Mitochondrial Phospholipid Hydroperoxide Glutathione Peroxidase Plays a Major Role in Preventing Oxidative Injury to Cells*

Phospholipid hydroperoxide glutathione peroxidase (PHGPx) is synthesized as a long form (L-form; 23 kDa) and a short form (S-form; 20 kDa). The L-form contains a leader sequence that is required for transport to mitochondria, whereas the S-form lacks the leader sequence. A construct encoding the leader sequence of PHGPx tagged with green fluorescent protein was used to transfect RBL-2H3 cells, and the fusion protein was transported to mitochondria. The L-form of PHGPx was identified as the mitochondrial form of PHGPx and the S-form as the non-mitochondrial form of PHGPx since preferential enrichment of mitochondria for PHGPx was detected in M15 cells that overexpressed theL-form of PHGPx, whereas no similar enrichment was detected in L9 cells that overexpressed the S-form. Cell death caused by mitochondrial injury due to potassium cyanide (KCN) or rotenone (chemical hypoxia) was considerably suppressed in the M15 cells, whereas the L9 cells and control RBL-2H3 cells (S1 cells, transfected with the vector alone) succumbed to the cytotoxic effects of KCN. Flow cytometric analysis showed that mitochondrial PHGPx suppressed the generation of hydroperoxide, the loss of mitochondrial membrane potential, and the loss of plasma membrane integrity that are induced by KCN. Mitochondrial PHGPx might prevent changes in mitochondrial functions and cell death by reducing intracellular hydroperoxides. Mitochondrial PHGPx failed to protect M15 cells from mitochondrial injury by carbonyl cyanide m-chlorophenylhydrazone, which directly reduces membrane potential without the generation of hydroperoxides. M15 cells were more resistant than L9 cells to cell death caused by direct damage to mitochondria and to extracellular oxidative stress. L9 cells were more resistant totert-butylhydroperoxide than S1 cells, whereas resistance to t-butylhydroperoxide was even more pronounced in M15 cells than in L9 cells. These results suggest that mitochondria might be a target for intracellular and extracellular oxidative stress and that mitochondrial PHGPx, as distinct form non-mitochondrial PHGPx, might play a primary role in protecting cells from oxidative stress.

Mitochondria are a major physiological source of reactive oxygen species (ROS), 1 which can be generated during mitochondrial respiration (1). Superoxide radicals, formed by minor side reactions of the mitochondrial electron transport chain or by an NADH-independent enzyme, can be converted to H 2 O 2 and to the powerful oxidant, the hydroxyl radical (2). Thus, mitochondria are continually exposed to ROS that cause peroxidation of membrane lipids, cleavage of mitochondrial DNA, and impairment of ATP generation, with resultant irreversible damage to mitochondria. Mitochondrial dysfunction might contribute to the pathogenesis of various human neurodegenerative disorders, such as Parkinson's, Alzheimer's, and Huntington's diseases, amyotrophic lateral sclerosis, stroke, epilepsy, aging, and the AIDS dementia complex (3)(4)(5). However, ROS don't have exclusively toxic effects; low levels of ROS generated in mitochondria can act as signaling molecules under physiological conditions. ROS produced in mitochondria can activate transcription factors, such as NFB and AP-1 (6), and can function as signals in apoptosis that is induced by TNF-␣ (7), ceramide (8), and chemical hypoxia (9).
The production of ROS in mitochondria is strictly regulated by mitochondrial antioxidant enzymes that include phospholipid hydroperoxide glutathione peroxidase (PHGPx), classical glutathione peroxidase (cGPx), and Mn-superoxide dismutase (Mn-SOD). The importance of antioxidant enzymes in mitochondria is indicated by the fact that knock-out mice without a gene for Mn-SOD suffer from catastrophic effects (10). By contrast, knock-out mice without a gene for cGPx are quite vigorous. Some of these tissues remain very resistant to oxidative stress even though GPx is the only antioxidant enzyme that is known to reduce the H 2 O 2 produced by Mn-SOD in mitochondria since mitochondria in most mammalian cells lack catalase activity (11). Two types of GPx, namely cGPx and PHGPx, are located in mitochondria. PHGPx is the only known intracellular antioxidant enzyme that can directly reduce peroxidized phospholipids (12) and cholesterol (13) in membranes. Therefore, PHGPx that can reduce H 2 O 2 , rather than cGPx, is thought to contribute to the enzymatic defenses against oxidative damage to mitochondria (14). However, the PHGPx in mitochondria has not been fully characterized.
We previously cloned a cDNA for PHGPx from the rat (15,16) and demonstrated that a short 20-kDa (S-form) and a long 23-kDa (L-form) form of PHGPx were translated from the cDNA, which included two potential sites for the initiation of translation in vitro (16). We showed that the L-form included a leader sequence and was selectively imported into the mitochondria of rat liver by an import system in vitro (16). Stable transformants of rat basophile leukemia 2H3 (RBL-2H3) cells, in which the S-form of PHGPx was overexpressed, were resistant to the cell death caused by a radical initiator or oxidized lipids (17). The S-form of PHGPx markedly inhibited the production of leukotrienes by 5-lipoxygenase by preventing production of intracellular hydroperoxides around the nucleus (18).
In the present study, RBL-2H3 cells that overexpressed the L-form of PHGPx were established and compared with those that overexpressed the S-form in an attempt to estimate the functional roles of the two types of PHGPx in protection against intracellular and extracellular oxidative stress. The L-form of PHGPx was more effective than the S-form in preventing cell death that was caused by ROS generated in mitochondria and by exogenously added hydroperoxides.
Construction of Plasmids-A BamHI fragment of pRPHGPx4 (16) was subcloned into pSR␣, as the expression vector, to construct pSR␣-L-form PHGPx that encoded the L-form of PHGPx (19). S-probe and L-probe were made from pRPHGPx4 by polymerase chain reactions for construction of S-GFP and L-GFP. The primers for construction of the S-probe, in which the cDNA encoded the 42 amino acids from the first residue of the S-form of PHGPx, were 5Ј-ACATAAGCTTGCTGGCAC-CATGTGTGCA-3Ј and 5Ј-ATTAGGTACCGGCCACGTTGGTGACGAT-3Ј. The primers for construction of the L-probe, in which the cDNA encoded the 32 amino acids from the first residue of the L-form of PHGPx, were 5Ј-ATTTAAGCTTCCGGCCGCCGAGATGAGC-3Ј and 5Ј-ATTAGGTACCGCGGGATGCACACATGGT-3Ј. The BamHI and KpnI fragments of S-probe and L-probe were inserted between the BamHI and KpnI sites of the GFP expression vector (pCMX-SAP/Y145 F) that had been constructed by Ogawa et al. (20).
Cell Culture and Transfection-We used the previously established control line of cells (S1 cells) and L9 cells that overexpressed the S-form (non-mitochondrial) of PHGPx (17). M15 cells, which overexpressed the L-form (mitochondrial) of PHGPx, were established by the transfection of RBL-2H3 cells with pSR␣-L-form PHGPx and pSV2neo by electroporation, as described previously (17). A suspension of RBL-2H3 cells (1 ϫ 10 7 cells/0.25 ml) was transferred to an electroporation cuvette (0.4-cm gap; Bio-Rad) with a total of 20 g of linearized DNA, which consisted of 18 g of each expression vector and 2 g of pSV2neo, used to confer resistance to G418 (Geneticin; Life Technologies, Inc.) (21). A potential difference of 250 V at 500 microfarads was applied at room temperature with a Gene Pulser II (Bio-Rad), and cell culture was reinitiated after a 10-min recovery period. Selection for resistance G418 (1 mg/ml) was initiated after 24 h, and cells were subsequently exposed to G418 at 0.5 mg/ml for 2 weeks. Individual G418-resistant colonies were isolated with cloning cylinders. Levels of expression of PHGPx were determined by immunoprecipitation with antibodies against PHGPx, and cells that overexpressed the L-form of PHGPx were isolated. Control cells and cells that overexpressed the L-form or the S-form of PHGPx were cultured in Dulbecco's modified Eagle's medium that contained 5% fetal calf serum and 0.5 mg/ml G418.
Subcellular Fractionation of Cells-Cells were labeled with 140 nCi/ml [ 75 Se]sodium selenite (3126 Ci/g; MURR) for 96 h to determine the distribution of PHGPx and cGPx in cells. Confluent cells in 225-cm 2 culture flasks were washed three times with phosphate-buffered saline (PBS) and harvested by treatment with trypsin. The cell suspension was centrifuged at 700 ϫ g for 5 min at room temperature and then [ 75 Se]sodium selenite-labeled cells were fractionated as described previously (18). The cell pellet was suspended in sucrose buffer (0.25 M sucrose, 1 mM EDTA, 3 mM imidazole, and 0.1% (v/v) ethanol, with leupeptin, antipain, chymostatin, and pepstatin A added at a final concentration of 10 g/ml each and phenylmethylsulfonyl fluoride at a final concentration of 100 g/ml, pH 7.2) and centrifuged at 700 ϫ g for 10 min at 4°C. Pelleted cells were resuspended in the same buffer at approximately 1.5 ϫ 10 7 cells/ml and homogenized with a Teflon/glass Potter-Elvehjem homogenizer. A nuclear fraction (pellet) and a postnuclear fraction (supernatant) were prepared by centrifugation at 700 ϫ g for 10 min. The nuclear fraction was suspended in 200 l of the sucrose buffer. Mitochondrial, microsomal, and cytosolic fractions from the postnuclear fraction were obtained by differential centrifugation as described by de Duve et al. (22). Each subcellular fraction was examined by standard enzymatic assays for activities of cytochrome c oxidase (a mitochondrial marker), NADPH-cytochrome c reductase (a microsomal marker), and lactate dehydrogenase (a cytosolic marker) as reported previously (23,24). The distribution of histone H1, as a nuclear marker, was determined by immunoblotting with polyclonal antibodies against histone H1 (25). The purity of each subcellular fraction of M15 cells was determined according to our previous paper in which the the purity of each organelle of control cells and non-mitochondrial PHGPx overexpressing cells including S1 and L9 cells has been determined (18). The purity of each subfractionation in M15 cells was the same as those in S1 and L9 cells. Cytochrome c oxidase was distributed in nuclear (7.9%), mitochondrial (75.6%), microsomal (3.5%), and cytosolic (12.9%) fractions of M15 cells. NADPH-cytochrome c reductase activities were found in nuclear (5.9%), mitochondrial (16.7%), microsomal (68.3%), and cytosolic (9.1%) fractions of M15 cells.
The nuclear, mitochondrial, and microsomal fractions were solublized in 200 l of 0.4% Triton X-100 in PBS for 2 h at 4°C, and each solution was centrifuged at 100,000 ϫ g for 1 h at 4°C. The supernatants were supplemented into 400 l each of PBS and subjected to immunoprecipitation with antibodies against PHGPx and cGPx, as described previously (17). Immunoprecipitated proteins were separated by SDS-polyacrylamide gel electrophoresis (12.5% polyacrylamide) under non-reducing conditions. Gels were stained, dried, and subjected to autoradiography. Total levels of PHGPx and cGPx were calculated from results of scanning densitometry after autoradiography with a Bio-Imaging Analyzer (BAS2000; Fuji Film, Tokyo).
Activities of PHGPx and cGPx were measured after the fractionation of cytosol and mitochondria from each cell line (1.5 ϫ 10 8 cells). Mitochondrial fraction was sonicated and centrifuged at 10,000 ϫ g for 10 min at 4°C. The supernatants obtained from mitochondrial fraction and cytosolic fraction were used for assays of PHGPx and cGPx activities. PHGPx activity was determined by using phosphatidylcholine hydroperoxide (PCOOH) as the substrate according to the previous paper (18). Activity of cGPx was determined by using hydrogen peroxide as the substrate (18). The total activity of SOD was measured in terms of the percentage inhibition of the formation of superoxide by the xanthine-xanthine oxidase system (26). Mn-SOD activity was measured in the presence of 5 mM KCN, and Cu,Zn-SOD activity was calculated by subtraction of the activity of Mn-SOD from the total SOD activity.
Distribution of GFP Fusion Proteins in Cultured Cells-RBL-2H3 cells were transfected with plasmids that encoded GFP, L-GFP, or S-GFP by electroporation, as described above. Transfected cells were cultured on coverslips in 35-mm dishes in 2 ml of Dulbecco's modified Eagle's medium that contained 5% fetal calf serum at 37°C in an atmosphere of 5% CO 2 in air. After 24 h, cells were fixed for 20 min on coverslips with 4% formaldehyde and washed with Hanks' balanced salts solution. The fluorescence of cells was monitored and photographed with an Axiovert 135M inverted microscope (Carl Zwiss, Germany) equipped with a Planapochromat 63 ϫ objective and a filter pack appropriate for GFP fluorescence.
GFP and mitochondria were simultaneously detected in the same cells by the double staining with GFP fluorescence and a monoclonal antibody of Cy3-conjugated anti-cytochrome c oxidase subunit IV that was a specific probe for the mitochondrial staining (27). Fixed cells were washed with phosphate-buffered saline (PBS) and were blocked with PBS containing 2% BSA at 25°C for 30 min. The cells were incubated with 2 g/ml mouse anti-cytochrome c oxidase subunit IV monoclonal antibodies diluted with 2% BSA-PBS at 25°C for 2 h. Then the cells were washed with PBS and incubated with Cy3-conjugated goat antimouse IgG (Amersham Pharmacia Biotech) diluted to 10 g/ml with PBS containing 2% BSA at 25°C for 1 h. Fluorescence of GFP and Cy3 in the same cells was monitored and photographed with an appropriate filter pack.
Cell Viability-S1, L9, and M15 cells were plated at 0.5 ϫ 10 5 cells/well in flat-bottomed 96-well culture plates and cultured for 24 h. Individual transformants were exposed to indicated doses of KCN, rotenone, CCCP, oligomycin, or t-BuOOH for appropriate periods. The LDH release assay was used for the determination of the cell viability, as described elsewhere (17). In one series of experiments, cells were incubated for 12 h prior to exposure to KCN with 0.5 mM buthionine sulfoxamine (BSO) for depletion of GSH.
Flow Cytometric Analysis-Changes in the integrity of plasma membrane and in the mitochondrial membrane potential were examined by monitoring staining with propidium iodide (PI) and Rh123, respectively. After treatment with KCN, cells were stained with PI (5 mg/ml) and Rh123 (1 mg/ml) for 10 min. We also used an oxidation-sensitive fluorescent probe, 5,6-carboxy-2Ј,7Ј-dichlorofluorescein-diacetate (DCFH-DA), to assess levels of intracellular peroxides, as follows. Cells were washed with PBS and incubated with 2.5 M DCFH-DA in PBS for 15 min. DCFH-loaded cells were incubated with or without 25 mM KCN for the times indicated. The intensity fluorescence from PI, Rh123, and dichlorofluorescein (DCF) in cells was analyzed with a flow cytometer (EPICS® Elite Flow cytometer; Coulter, Hialeah, FL).
Analysis of Cellular Levels of ATP-Cellular levels of ATP were determined by the luciferin-luciferase method using a kit from Sigma (29).
Fluorescence Measurements of Lipid Peroxidation-Cells (1 ϫ 10 6 cells) were loaded with 20 mM cis-parinaric acid for 1.5 h at 37°C and then washed with PBS. The loaded cells were treated with 25 mM KCN for the times indicated. After incubation, total lipids were extracted as described by Bligh and Dyer (30). Fluorescence of total lipids was monitored with a spectrofluorometric detector (RF-550; Shimazu Co. Ltd., Tokyo) with excitation at 303 nm and emission at 416 nm.
Fluorescence Measurements of GSH in Mitochondria and Cytosol-Amounts of GSH in mitochondria and cytosol were measured according to the previous paper with a slight modification (28). In brief, S1, L9, and M15 cells (each 2 ϫ 10 7 cell) were fractionated into cytosol and mitochondria. Cytosol and mitochondria dispersed with the sonication were precipitated by the addition of trichloroacetic acid at a final concentration of 5%. After centrifugation at 10,000 ϫ g for 10 min, GSH in the supernatant was converted to fluorescent derivative. The reaction was started by the addition of 0.5 ml of 0.02% ammonium 7-fluorobenzo-2-oxa-1,3-diazole-4-sulfonate (SBD-F) in 0.25 M borate buffer, pH 10.5, which contained 5 mM EDTA and 1% tri-n-butyl phosphine. The reaction mixture kept at 60°C for 30 min, and the reaction was terminated by the addition of 50 l of 4 N HCl, and fluorescent thiol derivatives were separated by reversed-phase high pressure liquid chromatography (TSK-gel ODS; TOSOH Co. Ltd., Japan). The mobile phase was 0.1 M citrate buffer, pH 4.0, tetrahydrofuran/acetonitrile (94.8:0.2:5). The fluorescent thiol derivatives were monitored with emission at 516 nm and excitation at 384 nm during elution at a flow rate of 1.0 ml/min.
Quantitation of Proteins-Concentrations of protein were deter-mined with the BCA protein assay reagent (Pierce), with bovine serum albumin (BSA) as the standard.
Expression of Results-All data from assays in which the number of replicates was three or more are expressed as mean values Ϯ S.D.

Sorting of Green Fluorescent Protein Tagged with the Leader Sequence of PHGPx into the Mitochondria of RBL-2H3 Cells-
The L-form of PHGPx contains a leader sequence, but the S-form does not (Fig. 1). Chimeric proteins that included green fluorescent protein (GFP) were expressed in RBL-2H3 cells in order to determine whether the leader sequence of the L-form could serve to target GFP to the mitochondria of living cells. One fusion protein consisted of GFP with the leader sequence of 32 amino acids from the first residue of L-form (L-GFP) (Fig.  1B). The other was a fusion protein of GFP with 42 amino acids from the first residue of S-form (S-GFP) (Fig. 1B). Expression vectors containing cDNA that encoded GFP, L-GFP, or S-GFP were used to transfect RBL-2H3 cells by electroporation and then the intracellular localization of fluorescence due to GFP was monitored with a fluorescence microscope 24 h later (Fig.  2). Fluorescence was diffusely distributed in cells that expressed S-GFP or GFP (Fig. 2, A and B). By contrast, discrete regions with strong fluorescence were observed in cells that expressed L-GFP (Fig. 2C). GFP and mitochondria in the L-GFP-transfected cells were simultaneously visualized by the double staining with GFP fluorescence and a monoclonal antibody of Cy3-conjugated anti-cytochrome c oxidase subunit IV (Fig. 2, C and D). The profile of fluorescence due to L-GFP was identical to that of mitochondrial cytochrome c oxidase. Efficient import of GFP with the leader sequence of L-form into mitochondria indicates that the leader sequence at the amino terminus of mitochondrial PHGPx is the signal for targeting to mitochondria.

Subcellular Localization of PHGPx and cGPx in RBL-2H3 Cells That Overexpressed the L-form and the S-form of
PHGPx-RBL-2H3 cells were transfected by electroporation with cDNAs that encoded the L-form and the S-form of PHGPx (Fig. 1A). Two types of transformant that stably expressed substantial levels of PHGPx were isolated after appropriate selection. M15 cells strongly expressed the L-form of PHGPx with the leader sequence and L9 cells expressed the S-form of PHGPx. The control line of cells (S1) had been transfected with the expression vector without an insert. The three kinds of transformants were labeled with [ 75 Se]sodium selenite for 4 days for determination of the amounts of PHGPx and cGPx ( Table I). The total amounts of PHGPx in L9 and M15 cells were 4 and 3.5 times higher than that in S1 cells, respectively. No significant differences in total respective amounts of cGPx, Cu,Zn-SOD, and Mn-SOD were detected among L9, M15, and S1 cells. Fig. 3 shows the subcellular distribution of 75 Se-labeled PHGPx and cGPx in L9, M15, and S1 cells. In S1 cells, PHGPx was more concentrated in the mitochondria than in the cytosolic and microsomal fractions (Fig. 3, A and C). Levels of PHGPx were significantly elevated in the cytosolic, microsomal, and nuclear fractions in L9 cells. The amount of PHGPx in the cytosolic fraction from L9 cells was 4 times higher than that from S1 cells, but the amounts of PHGPx in the mitochondrial fractions from L9 and S1 cells were similar. The amount of PHGPx in the mitochondrial fraction from M15 cells was twice that from S1 cells. The overexpression of L-form PHGPx caused the enhancement of the activity of PHGPx in mitochondria. Specific activity of PHGPx in mitochondria of M15 cells was 272 Ϯ 12 pmol/min/mg, whereas specific activities in S1 and L9 were 156 Ϯ 18 and 156 Ϯ 37 pmol/min/mg, respectively. The activities of cytosolic PHGPx in S1, L9, and M15 cells were 4.0 Ϯ 1.6, 16 Ϯ 1.9, and 8.2 Ϯ 1.7 pmol/min/mg, respectively. These results show that the leader sequence of the L-form of PHGPx is the targeting signal for transport to mitochondria, whereas the S-form PHGPx is widely distributed in various organelles. The L-form of PHGPx can be considered to be the mitochondrial PHGPx and the S-form to be the non-mitochondrial PHGPx.
In three types of cells, cGPx was localized exclusively in the cytosolic fraction (Fig. 3, B and D). In S1 cells, the amount of cGPx in the mitochondria was lower than that of PHGPx. No significant changes of cGPx activities in cytosol and mitochondria were found by the overexpression of PHGPx (data not shown).
Resistance of Transformant Cells to Oxidative Damage to Mitochondria-The sensitivity of mitochondria in M15, L9, and S1 cells to injury was evaluated by exposing cells to KCN, an inhibitor of the respiratory chain (chemical hypoxia) (Fig. 4). The viability of S1 cells and L9 cells decreased rapidly in a time-and dose-dependent manner, and only about 20% of cells remained viable after exposure to 25 mM KCN for 6 h (Fig. 4, A  and B). By contrast, M15 cells were much more resistant to cell death caused by KCN. The LD 50 of KCN for M15 cells was approximately 30 mM, whereas LD 50 for L9 and S1 cells was 20 mM. These results indicated that mitochondrial damage by KCN could be prevented to some extent by the overexpression of PHGPx in mitochondria. However, overexpression of nonmitochondrial PHGPx did not have such a protective effect.
Effects of KCN on the viability of S1, L9, and M15 cells that had been depleted of GSH were examined to estimate whether or not the resistance to KCN of M15 cells had resulted from overexpression of mitochondrial PHGPx activity. Buthionine sulfoximine (BSO), an inhibitor of the synthesis of glutathione, inhibits the activity of glutathione-dependent peroxidases, such as cGPx and PHGPx, by lowering the level of glutathione in cytosol and mitochondria of cells. Amounts of glutathione in cytosol and mitochondria of S1 cells were 35.6 Ϯ 2.8 and 18.25 Ϯ 1.6 nmol/mg protein, respectively. No significant changes of glutathione content were observed in the PHGPxoverexpressing cells. The levels of cellular glutathione were markedly reduced by the treatment with BSO. Amounts of glutathione in cytosol and mitochondria of S1 cells reduced to 3.39 Ϯ 2.8 and 2.80 Ϯ 1.32 nmol/mg protein by the treatment of BSO, respectively. In BSO-treated M15 cells, amounts of glutathione in cytosol and mitochondria were 3.38 Ϯ 1.33 and 2.75 Ϯ 1.69 nmol/mg protein, respectively. Decrease in the level of glutathione by BSO was also found in L9 cells at the same extent as that in M15 cells (data not shown). When M15 cells pretreated with BSO were exposed to KCN, the cells lost their resistance to KCN toxicity (Fig. 4C). These results confirm that resistance of M15 cells to KCN is due to the overexpression of PHGPx in mitochondria. Fig. 5 shows the effects of various reagents that interfere with mitochondrial function on the viability of the three lines of transformants (Fig. 5). Rotenone, oligomycin, and carbonyl cyanide m-chlorophenylhydrazone (CCCP) were used as an inhibitor of mitochondrial complex I, as an inhibitor of F 0 F 1 -ATPase, and as an uncoupler of oxidative phosphorylation,

FIG. 3. Subcellular localization of PHGPx and cGPx in PHGPx-overexpressing RBL-2H3 cells. S1 cells (white bars), L9 cells (gray bars), and M15 cells (black bars)
were incubated with 75 Se (sodium selenite; 0.14 mCi/ml) for 4 days. 75 Se-labeled cells were fractionated by differential centrifugation into a nuclear (Nu), a mitochondrial fraction (Mit), a microsomal fraction (Mic), and a cytosolic fraction (Cyt). Distributions of PHGPx and cGPx were determined by immunoprecipitation with antibodies against PHGPx (A and C) and against cGPx (B and D). Immunoprecipitates were analyzed by SDS-polyacrylamide gel electrophoresis (12.5% acrylamide) with subsequent autoradiography (C and D). Radioactivity (in relative units) was quantified with a Bio-Imaging Analyzer.

FIG. 4. Effects of KCN on the viability of PHGPx-overexpressing cells. S1 cells (closed circles), L9 cells (closed triangles)
, and M15 cells (closed squares) were exposed to 25 mM KCN for the times indicated (A). Individual cultures were exposed to the indicated concentrations of KCN for 6 h (B). Individual cultures were treated for 12 h with 0.5 mM BSO to deplete cells of intracellular GSH and were then exposed to the indicated concentrations of KCN for 6 h. After incubation, cell viability was determined from the extent of release of LDH. Viability was expressed as a percentage relative to the total LDH in the cells. The total LDH in cells was determined after lysis of cells with 0.2% Triton X-100. Data are the means Ϯ S.D. of results from four replicates in each case.
respectively. As shown in Fig. 5A, rotenone had a strong toxic effect on L9 and S1 cells, whereas M15 cells were more resistant to cell death caused by rotenone. By contrast, the sensitivity of M15 to mitochondrial injury due to oligomycin or to CCCP was the same as that of S1 and L9 cells (Fig. 5, B and C). These results indicate that mitochondrial PHGPx contributes to the protection of cells from cell death due to inhibitors of the respiratory chain but not from death due to the direct effects on membrane potential and reduction on the production of ATP. Non-mitochondrial PHGPx failed to protect cells from cell death that was caused by any impairment of mitochondrial function.

Inhibition of the Cyanide-induced Generation of Hydroperoxides by Mitochondrial PHGPx-The effects of KCN on intracellular levels of hydroperoxides in the individual transformants
were determined by flow cytometry using the oxidant-sensitive dye 2Ј,7Ј-dichlorofluorescein (Fig. 6). In S1 cells, we detected the rapid generation of hydroperoxides within 30 min (Fig. 6A). Hydroperoxides were produced in KCN-treated L9 cells to the same extent and with the same time course as in S1 cells (Fig.  6B). However, the production of hydroperoxides was considerably suppressed in KCN-treated M15 cells as compared with KCN-treated S1 and L9 cells (Fig. 6C).
Suppression of the KCN-induced Peroxidation of Lipids by Mitochondrial PHGPx-We next examined peroxidation of lipids in KCN-treated transformants. cis-Parinaric acid, which is a naturally fluorescent polyunsaturated fatty acid, was used as a sensitive indicator of lipid peroxidation in cells. A reduction in the intensity of fluorescence of cis-parinaric acid is an indicator of lipid peroxidation (31). Analysis of the fluorescence of cis-parinaric acid revealed that the fluorescence decreased within as little as 1 h after the start of exposure of cells to KCN. Fluorescence from S1 and L9 cells was reduced to 50% of the original value after treatment with KCN for 1 h (Fig. 7). By contrast, no reduction in fluorescence from M15 cells was observed after 1 h, an indication that lipid peroxidation was suppressed in M15 cells as compared with L9 cells.
Protection from Disruption of Mitochondrial Functions by Mitochondrial PHGPx-To assess changes in mitochondrial membrane potential (⌬) and in the integrity of plasma membranes in transformant cells, we performed flow cytometric analysis after double staining with rhodamine 123 (Rh123) and propidium iodide (PI) (Fig. 8). Rh123, a lipophilic cation, is selectively taken up by mitochondria, and uptake is directly proportional to mitochondrial ⌬. PI is imported into cells and binds to cellular DNA when the integrity of plasma membranes is lost. Cells that had not been treated with KCN were predominantly located in the PI-negative and high ⌬ field (PI(Ϫ)-⌬high) (Fig. 8, A, D, and G). New subsets of S1 cells appeared after treatment of S1 cells with KCN for 2 h; a large number of subsets was located in the PI(Ϫ)-⌬low field and a smaller subset was present in the PI(ϩ)-⌬low field (Fig. 8B). Most S1 cells lost membrane integrity upon exposure to KCN for 4 h and moved to the PI(ϩ)-⌬low field (Fig. 8C). Changes of the distribution of L9 cells upon treatment with KCN were expected to be identical to those of S1 cells. Indeed, L9 cells also shifted from the PI(Ϫ)-⌬low field to the PI(ϩ)-⌬low field upon exposure to KCN (Fig. 8, E and F). By contrast, M15 cells retained their mitochondrial membrane potential and remained in the PI(Ϫ)-⌬high field after the exposure to KCN for 2 h (Fig. 8H). Many M15 cells retained their mitochondrial membrane potential even after long term treatment with KCN (4 h) (Fig. 8I). These observations indicate that mitochondrial PHGPx protects cells by preventing loss of membrane potential and loss of membrane integrity, whereas non-mitochondrial PHGPx can't protect cells from the effects of KCN.
Rapid reductions in levels of cellular ATP were observed in S1 and L9 cells after 2 h of exposure to KCN, with further losses within the next 2 h (Table II). The level of ATP in M15 cells after 3 h of treatment with KCN was clearly much higher than in S1 and L9 cells. The decreases in levels of ATP corresponded closely to the loss of mitochondrial membrane potential in transformant cells upon treatment with KCN.
Resistance of Transformant Cells to Extracellular Oxidative Damage by t-BuOOH-Each line of transformants exhibited different sensitivity to extracellular oxidative damage by t-BuOOH (Fig. 9). Numbers of surviving S1 cells decreased rapidly and in a dose-dependent manner. L9 cells were more resistant to the cytotoxic effects of t-BuOOH than S1 cells, and M15 cells were even more resistant. In M15 cells, half-maximal killing with t-BuOOH occurred at less than 50 M of t-BuOOH. In contrast, the LD 50 of L9 cells was accounted for 32 M, although the total expressions of PHGPx in whole cells were the same between M15 and L9 cells. Thus, overexpression of mitochondrial PHGPx was more effective in protecting cells from extracellular oxidative injury than that of non-mitochondrial PHGPx.
Changes in Mitochondrial Functions Induced by t-BuOOH-Changes in mitochondrial functions induced by t-BuOOH were estimated by flow cytometric analysis (Fig. 10). Three subsets of S1 cells appeared after exposure to t-BuOOH for 1 h as follows: cells in the PI(Ϫ)-⌬high field; cells in the PI(Ϯ)-⌬moderate field; and cells in the PI(ϩ)-⌬low field. Thus, S1 cells lost mitochondrial membrane potential and also plasma membrane integrity, and they were located predominantly in the PI(ϩ)-⌬low field after treatment with t-BuOOH (Fig.  10A). Moderate decomposition of plasma membranes and the moderate loss of mitochondrial potential were observed in L9 cells. More L9 cells remained in the PI(Ϯ)-⌬moderate field than S1 cells (Fig. 10B). The membrane potential of M15 cells was retained, and cells with a low membrane potential were fewer than in the case of L9 cells (Fig. 10C). The population of M15 cells located in the PI(Ϯ)-⌬moderate field was also smaller than in the case of L9 and S1 cells. DISCUSSION Some proteins in mitochondria are initially synthesized in the cytoplasm as larger precursors, and then they are imported into the mitochondria where proteolytic cleavage yields the mature forms. The leader sequence of the 23-kDa precursor to PHGPx is the signal for import of this protein into the mitochondria by using in vitro import system (16). In the present study, we examined the transport of PHGPx into the mitochondria of living cells. Green fluorescent protein (GFP) has proved useful as a probe in an attempt to visualize the translocation of proteins in living cells. However, few studies have used GFP to study the transport of proteins into mitochondria in cells. Yano et al. (32) successfully visualized the translocation of a chimeric protein that consisted of GFP and the presequence of ornithine transcarbamylase among the organelles of COS 7 cells. We constructed a plasmid that encoded GFP protein linked to the leader sequence of PHGPx (L-GFP) to investigate the role of the leader sequence. L-GFP was correctly targeted to the mitochondria, whereas S-GFP, which lacked the leader sequence, was not detected in mitochondria (Fig. 2). Thus, the leader sequence of 23-kDa PHGPx (L-form) appears to be required not only as an import signal but also as a targeting signal. There are four known isozymes of glutathione peroxidases as follows: cytosolic GPx (cGPx), PHGPx, plasma GPx, and gastrointestinal GPx (33). PHGPx is unique among these isozymes in having a leader sequence for transport to mitochondria. cGPx is located in mitochondria but is not translated with a signal peptide.
The leader sequence of PHGPx is located between two different sites for initiation of translation of the cDNA for PHGPx. Pushpa-Rekha et al. (34) demonstrated that the gene for PHGPx has alternative transcription sites and that these sites result in two populations of mRNAs for PHGPx. They found that one mRNA had an upstream AUG codon that was primarily utilized in rat testis and was translated to yield the mitochondrial PHGPx (L-form). The other mRNA lacked the upstream AUG codon, and it encoded non-mitochondrial PHGPx (S-form), which was synthesized predominantly in somatic cells. In RBL-2H3 cells, the upstream initiation site appears to be used predominantly, with resultant transcription of mRNA for mitochondrial PHGPx, since the amount of PHGPx in mitochondria was higher than that in other organelles (Fig. 3). The mechanism for regulation of transcription from the two initiation sites in a single gene for PHGPx remains to be resolved.
A considerable amount of PHGPx was found in mitochondria when RBL-2H3 cells were transfected with a plasmid encoding the L-form of PHGPx (M15 cells). Levels of PHGPx were also elevated in cytosolic, microsomal, and nuclear fractions of M15 cells as compared with S1 cells (control transfected cells). Two possibilities can be considered to explain the increased amounts of PHGPx in organelles other than mitochondria in M15 cells. One possibility is that mRNA for the S-form of PHGPx is transcribed from the downstream initiation site, as well as from the upstream site, when the L-form of PHGPx is overexpressed. Alternatively, the signal peptide of PHGPx might be partially removed by proteolysis before import into mitochondria can be completed. The latter possibility can be rejected since the fusion protein that consisted of GFP with the signal peptide was efficiently transported into mitochondria (Fig. 2).
The overexpression of the mitochondrial type of PHGPx protected RBL-2H3 cells from cell death due to mitochondrial oxidative stress that resulted from exposure of cells to KCN and rotenone (chemical hypoxia). Neither S1 nor L9 cells were resistant to the cytotoxicity of KCN. Resistance of M15 cells was eliminated when the activity of PHGPx in these cells was inhibited by depletion of cytosolic and mitochondrial glutathi- one with buthionine sulfoximine (BSO). Thus, overexpression of mitochondrial PHGPx clearly contributed to protection from mitochondrial damage.
We next investigated how mitochondrial PHGPx interferes with the toxicity of KCN, an inhibitor of complex IV of the mitochondrial respiratory chain. KCN causes various types of damage to mammalian cells, inducing the rapid generation of ROS (35), the reduction of mitochondrial ⌬ (36), a decrease in cellular levels of ATP (36), and lipid peroxidation (37), for example. These phenomena developed in KCN-treated RBL-2H3 cells in a time-dependent manner. Flow cytometric analysis revealed that KCN induced the rapid generation of hydroperoxides in S1 cells within 30 min (Fig. 6A). Lipid peroxidation was induced within 1 h (Fig. 7). After the production of hydroperoxides, loss of mitochondrial membrane potential and a reduction in levels of intracellular ATP were observed from 2 to 4 h ( Fig. 8 and Table II). Such mitochondrial injury induced the loss of plasma membrane integrity (Fig. 8) and, finally, cell death at 8 h (Fig. 4). The overexpression of mitochondrial PHGPx prevented cell death caused by KCN.
Mitochondrial PHGPx hindered the generation of hydroperoxides, which was an early feature of cell damage. Mitochondrial PHGPx failed to prevent cell death in response to CCCP or oligomycin, both of which directly reduced the membrane potential and the level of cellular ATP without the production of hydroperoxides. Thus, mitochondrial PHGPx appeared to maintain the functions of mitochondria by reduction of intracellular hydroperoxides generated as a result of damage to the mitochondrial respiratory machinery.
PHGPx in mitochondria effectively reduced the H 2 O 2 generated in mitochondria that had been damaged by exposure to KCN since the extent of lipid peroxidation caused by H 2 O 2 was significantly reduced in KCN-treated M15 cells. In general, PHGPx reduces H 2 O 2 less effectively than cGPx. The present results suggest that PHGPx in mitochondria might limit local increases in concentrations of H 2 O 2 . By contrast, PHGPx in the cytosol might reduce H 2 O 2 that is dispersed in the cytosol much less efficiently.
M15 and L9 cells exhibited different levels of sensitivity to extracellular oxidative damage (Fig. 9). M15 cells were more  resistant to the cytotoxic effects of t-BuOOH than L9 cells even though the total activity of intracellular PHGPx was similar in both lines of transformed cells. These results suggest that mitochondria might be a primary target for t-BuOOH. Mitochondrial PHGPx prevents cell death by protecting the mitochondrial machinery from t-BuOOH (Fig. 10). Mitochondrial PHGPx not only prevents direct injury of mitochondria by KCN, but it also reduces the cytotoxicity of exogenously added t-BuOOH by protecting mitochondrial functions. Levels of ROS in mitochondria increase substantially in pathological situations, such as ischemia reperfusion (38). Several factors have been reported that protect mitochondria against oxidative damage. Mitochondrial SOD (Mn-SOD) is induced by TNF-␣, which initiates cell death through the elevation of levels of ROS in mitochondria (39). Induced Mn-SOD might protect cells from the toxic effects of TNF-␣ (40). Heat shock proteins are also induced to protect cells from stresses that include ROS. Overexpression of heat shock protein 70 results in resistance to the cytotoxicity of TNF-␣. Polla et al. (41) demonstrated that heat shock protein 70 prevented changes in mitochondrial membrane potential by H 2 O 2 , and they suggested that mitochondria might be selective targets for protective effects against the oxidative injury. Bcl-2, which is an anti-apoptosis protein, is located primarily on the outer membrane of mitochondria. Bcl-2 prevents cells from undergoing necrosis in response to inhibitors of the respiratory chain (chemical hypoxia) such as rotenone, antimycin A, and KCN (36) and from apoptosis in response to a variety of other stimuli (42,43). Bcl-2 prevents cell death by inhibiting the loss of mitochondrial membrane potential (36) and by inhibiting the release of cytochrome c, an activator of caspases (44,45), from the mitochondria to the cytosol. Exogenous H 2 O 2 is a potent inducer of the liberation of cytochrome c from damaged mitochondria, and Bcl-2 effectively prevents the toxic effects of exogenous H 2 O 2 that lead to cell death (43). These earlier results suggest that the generation of ROS in the mitochondria might be a key step in the initiation of cell death and that mitochondria are well placed to be sensors of oxidation damage to cells (46). PHGPx is localized predominantly at contact sites between the outer and inner membranes of mitochondria in the rat testes (47). The reduction of levels of ROS in mitochondria by PHGPx might also be associated with the defenses of the cell against apoptosis that is mediated by damages to mitochondria.
Recent reports indicate that ROS play an important role as mediators or modulators in cellular signaling pathways, such as the Ras signaling pathway (48), the mitogen-activated protein kinase pathway (49), and the activation of nuclear factor B (NFB) (50). ROS generated in mitochondria might be responsible for the activation of NFB (51) and of Sterile 20 (Ste 20), which is likely to be an oxidant-stress-responsive kinase-1 (SOK-1) (52). Brigelius-Flohe et al. (53) found that the activation of NFB by interleukin-1 was inhibited in ECV304 cells that overexpressed PHGPx. The present study also suggests that mitochondrial PHGPx might participate in the regulation of signal transduction pathways that are triggered by ROS in mitochondria. FIG. 9. Effects of t-BuOOH on the viability of PHGPx-overexpressing cells. S1 cells (closed circles), L9 cells (closed triangles), and M15 cells (closed squares) were exposed to the indicated concentrations of t-BuOOH for 2 h. Then viability was determined as described in the legend to Fig. 4. Data are means Ϯ S.D. of the results from four replicates in each case.
FIG. 10. Flow cytometric analysis of changes in mitochondrial membrane potential and the integrity of plasma membranes of PHGPx-overexpressing cells exposed to t-BuOOH. S1 cells (A), L9 cells (B), and M15 cells (C) were double-stained with Rh123 and PI after incubation with 50 M t-BuOOH for 60 min. The intensity of fluorescence from PI was plotted against that from Rh123. Similar results were obtained in three independent experiments.