Identification of a New Type of Mammalian Peroxiredoxin That Forms an Intramolecular Disulfide as a Reaction Intermediate

of PrxV. These results indicate that oxidation of PrxV by H 2 O 2 results in the formation of a disulfide between Cys 48 and Cys 152 , whereas Cys 73 -SH remains unoxidized.

mM EDTA] containing 2 mM DTT and were disrupted by pressure, and the resulting cell extract was centrifuged at 12,000 H g for 30 min. Streptomycin sulfate was added to the resulting supernatant to a final concentration of 1% (w/v), and, after 30 min at 4ºC, the mixture was centrifuged at 12,000 H g for 30 min. Solid ammonium sulfate was slowly added, at 4ºC with stirring, to the resulting supernatant until 80% saturation was achieved, after which the mixture was stirred for an additional 1 h. The resulting precipitate was collected by centrifugation at concentrator. The concentrated sample was reduced with 2 mM DTT for 10 min and applied to a Mono Q HR10/10 column (Pharmacia) that had been equilibrated with buffer B. The column was washed with the same buffer for 10 min. PrxV was detected in the flow-through material, and those fractions containing the protein were pooled, dialyzed against 2 liters of buffer A, and stored at -70ºC until use. The mutant C48S, C73S, and C152S proteins were prepared by a procedure similar to that used for the wild-type enzyme.
Assay of PrxV activity-The ability of PrxV to protect glutamine synthetase from by guest on March 24, 2020 http://www.jbc.org/ Downloaded from oxidative inactivation was measured as described previously (1), with a slight modification. The 25-µl reaction mixture, containing 0.5 µg of glutamine synthetase, 10 mM DTT, 3 µM FeCl 3 , 50 mM Hepes-NaOH (pH 7.0), and various concentrations of PrxV, was incubated at 37ºC for 10 min, after which 1 ml of γ-glutamyltransferase assay mixture was added and the remaining activity of glutamine synthetase was measured at 37ºC for 3 min. The Trx-dependent peroxidase activity of PrxV was measured as described (17) .
Subcellular fractionation-HeLa cells that had been grown to confluency in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum were washed twice with ice-cold phosphate-buffered saline. The washed cells were disrupted in a homogenization buffer [10 mM triethanolamine, 10 mM acetic acid (pH 7.4), 250 mM sucrose, 1 mM EDTA, 1 mM DTT, and aprotinin, leupeptin, pepstatin, and chymostatin each at a concentration of 10 µg/ml] by passing them 10 times through a 25-guage needle, followed by homogenization in a Dounce homogenizer. Nuclei and unbroken cells were removed by centrifugation of the homogenate at 1000 H g for 10 min. The resulting supernatant was further centrifuged sequentially at 14,000 H g for 30 min to separate organelles and at 100,000 H g for 1 hr to obtain plasma membrane and cytosolic fractions.
Transfection-NIH 3T3 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% calf serum, penicillin (100 U/ml), and streptomycin (100 U/ml), and were continuously passaged for 3 months after thawing. For transfection, the cells were plated at a density of 3 H 10 5 per 60-mm dish, allowed to recover for 24 h, and then transfected with the indicated plasmids with the use of Superfect (Qiagen). After 24 h, the cells were deprived of serum by incubation for an additional 18

RESULTS
Cloning and purification of recombinant PrxV-The amino acid sequence identity among the four human 2-Cys Prx (PrxI to PrxIV) enzymes is >70%, with the homology being especially marked in the regions surrounding the conserved NH 2 -and COOH-terminal Cys residues that correspond to Cys 52 and Cys 173 of PrxI (Fig. 1). The human 1-Cys Prx shares ~10% amino acid sequence identity with human 2-Cys Prx enzymes, but the sequence surrounding its NH 2terminal Cys (Cys 47 ) is highly homologous to those surrounding the corresponding Cys of 2-Cys Prx enzymes (Fig. 1).
In an attempt to identify new Prx enzymes, we searched a database of human ESTs for sequences homologous to the NH 2 -terminal conserved sequence (KGKYVVLFFYPLDFTFVCP) of 2-Cys Prx enzymes. A human EST clone (GenBank accession number, H26194) with a Cyscontaining sequence (KGKKGVLFGVPGAFTPGCS) that shares 52% identity (indicated in bold) with the search sequence was thus detected. With the use of the nucleotide sequence of this clone, 87 EST clones containing identical overlapping sequences were further identified. We aligned all of these sequences in advanced BLAST searches and then constructed a 767-bp nucleotide sequence that includes a Kozak consensus sequence for translation initiation, an open reading frame encoding a 162-amino acid polypeptide, a stop codon (TGA) in the same frame, and a 135-bp 3' untranslated sequence containing a poly(A) tract. The newly identified putative Prx enzyme encoded by this nucleotide sequence was designated PrxV for reasons described below. Similar searches of a mouse EST database with the same NH 2 -terminal conserved sequence of 2-Cys Prx enzymes also revealed a clone (GenBank accession number, AA472012) with a Cys-containing sequence that is identical to the corresponding portion of the human PrxV sequence. Identification of clones containing overlapping sequences yielded an open reading frame for a 162-amino acid protein that shows 95% sequence identity to human PrxV (Fig. 1).
Human PrxV is ~10% identical to human 2-Cys and 1-Cys Prx enzymes (Fig. 1). The COOH-terminal region of PrxV is smaller than those of 2-Cys Prx enzymes and lacks the conserved sequence containing the COOH-terminal Cys of the latter enzymes. Both human and mouse PrxV sequences contain Cys residues at positions 73 and 152 in addition to the conserved Cys 48 . However, the sequences surrounding Cys 73 and Cys 152 are not homologous to those surrounding the COOH-terminal conserved Cys residue of 2-Cys Prx enzymes, and the distances between Cys 48 and these other two Cys residues are substantially smaller than the 120 to 123 residues that separate the two conserved Cys residues in typical 2-Cys Prx enzymes. PrxV shows 30% sequence identity to a Saccharomyces cerevisiae Prx known as type II Trx-dependent peroxidase (18) or peroxisomal membrane protein 20 (PMP20) (19) or alkyl hydroperoxide reductase (20) ; however, the yeast protein does not contain Cys residues corresponding to Cys 73 and Cys 152 of PrxV.
Human PrxV was expressed in E. coli and purified to homogeneity. The purified protein appeared as a single band with an apparent molecular size of 17 kDa on SDS-PAGE under reducing conditions (not shown), consistent with the size of 17,030 Da calculated from the predicted amino acid sequence. PrxV that had been oxidized with H 2 O 2 was also detected as a monomer by SDS-PAGE under either reducing or nonreducing conditions ( Fig. 2A), suggesting by guest on March 24, 2020 http://www.jbc.org/ Downloaded from that the protein does not form a disulfide-linked dimer on oxidation by H 2 O 2 . HPLC of PrxV on a gel filtration column in the presence of a buffer containing DTT yielded a peak at a position corresponding to 34 kDa (Fig. 2B), suggesting that PrxV exists as a dimer in its native state.
Trx-dependent peroxidase activity of PrxV-We investigated whether the reducing equivalents required for the presumed peroxidase activity of PrxV could be provided by the Trx system (Trx, TrxR, and NADPH) or the Grx system (Grx, GSH, GSH reductase, and NADPH).
The rate of H 2 O 2 degradation was measured by monitoring the decrease in A 340 attributable to the oxidation of NADPH. PrxV catalyzed the H 2 O 2 -dependent oxidation of NADPH in the presence of the Trx system (Fig. 3); the oxidation of NADPH required all three protein components (PrxV, Trx, and TrxR), being negligible in the absence of any one of the three. In contrast, the Grx system did not support the H 2 O 2 -dependent oxidation of NADPH by PrxV. Increasing the concentrations of Grx, GSH, and GSH reductase severalfold relative to those specified in the legend to Fig. 3 did not affect the inability of the Grx system to support the peroxidase activity of PrxV (data not shown). The functional efficacy of the Grx and GSH reductase preparations was demonstrated as described previously (21). These results suggest that PrxV receives reducing equivalents readily from Trx but not from Grx or from millimolar concentrations of GSH. PrxV also reduced t-butyl hydroperoxide in the presence of the Trx system with initial rates similar to that apparent for H 2 O 2 reduction (data not shown).
Kinetic parameters for PrxV catalysis were determined by measuring the initial rates of NADPH oxidation at various concentrations of Trx and H 2 O 2 . Lineweaver-Burke plots (not shown) revealed that the K m values of PrxV for Trx and H 2 O 2 were 1 µM and <20 µM, respectively, and that the V max at 37°C was 2 or 2.8 µmol/min per milligram of protein for the experiments in which the concentrations of Trx and H 2 O 2 , respectively, were varied.
Peroxidase activity of Cys mutants of PrxV-To study the catalytic role of the Cys residues of PrxV, we replaced each of the three residues at positions 48, 73, and 152 individually with serine, thereby generating C48S, C73S, and C152S mutant enzymes, respectively. The mutant proteins were expressed in E. coli and purified to homogeneity (Fig. 4A). Measurement of Trx-dependent peroxidase activity toward H 2 O 2 revealed that the activity of the C73S mutant was similar to that of the wild-type enzyme, whereas no activity was detected with C48S and C152S proteins (Fig. 4B).
We also evaluated the peroxidase activities of the mutants by measuring their ability to protect glutamine synthetase from inactivation induced by a low concentration of H 2 O 2 produced by a mixed-function oxidation system comprising O 2 , DTT, and iron. In the presence of an electron donor such as DTT, iron catalyzes the reduction of O 2 to H 2 O 2 , which is further converted to hydroxyl radicals (OH • ) by the Fenton reaction (22). Glutamine synthetase possesses a binding site for divalent cations, at which bound iron catalyzes OH • production. The locally produced OH • results in oxidation and consequent inactivation of the enzyme (22). Given that oxidized Prx can be reduced by DTT, Prx prevents the inactivation of glutamine synthetase by the mixed-function oxidation system through removal of H 2 O 2 . We have previously used this glutamine synthetase protection assay for the detection of Prx activity, especially when we did not know the physiological source of the reducing equivalents (1,8). Consistent with the results of the Trx-dependent assay, C48S was completely inactive in the glutamine synthetase protection assay and the activity of C73S was equal to or slightly higher than that of wild-type PrxV (Fig.  4C). However, in contrast to the results obtained with the Trx-dependent assay, C152S protected glutamine synthetase, albeit less effectively than did the wild-type protein. These data suggest that Cys 48 is essential for both Trx-and DTT-dependent peroxidase activities of PrxV, that Cys 73 is not required for either of these two activities, and that Cys 152 is essential for Trx-dependent activity but not for DTT-dependent activity.
Formation of a disulfide bond between Cys 48 and Cys 152 of PrxV-We next examined the possibility that Cys 48 and Cys 152 form a disulfide bond during the catalytic cycle of PrxV. PrxV that had been oxidized by H 2 O 2 was digested with endopeptidase Lys-C, and the resulting peptides were fractionated by HPLC on a C 18 column (Fig. 5, upper panel). Reduction of a portion of the Lys-C digest with DTT before HPLC resulted in the disappearance of peak I that was detected with the unreduced sample (not shown), suggesting that peak I likely contained peptides linked by a disulfide. Treatment with DTT of the manually collected peak I fraction followed by reinjection into the HPLC column yielded peaks II and III (Fig. 5, middle panel).
Edman sequencing of the peptides corresponding to these latter two peaks yielded the sequences GVLFG and ALNVE, respectively, which match the five residues of the predicted Lys-C fragments containing Cys 48 and Cys 152 , respectively. We also prepared a Lys-C digest of oxidized PrxV that had been exposed to 5,5'-dithiobis-2-nitrobenzoic acid; fractionation of the digest by HPLC, with monitoring of elution of 5-thio-2-nitrobenzoic acid-labeled peptides by measurement of A 328 , revealed only one major labeled peak (peak IV) (Fig. 5, lower panel). The sequence of the peptide contained in peak IV was determined to be GVQVVA, which matches  Tables I   and II, respectively. Similar to other Prx isoforms, PrxV was expressed in almost all tissues and cell lines examined; it is especially abundant in kidney and kidney-derived KNRK cells, contributing as much as 0.13% of total soluble protein.
We next investigated the subcellular localization of PrxV by immunoblot analysis. HeLa cell homogenates were separated into organellar, plasma membrane, and cytosolic fractions.
PrxV was detected in organellar and cytosolic fractions in a ratio of ~2:1 but was not detected in the plasma membrane fraction (Fig. 7A). The molecular size of PrxV detected in cytosolic and organellar fractions was identical at 17 kDa. Further fractionation of the organellar fraction on Nycodenz gradients (23) revealed that the distribution of PrxV was similar to that of the mitochondrial protein PrxIII, but not to that of the peroxisomal protein catalase, suggesting that most PrxV in the organellar fraction is present in mitochondria (data not shown). The presence of PrxV in peroxisomes was further investigated by immunoblot analysis of a peroxisomal fraction prepared from guinea pig liver. PrxV and catalase, but neither PrxIII nor the cytosolic enzyme PrxII, were detected in this fraction (Fig. 7B), which has been previously characterized by Webber and Hajra (24).

Peroxidase activity of PrxV in intact cells-Stimulation of NIH 3T3 cells with PDGF or
TNF-α increases the intracellular concentration of H 2 O 2 , which can be monitored with the oxidation-sensitive fluorescent probe 2',7'-dichlorofluorescein diacetate and confocal microscopy . To determine whether PrxV is a physiologically relevant peroxidase in cells, we transiently transfected NIH 3T3 cells, which contain a relatively low amount of endogenous PrxV (Table   II), with expression vectors encoding wild-type or C48S mutant proteins. Overexpression of the PrxV proteins was confirmed by immunoblot analysis (Fig. 8A). Exposure of cells transfected with the empty vector to TNF-α (15 ng/ml) for 10 min or PDGF (10 ng/ml) for 5 min resulted in 4.5-and 3.5-fold increases, respectively, in the fluorescence of 2',7'-dichlorofluorescein (DCF) (Fig. 8B). Expression of wild-type PrxV, but not that of C48S, markedly inhibited both the TNFαand PDGF-induced increases in DCF fluorescence.
The increase in intracellular H 2 O 2 results in the activation of JNK (25,26). We therefore examined the effect of PrxV overexpression on the activation of HA-tagged JNK. TNF-α induced a twofold increase in JNK activity in NIH 3T3 cells that had been transfected with the empty PrxV vector, and this effect of TNF-α was partially inhibited by expression of wild-type PrxV, but not by expression of C48S (Fig. 9).

DISCUSSION
We have shown that PrxV catalyzes the reduction of H 2 O 2 both in vitro and in vivo, and that Trx is likely the specific donor of the reducing equivalents required for the reduction reaction. Neither Grx nor GSH was able to support the peroxidase activity of PrxV. The simplest reaction mechanism for PrxV that is compatible with the observations that both Cys 48 10B). Such a mechanism would explain why the C152S mutant protects glutamine synthetase from oxidation by the DTT-containing mixed-function oxidation system, albeit with an efficiency lower than that of the wild-type protein.
The mechanisms of 2-Cys and 1-Cys Prx enzymes are shown in Fig. 10C and 10D, respectively. The four mammalian members (PrxI to PrxIV) of the 2-Cys Prx subgroup form intermediates that contain an intermolecular disulfide between the NH 2 -and COOH-terminal Cys residues that correspond to Cys 52 and Cys 173 , respectively, of human PrxI. The two disulfide-forming Cys residues of 2-Cys Prx enzymes are separated by 121amino acids, whereas those of PrxV are separated by 104 residues. Moreover, the amino acid sequence surrounding Cys 152 of PrxV does not resemble that surrounding Cys 173 of PrxI. In human 1-Cys Prx, the NH 2terminal Cys (Cys 47 ) is the site of oxidation by H 2 O 2 , but the resulting Cys-SOH cannot form a disulfide because there is no other Cys-SH nearby. Although the physiological source of the reducing equivalents for the regeneration of Cys 47 -SH is not known, 2 DTT is able to support the regeneration in vitro. In this respect, the reaction mechanism of 1-Cys Prx resembles that of the C152S mutant of PrxV. The crystal structures of oxidized 1-Cys Prx (14) and PrxII (15)  The catalytic efficiency of Prx V is similar to those of 2-Cys Prx enzymes: the V max of 2.0 to 2.8 µmol/min per milligram of protein for PrxV is smaller than those of 6 to 13 µmol/min per milligram of protein for Prx I, PrxII, and PrxIII, but the K m for Trx of 1 µM for PrxV is also smaller than those of 3 to 6 µM for the 2-Cys Prx enzymes (12). The K m for H 2 O 2 is <20 µM for PrxV and 2-Cys Prx enzymes. The catalytic efficiency of 1-Cys Prx has not been evaluated because its physiological donor of reducing equivalents is not known.
While the present study was in progress, PrxV was identified as proteins designated PMP20 and antioxidant enzyme B166 (AOEB166). Yamashita et al. (19) cloned human and mouse PMP20 cDNAs as the result of a search for homologs of the yeast PMP20 protein, and Knoops et al. (27) cloned human and rat AOEB166 cDNAs as the result of an attempt to characterize a 17-kDa bronchoalveolar protein. Mammalian PMP20 proteins consist of 162 amino acids and contain a peroxisomal targeting sequence (Ser-Gln-Leu) at the COOH-terminus, as does PrxV (Fig. 1). The peroxisomal localization of PMP20 was demonstrated by expressing HA-tagged PMP20 in HeLa cells and immunostaining with antibodies to the HA tag. Knoops et al. observed that the AOEB166 cDNA contains two potential initiation sites in the same reading frame, the use of one of which would result in the production of a 162-residue protein identical to PrxV (PMP20), and the use of the other would generate a polypeptide of 214 residues. The 52 amino acid residues at the NH 2 -terminus of the longer polypeptide were shown to constitute a mitochondrial presequence that is capable of importing a fusion protein of AOEB166 and green fluorescent protein into mitochodria (27). Yamashita et al. (19) and Knoops et al. (27) demonstrated that the bacterially expressed 162-residue PMP20 (AOEB166) protein is able to protect glutamine synthetase from the DTT-containing mixed-function oxidation system.
Recognizing that AOEB166 is homologous (25 to 35% sequence identity) to yeast and bacterial members of the Prx family and that four mammalian Prx enzymes had been identified previously, Knoops  PrxV is an appropriate designation because the protein is a Trx-dependent enzyme whose function is dependent on two Cys residues.
With the use of immunoblot analysis with antibodies specific for PrxV, we estimated the amounts of PrxV in various rat tissues and cultured mammalian cells and compared them with Mitochondria and peroxisomes are equipped with PrxIII and catalase, respectively, to protect against the toxic effects of H 2 O 2 . PrxV can now be added to the known antioxidant enzymes that protect these two oxidant-generating organelles. As predicted from the long mitochondrial targeting sequence at its NH 2 -terminus, PrxIII is synthesized in the cytosol as a preprotein that is converted to the mature form in mitochondria (28). The reducing equivalents required for the reactions of PrxIII and PrxV are likely provided by the recently discovered mitochondriaspecific proteins Trx (29) and TrxR (30), both of which are also synthesized in the cytosol with mitochondrial targeting sequences . Whether peroxisomes also contain a specific Trx system for PrxV remains to be determined. PrxI, PrxII, and 1-Cys Prx are localized predominantly to the cytosol, and PrxIV, which contains a typical signal sequence of secretory proteins at its NH 2terminus, is secreted outside of cells (11).
In many mammalian cell types, H 2 O 2 is produced in response to a variety of extracellular stimuli that include TNF-α (31) and PDGF (32). This receptor-mediated generation of H 2 O 2 in the cytoplasm has been linked to various intracellular signaling events such as the activation of mitogen-activated protein kinases (32)and the triggering of apoptosis (33,34) . Specific inhibition of such H 2 O 2 accumulation prevents these receptor-mediated signaling events. Overexpression of wild-type PrxV, but not that of the peroxidase-defective mutant C48S, inhibited H 2 O 2 accumulation induced by TNF-α or PDGF as well as the TNF-α-induced activation of JNK in NIH 3T3 cells, suggesting that cytosolic PrxV likely participates in the signaling pathways of these extracellular stimuli.