in cells. However, these acidic Prx enzymes were not characterized in detail. We have now investigated the mechanism of human Prx I inactivation by H 2 O 2 .
Here, we demonstrate that the enzyme inactivation and concomitant acidic shift of Prx on two-dimensional gels are due in fact to the conversion of the active-site cysteine to Cys-SO 2 H. Furthermore, we observed that only those Prx molecules actively engaged in the catalytic cycle are vulnerable to oxidative inactivation.

MATERIALS AND METHODS
Preparation of Recombinant Proteins-The construction of a bacterial expression vector for human Prx I (pETprxI-WT) has been described (15). Two Prx I mutants in which Cys 51 and Cys 172 were individually replaced by serine residues (C51S and C172S, respectively) were generated by standard PCR-mediated site-directed mutagenesis with pETprxI-WT as the template and complementary primers containing a single-base mismatch that converts the codon for Cys to one for Ser. The final mutated PCR products were ligated into the pET vector to generate pETprxI-C51S and pETprxI-C172S. Escherichia coli BL21(DE3) competent cells (Novagen) were transformed with pETprxI-WT, pETprxI-C51S, or pETprxI-C172S; cultured at 37°C overnight in 100 ml of LB medium supplemented with ampicillin (100 g/ml); and then transferred to 10 liters of fresh LB medium in a Microferm fermentor (New Brunswick Scientific). When the absorbance of the culture at 600 nm reached 0.6 -0.8, expression was induced by isopropyl-␤-Dthiogalactopyranoside at a final concentration of 0.4 mM. After incubation for an additional 3 h, the cells were collected by centrifugation, frozen in liquid nitrogen, and stored at Ϫ70°C until used. The recombinant proteins were purified as described (30).
Prx Assay-NADPH oxidation coupled to the reduction of H 2 O 2 was monitored at 30°C as a decrease in A 340 using a Hewlett-Packard Model 8453 UV-visible spectrophotometer equipped with a thermostable cell holder and a multicell transport. The reaction was initiated by addition of the indicated concentration of Prx I to a 200-l reaction mixture containing 50 mM Hepes-NaOH (pH 7.0), 1 mM EDTA, 0.8 M TrxR, and the indicated concentrations of H 2 O 2 , NADPH, and Trx.
Cell Culture-HeLa S3 cells were adapted to suspension growth in Spinner minimum essential medium (Quality Biological, Inc.) supplemented with 10% (v/v) fetal bovine serum (Invitrogen). The cells were grown to a density of 1 ϫ 10 6 cells/ml and maintained by dilution with fresh complete medium every 4 days.
Sample Preparation, Two-dimensional PAGE, and Immunoblot Analysis-HeLa S3 cells were rinsed three times with ice-cold phosphate-buffered saline and lysed in lysis buffer (8 M urea, 4% CHAPS, and 40 mM Tris base) by sonication in a sonic bath three to four times for 30 s. After removal of insoluble materials by centrifugation at 14,000 ϫ g for 30 min, cell lysates were mixed with 10 volumes of rehydration buffer (8 M urea, 2% CHAPS, 0.5% immobilized pH gradient buffer, 20 mM DTT, and 0.005% bromphenol blue) and loaded onto immobilized pH gradient strips (pH 3-10, nonlinear). Isoelectric focusing on an IPGPhor isoelectrofocusing unit (Amersham Biosciences) and preparation (reduction and alkylation) of the immobilized pH gradient strips for the second-dimension SDS-PAGE were carried out according to the procedures recommended by the manufacturer. SDS-PAGE was conducted on 12% gels using an Amersham Biosciences SE600 vertical unit, and the protein spots were visualized by staining with silver nitrate. For immunoblot analyses of Prx enzymes, proteins on twodimensional gels were transferred electrophoretically to a nitrocellulose membrane, and the membrane was incubated with rabbit antibodies to Prx I. Immune complexes were detected with horseradish peroxidaseconjugated secondary antibodies and enhanced chemiluminescence reagents (Amersham Biosciences or Pierce). Silver-stained two-dimensional gels were scanned with a Molecular Dynamics SI personal densitometer.
Electrospray Ionization Mass Spectrometry (ESI-MS) Analysis of Prx I Proteins and Sequence Analysis of Tryptic Peptides-A Finnigan MAT LCQ electrospray ion-trap mass spectrometer was used for analysis of Prx I proteins and sequence analysis of Cys 51 -containing peptides. Dried HPLC fractions dissolved in 75% acetonitrile and 0.1% acetic acid were introduced at 1 l/min using a syringe pump. Data were collected for positive ions at 300 -2000 m/z using the following settings: capillary temperature, 215°C; maximum ion inject time, 100 ms; full scan target, 9 ϫ 10 7 ; and three microscans/full scan. For proteins, ESI-MS spectra were obtained and used to calculate the masses of proteins by deconvolution using Finnigan MAT BioWorks software. For sequence analysis of peptides, once the parent ions were identified, the mass spectrometer was set up to obtain collision-induced dissociation MS/MS spectra of the parent ions. The instrument for MS/MS spectra was configured as follows: capillary temperature, 215°C; maximum ion inject time, 100 ms; full scan target, 2 ϫ 10 7 ; three microscans/full scan; mass window, 2.5; and collision energy, used within a range of 0 -50%.
MALDI-TOF-MS-Mass spectrometry analysis of the tryptic peptides was performed on a Voyager-STR MALDI-TOF instrument (Per-Septive Biosystems, Framingham, MA) equipped with a nitrogen laser. Samples were dissolved in 50% acetonitrile and 5% formic acid, mixed with 10 mg/ml 3,5-dimethoxy-4-hydroxycinnamic acid (sinapinic acid; Aldrich) in 0.1% trifluoroacetic acid and 70% acetonitrile, and spotted on a sample plate. The spectra were obtained in the positive-ion reflector or linear mode with delayed extraction under standard conditions. Spectra were analyzed using DataExplorer software (PerSeptive Biosystems). Standard peptides were used for calibration of peptides, whereas apomyoglobin and carbonic anhydrase were used as internal standards for mass scale calibration of proteins.

Prx I Inactivation in Vitro-
The peroxidase activity of human Prx I was monitored by following the decrease in A 340 attributable to the oxidation of NADPH in a reaction mixture containing NADPH, Trx, TrxR, and varying concentrations of H 2 O 2 . The initial rate of NADPH oxidation (the slope at t ϭ 0) was independent of H 2 O 2 at saturating concentrations (0.1-1 mM) (K m for H 2 O 2 Ͻ 20 M (31)) ( Fig. 1A). However, the rate decreased with time, and the higher the H 2 O 2 concentration, the faster the rate of decrease. The H 2 O 2 concentration dependence of this decrease is shown more quantitatively by plotting the rate of NADPH oxidation (the first derivative of the NADPH oxidation curve) versus time (Fig. 1B). Replenishment of Prx I was shown to restore the enzyme rate, indicating that the markedly decreased NADPH oxidation rate was not attributable to exhaustion of substrate or to product inhibition (data not shown). Reactions containing 200 M H 2 O 2 were stopped at 0, 60, and 150 s, and the resulting samples were subjected to two-dimensional PAGE analysis. A single spot at a position corresponding to a pI value of 8.1 (the theoretical pI of Prx I is 8.2) was observed at 0 s. At 60 s, however, a new spot whose intensity was similar to that of the original spot appeared at a more acidic position (pI 7.6). At 150 s, the intensity of the acidic spot was enhanced at the expense of the original spot. The increasing proportion of the acidic species compared with the original spot is consistent with the notion that the 50 and 80% decreases in rate observed at 60 and 150 s, respectively, are related to the conversion of Prx I to a more acidic derivative.
Selective Hyperoxidation of Cys 51 -SH to Cys 51 -SO 2 H by H 2 O 2 -To identify the modification responsible for the acidic shift, Prx I was oxidized for 5 min in the presence of 1 mM H 2 O 2 and 10 mM DTT. Reduced and oxidized enzymes were separated from the reaction mixture by reverse-phase HPLC ( Fig.  2A). The molecular masses of the separated proteins were measured by ESI-MS (Fig. 2B). The masses calculated from m/z and total charge of multiply charged ions of the reduced and oxidized enzymes were 21,979 and 22,011 Da, respectively. The mass of the reduced enzyme was in good agreement with the theoretical mass of 21,979.2 Da. The difference of 32 mass units between the reduced and oxidized Prx I enzymes suggests the presence of two additional oxygen atoms in the oxidized species.
To determine the site of oxidation, the reduced and oxidized Prx I enzymes separated in Fig. 2A were digested with trypsin, and the resultant peptides were fractionated by reverse-phase HPLC (Fig. 2C). The HPLC elution profiles of the reduced and oxidized Prx I peptides were nearly identical except between 48 and 56 min (Fig. 2C). Major peptide peaks from the reduced and oxidized proteins were collected and analyzed by MALDI-TOF-MS, enabling the assignment of 12 such peaks to defined fragments of Prx I. In addition to the disulfide-forming Cys 51 and Cys 172 residues, Prx I contains two additional cysteine residues at positions 70 and 82. In both the reduced and oxidized states, the Cys 172 -containing peptide encompassing residues 168 -189 eluted from the HPLC column at 26.1 min. Peptides containing both Cys 70 and Cys 82 were also unchanged: in both states, the peptides encompassing residues 67-91 (representing one missed trypsin cleavage site at Lys 67 ) and residues 68 -91 eluted at 35.5 and 36.8 min, respectively. However, retention of Cys 51 -containing peptides was altered by oxidation. In the reduced state, the peptides encompassing residues 27-61 (representing two missed trypsin cleavage sites at Lys 34 and Lys 36 ), residues 35-61 (representing one missed cleavage site at Lys 36 ), and residues 37-61 eluted at 50.8, 51.7, and 55.1 min, respectively (Fig. 2D). In the oxidized protein, however, none of these peaks corresponding to the three Cys 51containing peptides were observed. Instead, several peaks eluted near the expected positions. When the peaks at 49.7, 50.4, and 53.0 min were analyzed by MALDI-TOF-MS, each of them displayed a dominant and a minor ion separated by 16 mass units (Fig. 2D). Furthermore, the dominant ions derived Mass spectral analysis of the tryptic peptides suggested that peroxide oxidation of Prx I Cys 51 yielded mainly Cys 51 -SO 2 H, but also Cys 51 -SO 3 H to a lesser extent. In Fig. 2, Prx I was digested with trypsin for only 3 h before HPLC analysis. However, when oxidized Prx I was digested overnight with trypsin, Cys 51 -SO 3 H-containing peptides became the major products (data not shown). Furthermore, purified Cys 51 -SO 2 H-containing peptides were slowly autoxidized to Cys 51 -SO 3 H-containing peptides upon exposure to air (data not shown). However, longer exposure of intact Prx I to either H 2 O 2 or air did not increase cysteic acid content (data not shown). Therefore, it appears that the formation of cysteic acid derivatives is the result of the air oxidation of Cys-SO 2 H-containing peptides. The sensitivity of Cys-SO 2 H to air has been reported (32).
Effect of Mutation of Cys 51 and Cys 172 on the Hyperoxidation of Prx I-Cysteines 51 and 172 were individually replaced by serine residues to generate the C51S and C172S mutant enzymes, respectively. The wild-type and mutant enzymes were subjected to two-dimensional PAGE analysis before and after incubation with a peroxidase reaction mixture containing H 2 O 2 , NADPH, Trx, and TrxR (Fig. 3). As observed in Fig. 1C, the wild-type enzyme migrated predominantly to the more acidic position upon incubation with H 2 O 2 , suggesting that Cys 51 -SH was hyperoxidized to Cys-SO 2 H. As expected, no such shift was observed for C51S. C172S, which is catalytically inactive because Cys 172 -SH is required for disulfide formation by Cys 51 -SOH, did not undergo hyperoxidation. In contrast, when reducing equivalents were provided by DTT rather than by the Trx system (Trx, TrxR, and NADPH), C172S became hyperoxidized just as the wild-type enzyme did, whereas C51S was still resistant to hyperoxidation. As previously shown with a yeast Prx mutant in which the C-terminal conserved Cys was changed to Ser (1), C172S is fully active when supported by DTT even though it lacks peroxidase activity in the presence of the Trx system (data not shown). Such catalysis is possible because a small diffusible thiol molecule like DTT can replace Cys 172 -SH in the formation of a disulfide with Cys 51 -SOH. The disulfide is subsequently reduced by DTT. These results indicate that oxidation of Cys 51 -SH to Cys 51 -SO 2 H occurs only when Prx is engaged in the catalytic cycle.
Effect of Trx Concentration on the Hyperoxidation of Prx I-To further examine whether the accumulation of the hyperoxidized protein requires continuous passage through the catalytic cycle, Prx I was inactivated by incubation with H 2 O 2 , NADPH, TrxR, and varying concentrations of Trx. An aliquot of the reaction mixture was removed at various times and assayed for peroxidase activity by measuring NADPH oxidation coupled to H 2 O 2 reduction. For each Trx concentration, the initial rate of NADPH oxidation was plotted against duration of inactivation to yield rates of enzyme inactivation (Fig. 4A). Prx I inactivation proceeded slowly in the absence of Trx and increased gradually with increasing Trx concentrations, reaching saturation at 4 M Trx. The Trx dependence of inactivation is displayed by plotting the residual activities measured after a 3-min incubation with the inactivation reaction mixture against Trx concentration. Half-maximal inactivation occurred at Ͻ0.5 M Trx. When either TrxR or NADPH was omitted from the inactivation mixture, inactivation essentially ceased, as observed in the absence of Trx (data not shown). These results suggest that active catalysis renders Prx I molecules susceptible to peroxidation.
Prx  In the presence of 45 milliunits of glucose oxidase, Prx I inactivation, reflected by the rate of NADPH oxidation (first derivative of the NADPH oxidation curve), occurred gradually during the first 1100 s, at which point NADPH was nearly depleted (Fig. 5A). The H 2 O 2 concentration remained below the limit of detection (Ͻ1 M) until NADPH depletion and increased rapidly thereafter. Both the rate of Prx I catalysis (NADPH consumption) and the rate of Prx I inactivation prior to NADPH depletion increased in a glucose oxidase-dependent manner when the amount of glucose oxidase was increased to 60 and 80 milliunits (Fig. 5, B and C). However, the concentration of H 2 O 2 remained below the sensitivity limit of analysis until just before the point of NADPH depletion, as with reaction conditions obtained with 45 milliunits of glucose oxidase. The NADPH oxidation rate (v) was fit to an equation describing the rate of the inactivation reaction as v ϭ v 0 Ϫkit to yield an initial rate (v 0 ) and inactivation rate constant (k i ) ( Table II). The peroxidase catalytic rate constant k cat in Table II (Table II). The average k i /k cat was 7.2 ϫ 10 Ϫ4 , indicating that 0.072% of Prx I molecules undergo inactivation per turnover when H 2 O 2 is maintained at a steady-state level. When 100 milliunits of glucose oxidase were added, the concentration of H 2 O 2 steadily increased, and the rate of inactivation also increased steadily before the rapid phase of inactivation (Fig. 5D). Therefore, these data were not used to calculate kinetic parameters. Prx Oxidation in Cells Exposed to H 2 O 2 -We examined whether Prx I molecules become hyperoxidized in cells exposed to H 2 O 2 . When HeLa cells were incubated for 30 min with various concentrations of H 2 O 2 , the dose-dependent conversion of Prx I to a more acidic spot was apparent (Fig. 6A). Mixing experiments showed that the acidic Prx I protein from H 2 O 2treated cells comigrated precisely with Prx I oxidized in vitro (Fig. 6B). Small quantities of the acidic Prx I form were con-  sistently observed in cells that had not been exposed to H 2 O 2 , suggesting that hyperoxidation also occurs in cells grown under normal conditions. To characterize the Cys oxidation state of Prx I in H 2 O 2 -treated cells, oxidized Prx I was purified from HeLa cells that had been exposed to 200 M H 2 O 2 for 30 min by subjecting the cell extracts to sequential HPLC on TSK DEAE-5PW, TSK heparin-5PW, and Vydac C 18 columns (data not shown). Reduced Prx I was also purified similarly from HeLa cells not exposed to H 2 O 2 . ESI-MS analysis revealed the difference of 32 mass units between the reduced and oxidized Prx I forms, suggesting that one of the Prx I Cys residues is in the sulfinic acid state in H 2 O 2 -treated cells (data not shown).

DISCUSSION
The catalytic cycle using Prx I as the archetype member of 2-Cys Prx enzymes is outlined in Fig. 7A. This model was initially proposed on the basis of studies with yeast Prx (1,14) and later substantiated by observations made with other 2-Cys Prx enzymes, including mammalian Prx V (13), bacterial AhpC (33), and trypanosomal TXNPx (34). During catalysis, Cys 51 -SH is selectively oxidized because this N-terminal conserved cysteine exists as a thiolate anion, whereas the other cysteines remain protonated at neutral pH. Cys 51 -SOH resulting from quantitative oxidation by H 2 O 2 then reacts with the C-terminal conserved Cys 172 -SH of the other subunit to form a disulfide. This intersubunit disulfide is reduced by Trx to complete one cycle of catalysis.
Sulfenic acids are usually very unstable and highly reactive; they readily react with any accessible thiol to form a disulfide or are prone to further oxidation to sulfinic acids or sulfonic acids (35)(36)(37). Nevertheless, stable intermediates containing Cys-SOH have been identified in various enzymes, including bacterial NADH peroxidase (32,38,39) and mammalian 1-Cys Prx (Prx VI) (19) using chemical methods, NMR, and x-ray crystallography. The major factor underlying the stability of the sulfenic acid intermediate in these proteins is the absence of a proximal partner Cys-SH. Stability is also attributable to the inaccessibility of Cys-SOH and participation of the sulfenate ion (Cys-SO Ϫ ) in charge or hydrogen-bonding interactions with other residues. For example, the x-ray structure of oxidized mammalian Prx VI revealed that Cys 47 -SO Ϫ not only has no neighboring Cys-SH, but also is stabilized in a deep narrow pocket through interactions with two positively charged residues (Arg 132 and His 39 ) and hydrogen bonding with a water molecule (19). Arg 132 and His 39 are likely the same residues that promote ionization of Cys 47 -SH in the reduced enzyme. In particular, Arg 132 of Prx VI, which is equivalent to Arg 127 of Prx I, is conserved in Prx enzymes from all species (2,3).
Our data presented here reveal that the Cys 51 -SOH intermediate of Prx I is occasionally oxidized to Cys-SO 2 H. Unlike disulfide and sulfenic acid intermediates, Cys-SO 2 H cannot be Ϫkit to yield the indicated lines (solid lines), from which the initial rates (v 0 ) and inactivation rate constants (k i ) listed in Table  II were derived. Data obtained during the first 100 s were not included because of the disturbance following addition of glucose oxidase and the possibility that the level of H 2 O 2 had not reached a steady state. The concentration of H 2 O 2 (E) was monitored during the NADPH oxidation reaction according to the procedure of Gay and Gebicki (44). The reaction mixture (18 l) was removed at the indicated times and added to a 200-l peroxide assay mixture containing 0.25 mM ferrous ammonium sulfate, 0.25 mM xylenol orange, and 110 mM perchloric acid. After a 30-min incubation at room temperature, A 560 of the ferric xylenol complex was measured against a blank, which contained all the components except glucose oxidase. H 2 O 2 concentrations were estimated from a standard curve generated with known concentrations of H 2 O 2 . b v 0 (initial rate of NADPH oxidation) and k i (inactivation rate constant) obtained as described for Fig. 5. reduced by Trx or DTT. Consequently, hyperoxidation leads to accumulation of inactivated enzyme, which is responsible for the gradual decay of Prx-dependent NADPH oxidation shown in Figs. 1, 4, and 5. The rate of enzyme inactivation increased when H 2 O 2 concentration was increased in the range of 0.1-1 mM (Fig. 1), despite an observed K m for H 2 O 2 of Ͻ20 M (a definite K m value could not be determined because the changes in A 340 were not large enough to allow evaluation of initial rates) (31). These observations indicate that although the initial oxidation of Cys 51 -SH is achieved by H 2 O 2 attracted to the active-site pocket with an affinity constant of Ͻ20 M, subsequent oxidation to Cys-SO 2 H depends upon collision with unbound or more weakly associated H 2 O 2 molecules. Nonetheless, as shown in Fig. 4, incubation with 1 mM H 2 O 2 alone did not cause significant inactivation. Prx I inactivation appeared to occur exclusively during active passage through the catalytic cycle, and increased catalytic rates were associated with increased rates of inactivation. One possible explanation for the dependence of inactivation on catalytic rate is that Cys 51 is susceptible to hyperoxidation only in the sulfenic acid state. Once Cys 51 -SH has established a disulfide with Cys 172 , this disulfide is no longer vulnerable to oxidation. Therefore, when enzyme molecules are found in the disulfide state with higher frequency because of a limited supply of reducing equivalents (Trx), the likelihood of hyperoxidation diminishes.
However, immunity of disulfide linkages to H 2 O 2 cannot account for the observation that hyperoxidation of the C172S mutant occurred in the presence of H 2 O 2 plus DTT, but not in the presence of H 2 O 2 alone (Fig. 3). Cys 51 -SH of the mutant protein was converted to the sulfenic acid state by H 2 O 2 , as evidenced by the fact that the C172S enzyme demonstrated peroxidase activity in the presence of DTT. C172S could function as a peroxidase in the presence DTT or 2-mercaptoethanol because these small molecules can donate thiol groups to disulfide linkages with Cys 51 -SOH and also reduce the resultant disulfides (Fig. 7B). Thus, C172S is similar to Prx VI, which contains only one conserved cysteine in the N-terminal domain and is catalytically active in the presence of DTT, but not Trx (18). The sulfenic acid oxidation state of the N-terminal conserved cysteine within oxidized Prx VI was demonstrated by x-ray structural analysis (19) and in an oxidized AhpC mutant in which the C-terminal Cys residue was replaced by Ser using a chemical method (33). The resistance of Cys 51 -SOH to further oxidation in the presence of H 2 O 2 alone suggests that the sulfenyl moiety is either shielded from H 2 O 2 or stabilized by surrounding residues such that its exposure to further oxidation is diminished. Perhaps nucleophilic attack of the sulfur atom of Cys 51 -SOH by either Cys 172 or DTT promotes structural shifts that expose and/or destabilize the sulfenyl moiety, rendering it reactive with H 2 O 2 . This would be consistent with the observed catalysis-dependent susceptibility of Cys 51 -SOH in both the wild-type and C172S mutant enzymes. A model invoking major structural changes finds strong support in the known three-dimensional structures of Prx enzymes.
The three-dimensional structures of four 2-Cys Prx enzymes (Prx I, Prx II, TXNPx, and AhpC) and one 1-Cys Prx enzyme (Prx VI) have been determined (19, 24 -27). Direct structural comparisons of the same Prx enzyme differing only in oxidation state at the N-terminal conserved cysteine are not available. However, examination of the structures reveals that in reduced enzymes, the sulfur atoms of the N-and C-terminal conserved cysteine residues are too far apart to react with each other (3,25,26). Thus, disulfide formation requires significant conformational changes that include unwinding of the active-site N-terminal helix and the movement of four loops (3,(25)(26)(27). Particularly valuable information was provided by the x-ray structure of Prx II that was purified from aged erythrocytes, in which Cys 51 is found in the sulfinic acid state (25). The sulfinic acid moiety is buried within the active-site pocket, whereas the thiol of Cys 172 is partially exposed, and the sulfur-oxyacid makes a salt bridge with Arg 127 , the same residue that interacts with the thiolate anion of Cys 51 in reduced form. As in the reduced enzyme, Cys 51 -SO 2 Ϫ and Cys 172 -SH are found far apart, with their sulfur atoms separated by ϳ13 Å (25). Other structural features also indicate that the sulfinic enzyme closely resembles its reduced form (25). On the basis of this information, we can predict that the sulfenic acid moiety of Prx I Cys 51 , also a sulfur-oxyacid group, is likewise shielded from solvent and stabilized by forming a salt bridge with Arg 127 such that it is impervious to further oxidation to sulfinic acid. To accommodate disulfide formation between Cys 51 -SOH and either Cys 172 or DTT, parts of the Prx protein must undergo major rearrangements (40). These conformational changes that necessarily accompany the catalytic cycle are likely to deprive the sulfenic acid moiety of shielding and stabilization, increasing its susceptibility to further oxidation.
Hyperoxidation to the sulfinic acid state is not a process that occurred only in the presence of high nonphysiological levels of H 2 O 2 . This hyperoxidation also took place, albeit slowly, when H 2 O 2 levels were maintained at low steady-state concentrations (Ͻ1 M) using glucose and glucose oxidase (Fig. 5). The limited sensitivity of the available analytic methods precluded measurement of H 2 O 2 concentration in the peroxidase reaction. Nevertheless, H 2 O 2 concentrations were expected to be constant over the time course of the reaction, independent of glucose oxidase concentration, as long as elimination by Prx I was faster than production by glucose oxidase. Under such steady-state conditions, the rate of Prx I inactivation increased as the rate of H 2 O 2 production by glucose oxidase (i.e. the rate of H 2 O 2 elimination by Prx I) increased. This is consistent with the notion that Prx I molecules are inactivated while engaged in the catalytic cycle and that the likelihood of inactivation increases as the enzyme rate increases. On the basis of data obtained at three different glucose oxidase concentrations, we estimated that 0.072% of Prx I molecules become hyperoxidized during each round of catalysis.
Our two-dimensional PAGE analysis showed a small spot of hyperoxidized Prx I from lysates derived from cells grown in the absence of H 2 O 2 (Fig. 6), indicating that the hyperoxidation of Prx I takes place in cells cultured under normal conditions. Hyperoxidation increased rapidly when increasing amounts of H 2 O 2 were applied to the cultures. While our studies were underway, Rabilloud et al. (41) reported that treatment of several different cells with alkyl hydroperoxide causes an acidic shift of Prx II and Prx III on two-dimensional gels. From the results of in-gel tryptic digestion followed by mass spectral analysis, they concluded that the shift is attributable to the oxidation of the N-terminal conserved cysteine to a sulfonic acid derivative. However, our results suggest that the acidic shift is associated with the formation of a sulfinic acid derivative, not a sulfonic acid derivative. We found that the sulfinic acid oxidation state in the intact enzyme is stable, probably because the sulfinic acid group is stabilized through a salt bridge to Arg 127 . When the sulfinic Prx protein was digested with trypsin, the sulfinic acid group was slowly autoxidized to the sulfonic acid state, suggesting that the sulfonic acid state reported by Rabilloud et al. (41) might be due to a stronger oxidation potential of alkyl hydroperoxide or an artifact of air oxidation in the course of two-dimensional PAGE and in-gel digestion. The latter suggestion is also supported by the fact that Prx II isolated from aged erythrocytes was in the sulfinic acid oxidation state (25).
Other 2-Cys Prx enzymes are as sensitive to hyperoxidation as Prx I (29,41). What might be the significance of Prx inactivation during catalysis? Apoptosis is one realm in which such inactivation could assume a regulatory role. Prx enzymes have been shown to protect various cell types from apoptosis induced by exogenous factors such as cytokines, serum deprivation, ceramide, H 2 O 2 , UV irradiation, and certain cancer drugs (42,43). These diverse stimuli trigger the production of reactive oxygen species that damage lipids, protein, and DNA, eliciting apoptosis as a consequence. Hyperoxidized Prx accumulates at the expense of active Prx as seen in H 2 O 2 -treated cells (Fig. 6). Accumulation of hyperoxidized Prx enzymes has also been shown in tumor necrosis factor-␣-treated cells (41). Meanwhile, the turnover rate of Prx enzymes in H 2 O 2 -treated cells remains as slow as that in untreated cells. 2 Thus, oxidation-mediated inactivation of Prx enzymes would inevitably disrupt the balance of production and elimination of H 2 O 2 , leading to H 2 O 2 accumulation. Enzyme inactivation proceeds in proportion to the number of catalytic cycles, and inactivated enzyme molecules would accumulate once the inactivation rate crosses a certain threshold. Thus, the fraction of inactive Prx molecules could serve as a record of the cell's history of exposure to oxidative stress. Cells that are subjected to prolonged oxidative stress are likely to suffer irreparable damage, necessitating programmed cell death. One speculation is that accumulation of hyperoxidized Prx enzymes at the expense of active enzyme molecules and the resulting accumulation of H 2 O 2 might trigger damage-induced programmed cell death when integrated with other cellular signals. This hypothesis obviously demands many more studies, including those addressing the fate of inactivated Prx enzymes.