Yeast lacking Cu-Zn superoxide dismutase show altered iron homeostasis. Role of oxidative stress in iron metabolism.

Saccharomyces cerevisiae lacking copper-zinc superoxide dismutase (sod1) shows a series of defects, including reduced rates of aerobic growth in synthetic glucose medium and reduced ability to grow by respiration in glycerol-rich medium. In this work, we observed that addition of iron improved the respiratory growth of the sod1 mutant and in glucose medium total intracellular iron content was higher in the sod1 mutant than in wild type cells. Transcription of the high affinity iron transporter gene, FET3, was enhanced in the sod1 mutant, suggesting that iron transport systems were up-regulated. An sod1/fet3 double mutant showed increased sensitivity to oxygen and increased transcription of FET4, an alternative, low affinity, iron transporter. We propose that this increased iron demand in the sod1 mutant may be a reflection of the cells' efforts to reconstitute iron-sulfur cluster-containing enzymes that are continuously inactivated in conditions of excess superoxide.

Superoxide dismutases (SODs) 1 play a protective role against the toxicity of reactive oxygen derivatives by catalyzing the dismutation of superoxide radicals (O 2 . ) to hydrogen peroxide (H 2 O 2 ) and dioxygen (1)(2)(3). Saccharomyces cerevisiae, like other eukaryotes, contains two intracellular SODs, one containing manganese and the other copper and zinc in their respective active sites. The MnSOD (product of the SOD2 gene) is localized in the mitochondrial matrix and CuZnSOD (product of the SOD1 gene) is distributed in the cytoplasm, nucleus, and other compartments. Neither SOD1 or SOD2 is strictly essential; however, the loss of SOD1 has dramatic phenotypic consequences in yeast (for review, see Ref. 4). Yeast strains lacking CuZnSOD (sod1) show several defects during aerobic growth, including reduced rates of growth, auxotrophies for lysine and methionine or cysteine, poor growth by respiration (in glycerol or ethanol), higher rates of spontaneous mutation, and more rapid loss of viability in stationary phase (4 -7). Knowledge about the exact biochemical causes of these defects is limited, but most genetic suppressors of the sod1 deletion mutation are involved in metal metabolism (8 -11). Two suppressor mutations, pmr1 and bsd2, were isolated by their ability to rescue the lysine and methionine auxotrophies of the sod1/sod2 double mutant. The PMR1 gene encodes a P-type ATPase involved in transport of calcium and manganese into the Golgi apparatus (8); pmr1 mutants accumulate manganese in the cytoplasm. BSD2 is a gene whose deletion causes an increase in the intracellular copper levels (9). The metabolic deficiencies of the sod1/sod2 mutant are also rescued by overexpression of the ATX1 gene, which encodes a small cytoplasmic copper carrier (11), and by overexpression of the ATX2 gene, which results in increased intracellular manganese (10). Both manganese and copper ions are scavengers of superoxide in vitro (12,13), and it is believed that a similar activity in vivo may be the mechanism by which all these mutations suppress the defects of sod1/sod2 yeast.
Recently, three mutations that suppress defects in the sod1 single mutant have been identified (14). Two of them (ssq1 and jac1) are mutations in genes for mitochondrial chaperone proteins (hsp70 and hsp20, respectively). The third suppressor, nfs1, is thought to function in the assembly of iron-sulfur clusters. All three show impaired respiration, decreased oxygen consumption, and reduced aconitase and succinate dehydrogenase activities.
Superoxide radical has been shown to inactivate certain [4Fe-4S] cluster-containing enzymes by oxidizing one iron, causing its release from the cluster (1,15). This process leads to both enzyme inactivation and further oxidative damage of other cellular components, as "free" iron can promote, via the Fenton reaction, the formation of hydroxyl radicals ( ⅐ OH) (16,17). Thus, superoxide dismutase both protects the [4Fe-4S] enzymes from inactivation and prevents the accumulation of excess intracellular iron.
It is not clear whether the inactive 3Fe-4S form of the cluster can be reactivated in vivo by reinsertion of the single iron ion, or whether the whole cluster must be removed and replaced. In Escherichia coli, there is evidence suggesting that under certain conditions the single iron can be replaced. A mutation that increases the intracellular concentration of an Fe 2ϩ chelator is an effective suppressor of multiple defects of sod Ϫ mutants and may function by providing an abundant supply of Fe 2ϩ that can be used to convert [3Fe-4S] to [4Fe-4S] clusters (18).
Because iron can play an important role in oxidative stress and yet is essential for cellular growth, its uptake is tightly regulated in both yeast and mammalian systems (19,20). In S. cerevisiae, two plasma membrane proteins, Fre1p and Fre2p, reduce Fe 3ϩ to Fe 2ϩ , a function that is critical for iron uptake (20,21). In iron-replete conditions, Fe 2ϩ is then transported by Fet4p, the low affinity iron transporter (22). Under low iron conditions, a high affinity iron transport system is used that proceeds via the coupled functions of the cell surface Fet3p multicopper oxidase, which catalyzes the conversion of Fe 2ϩ to Fe 3ϩ , and the Ftr1p permease, which transports the iron (23)(24)(25)(26). The pathways of iron transport and delivery inside the cell are as yet unknown.
In this work we explore iron metabolism of sod1 mutant yeast. We observed that addition of iron to the media improved the respiratory growth of sod1 mutants and that the mutant yeast under normal growth conditions accumulate excess iron and show increase expression of iron transport genes. We propose that these alterations in the iron status of the sod1 mutant may be a reflection of the cells' efforts continually to reconstitute [4Fe-4S] cluster-containing enzymes that have been inactivated by excess superoxide.

EXPERIMENTAL PROCEDURES
Yeast Strains-The strains used in this study are shown in Table I. All are derived from EG103, except DTY3 and EG151 (5,(27)(28)(29). For the fet3 disruptions, plasmid YIpfet3-2::HIS3 (23) was used to transform EG103 and EG118 yeast strains. Southern blot analysis was performed to verify correct insertion of the HIS3 fragment in the FET3 gene (data not shown). A multicopy FET3::lacZ fusion plasmid pAR1 (URA3 ϩ , Amp R ) was kindly provided by Dr. J. Kaplan was used to transform DTY3 (Y144) and EG151. Yeast strain DTY3 was provided by Dr. D.

Thiele.
Media and Growth Conditions-The cells were grown either in SDC with 2% dextrose (glucose) as described (30), except with a 4-fold increase in the supplements Leu, His, Trp, Met, Ura, and Ade, YPD (1% yeast extract, 2% peptone, and 2% dextrose), or in YPG (1% yeast extract, 2% peptone, and 3% glycerol) (30,31). For routine pregrowth, strains lacking CuZnSOD activity were cultured in microaerophilic conditions (5% oxygen) using CampyPaks (BBL) for plates or, for liquid cultures, 5 ml of medium in a 16 ϫ 100-mm culture tube. For flask cultures, low aeration was achieved by decreasing the surface to volume ratio and/or reducing the shaking rate to 100 rpm. High aeration for experimental samples was achieved by using a flask volume/medium volume ratio of 5:1 (typically 50 ml of medium in a 250-ml flask) shaking at 200 rpm. For anaerobic conditions, cells were inoculated into 250-ml flasks containing 50 ml of degassed SDC medium and grown with shaking at 200 rpm under a slow stream of nitrogen. Samples grown in YPG were inoculated at 1-2 ϫ 10 6 cells/ml and cultured for 18 h. Growth was followed by monitoring the turbidity at 600 nm (A 600 ); an A 600 of 1 corresponds to 1 ϫ 10 7 cells/ml. All growth was at 30°C.
For the experiments shown in Fig. 2, cells were grown in rich YPD liquid medium under low aeration for 18 h, diluted to 10 7 , 10 6 , and 10 5 cells/ml, and 5-l aliquots from each dilution were spotted onto YPD plates. Plates were incubated at 30°C under aerobic (20% oxygen) or microaerobic (5% oxygen) conditions, and growth was observed after 3 days.
Oxygen Consumption-Cellular oxygen uptake was measured at 30°C in a 3-ml stirred chamber using a YSI model 53 biological oxygen monitor (Yellow Springs Instruments) following the manufacturer's directions. Cells were cultured in YPG with or without 300 M iron added in flasks under high aeration as described above. (Each overnight culture was split into with and without iron flasks.) For the oxygen uptake measurement, cells were resuspended at 4 -5 ϫ 10 7 cells/ml in the same (spent) culture medium, and conditions that resembled the flask envi-ronment (30°C and stirring) were maintained in the chamber.
Metal Analysis by ICP-Mass Spectrometry-Yeast cultures (25 ml in a 125-ml flask or 50 ml in a 250-ml flask) were grown in the indicated liquid medium for 18 h, collected by centrifugation, and washed twice with 10 mM EDTA and once with metal free water. Cell pellets (1.5 ϫ 10 8 cells) were digested in 1 ml of 3% nitric acid (HNO 3 ) at 98°C overnight. After digestion, the samples were diluted to 5 ml with metalfree water and analyzed using a Perkin Elmer ICP-MS (inductively coupled plasma-mass spectrometer) to determine the metal content (32). All samples were measured 10 times, and the experiments were repeated at least twice. The data are expressed as picomoles of metal/ 10 7 cells. Alternatively, measurements were made using a Thermo Jarrell Ash Iris 1000 ICP-AE (inductively coupled plasma-atomic emission) instrument. S1 Analysis of RNA-RNA was measured using S1 nuclease analysis as described (33), using oligonucleotide probes for FET3 or FET4, with calmodulin (CMD) as an internal control (32): FET3, 5Ј-CCGCTTGTGC TAGCGAGAGC ATCGAGAAAA GCAAAACGGC TATAGAGAGC AAA-GCGTCGA CTG-3Ј; FET4, 5Ј-CCGAATTCTT CGTACTGTTT GCAGT-CAACA GTAGGTGCTC TATGATGAAC GTCAGGCCTA GCACCCA-AGC C-3Ј; CMD, 5Ј-GGGCAAAGGC TTCTTTGAAT TCAGCAATTT GTTCTTCGGT GGAGCC-3Ј. Briefly, probes were labeled at the 5Ј end with [␥-32 P]ATP using T4 polynucleotide kinase. Yeast were grown in SDC for 18 h, and total RNA was isolated and hybridized overnight at 55°C with either the FET3 or FET4 probe and the CMD probe. The RNA was incubated with 50 units of S1 nuclease and then applied to a 8% non-denaturing polyacrylamide gel. The gel was dried onto 3MM Whatman paper and submitted to PhosphorImager quantitation.

Iron Improves the Respiratory Growth of the sod1 Mutant-
Providing glycerol as the sole carbon source forces yeast to grow by respiration. Addition of iron (0.3-1 mM) to rich glycerol media improved the ability of the sod1 mutant to grow by respiratory metabolism (Fig. 1), indicating a higher iron requirement for optimal growth in the mutant. In order to further explore the relationship between iron metabolism and the mechanisms by which cells compensate for the loss of the sod1 gene, we constructed an isogenic family of strains missing the high affinity iron transporter Fet3p as well as SOD1. As expected, the fet3 mutant showed a respiratory growth defect in low iron medium, which could be suppressed by iron supplementation ( Fig. 1 and data not shown). However, treatment of the sod1/fet3 double mutant with iron did not improve its respiratory growth. This finding may indicate that Fet4p-mediated (low affinity) iron uptake is not powerful enough to provide enough iron to overcome the respiratory defect in the sod1/fet3 strain. Alternatively, the sod1/fet3 strain may have an even higher iron requirement for respiratory growth than either the sod1 and fet3 single mutant.
In order to test a more direct indicator of respiration, we measured oxygen consumption in cultures grown in normal and iron-supplemented media (YPG). Under normal iron conditions, oxygen consumption in the sod1 and fet3 mutants was lower than that of wild type, but it reached wild type levels when 0.3 mM iron was included in the medium, suggesting that respiratory function was restored by treatment with iron (Table II). We also tested the ability of iron to rescue the characteristic aerobic lysine and methionine auxotrophies of sod1 mutants. While addition of iron to SDC medium without methionine did not allow sod1 mutant to grow, addition of iron to SDC medium without lysine did somewhat improve the growth of the mutant (data not shown). This result supports our belief that the lysine defect results from inactivation of the iron-sulfur cluster enzyme homoaconitase, 2 but the effect was not very dramatic and was not pursued further.
We examined the growth of our family of strains under aerobic (20% oxygen) and microaerobic conditions (5% oxygen) on YPD plates with normal iron levels (Fig. 2). As expected, low oxygen improved the growth of the sod1 mutant as well as the sod1/fet3 mutant. But, interestingly, the sod1/fet3 mutant was more sensitive to oxygen under either oxygen condition than the sod1 strain.
The sod1 Mutant Displays Altered Intracellular Iron Levels in Glucose Medium but Not under Respiratory Conditions-In order to determine whether the intracellular iron content was altered in the sod1 mutant, we analyzed the total iron content in our strains. Fig. 3 shows that intracellular iron contents are similar in sod1 and wild type cells when they were grown in rich glycerol media. As expected, the fet3 mutant strain showed decreased intracellular iron content, since only the low affinity iron transporter can contribute to iron uptake (32,37). Surprisingly, the sod1 mutant and the sod1/fet3 double mutant accumulated the same amount of iron per cell as wild type cells. This finding may reflect an increased iron requirement due to the addition of the sod1 mutation; in the double mutant, and to a lesser extent in the sod1 single mutant, growth may be slowed until sufficient iron accumulates, while the single fet3 mutant, being oxygen resistant, is perhaps able to grow with less total cellular iron. This idea is supported by the growth data shown in Fig. 1; it can be seen that the sod1 and sod1/fet3 strains grew more slowly.
When cells were grown in glucose minimal medium (SDC) to late log phase, slightly different results were obtained. In agreement with the idea that the sod1 mutant requires extra iron, Fig. 4A shows that sod1 mutant contains more intracellular iron than the wild type strain. The sod1/fet3 double mutant shows the same amount of intracellular iron as does the sod1 mutant, which supports the idea that the sod1 mutation increases the cellular demand for iron.
Most investigators accept that the major source of intracellular superoxide is mitochondrial respiration (2,4,38). Indeed, reduced superoxide production in respiration-incompetent strains, such as coq3 and atp2 mutants, partially rescues the air-dependent lysine and methionine auxotrophies of the sod1 mutant (7). However, we observed that an sod1/coq3 double mutant still showed increased intracellular iron content as compared with an SOD1 ϩ /coq3 strain (Fig. 4B). This indicated to us that the effect was not caused simply by the cells respiring more and needing more iron for synthesis of respiratory enzymes and cytochromes, but that some other iron-requiring system(s) were involved.
It should be noted that the presence of the coq3 mutation in the sod1 strain did not fully restore a wild type phenotype, even  1. Iron improves respiratory growth of sod1 mutants. WT yeast (EG103) and mutant derivatives sod1 (EG118), fet3 (JF242), and sod1/fet3 (JF244) were grown in rich YPG medium containing the non-fermentable carbon source glycerol, with or without added ferric chloride (white bars, no added iron; light gray bars, 0.3 mM iron; dark gray bars, 1 mM iron). After an 18-h incubation with high aeration, aliquots were collected and the culture density was determined. The averages of two independent experiments are shown; error bars represent standard deviation. FIG. 2. Oxygen-dependent inhibition of growth of sod1 mutants under low iron conditions. Cultures of WT yeast (EG103) and mutant derivatives sod1 (EG118), fet3 (JF242), and sod1/fet3 (JF244) were grown in rich YPD liquid medium under low aeration for 18 h; dilutions were made; 5-l aliquots containing 5 ϫ 10 4 , 5 ϫ 10 3 , and 5 ϫ 10 2 cells were plated onto YPD plates; and incubation was carried out under aerobic (20% oxygen) or microaerobic (5% oxygen) conditions for 3 days.
FIG. 3. Total intracellular iron content in sod1, fet3, and sod1/ fet3 mutants under respiratory conditions. Cells were grown in YPG medium under high aeration for 18 h, and prepared for metal analysis as described under "Experimental Procedures." This experiment was repeated twice. though oxygen consumption was severely reduced (7). Thus, it is likely that superoxide is still produced in this strain in sufficient quantities to cause mischief and that the iron accumulation we see is attributable to superoxide-related toxicity.
We measured iron levels in anaerobically grown cultures of wild type and sod1 mutant yeast in order to test further the possibility that the accumulation of excess iron is related to oxidative stress. Accumulation of excess iron did not occur in anaerobically grown cultures of sod1 mutant yeast, indicating that this phenotype, like the other characteristics of the sod1 yeast, is related to oxidative stress (data not shown).
sod1 Mutant Shows Increased FET3 Gene Transcription-To test directly whether changes in the expression of the iron transport systems occurred, we analyzed expression of the FET3 and FET4 genes. Results of S1 nuclease protection experiments, shown in Fig. 5A, indicate that transcripts of the high affinity iron uptake gene, FET3, were increased approximately 2-fold in the sod1 mutant. The level of the FET4 transcript remained unaltered (data not shown). Similar results were obtained using a FET3 promoter-␤-galactosidase fusion construct (Fig. 5B). In the sod1/fet3 double mutant, there was an increase in the expression of the gene for the low affinity iron transporter, FET4 (Fig. 5C). This result is in agreement with the data presented in Fig. 4A showing that the sod1/fet3 mutant contains high intracellular iron levels; in this case, we infer that iron is transported by Fet4p. DISCUSSION Because of the close interrelationship between oxygen toxicity and the toxicity of redox active metals, we set out to explore the relationship between iron and oxidative damage in yeast lacking CuZnSOD. Iron is thought to exert a toxic effect on cells by catalyzing the formation of hydroxyl radical from hydrogen peroxide via the Fenton reaction (16,17). In vitro, and possibly in vivo as well, superoxide can be involved in this reaction as the reducing agent for the iron.
On the other hand, since superoxide participates in the Fenton reaction mainly as a reducing agent, it is hard to imagine that in the highly reducing conditions that prevail inside the cell, this reaction could be the main cause of superoxide toxicity. Recently, a more likely mechanism has been advanced to explain superoxide-related damage, which better explains our results. Superoxide radical is known to oxidize exposed [4Fe-4S] clusters in certain enzymes, leading to inactivation of the enzyme and "freeing" of iron (15,39,40). "Free" iron may then become available to participate in Fenton chemistry. For example, it has been shown in E. coli that DNA damage and mutagenesis are attributable to such a process (41). In this model superoxide is directly involved in enzyme inactivation and indirectly, but crucially, involved in oxidative damage, providing not only a possibly redundant reducing agent (superoxide), but also the catalytic "free" iron ion necessary for the Fenton reaction to occur. Thus, if Fenton chemistry was primarily responsible for the phenotype of sod1 yeast, added iron would be expected to be toxic to sod1 mutant yeast. On the other hand, if inactivation of iron-sulfur cluster enzymes was more important in causing the phenotype, we would expect added iron to be beneficial to these strains, which was what we observed.
We attribute the rescue of sod1 yeast by excess iron to its ability to meet the increased iron demand by the systems involved in reconstituting [4Fe-4S] cluster prosthetic groups. This conclusion is in accord with recent work, which showed that iron improved growth of sodA/sodB mutant E. coli by reactivating Fe-S cluster-containing enzymes (42). In our yeast experiments, if iron is present in limited amounts and is used to repair Fe-S clusters, less iron may be available for heme synthesis leading to deficiencies in cytochromes and thus respiratory deficits, as well as decreases in the activities of the specific were grown under high aeration in SDC liquid medium, as described under "Experimental Procedures." Total mRNA was extracted and subjected to S1 nuclease protection analysis using a probe for FET3 and an internal control probe for calmodulin (CMD). This experiment was repeated three times with similar results. A representative experiment is shown. B, WT (DTY3) yeast and sod1 mutant derivative EG151, both harboring the FET3::lacZ fusion plasmid, were grown under high aeration in SDC-Ura medium, and ␤-galactosidase activity was measured. Averages from four independent experiments are shown with error bars representing standard deviation. C, WT (EG103) yeast and mutant derivatives sod1 (EG118), and sod1/fet3 (JF244) were grown as above. FET4 transcription was analyzed by S1 protection assay using probes to FET4 and the internal control probe for CMD.
Fe-S cluster enzymes affected by excess superoxide. These effects may combine to severely diminish growth in glycerol.
The best studied enzyme with an exposed [4Fe-4S] cluster is aconitase (40), which catalyzes the interconversion of citrate and isocitrate and participates in the tricarboxylic acid cycle in the matrix of the mitochondria, but it is not the only potential target. Other likely candidates in yeast include homoaconitase, which converts homocitrate to homoisocitrate as part of the lysine biosynthetic pathway and is located in the intermembrane space of the mitochondria, 2 and 3-isopropylmalate dehydratase, which catalyzes an early step in leucine biosynthesis (40). Although it is not present in yeast, the iron-responsive element binding protein, or cytoplasmic aconitase, of mammalian cells is also a member of this family.
Iron deprivation activates a high affinity iron uptake system (23,37). This induction is mediated by increased transcription of the genes involved, including FET3, via the Aft1 transcription factor (37,43). We observed that the sod1 mutant showed higher levels of transcription of the FET3 gene (Fig. 5). Since the FET3 gene is induced by iron scarcity, we can infer that sod1 strain feels iron-deprived. The fact that the sod1 mutant nevertheless showed increased intracellular iron can be explained if the excess intracellular iron is in an inaccessible state, unavailable to the iron sensing and utilization machinery. Such a phenomenon has been reported recently in yfh1 and atm1 mutants in yeast (44,45).
Although high amounts of iron accumulate in the mitochondria of yfh1 and atm1 mutants, the cells feel iron-starved and continue to import more iron. Strains carrying either of these mutations spontaneously go rho Ϫ (i.e. lose their mitochondrial DNA and become unable to respire) at a very high rate (44,46). This loss was attributed to elevated oxidative damage due to the presence of excess iron in the mitochondria (44,45). In subsequent work, these proteins were shown to be involved in iron export from the mitochondria (47,48). In the sod1 mutant strain there is no obvious increase in the rate of rho Ϫ formation, or indeed of any kind of petite formation, so the iron that accumulates in this mutant is probably not in mitochondria, or (less likely) it is in an unreactive form. Recently, other workers have shown altered vacuolar morphology in sod1 mutants and suggested that iron may play a role in this phenomenon (36). Thus, the vacuole is another potential, perhaps more likely, location for the excess iron in our sod1 strains. It will be, of course, of great interest to determine where the excess iron is located in sod1 mutants, and we are currently proceeding in this direction.
Inactivation of the FET3 gene increased the severity of the sod1 phenotype (Figs. 1 and 2). In our experiments, both the sod1 and the sod1/fet3 mutant required increased iron (Figs. 1, 2, and 4A). However, the sod1/fet3 double knockout was more sensitive to oxygen than sod1 (Fig. 2). Recently it was shown that, in fet3 mutants, transcription of a low affinity iron transporter gene, FET4, was increased (32). Fet4p is a less specific metal ion transporter, transporting copper, cadmium, and cobalt as well as iron (22). The sod1/fet3 cell line showed an increase in FET4 transcription (Fig. 5C), but similar intracellular iron content compared with the sod1 mutant. The severity of the phenotype of the sod1/fet3 strain may reflect the reduced ability of the FET4 pathway to keep up with the sod1-induced iron demand, or it may reflect toxicity due to increased levels of other metal ions brought in by the less specific FET4 pathway.
Overall, we have observed an increased iron requirement in sod1 mutants and a beneficial effect of iron treatment on these cells. The fact remains, however, that excess iron could very well have simultaneous beneficial and harmful effects. The phenotype we observed in the sod1 mutants likely reflects a tension between harmful and beneficial effects of iron that in these experiments tended toward the beneficial. This dichotomy may explain why the phenotype we observed was not stronger, and it is another example of the basic principle that biology is a balancing act.