Superoxide Dismutase 1-mediated Production of Ethanol- and DNA-derived Radicals in Yeasts Challenged with Hydrogen Peroxide

Peroxiredoxins are receiving increasing attention as defenders against oxidative damage and sensors of hydrogen peroxide-mediated signaling events. In the yeast Saccharomyces cerevisiae, deletion of one or more isoforms of the peroxiredoxins is not lethal but compromises genome stability by mechanisms that remain under scrutiny. Here, we show that cytosolic peroxiredoxin-null cells (tsa1Δtsa2Δ) are more resistant to hydrogen peroxide than wild-type (WT) cells and consume it faster under fermentative conditions. Also, tsa1Δtsa2Δ cells produced higher yields of the 1-hydroxyethyl radical from oxidation of the glucose metabolite ethanol, as proved by spin-trapping experiments. A major role for Fenton chemistry in radical formation was excluded by comparing WT and tsa1Δtsa2Δ cells with respect to their levels of total and chelatable metal ions and of radical produced in the presence of chelators. The main route for 1-hydroxyethyl radical formation was ascribed to the peroxidase activity of Cu,Zn-superoxide dismutase (Sod1), whose expression and activity increased ∼5- and 2-fold, respectively, in tsa1Δtsa2Δ compared with WT cells. Accordingly, overexpression of human Sod1 in WT yeasts led to increased 1-hydroxyethyl radical production. Relevantly, tsa1Δtsa2Δ cells challenged with hydrogen peroxide contained higher levels of DNA-derived radicals and adducts as monitored by immuno-spin trapping and incorporation of 14C from glucose into DNA, respectively. The results indicate that part of hydrogen peroxide consumption by tsa1Δtsa2Δ cells is mediated by induced Sod1, which oxidizes ethanol to the 1-hydroxyethyl radical, which, in turn, leads to increased DNA damage. Overall, our studies provide a pathway to account for the hypermutability of peroxiredoxin-null strains.

Living organisms are constantly exposed to oxygen-and nitrogen-derived reactive species that are produced by normal metabolic activity and in response to external stimuli. To pro-tect themselves against the toxicity of these species, aerobic organisms have evolved a range of defense mechanisms (1). Among these, a family of cysteine-based peroxidases, currently named peroxiredoxins, has attracted considerable attention because of its ubiquity and versatility. These enzymes have been shown to detoxify hydrogen peroxide, organic peroxides and peroxynitrite through oxidation of their reactive cysteine residues, which are recycled back by reducing equivalents provided by thioredoxin and other thiol-electron donors (reviewed in Refs. [2][3][4][5]. In addition to detoxifying peroxides, specific peroxiredoxins have been shown to act as molecular chaperones (6,7) and to play roles in regulating hydrogen peroxide-mediated cell signaling events (3,8,9).
Many organisms have multiple peroxiredoxins. Six peroxiredoxins have been identified in human cells, and five in the yeast Saccharomyces cerevisiae. Peroxiredoxin isoforms are distributed to different locations within the cell, and two of the yeast peroxiredoxins, Tsa1 and Tsa2, are cytosolic (10). Tsa1 was the first peroxiredoxin identified in eukaryotes. It is expressed constitutively and corresponds to 0.7% of total soluble protein content in this species. In contrast, the levels of Tsa2 are very low under normal conditions, but are highly induced upon treatment of the yeast with peroxides (11,12). Tsa1 and Tsa2 share 86% identity in their amino acid sequence, and both react with hydrogen peroxide with second order rate constants similar to those of hemoproteins, such as catalase (k ϳ 10 7 M Ϫ1 s Ϫ1 ) (13). Deletion of the TSA1 gene increases the expression of Tsa2 and other peroxiredoxin isoforms (10,14,15), suggesting that these peroxidases have both redundant and non-redundant physiological functions.
The deletion of one or all of the peroxiredoxin genes in S. cerevisiae is not lethal, probably because of the resulting increase in the levels of other antioxidant enzymes, such as catalase, cytochrome c peroxidase, and Sod1, that has been demonstrated by transcriptional and proteomic analysis (10,15,16). A problem arising from this compensatory response, however, is the mutator phenotype of the peroxiredoxin-null strains that display increased spontaneous mutation rates and accumulated gross chromosomal rearrangements (16 -19). These consequences are particularly conspicuous in TSA1-null cells, suggesting a major contribution of Tsa1 to genome stability. Such protection has been attributed to the ability of Tsa1 to reduce the levels of reactive oxygen species that would other-wise oxidize DNA leading to mutations, chromosomal rearrangements and cell death. These genetic and lethality studies, however, did not provide mechanistic insights on how deletion of peroxiredoxin genes leads to DNA damage nor to its molecular nature (16 -19).
Here, contrary to the common notion, we show that deletion of both TSA1 and TSA2 genes in S. cerevisiae increases resistance to toxic concentrations of hydrogen peroxide. In parallel, hydrogen peroxide consumption and ethanol-and DNA-derived radical formation are increased mainly through higher expression of the enzyme Sod1.

EXPERIMENTAL PROCEDURES
Materials-All chemicals were purchased from Sigma-Aldrich, Merck, or Fisher and were analytical grade or better. Yeast nitrogen base (20) was from Difco and media supplements (amino acids, adenine and uracil) were from Sigma-Aldrich or Synth. Desferrioxamine was purchased from Novartis. Bathocuproine disulfonic acid and [2-13 C]ethanol were from Sigma-Aldrich. Hydrogen peroxide concentration was determined spectrophotometrically at 240 nm (⑀ 240 ϭ 43.6 M Ϫ1 cm Ϫ1 ) (21). All solutions were prepared with water purified in a Millipore Milli-Q system. All buffers were treated with Chelex-100 to remove trace amounts of metal ion contaminants prior to use.  (22) of the BY47411 strain with the YEp600 plasmid containing one copy of the human sod1 gene with its own promoter. This plasmid was extracted from EGy118 yeasts provided by Dr. Edith B. Gralla. The strains were grown in complete synthetic media (yeast nitrogen base) supplemented with 2% glucose, amino acids, adenine, and uracil under shaking at 300 rpm, 30°C (11). For all experiments, strains were grown overnight, diluted to A 600 nm ϭ 0.2, and collected as soon as they reached mid-log phase (A 600 nm ϭ 0.8). Growth curves under fermentative conditions were obtained for the four strains by diluting overnight cultures to A 600 nm ϭ 0.2 and by monitoring the optical density at 2-h intervals.

S. Cerevisiae Strains and Growth
Viability Assay-The viability of the cells was monitored by the number of colony forming units. Cell cultures at mid-log phase of growth were washed with 0.1 M phosphate buffer containing 0.1 mM DTPA, 3 pH 7.4, and resuspended in growth media to a density of 5 ϫ 10 7 cells/ml. Half of the cell suspension was treated with 1 mM hydrogen peroxide under shaking at 350 rpm for 30 min at 30°C; the other half was treated with the same volume of buffer and incubated as above. The cell suspensions were then serially diluted, and 100-l aliquots of 1/10 4 and 1/10 5 dilution were plated onto complete synthetic media supplemented with 2% glucose and 2% agar. The plates were incubated at 30°C for 2 days, and the colony forming unit was number counted. The results shown are expressed as the percentage of colony forming units of hydrogen peroxide-treated cultures in relation to the corresponding controls.
Hydrogen Peroxide Consumption-Hydrogen peroxide consumption was determined by the ferrous oxidation xylenol orange assay (FOX) (23). Cells at mid-log phase were treated with 1 mM hydrogen peroxide as above. At the specified times, 50-l aliquots were diluted with 500 l of growth media and mixed with 950 l of FOX solution (100 M xylenol orange, 250 M ammonium ferrous sulfate, 100 mM sorbitol, and 25 mM sulfuric acid). After 30-min incubation at room temperature, the absorbance of the samples was read at 560 nm. Calibration was performed with a standard solution of hydrogen peroxide.
Detection of Ethanol-derived Radicals-Production of the 1-hydroxyethyl radical formed from the oxidation of ethanol, a glucose metabolite, was detected by EPR spin-trapping experiments with POBN (24,25). Cells (5 ϫ 10 7 cells/ml) in growth media containing glucose were preincubated with 90 mM POBN for 5 min and 1 mM hydrogen peroxide was added under shaking at 30°C. After 30 min, aliquots were transferred to flat cells, and the EPR spectra were scanned. In some experiments, cells were incubated with POBN as above, and 171 mM [2-13 C]ethanol was added together with hydrogen peroxide. In other experiments, cells in growth media containing glucose were preincubated with 2 mM desferrioxamine, 2 mM bathocuproine disulfonic acid, or with 10 mM diethyldithiocarbamate for 30 min or 1 h before addition of the components specified above. Also, cells in media alone were heated at 90°C for 20 min and brought to 30°C before addition of glucose, POBN, and hydrogen peroxide. In the case of purified enzymes, 10 M catalase, horseradish peroxidase, cytochrome c, or bovine Sod1 was incubated with 90 mM POBN, 1 mM hydrogen peroxide, 0.1 mM DTPA, and 171 mM ethanol in 100 mM phosphate buffer, pH 7.4, at 30°C. After 30-min incubation, the samples were transferred to flat cells, and the EPR spectra were scanned. In the case of Sod1, kinetic experiments were also performed. EPR spectra were recorded at room temperature (25 Ϯ 2°C) on a Bruker EMX spectrometer equipped with an ER4122 SHQ 9807 high sensitivity cavity. Radical adduct quantification was performed by double integration of the EPR spectra and comparison with a standard solution of 4-hydroxy-2,2,6,6-tetramethylpiperidinyloxy. Computer simulation analyses of some spectra were performed by using a program written by Duling (26).
Chelatable Iron Ion Measurements-Chelatable iron levels in the strains were determined by low temperature EPR after chelation with desferrioxamine (27,28). WT and tsa1⌬tsa2⌬ cells were collected at mid-log growth phase. Approximately 10 9 cells of each strain were resuspended in 10 ml of fresh growth medium without glucose, but containing 2 mM desferrioxamine. After 30-min incubation at 30°C, cells were collected and washed with 10 ml of cold Tris-HCl buffer (20 mM), pH 7.4. The cell pellet was resuspended in 400 l of the same buffer con-taining 10% glycerol, transferred to a 1-ml disposable syringe, and frozen in liquid nitrogen. The samples were extruded from the syringe into a finger-tip Dewar flask containing liquid nitrogen and examined by EPR at 77 K in the region of g ϳ4.0 (28). The concentration of the desferrioxamine-iron complex present in the cell suspensions was obtained by double integration of the EPR signal and comparison with a standard curve constructed with known concentrations of the iron(III)-desferrioxamine complex. This was prepared by mixing different concentrations of ferrous ammonium sulfate with 2 mM desferrioxamine; the complex concentration was determined spectrophotometrically (⑀ 430 ϭ 2865 M Ϫ1 cm Ϫ1 ) (29).
Total Iron, Copper, and Zinc Ion Measurements-Levels of iron, copper, and zinc ions in each strain were determined by atomic absorption spectrometry. Approximately 10 9 cells from each strain were washed in metal-free water, resuspended in a solution of nitric acid, hydrogen peroxide, and water (2:1:3, v/v), and digested in a microwave oven. The metal contents of the digested samples were determined in an Analytikjena AG, AAS ZEEnit 60 instrument. Iron, copper, and zinc were detected at 324.8, 248.3, and 213.9 nm, respectively (30).
Total Peroxide Measurements-Levels of total peroxides in WT and tsa1⌬ tsa2⌬ strains were determined by the FOX assay (23) as adapted to yeasts (31). Cells (5 ϫ 10 7 cells/ml) of each strain were washed and resuspended in growth media and treated or not with 1 mM hydrogen peroxide. After 30-min incubation, the cells were washed twice and resuspended in 100 l of 50 mM phosphate buffer. An equal volume of glass beads (425-600 m) was added, and the samples were vortexed (1 min) and ice-cooled. After addition of 900 l of cool methanol containing 4 mM butylated hydroxytoluene, the samples were submitted to two cycles of vortexing (1 min) and ice cooling. Next, the samples were centrifuged (1000 ϫ g, 10 min). The supernatants were collected (100 l) and mixed with 900 l of the FOX reagent (100 M xylenol orange, 250 M ammonium ferrous sulfate, 25 mM sulfuric acid, and 4 mM butylated hydroxytoluene in 90% (v/v) methanol). After 30-min incubation at room temperature, the absorbance of the samples was read at 560 nm. According to previous calibrations performed with standard peroxides, a mean apparent extinction coefficient of 4.5 ϫ 10 4 M Ϫ1 cm Ϫ1 was employed (31).
Expression of Sod1-The level of Sod1 expression in WT and tsa1⌬ tsa2⌬ strains was determined by Western blot analysis. To this end, cell extracts of the strains were obtained as previously described (11). Briefly, cells collected at mid-log growth phase were washed and resuspended in 50 mM Hepes buffer containing 50 mM NaCl, 2 g/ml leupeptin, and 1 g/ml of pepstatin. Next, an equal volume of glass beads (425-600 m) was added. After 2 cycles of vortexing (6 min) and ice-cooling (6 min), the samples were centrifuged (16,000 ϫ g, 5 min). The supernatants were collected and centrifuged again (16,000 ϫ g, 30 min) to remove precipitated material. The supernatants are referred to as cell extracts, and their protein contents were determined by the Bradford method with a Bio-Rad Kit. Cell extracts (40 g of protein) were submitted to electrophoresis (12% acrylamide gel) and transferred onto a nitrocellulose membrane (Hybond-D Extra, Amersham Biosciences). After washing (3 times, 10 min) with TBS (50 mM Tris-HCl, pH 7.5, 150 mM NaCl), the membrane was blocked with TBST (TBS plus 0.05% Tween 20) containing 5% nonfat milk, at room temperature for 90 min. The primary antibody was anti-human Sod1 (Calbiochem) prepared in TBST containing 0.1% nonfat milk (1:1.500 dilution). Excess antibody was removed by washing three times with TBST. The membrane was then incubated with the secondary antibody (anti-sheep IgG, peroxidase conjugated, Calbiochem) prepared in TBST containing 0.1% nonfat milk (1:7,000 dilution). After 1-h incubation, the membrane was washed three times with TBST and treated for 5 min with the solutions from the kit SuperSignal West Pico Chemiluminescent Substrate (Pierce) and exposed to an x-ray film. Relative quantification of the bands was performed by densitometry (ImageQuaNT V5.2, Molecular Dynamics).
Superoxide Dismutase Activity-Superoxide dismutase activity in WT and tsa1⌬ tsa2⌬ cell extracts was determined by native PAGE staining (32) and inhibition of cytochrome c reduction (33,34). After electrophoresis of cell extracts (40 g of protein) under non-reducing conditions, the gel was incubated in the dark for 20 min with a solution composed of 0.25 mg/ml nitro blue tetrazolium plus 0.1 mg/ml riboflavin. Then, a solution of 10 mg/ml N,N,NЈ,NЈ-tetramethylethylenediamine was added, and the gel was kept under shaking and light until bands became visible (32). Relative quantification of the bands was performed by densitometry (ImageQuaNT V5.2). The cytochrome c reduction assay was performed as previously described (33,34). The incubations contained 100 M xanthine and xanthine oxidase in amounts that cause an absorbance change of 0.025/min, 1 unit of catalase ,and cell extracts (0 -50 g protein) in 100 mM phosphate buffer, pH 7.4, 30°C.
Bicarbonate-dependent Peroxidase Activity-This activity was evaluated in whole cells by EPR monitoring of the oxidation of the spin-trap 5,5-dimethylpyrroline-N-oxide (DMPO) to DMPO/ ⅐ OH radical adduct was monitored (35,36). Cells (5 ϫ 10 7 cells/ml) of each strain were resuspended in 100 mM phosphate buffer containing 0.1 mM DTPA, pH 7.4, and incubated with 80 mM DMPO at room temperature for 5 min. After addition of 25 mM sodium bicarbonate and 1 mM hydrogen peroxide, the samples were incubated at 30°C for 15 min and transferred to flat cells, and their EPR spectra were scanned at room temperature as described above. Quantification of radical adduct was performed by double integration of the EPR spectra and comparison with a standard solution of 4-hydroxy-2,2,6,6-tetramethylpiperidinyloxy.
Detection of DNA-derived Radicals-Formation of DNAderived radicals was monitored by immuno-spin trapping employing DMPO and the antibody against oxidized DMPO adducts (37). Cells (5 ϫ 10 7 cells/ml) of each strain were washed, resuspended in fresh media, and treated with 100 mM DMPO and 1 mM hydrogen peroxide at 30°C. After 30-min incubation, the cells were washed with distilled water (2 times) and resuspended in 800 l of a buffer to digest cell walls (0.9 M sorbitol, 0.1 M EDTA, 50 mM dithiothreitol, and 3.5 g/ml zymolyase 20T) (38). After 2-h incubation at 37°C, samples were centrifuged (1,000 ϫ g, 10 min), and the spheroplasts obtained were resuspended in lysis buffer (1% SDS, 100 mM NaCl, 25 mM DTPA in 10 mM Tris-HCl, pH 8.0). From this point on, DNA extraction and transfer onto nitrocellulose membrane followed the previously described protocol (37). DNA extracted from WT cells (100 ng) treated with 50 mM DMPO, 1 mM CuCl 2 , and 20 M hydrogen peroxide in phosphate buffer, pH 7.4, for 30 min at 30°C was employed as a positive control. DNA radicals/DMPO nitrone adducts blotted onto the nitrocellulose membrane were detected with the anti-DMPO nitrone adduct antibody provided by Dr. Ronald P. Mason (37). First, the membrane was blocked for 1 h with phosphate-buffered saline buffer (0.725 M NaCl, 2.7 mM KCl, 80 mM Na 2 HPO 4 and 14 mM KH 2 PO 4 ) containing 3% nonfat milk. After washing for 10 min with phosphate-buffered saline supplemented with 0.05% nonfat milk and 0.1% Tween 20, the membrane was incubated with the anti-DMPO/adduct antibody for 1 h. The antibody was prepared in phosphate-buffered saline supplemented with 0.05% nonfat milk and 0.1% Tween 20 (1:10,000 dilution). Excess antibody was removed by three 10-min washes with the same buffer, and the membrane was incubated 1 h with the secondary antibody (anti-rabbit IgG, peroxidase-conjugated). After one phosphate-buffered saline wash, the membrane was incubated with the solutions from the kit SuperSignal West Pico Chemiluminescent Substrate (Pierce Biotechnology) for 5 min and exposed to an x-ray film. Relative quantification of the bands was performed by densitometry (ImageQuaNT V5.2).
Detection of DNA Adducts-To probe for the addition of glucose metabolites to yeast DNA, we monitored incorporation of 14 C at the DNA by the strains grown in media supplemented with [ 14 C]glucose (Schwarz/Mann; 1 mCi/ml and 230 mCi/ mM). Cells (2 ϫ 10 8 ) harvested at the exponential growth phase were resuspended in 500 l of fresh media containing four times less glucose than usual and supplemented with 50 Ci of [ 14 C]glucose (39). These samples were treated with 1 mM hydrogen peroxide for 30 min at 30°C, in the absence or presence of POBN (90 mM). DNA from the samples was extracted as above. DNA (100 ng) from each experimental condition was transferred to filter paper that was incubated with scintillation liquid. Radioactivity was measured in a liquid scintillation analyzer (166 TR, Packard).

RESULTS
tsa1⌬tsa2⌬ Cells Are More Resistant to Hydrogen Peroxide but Metabolize It Faster Producing Ethanol-derived Radicals-As previously reported (11,14,16), tsa1⌬, tsa2⌬, and tsa1⌬tsa2⌬ yeast cells were shown to be viable and to grow similarly to the WT strain in fermentative media (data not shown). However, surprisingly, tsa1⌬tsa2⌬ cells were significantly more resistant than WT cells to toxic concentrations of hydrogen peroxide (Fig. 1A). No significant differences were observed between the WT and tsa1⌬ and tsa2⌬ cells. The unexpected resistance of the double mutant cells to hydrogen peroxide led us to explore this phenomenon in more detail. First, the consumption of 1 mM hydrogen peroxide by the strains was compared. No significant differences were observed in the rates of hydrogen peroxide consumption by the WT and single mutants (data not shown). The tsa1⌬tsa2⌬ strain, however, consumed hydrogen peroxide considerably faster than the WT strain (Fig. 1B), suggesting an increased capacity to metabolize hydrogen peroxide (Fig. 1A).
Hydrogen peroxide consumption by yeast cells may produce radical intermediates (25). Thus, radical production by the strains while metabolizing 1 mM hydrogen peroxide was compared by EPR spin-trapping experiments with POBN. After 30 min of incubation, radical production was marginal except for tsa1⌬tsa2⌬ cells that yielded a clear EPR signal whose parameters (a N ϭ 15.8 G; a H ϭ 2.6 G) are characteristic of the POBN/ ⅐ 1-hydroxyethyl radical adduct (POBN/ ⅐ CHOHCH 3 ) (Fig. 2) (24,25). To confirm radical identity, we performed parallel experiments where 171 mM [2-13 C]ethanol was added together with hydrogen peroxide to the incubations containing tsa1⌬tsa2⌬ cells, growth media, glucose, and POBN. In these cases, the 6-line spectrum detected in the presence of glucose alone was substituted by a 12-line spectrum (Fig. 2). Computer  FEBRUARY 27, 2009 • VOLUME 284 • NUMBER 9 simulation showed that the latter is an EPR composite spectrum of the POBN/ ⅐ CHOHCH 3 and POBN/ ⅐ 13 CHOH 13 CH 3 radical adduct (a N ϭ 15.5 G; a H ϭ 2.6 G; a 13C ϭ 4.2 G) (24) in relative yields of 30 and 70%, respectively (Fig. 2). Detection of the hyperfine splitting of 13 C from ethanol unambiguously proves that the main radical produced by tsa1⌬tsa2⌬ cells challenged with hydrogen peroxide is the 1-hydroxyethyl radical.

Ethanol-and DNA-derived Radicals in TSA1TSA2-null Cultures
The spin-trapping experiments were repeated several times, and the results were consistent. Considerable yields of the POBN/ ⅐ 1-hydroxyethyl radical adduct were obtained only in cultures of the tsa1⌬ tsa2⌬ strain and were dependent on all incubation components, that is, yeast cells, glucose, and hydrogen peroxide (Fig. 2). Thus, despite being more resistant to hydrogen peroxide, the tsa1⌬tsa2⌬ strain removes it faster and produces the 1-hydroxyethyl radical.
Routes for 1-Hydroxyethyl Radical Formation-The source of the 1-hydroxyethyl radical in yeast cultures metabolizing glucose should be its metabolite ethanol. Indeed, fermentative yeasts accumulate ethanol, because glucose represses respiration and mitochondrial biogenesis through several signaling pathways (40). To produce the 1-hydroxyethyl radical, ethanol has to be oxidized by the hydroxyl radical or by a hydroxyl radical-like oxidant (Reaction 1), whose production is likely to depend on Fenton chemistry, that is, on hydrogen peroxide decomposition by redox-active transition metal ions (Reaction 2) (24,41,42).
Although intracellular levels of transition metal ions are tightly controlled, regulatory mechanisms can be altered by stressed conditions such as those likely to result from deletion of genes encoding antioxidant enzymes, including Tsa1 and Tsa2. Redox-active iron ions in cells become more accessible to chelators, and their levels can be estimated by complexation with desferrioxamine and EPR analysis (see "Experimental Procedures") (27,28). The levels of the iron(III)-desferrioxamine complex, however, were higher in WT than in tsa1⌬tsa2⌬ cells (1.00 Ϯ 0.05 and 0.28 Ϯ 0.03 g/g of cell pellet, respectively) ( Fig. 3, inset, and Table 1). Relevantly, parallel experiments showed that preincubation of tsa1⌬ tsa2⌬ cells with 2 mM desferrioxamine, or 2 mM bathocuproine before the addition of hydrogen peroxide did not prevent POBN/1-hydroxyethyl radical adduct production, although its yield decreased (ϳ30%) (Fig. 3). Total iron content in WT and tsa1⌬tsa2⌬ cells was determined by atomic absorption spectrometry, and the same trend of chelatable iron ion levels was observed. Total iron content was higher in WT than in tsa1⌬tsa2⌬ cells (Table 1). In contrast, total copper content was higher in tsa1⌬tsa2⌬ cells, whereas total zinc content was similar. Although copper content is higher in tsa1⌬tsa2⌬ cells, chelators of copper I and II (1) inhibited radical formation to ϳ70% of control levels (Fig. 3). Taken together, these results ( Fig. 3 and Table 1) argue against iron and copper ion-mediated Fenton chemistry being the main route for 1-hydroxyethyl radical formation (Reactions 1 and 2).
Because peroxiredoxins are important to detoxify organic peroxides, deletion of their genes may lead to accumulation of organic peroxides and their derived peroxyl radicals (ROO ⅐ ), which can oxidize ethanol to the 1-hydroxyethyl radical (k ϳ 1 ϫ 10 4 M Ϫ1 s Ϫ1 ) (43). However, levels of total peroxides present in WT and tsa1⌬tsa2⌬ cells were similar under basal conditions or after treatment with hydrogen peroxide (Table 1). Thus, organic peroxides and their derived radicals are unlikely to be major sources of the 1-hydroxyethyl detected in tsa1⌬tsa2⌬ cells.
On the other hand, the higher content of copper in tsa1⌬tsa2⌬ cells compared with WT cells led us to consider the possibility of increased expression of Sod1 providing a route for ethanol oxidation to the 1-hydroxyethyl radical. In fact, deletion of the TSA1 gene has previously been shown to increase the mRNA levels of Sod1, catalase, and cytochrome peroxidase among other enzymes (10,15,16). In addition, Sod1 displays a peroxidase activity, which may oxidize ethanol to 1-hydroxyethyl radical, although this process has been considered inefficient (35,44). These previous studies, however, were performed in the presence of bicarbonate buffer, which modulates Sod1 peroxidase activity to produce the carbonate radical (35, 36, 45), which is not an efficient oxidant of alcohols (46). Thus, it became relevant to compare the ability of purified enzymes, such as bovine Sod1, catalase, cytochrome c, and horseradish peroxidase, to oxidize ethanol to the 1-hydroxyethyl radical in the presence of hydrogen peroxide in phosphate buffer con-

TABLE 1 Basal levels of desferroxamine-chelatable iron ion (Fe(III-DF), total iron, copper, and zinc ions, and of total peroxides in WT and tsa1⌬tsa2⌬ cells
Chelatable iron ions were determined by low temperature EPR; total iron, copper and zinc ion contents were determined by atomic absorption spectrometry; and total peroxides were determined by the FOX assay (see "Experimental Procedures").
The shown values are expressed as micrograms of ion/g of cell pellet and nanomoles of peroxide/g of cell pellet, and correspond to the mean Ϯ S.D. values obtained from three different experiments.  FEBRUARY 27, 2009 • VOLUME 284 • NUMBER 9

Ethanol-and DNA-derived Radicals in TSA1TSA2-null Cultures
taining DTPA. Considerable yields of the POBN/ ⅐ 1-hydroxyethyl radical adduct were produced only in the presence of Sod1 (Fig. 4). This was confirmed by the 12-line spectrum produced in incubations containing [2-13 C]ethanol, which is characteristic of the POBN/ ⅐ 13 CHOH 13 CH 3 radical adduct (a N ϭ 15.5 G; a H ϭ 2.6 G; a 13C ϭ 4.2 G) (24). The radical adduct detected in the presence of horseradish peroxidase is an oxidation product of the spin trap POBN as demonstrated by control experiments in the absence of ethanol (data not shown). Production of the POBN/ ⅐ 1-hydroxyethyl radical adduct by Sod1/hydrogen peroxide was shown to be ethanol-and time-dependent (Fig. 4B). This indicates that ethanol oxidation to the 1-hydroxyethyl radical depends mainly on the peroxidase activity of Sod1 and not from Fenton chemistry resulting from enzyme inactivation and copper liberation from its active site (47). These results confirm that biological oxidation of ethanol to the 1-hydroxyethyl radical is likely to be mediated mainly by the hydroxyl radical and hydroxyl radical-like oxidants (24,41,42). Indeed, the peroxidase activity of Sod1 produces a hydroxyl-like oxidant, which is able to oxidize ethanol, whereas catalase, horseradish peroxidase, and cytochrome c use hydrogen peroxide to oxidize substrates through ferryl states that are not efficient one-electron oxidants of ethanol (Fig. 4).
To further establish a role for Sod1 in the production of 1-hydroxyethyl radicals by the tsa1⌬tsa2⌬ strain, cells were pretreated with 10 mM diethyldithiocarbamate, a Sod1 inhibitor and copper chelator (48), or heated to promote enzyme inactivation. Cells submitted to both treatments failed to produce POBN/ ⅐ 1-hydroxyethyl radical adduct upon incubation with glucose, hydrogen peroxide, and POBN, in contrast with untreated cells (Fig. 5). These results indicate that 1-hydroxyethyl radical formation depends on an enzymatic activity, most likely the Sod1 peroxidase activity (Fig. 4). This conclusion was reinforced by the 5-fold higher level of Sod1 expression in the tsa1⌬tsa2⌬ cells compared with the WT cells (Fig. 6A). This higher expression was confirmed by comparing the superoxide dismutase activity of the extracts monitored by both native PAGE stain (Fig. 6B) and inhibition of cytochrome c reduction (Fig. 7A) (32)(33)(34). Likewise, the bicarbonate-dependent peroxidase activity, characteristic of the Sod1 enzyme (35,36,45), was higher in tsa1⌬tsa2⌬ cells (Fig. 7A, inset). Next, we transfected WT yeast to overexpress human Sod1. These cells displayed superoxide dismutase and bicarbonate-dependent peroxidase activity similar to those of the tsa1⌬tsa2⌬ cells (Fig.  7A). Likewise, they produced similar yields of the 1-hydroxyethyl radical adduct (Fig. 7B), further demonstrating the role of Sod1 in radical production.
Independent of the employed methodology, the increase in superoxide dismutase and bicarbonate-dependent peroxidase activity in tsa1⌬tsa2⌬ cells compared with WT cells was roughly the same (ϳ2-fold) (Figs. 6B and 7A). However, the increase in Sod1 contents was higher (ϳ5-fold) (Fig. 6A) indicating that not all the expressed enzyme is active, probably because total copper content increased less (Ͻ2-fold) ( Table 1) than overall Sod1 expression (49).
Production of DNA-derived Radicals and Adducts-Like other alkyl radicals, the 1-hydroxyethyl radical is known to attack DNA to produce DNA-derived radicals and DNA-8-alkylguanine adducts (24,50,51). To monitor the production of DNA adducts under our experimental conditions, [ 14 C]glucose was added to the cultures, and the amount of 14 C incorporated to DNA ([ 14 C]DNA) was determined after 30-min incubation. As shown in Fig. 8A, tsa1⌬tsa2⌬ cells treated with hydrogen peroxide contained about three times more [ 14 C]DNA than WT cells. Relevantly, [ 14 C]DNA production in tsa1⌬tsa2⌬ cells was inhibited by the spin-trap POBN (ϳ30%) indicating that a considerable fraction of 14 C incorporated into DNA comes  from a radical metabolite of glucose, most likely, the 1-hydroxyethyl radical (Fig. 2). Attack of radicals on DNA was also indicated by parallel experiments with the antibody anti-DMPO, which reveals the amount of DNA-derived radicals that are  trapped by DMPO (37). Indeed, treatment of tsa1⌬tsa2⌬ cells with hydrogen peroxide produced about two times as many DNA-derived radicals as WT cell treatment (Fig. 8B). Taken together, these results associate the higher production of the 1-hydroxyethyl radical by tsa1⌬tsa2⌬ cells (Fig. 2) with increased damage to their DNA (Fig. 8).

DISCUSSION
Our results demonstrated that deletion of both the TSA1 and TSA2 genes in S. cerevisiae significantly increases cellular resistance to toxic concentrations of hydrogen peroxide while increasing its consumption rate (Fig. 1). These results contrast with previous studies reporting that tsa1⌬, tsa2⌬, and tsa1⌬tsa2⌬ strains are more susceptible than the WT strain to hydrogen peroxide (14,16). In another study employing various experimental conditions, tsa2⌬ cells were shown to be more resistant than WT (11), although to a lesser extent than reported here for the double mutant. These tolerance assays, however, were performed by spot tests, in which the solid media may affect hydrogen peroxide metabolism by the strains. In synthetic liquid media, the resistance of tsa1⌬tsa2⌬ cells to 1 mM hydrogen peroxide (Fig. 1) suggests that catalase, which is induced in these cells (15,16), is the main enzyme responsible for decomposing these toxic oxidant concentrations. In agreement, the consumption of hydrogen peroxide by cell extracts of the tsa1⌬ and tsa2⌬ strain was shown to be almost completely inhibited by azide (11). However, our results indicate that part of the hydrogen peroxide consumption by tsa1⌬tsa2⌬ cells (Fig. 1B) is also mediated by induced Sod1 (Figs. 5-7). Although catalases are efficient enzymes with specific activity values around 17,750 units/nmol (52), their levels in extracts of WT and tsa1⌬ and tsa2⌬ cells in exponential growth under fermentative conditions (high glucose concentrations) are low (11). Indeed, catalase activity values ranging from 0.38 to 1.3 units/mg have been reported for yeast extracts (11,53), which corresponds to ϳ5 ϫ 10 Ϫ5 nmol catalase/mg. In contrast, Sod1 is an abundant enzyme, and its level in extracts of tsa1⌬tsa2⌬ cells can be estimated as ϳ7 ϫ 10 Ϫ1 nmol Sod1/mg from the data in Fig. 7A (67 units/mg of protein) and the specific activity reported for the bovine liver enzyme (ϳ90 units/nmol) (54).
Thus, the higher level of Sod1 in tsa1⌬tsa2⌬ cells (Figs. 6 and 7) should favor the occurrence of its peroxidase activity (Fig. 4), which oxidizes the glucose metabolite ethanol to the 1-hydroxyethyl radical (Fig. 2). This radical, direct or indirectly, leads to the production of DNA-derived radicals and adducts (Fig. 8). Thus, our studies provide a pathway to account for the hypermutability of peroxiredoxin-null yeast cells while performed with cells challenged with toxic concentrations of hydrogen peroxide (1 mM). This high concentration was probably required to make it possible to detect significant differences among the strains regarding hydrogen peroxide resistance and consumption (Fig. 1) and radical production (Figs. 2 and 8B). Indeed, the difficulties involved in the precise quantification of oxidants and radicals in cells and cell cultures are well known (36,41,42,46). In unstressed cells, it has been reported that the mutation rates of the studied strains follow the order WT Ϸ tsa2⌬ Ͻ tsa1⌬ Ͻ tsa1⌬tsa2⌬ (16). This trend can be considered similar to the one observed here, if the sensitivities of the method-ologies available to measure mutation rates and radical production are taken into account.
Generation of the 1-hydroxyethyl radical in tsa1⌬tsa2⌬ cells challenged with hydrogen peroxide was attributed mainly to the peroxidase activity of the enzyme Sod1, although the participation of Fenton chemistry to a small extent (ϳ30%) is apparent (Fig. 3). 1-Hydroxyethyl radical might be formed through hydrogen abstraction by peroxyl radicals, but no quantitative evidence is this direction was obtained (Table 1). In contrast, the major role of Sod1 was established by several lines of evidence (Figs. 2-7 and Table 1). Among them, the similar properties of WT overexpressing human Sod1 and tsa1⌬tsa2⌬ cells in regard to the superoxide dismutase activity, bicarbonatedependent peroxidase activity, and 1-hydroxyethyl radical production were particularly convincing (Fig. 7).
The results demonstrated that increased 1-hydroxyethyl radical production by tsa1⌬tsa2⌬ cells increased damage to their DNA with formation of DNA-derived radicals and adducts. Both of these lesions can be promoted directly by the 1-hydroxyethyl radical (24,50,51). In fact, the inhibition of [ 14 C]DNA production by the spin-trap POBN argues for a direct participation of the 1-hydroxyethyl radical in forming DNA adducts. At this point, however, we cannot exclude the formation of DNA adducts by 1-hydroxyethyl radical metabolites, such as acetaldehyde (55)(56)(57). Likewise, the formation of DNA-derived radicals may result from secondary radicals arising from processes triggered by the 1-hydroxyethyl radical, such as lipid peroxidation (42,58). It is noteworthy that production of the 1-hydroxyethyl radical close to DNA is feasible, because Sod1 is mainly a cytosolic enzyme but is also found in the nucleus, peroxisomes, and mitochondrial intermembrane space of eukaryotic cells (59). Here, we measured overall expression and activity of Sod1, which was augmented in tsa1⌬tsa2⌬ cells even in the absence of a hydrogen peroxide challenge (Fig. 6). Thus, the molecular events proposed here may also occur in unstressed cells. Relevantly, a higher yield of DNA adducts was obtained in unstressed tsa1⌬tsa2⌬ compared with unstressed WT cells, although the difference was not statistically significant (Fig. 8A). The peroxidase activity of Sod1 is increased in the presence of carbon dioxide to produce the carbonate radical, which is also able to produce DNA radicals (60,61). Consequently, the peroxidase activity of Sod1 may also be involved in promoting DNA mutations in peroxiredoxin-null yeast cells grown aerobically (18).
The signaling pathway responsible for the elevated levels of Sod1 in tsa1⌬tsa2⌬ cells is not known but probably involves Yap1 and Skn7 regulons, because the SOD1, TSA1, and TSA2 genes are under their control (62). On the other hand, deletion of both TSA1 and TSA2 genes is likely to result in higher steadystate levels of hydrogen peroxide, which, in turn, activate both Yap1 and Skn7 leading to increased Sod1 expression. Another possibility is that induction of SOD1 could be a response of yeast to increased levels of copper ions (Table 1), probably mediated by Ace1 transcriptional factor (63). Other transcriptional regulators such as Hap1 might also be involved in the induction of Sod1 in tsa1⌬tsa2⌬ cells.
Sod1 is considered one of the most important antioxidant defenses because it efficiently catalyzes the dismutation of superoxide anion into hydrogen peroxide plus molecular oxygen (59). For instance, the sod1⌬ yeast mutant presents aerobic spontaneous mutation rates that are 3-to 4-fold higher than for the WT strain (63,64). On the other hand, higher expression levels of Sod1 in mammalian cells have been frequently associated with deleterious effects (65), including increased DNA damage in cells defective for DNA repair (66 -69). Although many investigators attribute these deleterious effects to Sod1 increasing cellular steady-state concentrations of hydrogen peroxide, they are most likely due to the enzyme competing with other metalloproteins for metal cofactors and to the extraneous activities of Sod1 (65). Indeed, Sod1 displays catalytic activities other than superoxide dismutation, such as superoxide oxidase, peroxynitrite synthase, thiol oxidase, and peroxidase activity (35,36,44,45). Among these, our studies of WT and tsa1⌬tsa2⌬ yeast cells challenged with hydrogen peroxide emphasize the importance of the peroxidase activity, because the mutant cells presented increased levels of DNA damage (Fig. 8) in parallel with increased levels of 1-hydroxyethyl radical formation (Fig. 2), Sod1 expression (Fig. 6), and peroxidase activity (Fig. 7A). Other extraneous Sod1 roles, such as activation and control of specific DNA repair mechanisms, were not examined here but deserve future consideration (69).
In summary, our studies examined the relative resistance of tsa1⌬tsa2⌬ cells to hydrogen peroxide and provided a sequence of molecular events whereby the compensatory response of S. cerevisiae to deal with TSA1 and TSA2 gene disruption leads to hypermutability of the strain. In parallel, they present a framework to examine some of the deleterious effects that have been frequently associated with the overexpression of Sod1 in cells and animal models.