Manganese Is the Link between Frataxin and Iron-Sulfur Deficiency in the Yeast Model of Friedreich Ataxia*

Friedreich ataxia is a human neurodegenerative and myocardial disease caused by decreased expression of the mitochondrial protein frataxin. Proteomic analysis of the mutant yeast model of Friedreich ataxia presented in this paper reveals that these cells display increased amounts of proteins involved in antioxidant defenses, including manganese-superoxide dismutase. This enzyme shows, however, lower activity than that found in wild type cells. Our results indicate that this lack of activity is a consequence of cellular manganese deficiency, because in manganese-supplemented cultures, cell manganese content, and manganese-superoxide dismutase activity were restored. One of the hallmarks of Friedreich ataxia is the decreased activity of iron/sulfur-containing enzymes. The activities of four enzymes of this group (aconitase, glutamate synthase, succinate dehydrogenase, and isopropylmalate dehydratase) have been analyzed for the effects of manganese supplementation. Enzyme activities were recovered by manganese treatment, except for aconitase, for which, a specific interaction with frataxin has been demonstrated previously. Similar results were obtained when cells were grown in iron-limited media suggesting that manganese-superoxide dismutase deficiency is a consequence of iron overload. In conclusion, these data indicate that generalized deficiency of iron-sulfur protein activity could be a consequence of manganese-superoxide dismutase deficiency, and consequently, it opens new strategies for Friedreich ataxia treatment.

Friedreich ataxia (FRDA) 3 is a human disease caused by decreased expression of the mitochondrial protein frataxin due to the presence of large expansions of a GAA triplet repeats in the first intron of the nuclear gene coding for the protein (1). FRDA patients suffer from multiple symptoms, including progressive gait and limb ataxia, hypertrophic cardiomyophathy, and diabetes mellitus (2). These alterations are thought to be the consequence of a mitochondrial defect related to iron metabolism. However, although a large number of studies link frataxin to iron metabolism, its precise function remains a matter of debate. Frataxin has been suggested to play a role in iron-sulfur cluster biosynthesis (3,4), heme biosynthesis (5,6), iron storage (7,8), electron transfer to ubiquinone (9), or holo-aconitase reconstitution (10,11). In hearts from FRDA patients, decreased activities of iron-sulfur-contain-ing proteins and increased mitochondrial iron accumulation have been reported (12). Furthermore, conditional knock-out mice in neuronal and cardiac tissues have been developed as an animal model for FRDA. Accordingly, this model displays iron accumulation and decreased activity of iron-sulfur proteins (13). Consequently, some therapeutic approaches under study suggest reducing the degree of iron accumulation by chelators or prevent its prooxidant effects by the use of antioxidants such as idebenone, a coenzyme Q analogue (2). Many studies on frataxin function arise from experiments performed with the budding yeast Saccharomyces cerevisiae (3,6,8,9) as frataxin and the yeast homologue Yfh1 are orthologues. Frataxin and Yfh1 are mitochondrial proteins; the S. cerevisiae mutant in the YFH1 gene (⌬yfh1) accumulates iron and shows decreased activities of iron-sulfur-containing enzymes (14); and finally, human frataxin complements ⌬yfh1 mutant strains (15).
Despite the great amount of data obtained with these different model systems, the analysis of the proteome of frataxin-deficient cells has not been addressed. Thus, we have focused on the effects of the absence of frataxin on the cellular proteome of yeast. One of the major findings is that, although the amount of the antioxidant enzyme manganese-superoxide dismutase (Mn-SOD) is increased, its activity is decreased in ⌬yfh1 cells. This deficiency contributes to lowered activities of several iron-sulfur-containing enzymes, one of the hallmarks of Friedreich ataxia.

EXPERIMENTAL PROCEDURES
Organisms and Culture Conditions-S. cerevisiae W303 and its isogenic null ⌬yfh1 derivative (MML298) were kindly provided by Dr. Enrique Herrero from the Universitat de Lleida (16). Strains BY4741 and Y02769 (⌬grx5 in BY4741 backgound) were provided by EURO-SCARF (Frankfurt, Germany). The YFH1 gene used in complementation studies was also obtained from the EUROSCARF collection. It carries upstream and downstream regulatory sequences and is cloned in pRS416, an ARS-CEN-URA vector. Yeast cells were grown in rich medium (1% yeast extract, 2% peptone) with either 2% glucose (YPD) or 3% glycerol (YPG) by incubation in a rotary shaker at 30°C. Raffinose synthetic medium contained 2% raffinose, 0.67% yeast nitrogen base (Difco), 0.25% ammonium sulfate, a mixture of amino acids, and the required auxotrophic supplements (17). The ⌬yfh1 mutant transformed with pRS416-YFH1 plasmid was maintained in synthetic medium without uracil. Before each experiment the strain was transferred to rich medium for 4 -5 generations.
Two-dimensional Gel Electrophoresis-Cells were resuspended in 25 mM Tris-HCl buffer, pH 8, plus 8 M urea and disrupted using glass beads. An equal volume of 8 M urea, 8% CHAPS, and 50 mM dithiothreitol was added to the lysed cells, and after centrifugation (12,000 rpm for 5 min) 10 l of the supernatant were diluted in 350 l of rehydration buffer (18). Isoelectric focusing was performed in immobilized pH gradient strips (3-10 NL; Bio-Rad). Second dimension SDS-PAGE was per-formed on 12.5% polyacrylamide gels. Gels were scanned in a GS800 densitometer (Bio-Rad) and analyzed with PDQuest software (Bio-Rad).
Analyses-Total cellular iron was determined under reducing conditions with bathophenanthroline sulfonate as chelator (19). Manganese content was assayed in a graphite furnace atomic absorption spectrophotometer. Superoxide ion levels were measured in a spectrofluorophotometer (RF-5000, Shimadzu). Exponentially growing cells were washed in water and resuspended in phosphate-buffered saline plus 0.1% glycerol and 5 g/ml the superoxide specific probe dihydroethidium (DHE) (Fluka) (20). The rate of oxidation of DHE was calculated from the rate of increase in fluorescence for 30 min (excitation 520 nm, emission 590 nm).
Western Blot Analysis-Cell extracts were separated in SDS-polyacrylamide gels and transferred to polyvinylidene difluoride membranes. Antibodies against superoxide dismutases were from Stress-Gene (SOD-111, Mn-SOD) and from Chemicon (AB1237, CuZn-SOD). Anti-Aco1 aconitase (from R. Lill, Marburg, Germany) and anti-succinate dehydrogenase (from B. Lemire, University of Alberta) antibodies were also used. A peroxidase-conjugated anti-rabbit antibody was used for detection. Image acquisition was performed in a ChemiDoc CCD camera (Bio-Rad).

Analysis of the Proteome of ⌬yfh1 Cells by Two-dimensional
Electrophoresis-Tissues affected in FRDA are those with elevated oxidative metabolism, like heart and nervous system (2). Thus, unless otherwise stated, our experiments were carried out on yeast cells grown in rich medium containing glycerol (YPG), because in this medium cells exhibit high respiration rates. It is worth mentioning that the ability of this strain to grow in a non-fermentable carbon source such as glycerol has been described previously (24 -26). Wild type and ⌬yfh1 cells were harvested during exponential phase and the crude extracts analyzed by two-dimensional gel electrophoresis (Fig. 1). Spots that repetitively showed at least 2-fold difference in protein amount between both strains were identified and are shown in Table 1. Similar results were found in cells grown on rich medium containing glucose (YPD) (Fig. 2). Cells deficient in YFH1 clearly exhibit increased amounts of enzymes involved in the oxidative stress response. Among these proteins, four thiol peroxidase isoenzymes have been identified. These enzymes eliminate H 2 O 2 and alkyl hydroperoxides with the use of thioredoxin and thioredoxin reductase as thiol-reducing donor (27), two proteins that are also induced in ⌬yfh1 cells. Two additional enzymes involved in the oxidative stress response that have been identified are both mitochondrial and cytosolic SODs. These enzymes catalyze the dismutation of superoxide into O 2 and H 2 O 2 (28). The increased expression of these eight proteins indicates an endogenous situation of oxidative stress in ⌬yfh1 cells. It is interesting to mention that no changes in the expression

TABLE 1 Proteins identified differentially expressed in wild type and ⌬yfh1 cells
Spots showing a minimum of 2-fold difference in protein abundance between wild type and ⌬yfh1 proteomes (see Figs. 1 and 2) were digested with trypsin and analyzed in an Applied Biosystems Voyager DE PRO MALDI-TOF. Proteins were identified by peptide mass fingerprinting with MASCOT. Percent protein coverage for each spot in the MASCOT analysis is shown. Fold induction (⌬yfh1 over wild type, mean of three independent experiments Ϯ S.E.) was estimated considering all the spots identified as the same protein. The values with the Ͼ symbol are estimated and indicate that a corresponding spot in the wild-type sample is too low to be used as a reference value.

Protein
Gene Analysis of the proteome of wild type and ⌬yfh1 cells by two-dimensional gel electrophoresis. W303 (wild type) and the isogenic derivative ⌬yfh1 (MML298) cells were grown in YPG medium, and 40 g of total cell lysates were separated by two-dimensional gel electrophoresis. Gels were silver-stained and analyzed with PDQuest software (Bio-Rad).
of antioxidant enzymes were observed by gene expression analysis of ⌬yfh1 mutant cells using DNA microarrays (25), even though the strain used had the same genetic background in both studies. A partial explanation to this fact is that, in the reference cited, many of the induced genes code for membrane proteins that are poorly resolved in twodimensional gels. Furthermore, lack of concordance between proteomic and transcriptomic analysis is not unusual (29), thus highlighting the importance of both approaches.

Manganese and Mn-SOD Deficiency in ⌬yfh1
Cells-Among the proteins induced in ⌬yfh1 cells, both SODs were found. This result was further confirmed by Western blotting using specific antibodies (Fig.  3A). Interestingly, when the enzymatic activity of Mn-SOD (coded by SOD2) was analyzed, we found that in ⌬yfh1 cells it was lower than in wild type cells (Fig. 3, A and B). The decreased Mn-SOD activity resulted in increased steady state levels of superoxide anion inside yeast cells. As shown in Fig. 3C, the rate of oxidation of DHE, a superoxide-specific probe, is increased by 2-fold in YFH1-deficient cells.
Trying to unravel the cause of Mn-SOD deficiency in ⌬yfh1 cells, we first analyzed the manganese content of wild type and ⌬yfh1 cells. As shown in Fig. 4A, manganese concentration in the latter was 20% of that found in wild type cells, suggesting that manganese transport could be altered in that strain. To further confirm that this altered phenotype is a consequence of YFH1 deficiency, we transformed the ⌬yfh1 strain with a plasmid containing the YFH1 gene with its own promoter (see "Experimental Procedures"), and transformants (⌬yfh1-YFH1) were analyzed for the altered parameters. As shown in Fig. 3A, Mn-SOD activity, as well as Sod1 and Sod2 protein levels, were restored. Also, ⌬yfh-YFH1 cells displayed manganese and iron concentrations close to those found in wild type cells (Fig. 4A).
Recovery of Mn-SOD Activity by Manganese Treatment-The results described above suggested that intracellular manganese and Mn-SOD activity in ⌬yfh1 could be reestablished by supplementing growth media with this metal. To this purpose, ⌬yfh1 cell cultures were treated with increasing concentrations of MnCl 2 . As shown in Fig. 4B, manganese levels were restored. Consistently, recovery of Mn-SOD activity in a dose-dependent manner was also observed (Fig. 3A). Thus, lack of Mn-SOD activity in ⌬yfh1 cells is a consequence of cellular manganese deficiency. In accordance with these results, a clear decay in superoxide anion levels was observed in ⌬yfh1 cells grown in manganese-supplemented media, and the values were indistinguishable from those of the  wild type cells (Fig. 3C). Nevertheless, as shown in Fig. 4C, manganese treatment had no significant effect on total cellular iron concentration, both in wild type or ⌬yfh1 cells.
Recovery of Iron-Sulfur Enzymes by Manganese Treatment-Frataxin has been suggested to play an important role in iron-sulfur cluster biosynthesis. This hypothesis is based on the decreased activities of ironsulfur enzymes found in frataxin-deficient cells (12) and a physical and genetic interaction with Isu1, a scaffold protein required for iron-sulfur clusters biogenesis (3,4). However, decreased activities of iron-sulfur enzymes have been observed in yeast lacking SOD2 (30) and in Mn-SOD knock-out mice (31). In this context, our results suggest that a decline in iron-sulfur enzyme activities in ⌬yfh1 cells could be a consequence of oxidative stress in mitochondria due to decreased Mn-SOD activity. To test this hypothesis, we analyzed activities of three ironsulfur mitochondrial enzymes (aconitase, glutamate synthase, and succinate dehydrogenase) and a cytosolic one (isopropylmalate dehydratase) in wild type and ⌬yfh1 cells grown in manganese-supplemented YPG (Fig. 5). Values of iron-sulfur enzyme activities in ⌬yfh1 cells are in agreement to those described by other authors (9,17,26,32) With the exception of aconitase, all other enzyme activities analyzed showed a clear recovery by manganese treatment, indicating that lack of Mn-SOD activity contributes to inactivating these enzymes. Also, these results suggest that YFH1 is not essential for iron-sulfur biosynthesis, except for aconitase. Such an exception can be explained by recent observations reporting that frataxin interacts with aconitase and that such interaction is crucial for reconstitution of the aconitase [4Fe4S] cluster (10,11). (17), it was found that succinate dehydrogenase, but not aconitase activity, could be rescued when ⌬yfh1 yeast cells were grown in iron-limited medium using bathophenanthroline sulfonate (BPS) as iron chelator. In such conditions, cellular iron accumulation is prevented. To test whether such treatment is also capable of restoring iron-sulfur enzymes other than succinate dehydrogenase, wild type and ⌬yfh1 cells were grown in raffinose synthetic medium supplemented with BPS (Fig. 6). The results obtained not only confirmed the findings reported by Foury but also showed that isoprpopylmalate dehydratase and Mn-SOD activities were recovered in such conditions. These are additional indications that decreased Mn-SOD activity leads to inactivation of iron-sulfur containing enzymes in ⌬yfh1 cells.

Recovery of Iron-Sulfur Enzymes and Mn-SOD in Iron-limiting Media-In a previous work published by Foury's group
Manganese and Mn-SOD Deficiency Are the Consequence of Iron Overload-As shown above, ⌬yfh1 cells displayed low levels of manganese that caused decreased activities of several iron-sulfur containing enzymes. This phenotype was corrected by manganese supplementation. Nevertheless, iron accumulation, induction of Sod1 and Sod2 (Fig.  3A), and slow growth in YPG (data not shown) persisted in manganesetreated ⌬yfh1 cells, suggesting that they are not strictly a consequence of manganese deficiency. Then, could manganese deficiency be a consequence of iron accumulation caused by the absence of Yfh1? The experiments in iron-limited media support this hypothesis, because Mn-SOD activity is restored when iron accumulation is prevented in ⌬yfh1 cells (Fig. 6). To study whether this phenotype is a more general phenomenon and not restricted to the mutant strain used, we further checked this hypothesis by analyzing manganese content and SOD activity in a strain with altered iron homeostasis such as a yeast mutant lacking  GRX5. Mutants deficient in this glutaredoxin exhibit a strong iron accumulation, as well as decreased activities in aconitase and succinate dehydrogenase (16). As shown in Fig. 7A, manganese levels in this strain were 60% of those found in wild type cells, in agreement with the data obtained in ⌬yfh1 mutant strain. Also, ⌬grx5 cells exhibit an 80% decrease in Mn-SOD activity. In this case, this activity can also be restored by manganese supplementation (Fig. 7B). Taken together, the data obtained from both mutant strains clearly suggest that manganese deficiency would be a side effect of iron accumulation.

DISCUSSION
The precise role of frataxin in iron metabolism is a hot field of discussion. The most accepted hypothesis is that it cooperates in iron delivery to Isu1, a scaffold protein required for iron-sulfur biogenesis. However, it has been suggested that the role of frataxin would be non-essential in such process (26). Moreover, recent data support this hypothesis: (i) frataxin interacts much more strongly with ferrochelatase than with Isu1 (5); (ii) in Escherichia coli, IscA (homologous to Isa1 and Isa2 in yeast) is the iron donor to IscU (homologous to Isu1 in yeast), and frataxin homologues are not required for the biogenesis of iron-sulfur clusters (33); and (iii) a direct interaction between frataxin and aconitase has been reported (10,11). The results shown in this paper clearly indicate that frataxin is not essential for iron-sulfur cluster biogenesis (although required for full recovery of aconitase activity) and give a reasonable explanation of how the absence of frataxin leads to ironsulfur deficiency in the yeast model of FRDA. Iron overload would lead to reduced manganese levels that would result in impaired manganese incorporation into Sod2. Lack of Mn-SOD activity would result in increased levels of superoxide (Fig. 3C) that would inactivate ironsulfur-containing enzymes. The link between iron accumulation and Mn-SOD deficiency is clearly established with the experiments performed with BPS and ⌬grx5 cells. Also, a direct link between Mn-SOD activity and inactivation of iron-sulfur enzymes arises from experiments reported in Figs. 5 and 6. Results shown in Fig. 5 demonstrate the recovery of Mn-SOD and iron-sulfur containing enzymes by manganese treatment. In this context, it is worth mentioning that manganese itself has antioxidant properties that suppress growth defects in a ⌬sod1⌬sod2 mutant (34). However, this suppression required concentrations above 2 mM and was not observed at concentrations used in our work (50 -100 M). Consistently, when we treated a ⌬sod2 mutant with 100 M manganese, recovery of iron-sulfur enzymes was not observed (data not shown). Furthermore, as shown in Fig. 6, enzyme activities were recovered by reducing iron concentration (without manganese supplementation). Consequently, it is not likely that recovery of ironsulfur enzymes could be a consequence of the antioxidant properties of manganese independently of Mn-SOD recovery.
An important problem to solve in the future is how iron overload results in manganese deficiency. As a hypothesis, it should be mentioned that Smf2p, a high affinity manganese transporter, could be the link. Cells lacking SMF2, as those lacking YFH1, show decreased manganese content and limited Mn-SOD activity. These defects are corrected after manganese supplementation of growth media (23). Interestingly, Smf2 activity is decreased at the post-translational level by elevated iron levels and its expression is not regulated by the iron sensor Aft1 (35). Thus, in contrast to Aft1, Smf2 could be able to sense increased iron levels in ⌬yfh1 cells and lead to decreased high affinity manganese transport. It is worth mentioning that manganese deficiency linked to increased iron levels could be a general situation not only applying to FRDA. The results obtained with ⌬grx5 mutant cells indicate that these could be a more general phenomena occurring in iron overloading diseases. An important consequence is that oxidative stress triggered by an excess of iron would be potentiated by diminished SOD activities.
As mentioned earlier, our work supports that Yfh1 should be dispensable for iron-sulfur cluster biogenesis. However, how lack of Yfh1 induces iron accumulation in yeast is still an intriguing question. Iron accumulation in yfh1 mutants depends on constitutive activation of the iron sensor Aft1 (25), and it has been reported that this sensor responds to a signal connected to mitochondrial iron-sulfur cluster biogenesis (36). As shown in Fig. 5, manganese treatment restores the activity of some iron-sulfur enzymes, but iron overload persists. This indicates that the signal acting on Aft1 is not restored in these particular circumstances. Could this signal be related to aconitase? It should be kept in mind that in manganese-treated ⌬yfh1 cells, its activity is not restored, and according to two recent reports (10,11), reconstitution of the aconitase [4Fe4S] cluster depends on its interaction with frataxin. Also, in mammals, cytosolic aconitase (Irp-1) has a well known role as an iron sensor (37). In yeast, apo-aconitase binds to mitochondrial DNA where it could play a regulatory function in response to changing metabolic needs (38).
Another important conclusion from this work arises from the analy-  sis of the proteome of ⌬yfh1 cells. Induction of antioxidant defenses reveals that YFH-deficient cells are exposed to an endogenous oxidative stress. This would be the consequence of Mn-SOD deficiency but more importantly of iron overload, since the increased amounts of Sod1p and Sod2p persisted in manganese-treated ⌬yfh1 cells. The relevance of oxidative stress in FRDA is also a matter of discussion. Markers of oxidative stress in samples of FRDA patients have been extensively reported; for instance, in lymphoblasts, mitochondrial peroxides were increased and the GSH/GSSG ratio was reduced (39); in fibroblasts, the GSH/GSSG ratio decreased by 3-fold and protein-bound glutathione and actin glutathionylation were increased (40). On the other hand, the existence of oxidative stress in FRDA has been questioned by some reports that were unable to identify markers of oxidative stress, like protein-bound carbonyl groups (41) or malondialdehyde (42), in different FRDA models. But it should be kept in mind that specific targets depend on how oxidative stress is generated (43,44). This means that the absence of a specific marker of oxidative stress in one system does not preclude the presence of oxidative stress in such system. Finally, the results shown here open the question on whether manganese and deficient SOD activity are both related to FRDA. Alterations in SOD activity and expression have been reported in other FRDA models. Embryonic mouse cells treated with retinoic acid and frataxin antisense RNA show increased expression of Mn-SOD compared with control cells. Retinoic acid induces differentiation of embryonic cells into neurons, astroglia, and microglia (45). According to a recent report (46), fibroblasts and lymphoblasts from FRDA patients could have decreased antioxidant capacity. Decreased total SOD activity is observed in hearts from the mouse model of FRDA ataxia (47). In this model, Mn-SOD expression increases after birth and suddenly drops after 10 weeks (41). Could this drop in Mn-SOD expression be the consequence of manganese deficiency? Interestingly, Smf2p is homologous to DMT1, which transports iron as well as manganese in mammals (48). The existence of a manganese deficiency in FRDA patients could lead to a sudden change in FRDA therapies. Manganese is known to be essential for the development and function of the brain. Manganese-deprived rats are often ataxic (49). The reason is not known, but it is remarkable that several brain manganese-containing enzymes, such as glutamine synthetase, pyruvate carboxylase, or Mn-SOD, play a significant role in brain function and could be affected by manganese deficiency. Should manganese supplementation be considered in FRDA patients? Even if such deficiency is confirmed in mammals, this is a delicate point, since high environmental exposure to manganese has toxic effects. However, oral manganese intake is considered to be safe even at concentrations 5-fold above the recommended daily doses (50). Consequently, the results presented here open new strategies in FRDA therapies.