Inhibition of Fe-S cluster biosynthesis decreases mitochondrial iron export: Evidence that Yfh1p affects Fe-S cluster synthesis

Abstract

Decreased expression of Yfh1p in the budding yeast, Saccharomyces cerevisiae, and the orthologous human gene frataxin results in respiratory deficiency and mitochondrial iron accumulation. The absence of Yfh1p decreases mitochondrial iron export. We demonstrate that decreased expression of Nfs1p, the yeast cysteine desulfurase that plays a central role in Fe-S cluster synthesis, also results in mitochondrial iron accumulation due to decreased export of mitochondrial iron. In the absence of Yfh1p, activity of Fe-S-containing enzymes (aconitase, succinate dehydrogenase) is decreased, whereas the activity of a non-Fe-S-containing enzyme (malate dehydrogenase) is unaffected. Aconitase protein was abundant even though the activity of aconitase was decreased in both aerobic and anaerobic conditions. These results demonstrate a direct role of Yfh1p in the formation of Fe-S clusters and indicate that mitochondrial iron export requires Fe-S cluster biosynthesis.

Frataxin is a highly conserved protein found in all eukaryotes. The protein is encoded by a nuclear gene and is localized in the mitochondrial matrix. Deficits in frataxin protein result in Friedreich ataxia, a neurologic and cardiac disorder (1). Deletion of Yfh1p, the frataxin orthologue in yeast, leads to a respiratory defect resulting from excessive mitochondrial iron accumulation, which increases oxidant damage (2). The increase in mitochondrial iron is caused by malregulation of the mitochondrial iron cycle (3). Mitochondrial iron accumulation is only seen when cytosolic iron levels are high. Reduction in cytosolic iron due to either low-iron media (3, 4), deletion of genes required for high-affinity iron transport (3), or increased vacuolar iron transport (5), prevents excessive mitochondrial iron accumulation and will preserve respiratory activity in Δyfh1 cells.

In yeast, the proteins involved in Fe-S cluster synthesis are compartmentalized in the mitochondria where Fe-S clusters are synthesized for use both in mitochondria and for export to cytosolic proteins (6). Excessive mitochondrial iron is also seen as a result of deletion of genes required for synthesis of Fe-S clusters (for review, see ref. 6). For example, decreased expression of Nfs1p, the cysteine desulfurase (7, 8), Isu1p, a putative scaffolding protein (9), Yah1p, a putative ferredoxin (10), and Arh1p, a putative ferredoxin reductase (11), results in mitochondrial iron accumulation. In addition, reduced expression of Atm1p, a putative mitochondrial Fe-S exporter, also leads to accumulation of toxic levels of mitochondrial iron (7). The mechanism by which iron accumulates has not been elucidated. Herein, we describe that reduced expression of a critical enzyme involved in Fe-S synthesis, Nfs1p, results in mitochondrial iron accumulation through decreased mitochondrial iron export, similar to that seen in Δyfh1 cells (3). This result indicates that mitochondrial iron export depends on Fe-S cluster synthesis, suggesting that Yfh1p may play a role in Fe-S cluster synthesis. To test this hypothesis, we examined the affect of Yfh1p depletion on Fe-S cluster containing enzymes in cells grown in aerobic and anaerobic conditions.

Materials and Methods

Strains and Plasmids.

DY150 [MATa, ura3–52, leu2–3,112, trp1–1, his3–11, ade2–1, can1–100(oc)] and DY1457 [MATα ura3–52, leu2–3,112, trp1–1, his3–11, ade6, can1–100(oc)] were derived from the W303 strain of Saccharomyces cerevisiae. The METYFH1 strain (Δyfh1, pMET3YFH1[URA3]) was generated by crossing a Δyfh1 [MATa, ura3–52, leu2–3,112, trp1–1, ade2–1, can1–100(oc), yfh1:HIS3] strain with DY1457, as described (3). The expression of YFH1 was inhibited by growing cells in complete synthetic media (CM) supplemented with methionine (330 μg/ml).

A conditional NFS1 mutant strain (METNFS1) was generated by using an approach similar to that used to generate METYFH1 cells. A chromosomal copy of NFS1 was disrupted in a diploid strain, DY1640 [ura3–52/ura3–52, leu2–3,112/leu2–3,112, trp1–1/trp1–1, his3–11/his3–11, ade2–1/ade2–1, can1–100(oc)/can1–100(oc)] by homologous recombination by using a HIS3-containing construct and a PCR deletion approach (12). A plasmid was constructed with NFS1 under the control of the MET3 promoter with PCR primers, nfsfor (CGC GGA TCC GCG ATG TTG AAA TCA ACT GCT ACA) and nfsrev (cgc gaa ttc cgg tca ctt gtc atc gtc atc ctt gta atc atg acc tga cca ttt gta). The PCR product was ligated into a yeast expression vector (pTF63) with URA3 as a selectable marker. To make the plasmid referred to as pMET3NSF1, a MET3 promoter was then inserted at the 5′ end of the NFS1 sequence. The heterozygous Δnfs1/NFS1 diploid strain was transformed with the pMET3NFS1 plasmid. The transformed diploid was sporulated and the haploid METNFS1 (Δnfs1, pMET3NFS1[URA3]) was obtained.

A plasmid containing ACO1 with its own promoter (13) was the generous gift of W. Walden (University of Illinois, Chicago). The ACO1 gene was subcloned into a high-copy vector (pTF62). A high-copy CCC1 plasmid was prepared as described (5).

Analysis of mRNA by S1 Nuclease Assay.

Cells were harvested at midlog phase, and total RNA was extracted by vortexing with glass beads in phenol/chloroform/isoamyl alcohol (25:24:1) solvent. DNA oligonucleotides, with sequences complementary to FET3 and CMD1 transcripts (calmodulin, an internal control), were end-labeled with ³²P by T4 polynucleotide kinase (New England Biolabs). The FET3 oligo (no. 049) sequence used was: ggc gca att gga cat tgc gtc aag aag ggc aca ccg tcc ata gag gcg gtt ccg ttt tgg aag aga ccg tgg ggg cat. The CMD1 oligo (CMD1OLIGO) sequence used was: GGG CAA AGG CTT CTT TGA ATT CAG CAA TTT GTT CGG TGG AGC C. S1 analysis was performed as described (14). The ³²P-labeled oligos were hybridized with 25 μg of total RNA in Hepes buffer (38 mM Hepes, pH 7.6/0.3 M NaCl/1 mM EDTA/0.1% Triton X-100) at 55°C for at least 12 h. The reaction mixtures were digested with S1 nuclease, and the undigested double-strand duplexes were ethanol-precipitated and heat-denatured in formamide buffer. The DNA:RNA hybrids were separated on an 8 M urea/polyacrylamide gel in TBE buffer and quantified by phosphorimage analysis (STORM 860, Molecular Dynamics).

⁵⁹Fe Pulse-Chase and Subcellular Fractionation.

Mitochondrial iron efflux was examined as described (3, 5). As shown, mitochondria are found in fractions 3–9 and are well separated from other organelles.

Yeast Cell Lysate Extraction and Mitochondria Isolation.

Cells were disrupted by using glass beads as described (15) in the presence of a protease inhibitor “mixture” (Tablets, Complete, Roche Diagnostics) plus 1.0 mM PMSF. Samples of disrupted cells were taken for enzyme activity and Western blot analysis.

Mitochondria were isolated as described (16). The isolated mitochondria were used for enzyme assays and for Western blot analysis. For some experiments, mitochondria were incubated with 1.0% Triton X-100 and centrifuged at 15,000 × g for 10 min; the supernatant was used for enzyme assays. Densitometric analysis was performed on multiple exposures by using a MOLECULAR IMAGER FX (Bio-Rad) and QUANTITY ONE software (Bio-Rad).

Enzyme Assay.

Aconitase activity was measured as described (17). Malate dehydrogenase was measured in isolated mitochondria as described (18), as was succinate dehydrogenase (19).

Western Blot Analysis.

Samples (10 μg of protein of cell lysate, mitochondrial suspension, or Triton X-100-soluble mitochondrial fraction) were run on a SDS-12% polyacrylamide gel, transferred to nitrocellulose, and probed with either rabbit antibodies against aconitase [kindly provided by A. Dancis (University of Pennsylvania) and R. Lill (Institute für Zytobiologie, Universität Marburg, Germany)] or a mouse antiporin antibody (Molecular Probes), followed by a horseradish peroxidase-conjugated goat anti-rabbit or goat anti-mouse antibody. Membranes were developed by using a Chemiluminescence Renaissance Reagent (NEN Life Science).

Results

Effect of Yfh1p or Nfs1p on Mitochondrial Iron Export.

We demonstrated that, in the absence of Yfh1p, ⁵⁹Fe accumulated in mitochondria in marked excess over that seen in wild-type cells (3). Reintroduction of Yfh1p through regulated expression resulted in a loss of accumulated mitochondrial ⁵⁹Fe. To demonstrate that accumulated iron could be made bioavailable, we took advantage of the observation that transcription of the high-affinity iron transport system is inversely regulated by cytosolic iron. In response to low cytosolic iron levels, the iron-sensing transcription factor Aft1p translocates from the cytosol into the nucleus and activates transcription of FET3 and other genes that comprise the iron regulon (20). Accumulation of mitochondrial iron in Δyfh1 cells decreases cytosolic iron resulting in transcription of FET3 (2). In the absence of media iron, export of iron from mitochondria in Δyfh1 cells should lead to a decrease in FET3 transcription. When wild-type cells are incubated with high levels of the impermeable iron chelator bathophenanthroline sulfonate (BPS), a rapid induction of FET3 mRNA occurs [wild type (WT), CM to BPS], whereas little FET3 mRNA is transcribed when cells are incubated in iron-replete media (Fig. 1). In contrast, high levels of FET3 mRNA are seen in METYFH1 cells grown in iron-replete media in the presence of methionine, which represses the transcription of the MET3-regulated YFH1 (3). FET3 mRNA levels remain high when cells not expressing Yfh1p are incubated in BPS media. However, if YFH1 expression is permitted (by incubating cells in methionine-free media) and cells are incubated with BPS, then a transient decrease in FET3 mRNA occurs. We demonstrated that, on expression of YFH1, iron is exported from mitochondria (3). This result shows that the exported iron is bioavailable.

Fig 1.

Expression of Yfh1p affects FET3 transcription. Wild-type cells (transformed with a control vector) were grown in CM media and then transferred to CM media containing the iron-chelator BPS at concentrations high enough to chelate all available iron. Overnight cultures of METYFH1 cells grown in CM with methionine (330 μg/ml) were transferred to BPS-containing media in the presence (“off to off”) or absence (“off to on”) of methionine. At selected times cells were harvested, mRNA was isolated and analyzed by S1 nuclease analysis for both FET3 and CMD1 transcripts.

To examine if deletion of genes required for Fe-S cluster synthesis leads to mitochondrial iron accumulation through an effect on mitochondrial iron export, we generated a MET3-regulated NFS1. Nfs1p catalyzes the desulfation of cysteine, producing sulfide, a critical component in Fe-S cluster synthesis (28). A diploid heterozygote containing a chromosomal deletion of NFS1 was transformed with a plasmid that had NFS1 under the control of the MET3 promoter. The diploid was sporulated and the haploids were grown in methionine-free media. When haploids were plated to methionine-containing media (NSF1 “off”), two of the four spores were inviable. These cells, termed METNFS1, could be grown for 12–16 h in methionine-containing media without loss of cell viability. Decreased expression of Nfs1p leads to accumulation of ⁵⁹Fe in mitochondria (Fig. 2A). If NFS1 is then turned on (by growing cells in methionine-free media), a rapid loss of ⁵⁹Fe from mitochondria results, similar to that seen when Yfh1p is reintroduced into Yfh1p-depleted mitochondria.

Fig 2.

Expression of Nfs1p affects mitochondrial iron export and FET3 transcription. (A) METNSF1 and METYFH1 cells were grown overnight in the presence of methionine. The cells were incubated for 10 min with ⁵⁹Fe in the presence of ascorbate and then incubated in ⁵⁹Fe-free media for 12 h in the presence or absence of methionine. The cells were sphereoplasted, homogenized, and the membrane fraction applied to an Iodoxianol gradient. Radioactivity was determined in each fraction and was plotted as a percentage of total cell radioactivity. The mitochondria are found in fractions 3–9. (B) The experimental design was as described in the legend to Fig. 1 except that METNFS1 was studied.

To confirm that mitochondrial iron export results in bioavailable iron, we measured FET3 transcription in METNFS1 cells. As expected, in the absence of Nfs1p, the expression of FET3 mRNA was increased, and no change in FET3 mRNA was seen when cells were incubated in BPS media in the presence of methionine (Fig. 2B). Incubation of cells in BPS media with NFS1 turned on resulted in a transient decrease in FET3 mRNA. Because no external iron existed, the decrease in FET3 mRNA resulted from increased cytosolic iron due to increased mitochondrial iron export. The decrease in FET3 mRNA was transient; after a few hours in BPS media, FET3 mRNA levels increased. This increase reflects iron limitation due to low media iron, as suggested by the fact that if iron is present in the media and Nfs1p is expressed, FET3 mRNA levels remain low. These results demonstrate that loss of Nfs1p, which leads to loss of Fe-S cluster synthesis, affects mitochondrial iron export.

Loss of Yfh1p Affects Fe-S Cluster Biosynthesis.

Inhibition of respiratory activity through the generation of rho⁰ cells does not lead to mitochondrial iron accumulation as long as cells express Yfh1p (21). In preliminary studies, we observed that inhibition of heme synthesis does not lead to mitochondrial iron accumulation. Cumulatively, these observations show that alteration of mitochondrial iron export is a consequence of decreased Fe-S cluster synthesis, suggesting that Yfh1p plays a role in Fe-S cluster synthesis. Although this hypothesis had been suggested previously, limited supporting data were available (4, 22). To examine this hypothesis, we measured the activity of two mitochondrial Fe-S cluster enzymes (aconitase and succinate dehydrogenase) and a non-Fe-S cluster enzyme (malate dehydrogenase). Because Δyfh1 cells cultured in complete media lose their mitochondrial genome and are rho⁰, we generated a rho⁰ wild-type cell to use as a control. Loss of the mitochondrial genome in wild-type cells resulted in an increase in aconitase activity (Fig. 3A) and a decrease in both succinate dehydrogenase (Fig. 3B) and malate dehydrogenase (Fig. 3C). The level of malate dehydrogenase activity was affected little in a Δyfh1 deletion, whereas the declines were precipitous in the activity of the two Fe-S enzymes, aconitase and succinate dehydrogenase. In the absence of Yfh1p, aconitase activity is decreased (less than 4.0%), even though, as shown by Western blot analysis and quantified by densitometry, a substantial amount of aconitase protein exists (greater than 20% relative to porin, Fig. 3D).

Fig 3.

Yfh1p depletion affects the activity of aconitase. A Triton X-100 mitochondrial extract from wild-type, rho⁰, and Δyfh1 strains was assayed for aconitase activity (A), succinate dehydrogenase activity (B), and malate dehydrogenase activity (C). Aconitase and porin protein levels were determined by Western blot analysis on isolated mitochondria (D). The degradation of aconitase protein on Western blot analysis is denoted by an asterisk.

To demonstrate that the decrease in the activity of Fe-S enzymes was not a result of secondary effects arising during the culture of Δyfh1 cells, we examined aconitase activities in METYFH1 cells grown in the presence of methionine (YFH1 “off”) (Fig. 4A). Again, in the absence of Yfh1p, aconitase activity decreases (>90%), whereas no decrease (as determined by densitometry) occurs in aconitase protein and degradation products. When Yfh1p is not expressed, we often see a decrease in the full-size aconitase protein and the presence of what seem to be degradation products. No such degradation products were ever seen in control cells or when Yfh1p is expressed. As shown below, much less degradation of aconitase protein occurs under anaerobic conditions, suggesting that the presence of degradation products may result from oxidative damage.

Fig 4.

Aconitase activity in METYFH1 cells overexpressing ACO1 and CCC1. Aconitase activity (Upper) and protein (Lower) were assayed in METYFH1 cells (A), in METYFH1 cells transformed with a high-copy ACO1 plasmid or vector (B), in METYFH1 cells transformed with a high-copy CCC1 plasmid (C), or with both high-copy ACO1 and CCC1 plasmids (D). The cells were grown in the presence (Yfh1p “off”) or absence (Yfh1p “on”) of methionine for 15 h, a total cell extract prepared, and aconitase activity or protein determined. The degradation of aconitase protein on Western blot analysis is denoted by an asterisk.

Decreased aconitase activity due to lack of Yfh1p was confirmed by using cells overexpressing aconitase. Cells were transformed with either a high-copy ACO1-containing plasmid or a control vector. Wild-type cells transformed with pACO1 showed a 3-fold increase in aconitase activity compared with cells transformed with the control vector. In cells transformed with pACO1 but not expressing Yfh1p, the absence of aconitase activity was nearly complete. Indeed, we consistently observed that in the absence of Yfh1p, ACO1-transformed cells had less activity than vector-transformed cells. We do not have an explanation for this observation, except that it is not due to a general decrease in mitochondrial enzyme activities, because malate dehydrogenase activity is unchanged (see below). Further, examination of aconitase protein level by Western blot analysis revealed abundant protein (Fig. 4B). This result suggests that in the absence of Yfh1p, the loss of aconitase activity is not due to loss of aconitase protein but rather reflects the presence of an apoprotein.

Loss of Aconitase Activity Is Not Due to Oxidant Damage.

The loss of aconitase activity could result from decreased Fe-S cluster synthesis or from oxidant damage. Data suggest that aconitase is highly susceptible to oxidative damage (23), and studies have shown that loss of Yfh1p increases oxidative damage (2, 24). We used two approaches to test the possibility that decreased aconitase activity was caused by oxidant damage. In the first approach, we examined aconitase activity in METYFH1 cells overexpressing the vacuolar iron transporter Ccc1p. Overexpression of CCC1 lowers cytosolic iron by increasing vacuolar iron accumulation (25). Reduction in cytosolic iron in Δyfh1 cells prevents accumulation of toxic levels of mitochondrial iron and limits oxidant damage (5). When Yfh1p is expressed, measurable levels of aconitase activity are seen in cells overexpressing CCC1. However, the level of aconitase activity in CCC1-overexpressing cells is 60% lower than vector-transformed cells (Fig. 4C). We think that the decreased aconitase activity in CCC1-overexpressing cells reflects cytosolic iron depletion, which affects mitochondrial iron uptake. Overexpression of CCC1 does not, however, restore aconitase activity in the absence of Yfh1p. This result is shown more dramatically in cells that overexpress aconitase as a result of transformation with pACO1 (Fig. 4D). Transformation of pACO1 into CCC1-overexpressing wild-type cells leads to an increase in aconitase activity compared with vector-transformed cells. In the absence of Yfh1p, the level of aconitase activity is barely detectable. Loss of Yfh1p leads to a greater than 99% loss of activity, yet only a 50% decrease in protein occurs. We observed that a decrease occurs in the degradation of aconitase protein under conditions in which aconitase is overexpressed, when mitochondrial iron is limiting or under anaerobic conditions (see below). These observations suggest that aconitase degradation is saturable and may be dependent on iron-mediated oxidative damage.

The second approach used to study the affect of Yfh1p on aconitase activity in Δyfh1 cells used anaerobic conditions. Incubation of cells under anaerobic conditions prevents oxidative damage, as shown by the increased viability of Δsod1 cells under anaerobic as opposed to aerobic conditions (26). As Δyfh1 cells lose their mitochondrial genome during cultivation, we generated a rho⁰ cell line to provide a more accurate control. Wild-type rho⁰ cells grown under anaerobic conditions for several generations had measurable levels of aconitase (Fig. 5A), succinate dehydrogenase (a Fe-S protein) (Fig. 5C), and the non-Fe-S TCA enzyme malate dehydrogenase activity (Fig. 5B). Δyfh1 cells grown anaerobically had comparable levels of malate dehydrogenase activity but lower aconitase and succinate dehydrogenase activities. Western blot analysis of aconitase in Δyfh1 cells showed levels of aconitase protein similar to that seen in wild-type rho⁰ cells. The ratio of aconitase to porin as determined by densitometry was 4.6 for wild-type, 3.9 for rho⁰, and 5.3 for Δyfh1 Fig. 5D), whereas Δyfh1 cells had less than 20% aconitase activity. These results suggest that aconitase is in the apo form.

Fig 5.

Yfh1p depletion affects aconitase activity in cells grown under anaerobic conditions. Control cells (wild-type and wild-type rho⁰ cells) and Δyfh1 cells were grown in yeast extract peptone dextrose media under anaerobic conditions for 3 days. Mitochondria were isolated and a Triton X-100-soluble fraction was used to determine the activities of aconitase (A), malate dehydrogenase (B), and succinate dehydrogenase (C). (The increase in specific activities in this figure, compared with Fig. 4, is due to use of the more purified Triton X-100 extract to assay enzyme activities.) (D) The abundance of aconitase in isolated mitochondria was determined by Western blot analysis. The level of the mitochondrial protein porin was determined by Western blot analysis to demonstrate equal loading of total mitochondrial protein.

Overexpression of ACO1 in wild-type rho⁰ cells grown under anaerobic conditions leads to a 2-fold increase in aconitase activity (Fig. 6A). In the absence of Yfh1p, a dramatic decrease again occurs in aconitase activity, whereas aconitase protein levels are little affected (Fig. 6C). Densitometric analysis showed that pACO1-transformed Δyfh1 cells had 70% more aconitase protein than vector-transformed Δyfh1 cells yet had only 14% of the activity. Overexpression of ACO1 in the wild-type rho⁰ cells or in the Δyfh1 cells had little effect on malate dehydrogenase activity (Fig. 6B). Thus, in the absence of oxygen, loss of Yfh1p affects aconitase activity but not aconitase protein.

Fig 6.

Yfh1p depletion affects aconitase in cells overexpressing ACO1 grown under anaerobic conditions. Both wild-type rho⁰ cells and Δyfh1 cells transformed with a high-copy ACO1 plasmid (pACO1) or a control vector (V) were grown in CM media under anaerobic conditions for 3 days. Mitochondria were isolated, and the activities of aconitase (A) and malate dehydrogenase (B) were determined in the Triton X-100-soluble mitochondrial fraction. (C) Aconitase and porin protein levels were determined by Western blot analysis.

Discussion

Mitochondria house two important biosynthetic activities that require iron: synthesis of heme and Fe-S clusters. In higher eukaryotes, Fe-S groups may be synthesized in both mitochondria and cytosol (27). In yeast, the enzymes that perform this activity are compartmentalized in mitochondria (6). Iron must be imported into mitochondria for both heme synthesis and Fe-S synthesis, although the transporter(s) that effect this activity have not been identified. Both heme and Fe-S clusters must be exported from mitochondria. The ABC-transporter Atm1p is a reasonable candidate for the Fe-S cluster exporter, although the exact nature of the Fe-S substrate is unknown (23). Deletion of genes that encode proteins involved in Fe-S cluster synthesis and export result in excessive mitochondrial iron accumulation. The more essential the gene to Fe-S cluster synthesis, the greater the effect on mitochondrial iron accumulation. Depletion of Nfs1p, for example, results in a 20- to 30-fold increase in mitochondrial iron accumulation over that seen in wild-type cells (7, 8).

By using two assays, we demonstrated that excessive mitochondrial iron accumulation due to depletion of Nfs1p results from a decrease in mitochondrial iron export. Reintroduction of Nfs1p into Nfs1p-depleted cells leads to loss of accumulated iron, which is exported to the cytosol. These results suggest a mitochondrial iron cycle in which iron export is a regulated step. Because deletion of Atm1p also affects mitochondrial iron accumulation, but does not affect mitochondrial Fe-S cluster-containing proteins, a reasonable hypothesis is that the absence of a cytosolic Fe-S protein regulates mitochondrial iron export.

Alteration of mitochondrial iron export seems restricted to defective Fe-S cluster synthesis. Alteration of two other yeast mitochondrial functions, respiration (21) and heme synthesis (unpublished data), does not lead to increased mitochondrial iron accumulation. On the basis of this observation, the finding that depletion of Yfh1p also affects mitochondrial iron export suggests that Yfh1p is involved in Fe-S cluster synthesis. A role for Yfh1p and its mammalian orthologue frataxin in Fe-S cluster synthesis was suggested by studies indicating decreased activities of mitochondrial Fe-S-containing enzymes both in Δyfh1 yeast and in biopsies from patients with Friedreich ataxia (4, 22). Because protein levels were not reported, the reason for the decrease in activity was unclear. Lutz et al., studying Fe-S cluster assembly in isolated mitochondria, also suggested a role for Yfh1p in Fe-S cluster synthesis (28). Those studies demonstrated that apoYah1p (the precursor to ferredoxin), synthesized in vitro and imported into isolated mitochondria, could be converted into a Fe-S-containing holoprotein. A similar experiment showed that import of apoYah1p into mitochondria isolated from Δyfh1 cells did not lead to the assembly of Fe-S on Yah1p. A caveat to this experiment is that Δyfh1 mitochondria might be expected to have high levels of iron, which may affect Fe-S cluster formation or oxidize Fe-S clusters once assembled.

We tested the effect of Yfh1p depletion on aconitase activity and protein levels. Loss of Yfh1p invariably led to a decrease in aconitase activity. Decreased aconitase activity was not accompanied by a decrease in aconitase protein as assayed by Western blot analysis, suggesting that decreased activity resulted from an apoprotein lacking a Fe-S prosthetic group. The apoprotein seems highly susceptible to degradation. Overexpression of aconitase in wild-type cells led to a 2- to 3-fold increase in enzyme activity. In the absence of Yfh1p, overexpression of ACO1 did not lead to increased activity. In fact, the level of aconitase activity was even lower than the level of endogenous aconitase in Yfh1p-depleted cells. Again, aconitase protein was abundant, suggesting the absence of the Fe-S prosthetic group. Similar results were obtained under two conditions expected to lower mitochondrial iron and/or oxidant damage: overexpression of CCC1 and anaerobic conditions. The presence of the aconitase protein in the absence of aconitase activity suggests that the protein is in the apo form, indicating that Yfh1p may have a direct role in Fe-S cluster synthesis.

Most of the enzymes involved in Fe-S cluster synthesis are highly conserved between bacteria and eukaryotes (6). Yfh1p/frataxin is an exception, because it is not present in most bacteria. A gene that exists in bacteria with some homology to Yfh1p/frataxin is termed CYA. Deletion of CYA, however, has no phenotype. Our data suggest a role for Yfh1p in Fe-S synthesis. Because Yfh1p is not represented in the bacterial Fe-S cluster operon, its role in Fe-S cluster synthesis may reflect an adaptation specific to eukaryotes. In the absence of Yfh1p, Fe-S cluster biosynthesis can occur, albeit at a reduced level, suggesting that Yfh1p plays an important but not essential role in Fe-S cluster synthesis. One speculation is that Yfh1p may play a role in iron donation, as a type of mitochondrial metallo-chaperone, analogous to the copper chaperones (for review see ref. 25). High-affinity insertion of copper into target proteins is mediated by specific proteins. In the absence of those proteins, metal incorporation will occur, albeit inefficiently. In this regard, it is intriguing that data suggest the Yfh1p/frataxin is an iron-binding protein (29). Given a role of Yfh1p in Fe-S cluster synthesis and the ability to tease apart Fe-S cluster synthesis, either in vitro or in isolated yeast mitochondria, an experimental approach to determine the mechanism of Yfh1p action is now feasible. Additionally, genetic approaches now permit a search for cytosolic Fe-S proteins that may regulate mitochondrial iron export.

Note Added in Proof.

A recent paper (30) also demonstrates that Yfh1p is involved in Fe-S cluster synthesis.

Abbreviations

• BPS, bathophenanthroline sulfonate

• CM, complete synthetic media