Cysteine-independent Catalase-like Activity of Vertebrate Peroxiredoxin 1 (Prx1)*

Background: Peroxiredoxin (Prx) was previously known only as a Cys-dependent thioredoxin. Results: Cys-independent catalase-like activity was observed in two vertebrate Prx1 proteins. Conclusion: Prx1 possesses dual antioxidant activities with varied affinities toward H2O2. Significance: This discovery extends our knowledge on Prx1 and provides new opportunities to further study the biological roles of this family of antioxidants. Peroxiredoxins (Prxs) are a ubiquitous family of antioxidant proteins that are known as thioredoxin peroxidases. Here we report that Prx1 proteins from Tetraodon nigroviridis and humans also possess a previously unknown catalase-like activity that is independent of Cys residues and reductants but dependent on iron. We identified that the GVL motif was essential to the catalase (CAT)-like activity of Prx1 but not to the Cys-dependent thioredoxin peroxidase (POX) activity, and we generated mutants lacking POX and/or CAT activities for individually delineating their functional features. We discovered that the TnPrx1 POX and CAT activities possessed different kinetic features in reducing H2O2. The overexpression of wild-type TnPrx1 and mutants differentially regulated the intracellular levels of reactive oxygen species and p38 phosphorylation in HEK-293T cells treated with H2O2. These observations suggest that the dual antioxidant activities of Prx1 may be crucial for organisms to mediate intracellular redox homeostasis.

Peroxiredoxins (Prxs) are a ubiquitous family of antioxidant proteins that are known as thioredoxin peroxidases. Here we report that Prx1 proteins from Tetraodon nigroviridis and humans also possess a previously unknown catalase-like activity that is independent of Cys residues and reductants but dependent on iron. We identified that the GVL motif was essential to the catalase (CAT)-like activity of Prx1 but not to the Cys-dependent thioredoxin peroxidase (POX) activity, and we generated mutants lacking POX and/or CAT activities for individually delineating their functional features. We discovered that the TnPrx1 POX and CAT activities possessed different kinetic features in reducing H 2 O 2 . The overexpression of wild-type TnPrx1 and mutants differentially regulated the intracellular levels of reactive oxygen species and p38 phosphorylation in HEK-293T cells treated with H 2 O 2 . These observations suggest that the dual antioxidant activities of Prx1 may be crucial for organisms to mediate intracellular redox homeostasis.
Peroxiredoxins (Prxs) 5 are a family of ubiquitous antioxidant enzymes known to be involved in sensing and detoxifying hydrogen peroxide (H 2 O 2 ) and other reactive oxygen species (ROS) in all biological kingdoms (1)(2)(3). Mammalian Prxs also participate in the regulation of signal transduction by controlling the cytokine-induced peroxide levels (4 -6). Humans and other mammals possess six Prx isoforms, including four typical 2-cysteine (2-Cys) Prxs (Prx1-4), an atypical 2-Cys Prx (Prx5), and a 1-Cys Prx (Prx6) (7)(8)(9). The thioredoxin peroxidase (POX) activity is the hallmark of Prx proteins. In the case of Prx1-4, the conserved N-terminal peroxidatic Cys residue (Cys P -SH, corresponding to the Cys 51 in the mammalian Prx1) is oxidized by H 2 O 2 to cysteine sulfenic acid (Cys P -SOH) and then resolved by a reaction with the C-terminal resolving Cys 172 (Cys R -SH) in the adjacent monomer to form a disulfidebound Cys 51 and Cys 172 . The disulfide linkage is reduced by NADPH-dependent thioredoxin (Trx)/Trx reductase cycles to complete the Prx catalytic cycle in cells or by a reducing agent such as dithiothreitol (DTT) commonly used in assaying POX activity (10 -12). Alternatively, at least the Cys P -SH and Cys R -SH residues in Homo sapiens Prx1 (HsPrx1) can be glutathionylated in the presence of a small amount of H 2 O 2 and deglutathionylated by sulfiredoxin or glutaredoxin I. Cys P -SH may also be hyperoxidized in the presence of an excessive amount of H 2 O 2 to form reversible sulfinic acid (Cys P -SO 2 H), which can be slowly recycled by sulfiredoxin, or irreversible sulfonic acid (Cys P -SO 3 H), resulting in the loss of the POX activity and the formation of Prx1 decamers with protein chaperone function (13)(14)(15)(16)(17). Among these reactions, the rapid recycling of POX activity is responsible for the reduction of H 2 O 2 and other ROS, whereas the other two appear to be involved in the regulation of Prx functions (18).
Although Prxs can be oxidized in multiple ways, all these POX activities rely on the Cys-dependent peroxidation cycles. However, in the present study, we unexpectedly observed that the Prx1 from the green spotted puffer fish Tetraodon nigroviridis (TnPrx1) was able to reduce H 2 O 2 that was independent of the Cys peroxidation and in the absence of reducing agents. This Cys-independent activity observed in wild-type (WT) and site-mutated TnPrx1 proteins differs from the classic POX activity in Prxs but resembles the catalase-like activity, making Prx1 a dual antioxidant protein. For clarity, we have denoted Cys-dependent POX and Cys-independent CAT-like activities in TnPrx1 as TnPrx1-POX and TnPrx1-CAT, respectively. We determined the detailed kinetic features of the TnPrx1-CAT activity and identified that the 117 GVL 119 motif was essential to this activity. Using a human embryonic kidney 293T (HEK-293T) cell transfection system, we showed that the TnPrx1-CAT participated in the regulation of H 2 O 2 and H 2 O 2 -dependent phosphorylation of p38 in cells. Additionally, CAT activity was also confirmed in HsPrx1, suggesting that the Cys-independent Prx1-CAT activity is conserved from fish to mammals.

Cloning and Expression of Recombinant Prx1 Proteins-The
Prx1 open reading frames (ORFs) of T. nigroviridis (TnPrx1) and H. sapiens (HsPrx1) were amplified by RT-PCR from mRNA isolated from pufferfish kidney and HeLa cells (corresponding to GenBank accession numbers DQ003333 and NM_001202431, respectively) and cloned into the pET28a bacterial expression vector containing a His 6 tag at the N terminus as described (19). TnPrx1 mutants were generated by site-directed mutagenesis by replacing all three Cys residues (i.e. Cys 52 , Cys 71, and Cys 173 ) with Ser residues to eliminate POX activity (denoted by POX Ϫ CAT ϩ ), the 117 GVL 119 motif with 117 HLW 119 to eliminate the CAT-like activity (POX ϩ CAT Ϫ ), or both (POX Ϫ CAT Ϫ ) (see Table 1 for details on the genotypes of constructs). Recombinant Prx1 proteins were expressed in Escherichia coli and purified from the soluble fractions by nickel-nitrilotriacetic acid-agarose bead-based chromatography and eluted with elution buffer containing 250 mM imidazole or as specified (19). Purified Prx1 proteins were subjected to SDS-PAGE analysis and stained with Coomassie Brilliant Blue. The protein purities were determined by densitometry using 1D Image Analysis Software with a Kodak Gel Logic 200 Imaging System (Eastman Kodak Co.).
A reversible monomer-to-dimer transition system was established to evaluate the Cys-dependent formation of dimers in which the purified recombinant proteins in the form of mono-mers were first allowed to be oxidized to form dimers in air at 4°C, and then the resulting protein dimers were reduced to monomers by the treatment with DTT (50 mM or as specified) at room temperature for 10 min. The reduced and oxidized forms of Prx1 were detected by non-reducing SDS-PAGE.
Protein Structure Homology Modeling-TnPrx1 protein structure homology modeling was performed using rat Prx1 (Protein Data Bank code 1QQ2; 80% identity) as a template. Global alignment of various structural models was performed by using PyMOL to produce various structural model figures. The active site of TnPrx1 was predicted using an ␣ shape algorithm to determine potential active sites in three-dimensional protein structures in MOE Site Finder, and further mutation was designed to disturb the structure of the active site. Site-directed mutants of TnPrx1 were constructed using the overlapping extension PCR strategy. Primers used in the experiments are shown in Table 2. All constructed plasmids were sequenced to verify the correct gene insertion and successful mutation.
Enzyme Activity Assays-The reduction of H 2 O 2 by TnPrx1 and HsPrx1 was determined by a modified sensitive Co(II) catalysis luminol chemiluminescence assay as described (20). Briefly, the luminol-buffer mixture was composed by 100 l of luminol (100 mg ml Ϫ1 ) in borate buffer (0.05 M, pH 10.0) and 1 ml of Co(II)-EDTA (2 and 10 mg ml Ϫ1 , respectively, pH 9.0). Reactions started with mixing 50 l of proteins (50 g ml Ϫ1 ) with 50 l of a series of H 2 O 2 solutions (0 -500 M) for 1 min at 25°C followed by adding 1.1 ml of the luminol-buffer mixture to stop the reaction. The same amount of phosphate-buffered saline (PBS) was used to replace proteins in the control and for generating standard curves.
The intensity of emission was measured with an FB12 luminometer (Berthold Detection Systems, Pforzheim, Germany), and the maximum values were recorded. The kinetic parameters of Prx1 proteins were determined using the Michaelis-Menten and/or allosteric sigmoidal kinetic models. The production of oxygen was measured with an oxygen electrode (341003038/9513468, Mettler Toledo). The reaction was performed in 4 ml of 600 M H 2 O 2 solutions, and the measurement was started by addition of proteins, POX ϩ CAT ϩ dimers (0.32  Determination of Enzyme Properties-The effect of pH on Prx1-CAT activity was evaluated by detecting the reduction of H 2 O 2 in reactions carried out in 0.2 mM Na 2 HPO 4 , 0.1 mM citrate buffer for pH 2.0 -8.0 and 50 mM disodium pyrophosphate, NaOH buffer for pH 8.0 -11.0, respectively. The effect of temperature was tested between 0 and 70°C at pH 7.0. The thermal and pH stabilities were similarly assayed except that concentrated Prx1 proteins were first treated at 0°C for 1 h at various temperatures or 6 h under various pH conditions, and then their specific activity was determined under regular assay conditions (i.e. pH 7.0 at room temperature).
Specific activities were also assayed for iron-saturated proteins prepared by mixing proteins with FeCl 3 (1:100 molar ratio) followed by ultrafiltration (molecular mass cutoff at 10 kDa) to remove unbound iron. The role of iron in Prx1-CAT was further evaluated by iron chelation and rescue assays in which TnPrx1 proteins were treated with Tiron (4,5-dihydroxy-1,3-benzene disulfonic acid; 25 mM) and 2,2-dipyridy (50 mM) at 4°C overnight followed by ultrafiltration. Chelatortreated samples were then incubated with FeCl 3 (200 M or as specified) at 4°C overnight followed by ultrafiltration to remove unbound iron. The residual TnPrx1-CAT activities of iron-free and iron-rescued proteins were determined in standard reactions as described above. To confirm that iron was truly bound to TnPrx1, iron-rescued samples were subjected to extensive ultrafiltration with PBS, and the iron content was detected by inductively coupled plasma optical emission spectroscopy (Optima 8000DV, PerkinElmer Life Sciences) as described (21).

Effects of WT TnPrx1 and Mutants on Intracellular ROS Level and the Phosphorylation of p38 Mitogen-activated Protein
Kinase (MAPK)-The ORFs of WT and TnPrx1 mutants were subcloned into pCMV-Tag2B vector. HEK-293T cells were maintained at 37°C in Dulbecco's modified Eagle's medium (DMEM; Invitrogen) supplemented with 10% fetal bovine serum (FBS; Gibco) in the presence of 5% CO 2 . For transient transfection, cells were plated in 100-mm cell culture plates (1.4 ϫ 10 6 cells/plate), grown overnight, and transfected with 17 g of WT Prx1 or mutant plasmids using FuGENE reagent (Promega). Blank pCMV-Tag2B vector was used as a negative control. After 48 h of transfection, cells were washed with PBS and incubated with 2Ј,7Ј-dichlorodihydrofluorescein diacetate (200 M; Sigma) in serum-free medium at 37°C for 30 min to allow uptake by cells and intracellular cleavage of the diacetate groups by thioesterase. Cells were washed with PBS to remove free 2Ј,7Ј-dichlorodihydrofluorescein diacetate from the medium and counted by the trypan blue (0.4%) exclusion method. Viable cells were then plated into 96-well collagen-coated plates (2 ϫ 10 4 cells/well) and treated with H 2 O 2 at final concentrations between 0 and 850 M for 60 min.
Intracellular fluorescence signals of oxidized dichlorodihydrofluorescein at 0 and 1-h time points (T0 and T1) followed by H 2 O 2 treatment were measured with a Synergy H1 hybrid reader (BioTek) ( ex / em ϭ 485/525 nm). The relative fluorescence signal for each sample (RF sample ) was calculated using the following equations.
where ⌬F max represents the fluorescence signal increases from the sample treated with the highest concentration of H 2 O 2 .
For evaluating the effect of TnPrx1 constructs on the phosphorylation of intracellular p38 MAPK, transfected cells treated with H 2 O 2 (0 -1200 M) were collected and lysed followed by Western blot analysis using antibodies against p38 and phosphorylated p38, respectively (Cell Signaling Technol-

Primer Sequences (5-3) Application
. The immunoreactive bands were visualized using an enhanced chemiluminescence (ECL) system (Pierce). Statistical Analysis-All experiments were performed independently at least three times. Data are presented as the mean Ϯ S.D. A two-tailed Student's t test was used to assess statistical significance between experimental and control groups.

Results
Cys-independent CAT Activity in TnPrx1-Thioredoxin POX was previously the only known enzyme activity in Prxs that relied on NADPH-dependent oxidoreduction between Trx and Trx reductase to maintain the continuation of their POX activity. In the absence of Trx/Trx reductase/NADPH or a reducing agent (e.g. DTT), the reactions stop after the formation of a Cys P -Cys R disulfide bound in which one pair of Prx monomers may only reduce two H 2 O 2 molecules. Four to six H 2 O 2 molecules may be reduced when they are hyperoxidized without the formation of disulfide bounds.
Surprisingly, however, in the absence of a reducing agent, we observed that the recombinant WT TnPrx1 monomers (Ͼ99% purity in reduced status) were able to continuously reduce H 2 O 2 molecules (Fig. 1, A, B, and C), implying the presence of non-POX oxidoreduction activity in TnPrx1. Similar activity was observed when TnPrx1 was fully oxidized to form dimers (Fig. 1, B and D), confirming that the observed non-POX activity was independent of the status of Cys residues. The observed activity was not attributed to nonspecific background reactions as it was not observed in reactions containing no or denatured TnPrx1 (Fig. 1, C and D, first and last columns). Additionally, we also detected O 2 production (Fig. 1, E and F), and the calculated ratio between the reduced H 2 O 2 and the produced O 2 was 2.29:1, indicating that this activity was derived from a CAT-like activity (i.e. 2 H 2 O 2 3 2 H 2 O ϩ O 2 ) rather than the POX activity that only produces H 2 O.
To fully rule out the possibility that a trace amount of contaminating catalase from E. coli was present in the TnPrx1 preparations (despite Ͼ99% purity) and contributed to the activity, we prepared TnPrx1 proteins under different elution stringencies (i.e. imidazole at 150 -300 mM) to allow various impurities (i.e. containing various amounts of contaminants). We confirmed again that the activity was derived from TnPrx1 as it was correlated with the amount of TnPrx1 rather than with the level of impurity (Fig. 1, G and H).
The activity was iron-dependent as it could be inhibited by ferrous/ferric chelators Tiron and 2,2-dipyridy, and the addition of Fe 3ϩ could not only increase the activity of untreated TnPrx1 but also reverse the inhibition by chelators (Figs. 1H and 2A). Fe 3ϩ displayed low nanomolar level binding affinity with TnPrx1 (apparent K d ϭ 0.17 M) and an ϳ1:1 (metal:Prx1) stoichiometry (Fig. 2C). To confirm the iron-TnPrx1 binding, we directly evaluated the iron content of recombinant TnPrx1 proteins under various conditions. Proteins were subjected to extensive ultrafiltration to remove unbound iron. The molecular ratio between iron and untreated TnPrx1 protein was 0.64 Ϯ 0.002:1 (Fig. 2D). Treatment by chelators reduced the ratio to 0.08 Ϯ 0.003:1, whereas the addition of FeCl 3 (200 M) restored the ratio to 0.75 Ϯ 0.006:1 (Fig. 2D). These observations indi-cated that each TnPrx1 binds to one iron, and up to 75% of the recombinant TnPrx1 proteins were in the active form.
Additionally, the effect of other metals, including Mg 2ϩ , Ca 2ϩ , Cu 2ϩ , Mn 2ϩ , Co 2ϩ , Ni 2ϩ , and Zn 2ϩ on the CAT-like activity was tested, but no enhancement activity was observed (data not shown). The dependence on iron, but not on reducing agents and Cys residues, was characteristic of catalases, further confirming that the observed activity was not derived from the POX activity of Prxs. Instead, it resembled a catalase that was previously unknown to Prxs. However, TnPrx1 was insensitive to the inhibitors of typical CATs, such as DTT and the irreversible inhibitor 3-amino-1,2,4-triazole (Fig. 2, E and F), suggesting that Prx1 might represent a new class of CAT-like enzyme. Indeed, unlike typical CATs, TnPrx1 lacked the Soret absorbance peak unique to heme-containing moieties (data not shown), indicating that it is a hemeless metalloprotein rather than a heme-containing protein.
In the presence of DTT, WT TnPrx1 displayed Michaelis-Menten kinetics on low concentrations of H 2 O 2 (i.e. Ͻ100 M) (Fig. 3A). The K m value was 2.2 M, which was comparable with the K m values reported previously for Prx1-POX activities that were typically much lower than 20 M  Table 1). The data were in agreement with the notion that TnPrx1 possessed both POX and CAT activities as the activities with DTT (POX ϩ CAT) were higher than those without DTT (CAT only) by a relatively constant rate (i.e. 2.5 s Ϫ1 determined by a "Michaelis-Menten ϩ allosteric sigmoidal" model).
Because CAT activity was described for the first time in a Prx1 of fish origin, we wanted to know whether it was also present in mammalian Prx1. We expressed recombinant HsPrx1 and performed a similar assay with or without a reducing agent. Our data supported that HsPrx1 was also bifunctional by possessing POX and CAT activities with kinetic parameters comparable with those of TnPrx1 (i.e. KЈ app(ϪDTT) ϭ 347 M, n H(ϪDTT) ϭ 10.1 and KЈ app(ϩDTT) ϭ 342 M, n H(ϩDTT) ϭ 8.8, respectively) ( Fig. 3E and Table 1). Although Prxs from more species need to be examined to make a firm conclusion, the data here suggest that the CAT-like activity is likely conserved among vertebrate Prx1 from fish to mammals.
To further validate TnPrx1-CAT activity, we performed a site-directed mutagenesis and constructed a mutant by replacing all three Cys residues with Ser residues to completely eliminate its Cys-dependent POX activity. The resulting mutant (POX Ϫ CAT ϩ ) was unable to form dimers as expected (Fig. 3, G and H) but still capable of converting H 2 O 2 to O 2 (Fig. 1, E and  F). The POX Ϫ CAT ϩ mutant displayed virtually identical sigmoidal curves when assayed with and without DTT that resembled that of WT TnPrx1 without DTT as well as similar kinetic parameters (i.e. KЈ app(ϪDTT) ϭ 211 M, n H(ϪDTT) ϭ 3.7 and KЈ app(ϩDTT) ϭ 227 M, n H(ϩDTT) ϭ 3.3, respectively) ( Fig. 3B and Table 1). These observations confirmed that the observed TnPrx1-CAT activity was truly independent of the Cys residues and reducing agent.

Potential Active Site for the CAT-like Activity in TnPrx1-
The discovery of a previously unknown Prx1-CAT activity prompted us to search for the functional motif. By examining a previous reported structure of rat Prx1 (Protein Data Bank code 1QQ2) and homology-based modeling of TnPrx1, we observed a flexible loop consisting of six residues, Gly 117 , Val 118 , Leu 119 , Phe 127 (rat Prx1) or Tyr 127 (TnPrx1), Ile 142 , and Ile 144 , at the dimer interface in which an H 2 O 2 molecule could well fit into a pocket formed by the highly conserved 117 GVL 119 residues (Fig.  4). To test whether this pocket might contribute to the TnPrx1-CAT activity, we generated a TnPrx1 construct by replacing 117 GVL 119 with 117 HLW 119 (denoted by POX ϩ CAT Ϫ ) to alter the pocket structure. Indeed, the mutant POX ϩ CAT Ϫ lost CAT-like activity (i.e. no activity without DTT in the reactions) but retained only DTT-dependent POX activity that followed Michaelis-Menten kinetics characteristic of Prx1-POX activity (K m ϭ 4.15 M) (Fig. 3C and Table 1).
To further dissect individual TnPrx1-POX and Prx1-CAT activities, we generated a double mutation (POX Ϫ CAT Ϫ ) in which all Cys residues and 117 GVL 119 were replaced by Ser and 117 HLW 119 , respectively. As expected, this double negative mutant lost both POX and CAT activities and was unable to reduce H 2 O 2 regardless of whether DTT was present or not (Fig. 3D). Among all the mutants tested, POX Ϫ CAT ϩ also displayed the expected iron dependence in which iron chelators inhibited its activity that could be restored by adding iron (Fig.  2B), whereas the two CAT Ϫ mutants (i.e. POX ϩ CAT Ϫ and POX Ϫ CAT Ϫ ) only retained low activity (6% versus WT) that were unaffected by iron chelators and iron (data not shown). Additionally, the TnPrx1-CAT activity tolerated low temperature better than pH as it was able to retain virtually constant peak activity between 0 and 40°C but only retained peak activity at ϳpH 7.0 (Fig. 5).
Collectively, these observations confirm that TnPrx1 possesses both POX and CAT activities, and the residues 117 GVL 119 are critical to Prx1-CAT activity. TnPrx1-POX acted on H 2 O 2 with much higher affinity (K m ϭ 4.15 M) but had a relatively low maximal activity (k cat ϭ 0.23 s Ϫ1 ) with a wider range of H 2 O 2 levels (Table 1 and Fig. 3C), whereas Prx1-CAT acted on H 2 O 2 with lower affinity (KЈ (ϪDTT) ϭ 210.7 M) but had a much higher activity (k cat ϭ 2.3 s Ϫ1 ) ( Table 1).
Implication of TnPrx1-CAT in Regulating ROS Level and Signaling-The physiological roles of TnPrx1-CAT activity were investigated using a mammalian cell transfection system. First, we transfected HEK-293T cells to overexpress various TnPrx1 constructs and examined the effects in regulating intracellular ROS (iROS) in response to H 2 O 2 treatment. The expression of TnPrx1 constructs in transfected cells was confirmed by qRT-PCR (Fig. 6, A and B). We observed a general trend that cells overexpressing CAT ϩ proteins (i.e. WT and . D, molar ratio between TnPrx1 protein and bound iron determined by inductively coupled plasma optical emission spectroscopy. TnPrx1 was treated as specified followed by extensive washes with water by ultrafiltration prior to inductively coupled plasma. Bovine catalase and PBS were used as controls. E and F, effects of catalase inhibitors 3-amino-1,2,4-triazole (3-AT) and DTT on the CAT activity of WT TnPrx1 and mutants. Catalase was used as a positive control. Residual activities are expressed as the percent activity (versus untreated WT TnPrx1). Data are representative of at least three independent experiments. The error bars represent S.D., and statistical significances between experimental and control groups were determined by Student's t test. **, p Ͻ 0.01; ***, p Ͻ 0.001. POX Ϫ CAT ϩ ) had lower iROS levels than those expressing CAT Ϫ proteins (i.e. blank vector, POX ϩ CAT Ϫ , and POX Ϫ CAT Ϫ ) in response to the treatment with 150 -600 M exogenous H 2 O 2 (Fig. 6C).
Second, because H 2 O 2 is known to also function as a signaling molecule, particularly in regulating kinase-driven pathways (22), we tested whether Prx1-CAT-associated regulation of intracellular H 2 O 2 affected the phosphorylation of p38 that played a central role in the p38 MAPK signaling pathway. In HEK-293T cells transfected with blank or double negative (POX Ϫ CAT Ϫ ) plasmids, there were low background levels of phosphorylated p38 in the absence of H 2 O 2 stimulation (0 M) (Fig. 6D). The levels of phosphorylated p38 in cells treated with 225-1,200 M H 2 O 2 displayed a bell curve that peaked in the 525-900 M H 2 O 2 groups, which was comparable with previously reported data (23). When CAT ϩ constructs (i.e. WT and POX Ϫ CAT ϩ ) were overexpressed, a considerable delay of phosphorylation of p38 was observed as phosphorylated p38 was significantly (p Ͻ 0.05) up-regulated in cells challenged with H 2 O 2 , starting at 525 M and peaking at 900 -1,200 M.
Conversely, in cells overexpressing POX ϩ CAT Ϫ TnPrx1, no significant delay of p38 phosphorylation was observed as phosphorylated p38 was only significantly up-regulated in cells chal-   (Fig. 6D), a pattern similar to that of the blank or double negative group. Although further studies are needed to fully dissect the physiological roles of individual Prx1-POX and Prx1-CAT activities in cells and in vivo, these observations provide primary evidence on the involvement of the TnPrx1-CAT activity in regulating the ROS-mediated p38 signaling pathway when cells are incubated with high micromolar to low millimolar levels of H 2 O 2 .

Discussion
Eukaryotic cells contain a complex system to detoxify and regulate H 2 O 2 and other reactive oxygen species. These include small molecules, such as ascorbic acid, ␤-carotene, glutathione, and ␣-tocopherol, and various enzymes, such as superoxide dismutase (SOD), CAT, glutathione peroxidase (GPx), and Prx (24). Some of these enzymes or isoforms are mainly cytosolic (e.g. Prx1, Prx2, Prx5, Prx6, SOD1, and GPx1), whereas others may be compartmentalized (e.g. catalase in peroxisomes; SOD2 and Prx3 in mitochondria; and SOD3, GPx3, and Prx4 in plasma), which constitutes a precise antioxidant network for the defense against various oxidative stresses in the diverse cellular activities (4,25,26).
Cells are known to rely heavily on Prxs in scavenging H 2 O 2 and other ROS molecules. In fact, they are the third most abundant proteins in erythrocytes and represent 0.1-1% of total soluble proteins in other cells (4,7,24,26,27), and Prx1/2 knockout in mice may lead to the development of severe blood cell diseases (e.g. hemolytic anemia and hematopoietic cancer) (27,28). Prxs are widely distributed and have been found in animals, plants, fungi, protists, bacteria, and cyanobacteria, suggesting that they are a family of ancient proteins essential to a variety of critical cellular activities (29,30).
Prxs were previously recognized only as a family of thioredoxin POX for which the biochemical features and biological functions were subjected to extensive investigations (22). In the present study, we discovered that TnPrx1 and HsPrx1 are bifunctional by possessing both Cys-dependent POX and Cys-independent CAT-like activities, further extending our understanding of this important family of antioxidant proteins. The CAT-like activity in TnPrx1 was validated by the identification of the active site containing the GVL motif, which also enabled us to generate mutants lacking CAT and/or POX activity for dissecting their individual activities. Our data suggested that previously observed antioxidant activity in WT Prx1 (at least in some animals) was in fact a combined activity of Prx1-POX and Prx1-CAT. The alteration of GVL motif abolished TnPrx1-CAT activity but not TnPrx1-POX activity ( Fig. 3C and Table 1). This suggests that Prx1 contained two independent H 2 O 2 -binding sites in agreement with previous reports that the H 2 O 2 -binding site for the Cys-dependent POX activity was near the Cys 51 and Cys 172 residues but distant from the GVL site (31)(32)(33).
Catalases are heme-containing enzymes (34). Mammalian Prx1 was previously identified as heme-binding protein 23 kDa (HBP23) (10,35), and a bacterial 2-Cys peroxiredoxin alkyl hydroperoxide reductase C (AhpC) was also reported to be able to bind heme (36), although heme binding is non-essential to their functions. Our data indicated that TnPrx1-CAT activity was not heme-related but dependent on mononuclear iron. However, the exact iron-binding site remains to be determined. Sequence analysis indicates that Prx1 proteins from T. nigroviridis and mammals contain a 2-His-1-carboxylate facial triad-like motif (e.g. motif 81 HX 2 HX 36 E 121 in TnPrx1) that is conserved in mononuclear non-heme iron enzymes (37). Additionally, a Trp 87 residue is also present at the motif. Aromatic residues, particularly Trp and Tyr, are known to be enriched at the iron sites of iron proteins (38). The involvement of aromatic residues in redox catalysis and/or electron transfer is not yet fully understood, but their capability to mediate electron transfer reactions makes them most suitable for tunneling electrons to/from redox sites (38). Conversely, the putative facial triad is not in the immediate proximity of the GVL motif. Therefore, its involvement in iron binding and the mechanism of iron-mediated electron transfer for the Prx1-CAT activity needs to be verified by further structure-based analysis.
The identity and similarity of vertebrate Prx1 are over 77 and 88% (Table 3), respectively, and the active site of CAT activity is completely conserved among Prx1 proteins, suggesting that CAT activity may be a ubiquitous function of Prx1 family members. The confirmation of reductant-independent HsPrx1-CAT activity indicates that this new function is likely conserved at least in some vertebrates. Furthermore, 117 GVL 119 are conserved in Prx1-3, whereas 117 GVY 119 are found in Prx4. Although Prx5 and Prx6 share low similarity with Prx1-4, they have similar threedimensional structures (39), suggesting that CAT activity might be present in other Prxs, at least Prx1-3. It might also explain why some parasites and cyanobacteria do not contain catalase and GPx but have diverse Prx homologies (29,40).
The intracellular concentrations of H 2 O 2 and other ROS molecules in vivo are not precisely known but may range from sub-to lower micromolar levels in various prokaryotic and eukaryotic cells. However, intracellular H 2 O 2 levels may rise to the order of 100 M in phagocytes, and the transient H 2 O 2 levels may reach Ͼ200 M in brain cells (41,42). Moreover, appropriately stimulated polymorphonuclear leukocytes and monocytes can produce up to 1.5 nmol of H 2 O 2 in 10 4 cells/h (which is roughly equivalent to Ͼ350 -450 mM of H 2 O 2 if it is not removed and accumulated per hour given their cell sizes at ϳ330 and 420 fl) (43,44). In the present study, we have shown that Prx1 acts mainly (if not only) as a POX under a low level H 2 O 2 environment with high affinity and relatively low capac-ity (K m and k cat at ϳ2.23-4.15 M and ϳ0.23 s Ϫ1 , respectively) but as both POX and CAT when the H 2 O 2 level reaches ϳ50 M or higher in which the latter behaves as an allosteric enzyme with 10 times higher activity than the former (K m and k cat at ϳ210 M and 2.3 s Ϫ1 , respectively). In vitro transfection experiments also confirmed the notion as HEK-293T cells overexpressing WT TnPrx1 and mutant retaining CAT activity were capable of scavenging more iROS than those overexpressing mutants lacking CAT or both POX and CAT activities at low to middle micromolar H 2 O 2 levels (Fig. 6C). The levels of exogenous H 2 O 2 to produce significant effects on cellular activities such as the phosphorylation of p38 in cells transfected with various TnPrx1 mutants were ϳ375-1050 M or higher (Fig.  6D), which corresponds to ϳ50 -150 M intracellular H 2 O 2 based on the model predicting that intracellular H 2 O 2 concentrations are ϳ7-fold or even 10 -100-fold lower than that applied exogenously (42,45,46). The corresponding intracellular levels of H 2 O 2 fell within the levels for physiologically relevant signaling (i.e. 15-150 M) (46).
Collectively, these features enable Prx1 to function on a wider range of ROS concentrations than many other proteins in  (Table 2). -Fold changes of Prx1 and catalase transcripts are expressed relative to the catalase transcripts in the blank control (A) or to the transcripts of their own genes (B). C and D, effects of TnPrx1 constructs on intracellular ROS and ROS-mediated phosphorylation of p38 MAPK in transfected cells treated with exogenous H 2 O 2 as determined by dichlorodihydrofluorescein fluorescence assay and Western blot analysis, respectively. In the Western blot analysis, antibody to human GAPDH was used as a control (D, lower panel). Representative data from one of three or more independent experiments are shown. The error bars represent S.D., and statistical significances between experimental and control groups were determined by Student's t test. *, p Ͻ 0.05. p-p38, phosphorylated p38.
the cytosol in which Prx1-POX acts on sub-to lower micromolar iROS normally present in cells, whereas Prx1-CAT (probably along with GPx and classic CAT enzymes) acts on moderate to higher micromolar iROS concentrations that are present in certain types of cells (e.g. some brain and immune cells) and/or required for H 2 O 2 signaling (Table 1).  However, it is noticeable that, although TnPrx1 and HsPrx display CAT-like activity, their catalytic efficiencies are ϳ100fold smaller than those of regular CATs (i.e. k cat /K m Prx1-CAT at ϳ10 4 M Ϫ1 s Ϫ1 versus k cat /K m CAT at ϳ10 6 M Ϫ1 s Ϫ1 ), which raises the question of whether Prx1-CAT function is critical to organisms as a higher level of iROS may be quickly scavenged by regular CAT. Prx1 is a cytosolic protein, whereas native CATs are typically present in peroxisomes. Data mining the Multi-Omics Profiling Expression Database (MOPED) also reveals that human Prx1 is much more abundant than CAT in most cells/tissues (Fig. 7A). Therefore, we speculate that the CATlike activity in Prx1 and possibly in other Prxs may act as one of the first line of scavengers for cytosolic ROS. Prx-CAT may also play a more critical role in scavenging and/or regulating ROS in certain cells and tissues that are deficient or contain extremely low levels of CAT. For example, in human bone, oral epithelium, and retina, the CAT protein levels are 132-, 45-, and 36-fold less, respectively, than Prx1 (i.e. 13 versus 1,730, 55 versus 2,490, and 110 versus 4,020 ppm, respectively). Some cancer cells might also take advantage of the Prx1-CAT activity as the expressions of CAT are deficient or highly down-regulated in many of cancer cells (47), whereas those of Prx1 are up-regulated in cancer cells, including breast, lung, and urinary cancers and hepatocellular carcinoma (48). The down-and up-regulation of CAT and Prx1 were also clearly supported by comparing the MOPED protein expression profiles between cancer and non-cancer cells (Fig. 7). Additionally, we also confirmed by qRT-PCR that the mRNA level of CAT in HEK-293T cells was ϳ50 -200-fold less than that of Prx1 (Fig. 6, A and B).
The Prx-CAT function might also explain how some invertebrates lacking CAT and GPx regulate high levels of intracellular ROS. For example, some parasitic helminths (e.g. Fasciola hepatica and Schistosoma mansoni) and roundworms (e.g. filarial parasites) as well as some protozoa (e.g. Plasmodium sp.) are CAT-and GPx-deficient but possess highly expressed Prx genes (29,40).
In summary, we observed a CAT-like activity in the pufferfish and human Prx1 proteins that were independent of Cys residue and reductants but dependent on non-heme mononuclear iron. TnPrx1-CAT activity was capable of regulating intracellular ROS and the ROS-dependent phosphorylation of p38 in transfected HEK-293T cells. These newly discovered features extended our knowledge on Prx1 and provided a new opportunity to further dissect its biological roles.