Mitochondrial Control of Iron Homeostasis

Deletion of YFH1, the yeast frataxin homologue gene, elicits mitochondrial iron accumulation and alters cellular iron homeostasis. Here, we report a genome wide analysis of gene expression in a yfh1(ΔYFH1) deleted strain. Frataxin deficiency results in enhanced expression of some 70 genes including a set of genes, called the iron regulon, that are under the control of the iron-sensing transcription factor AFT1. Five newAFT1-dependent genes, YOR382w,YOR383c, YDR534c, YLR136c, andYLR205c were found. The first three genes presumably encode cell-wall glycosylphosphatidylinositol anchor proteins and exhibit a 30–100-fold increased expression. The triple deletion of these genes decreases efficiency in utilization of the iron of ferrioxamine B by the yeast cell. YLR136c bears homology to tristetraproline proteins, which are post-transcriptional regulators in mammalian cells. Deletion of YLR136c increases the mRNA levels of iron regulon members. YLR205c bears homology to heme oxygenases. Our data show that frataxin deficiency elicits iron mobilization from all iron sources in an AFT1-dependent manner. Wild-type and ΔYFH1 glycerol-grown cells exhibit similar high respiration rates, no mitochondrial iron accumulation, and high expression of the iron regulon, suggesting that under these conditions little iron is extruded from mitochondria. These data suggest that the activity of Yfh1p is not essential in cells grown on glycerol. This study has also revealed unexpected links between mitochondria and remote metabolic pathways since frataxin deficiency also enhances the expression of genes such as HSP30, that escape toAFT1 control. Finally, no oxidative stress gene is induced.

Friedreich's ataxia is a neurodegenerative disease, associated with cardiomyopathy and, in some cases, with diabetes (for a review, see Ref. 1). It is caused by mutations in a nuclear gene encoding frataxin, a small protein with no homology to proteins of known function (2). Frataxin has been conserved during evolution from Gram-negative bacteria to humans, and in eukaryotes it is located in mitochondria (3)(4)(5)(6). Deletion of the yeast frataxin gene (YFH1) elicits a 10 -15-fold increased mitochondrial iron accumulation in cells cultivated in glucoserich iron-replete medium (7), and moreover, the accumulated iron cannot be extruded from mitochondria (8). Loss of mitochondrial DNA has been observed at high iron concentrations (7)(8). In parallel, expression of the FET3 and FTR1 genes which encode the two subunits of high affinity iron transporter is increased (4,8). Analysis of the solution and crystal structures of human frataxin (9 -10) and its bacterial Cyay homologue (11) has revealed a novel protein fold consisting of a five-stranded antiparallel ␤ sheet packing against a pair of parallel ␣ helices. A patch of anionic residues at the surface could be involved in iron binding. It has recently been reported (12) that recombinant purified yeast frataxin is a soluble monomer that contains no iron but self-aggregates upon addition of ferrous iron as a high molecular weight product that sequesters more than 3000 iron atoms in an available form. Thus, frataxin might be an iron storage protein regulating iron homeostasis (12).
Iron is an essential cofactor for many reactions in the cell; in particular, it is present in cytochromes, iron-sulfur proteins, and ribonucleotide reductase. Moreover, mitochondria contain the protein machinery responsible for iron-sulfur cluster biosynthesis and export into the cytosol (for reviews, see Refs. 13 and 14). However, excess iron can generate via the Fenton reaction highly toxic-free radicals generating oxidative damage to the cell. Thus, cellular iron concentration must be tightly controlled. We have found that mitochondrial iron accumulation in a yfh1 deletion strain is strongly dependent on culture medium composition (7,15). In contrast with cells grown in glucose-rich medium, only a 2-3-fold increase in mitochondrial iron is observed in raffinose minimum medium, and no iron accumulation is detected in mitochondria from cells grown in glycerol-rich medium (see below). Mitochondrial iron overload in a ⌬YFH1 strain can be enhanced by increasing external iron concentration or can be avoided by preventing iron import into mitochondria (15,16).
Thus, we thought that loss of frataxin could alter the expression of a number of genes, involved either in iron metabolism or oxidative stress, or in any other unexpected metabolic pathway. To address this question, we have carried out a genome wide analysis of gene expression in wild-type and YFH1-deleted strains using DNA microarrays including all available Saccharomyces cerevisiae open reading frames.

EXPERIMENTAL PROCEDURES
Strains and Growth Media-S. cerevisiae strains were W303-1B (MAT␣ ura3 -52, leu2-3, 112, trp1-1, his3-11, 15, ade2-1) and its isogenic derivative W303-1B⌬YFH1 (MAT␣ ura3-52, leu2-3, 112, trp1-1, his3-11,15, ade2-1 yfh1⌬::Kan R ) (7). All deletion mutants were constructed in strain W303-1B and/or W303-1B⌬YFH1. Briefly, the deletion mutants were obtained as follows. AFT1 and FET3 deletion cassettes were constructed in a pSK vector by insertion of the URA3 gene between two polymerase chain reaction-amplified DNA fragments homologous to the 3Ј-and 5Ј-flanking regions of the AFT1 gene. YLR136c deletion cassette was obtained from strain BY4741 (17) by polymerase chain reaction amplification of the kanamycin resistance marker and yeast DNAflanking regions. YOR382w and YOR383c were disrupted in a single step using a loxP-kanMX-loxP gene disruption cassette (18). The kanMX module was excised by the Cre recombinase expressed in the presence of galactose and the same procedure was used to delete YDR534c in ⌬YOR382⌬YOR383 strain. All deletion strains were verified by polymerase chain reaction amplification of the 3Ј and 5Ј regions flanking the target gene using specific oligonucleotide primers. Deletion cassettes were used to transform W303 diploid and W303-1B haploid strains. Introduction of the ⌬YFH1 deletion in the ⌬YOR382⌬-YOR383⌬YDR534 strain was obtained by mating, zygote isolation, and tetrad dissection. The absence of the wild-type gene and presence of the deletion markers were verified by appropriate polymerase chain reaction amplifications using specific oligonucleotide primers.
RNA Preparation and Northern Blotting Analysis-Cells were grown to an OD of 0.6 -0.7, harvested in ice-cold water as quickly as possible and frozen in liquid nitrogen before use. Total RNA prepared by hot acidic phenol extraction was used for Northern blot analysis and was further purified for DNA microarray hybridizations by the RNeasy purification kit from Qiagen. For each sample, equal amounts of RNA (20 to 30 g) were loaded on 1.5% agarose-formaldehyde gels, blotted onto a nylon membrane, and hybridized with appropriate 32 P-labeled probes. Hybridizations were carried out using standard procedures.
cDNA Synthesis for Microarray Hybridizations-The reverse transcription reaction was performed in a final volume of 40 l. Up to 50 g of purified total RNA was combined to 8 l of 5X First Strand reaction buffer (Superscript II, Life Technologies), 3 l of AncT mRNA primer (T20VN, 100 pmol), 3 l of 20 mM dNTP minus dCTP (6.67 mM of each dATP, dGTP, dTTP), 1 l of 2 mM dCTP, 1 l of 1 mM Cy3-dCTP or Cy5-dCTP (Amersham Pharmacia Biotech), and 4 l of 0.1 M dithiothreitol. The mixture was incubated at 65°C for 5 min, then at 42°C for 5 min. 2 l of reverse transcriptase Superscript II (400 units) (Life Technologies) and 1 l of RNasin (Promega) were added to the reaction and incubated at 42°C for 2 h. The reaction was stopped by adding 7 l of a 5:2 mixture of 50 mM EDTA (pH 8), and 1 N NaOH and incubated for 20 min at 65°C. The unincorporated fluorescent dNTP, AncT primers, and salt were removed by precipitating the cDNA on ice for at least 30 min in the presence of 4 l of 0.5 M acetic acid, 4 l of 3 M sodium acetate, and 50 l of isopropyl alcohol. The fluorescent cDNA was dissolved in 2.5 l of water.
Microarray Hybridizations and Data Analysis-The yeast DNA chips manufactured by the Microarray Center, Ontario Cancer Institute, Toronto, Canada, are a generous gift from Dr. Bryan Mcneil in this Center. The microarray was placed in a sealed humidified chamber and prehybridized using 57 l of warmed DIG Easy Hybridization buffer (Rochem Molecular Biochemicals) and 3 l of 10 g/ml fragmented denatured salmon sperm DNA. A coverslip of 24 ϫ 50 mm was placed on the top and the microarray was incubated in a water-bath at 37°C for at least 2 h. The slides were then washed twice with 0.1 ϫ SSC at room temperature, briefly rinsed with bi-distilled water and dried by centrifugation at 500 rpm. For hybridization, 2.5 l of Cy3-cDNA (wild-type) and 2.5 l of Cy5-cDNA (⌬YFH1) were mixed, added to 55 l of DIG Easy Hybridization solution, and applied to the yeast DNA chips. The dual color hybridization was carried out at 37°C for 16 h. The glass slides were washed twice, 15 min each, with 0.1 ϫ SSC, 0.1% SDS washing solution prewarmed at 50°C, then transferred to 0.1 ϫ SSC at room temperature for 5 min and dried by centrifugation at 500 rpm. The hybridization signal was detected by scanning the microarrays using the GenePix 4000 laser scanner (Axon Instruments). The signal quantification was performed using the integrated GenePix software version 3.01. The Yeast Proteome (YPD), Saccharomyces cerevisiae Genome (SGD) and MIPS data bases have been used.

RESULTS
Cells were cultivated in glucose-rich medium and harvested in the early exponential phase of growth. The proportion of cytoplasmic petites in the culture medium was below 5% and could not influence the results. Microarray hybridization experiments were repeated three times using independent cultures. Only those genes whose expression was stimulated at least 2.5 times on average in the ⌬YFH1 strain are reported in Table I. This is certainly an underestimate of the number of genes up-regulated by frataxin loss as the three experiments gave appreciable fluctuations in the expression of several genes, probably reflecting tight dependence on subtle physiological conditions. Reliable identification of down-regulated genes was often limited by low expression levels in the wildtype strain. However, the expression of several genes was significantly reduced in the ⌬YFH1 strain, in particular GLT1 and LEU1 which possess iron-sulfur clusters (Table I).
Up-regulated Genes-The most widely represented class of up-regulated genes in a frataxin null mutant encodes proteins that localize, or are predicted to localize to the cell surface (Table I). More specifically, the expression of YOR382w, YOR383c, and YDR534c was increased 30 to 100 times as compared with wild-type strain (Table I). These data were confirmed by Northern blot analysis ( Fig. 1, lanes 1 and 2). The encoded proteins have no homology to other proteins even though they exhibit the typical features of cell-wall GPI-anchor glycoproteins (19). They are acidic, extremely rich in serine/ threonine residues, and their amino acid sequence is characterized by reiterated motifs. YOR383c and YDR534c are very similar except that their common motif is repeated four times in YDR534c and only twice in YOR383c.
The expression of all known members of the "iron regulon" was increased in a ⌬YFH1 strain (Table I and Fig. 1). The iron regulon includes the genes of the ferric reductases (FRE1, FRE2, and FRE3) (20,21), the high affinity iron uptake system (FET3/FTR1) (22,23), siderophore transporters (ARN1-4) (24 -27), vacuolar iron transporter (FET5/FTH1) (28), and copper delivery to the Fet3p oxidase (ATX1 and CCC2) (29 -31). All are under the control of the iron-sensing transcription factor, Aft1p (32). This protein recognizes a specific sequence in the promoter of the iron regulon genes and activates their transcription (32)(33)(34). Iron prevents binding of Aft1p to the target genes, and therefore, represses their expression. Alteration of iron homeostasis in the ⌬YFH1 strain is also associated with increased expression of several genes involved in heavy metal uptake or homeostasis (CTR2, ATX2, COT1, and SMF3) (35-37) that are not under AFT1 control (Table I). This is in agreement with the observation that ⌬YFH1 strains are more sensitive to copper, cobalt, and manganese than the wild-type strain. 2 The high increase in the expression of several genes of unknown function might be particularly interesting (Table I). HSP30 (Fig. 1) encodes a seven-transmembrane helix protein related to the opsin family (38 -39). YOR285w bears homology to sulfuryltransferases of the mitochondrial rhodanese protein family. YLR205c shares high amino acid sequence identity with heme oxygenases. These are enzymes associated with the endoplasmic reticulum that open the heme ring and release iron. YLR136c ( Fig. 1) bears homology to the mammalian tristetraproline (TIS) proteins which are characterized by CX 8 CX 5 CX 3 H repeats and play a role in mRNA stability (40 -42).
By comparison, the effect of ⌬YFH1 deletion on mitochondria was modest. However, the expression of the IDH1 and IDH2 1 The abbreviations used are: BPS, bathophenantroline sulfonate; GPI, glycosylphosphatidylinositol; TIS, tristetraproline. 2 F. Foury, unpublished data.  genes which encode isocitrate dehydrogenases was 3 to 4 times increased (Table I), in agreement with the observation that isocitrate dehydrogenase activity was substantially increased in ⌬YFH1 mitochondrial extracts (data not shown). The expression of the ISU1 and ISU2 genes which are involved in mitochondrial iron-sulfur cluster biosynthesis (43,44) was also enhanced by a factor of 2.5-3.5 in a ⌬YFH1 strain (Table I).
Surprisingly, the expression of the genes involved in the response to oxidative stress was not significantly increased. We have verified by Northern blotting (Fig. 1) that neither SOD1, which encodes the zinc-copper superoxide dismutase in the cytosol (45), nor SOD2, which encodes the mitochondrial manganese superoxide dismutase (46,47), were up-regulated.
The Iron Regulon Is Up-regulated in a Frataxin-deficient Strain-We addressed the question whether the most representative genes of unknown function identified by our microarray analysis, YOR382w, YOR383c, YDR534c, YLR136c, YLR205c and HSP30 belong to the iron regulon. A typical feature of the members of the iron regulon is that they are induced upon iron deprivation and are not expressed in a ⌬AFT1 strain (32)(33)(34).
We performed two types of Northern blot experiments. First, we used RNA from iron-starved wild-type cells grown in the presence of 100 M BPS, a nonpermeant iron chelator. YOR382w, YOR383c, YDR534c, and YLR136c were strongly induced in the presence of BPS while HSP30 was not (Fig. 1,  lanes 5 and 6). Second, AFT1 was deleted in YFH1/YFH1 and YFH1/⌬YFH1 diploid strains and tetrads were dissected. It has previously been reported that ⌬AFT1-deleted strains cannot grow on a respiratory carbon source such as glycerol (34). We found that strains W303-1B and W303-1B⌬YFH1 with an AFT1 gene deletion could grow slowly on glycerol medium. Optimal growth was restored by addition of iron in the culture medium (Fig. 2a). While in a glucose-rich medium containing 250 M iron 70% petite mutants accumulated in the ⌬YFH1 strain, less than 10% petite colonies were observed in the ⌬YFH1⌬AFT1 strain (Fig. 2b). However, the double mutant was still sensitive to high iron concentrations, so that optimal growth on glycerol medium was obtained for a narrow range of iron concentrations (Fig. 2a). Therefore, AFT1 deletion did not prevent mitochondrial iron accumulation in a ⌬YFH1 background provided enough iron was imported into the cell. Tetrad analysis showed that AFT1 deletion considerably decreased the expression of YOR382w and YLR136c both in wild-type and ⌬YFH1 strains (Fig. 3). Only a very slight decrease in HSP30 expression was observed (Fig. 3). These data show that YOR382w, YOR383c, YDR534c, and YLR136c belong to the iron regulon, in agreement with the presence of several copies of the consensus sequence recognized by Aft1p in their promoter. YLR205c is probably also under AFT1 control since the consensus sequence recognized by Aft1p is present twice in the promoter region at positions 582 and 292 upstream of the start codon. In contrast, HSP30 promoter does not contain this sequence and, therefore, is not directly regulated by AFT1.
The New Members of the Iron Regulon Play a Role in Iron Metabolism-All known iron regulon members play a role in iron mobilization, or copper delivery to Fet3p. Induction of YOR382w, YOR383c, and YDR534c upon iron deprivation is so great that it is difficult to imagine that these proteins have no role in iron mobilization. In addition to ferrous iron which is used by the high affinity iron uptake system (Fet3p/Ftr1p), S. cerevisiae can utilize siderophores synthesized by fungi and bacteria. Siderophores are imported by four siderophore transporters (Arn1-4p) with distinct substrate specificities (25)(26)(27), or their ferric iron can be released, reduced, and transported by the high affinity iron uptake system (25). A triple deletion strain ⌬YOR382⌬YOR383⌬YDR534 had no obvious growth defect (data not shown). When a ⌬FET3 mutation, which abolishes the high affinity iron uptake, was introduced in the triple deletion strain, the colonies grown in the presence of limiting amounts of a siderophore such as ferroxiamine B (2 M) were slightly smaller than in a simple ⌬FET3 strain (Fig. 4A). No difference was observed in the presence of saturating concentrations (100 M) of ferroxiamine B (data not shown). These data suggest that the loss of the three GPI-anchor proteins decreases siderophore-mediated iron utilization. We also observed that a ⌬YFH1 strain grew more poorly on 100 M ferroxiamine B than on 2 M of this siderophore (Fig. 4, B and  C, first row). This indicates that ferroxiamine B is an efficient iron source for the cell which becomes toxic at high concentrations in a frataxin-deficient strain. This growth defect was partially alleviated by the ⌬YOR382⌬YOR383⌬YDR534 deletion (Fig. 4C, first and second rows). In addition, while mitochondrial free iron reached 3.8 nmol/mg of protein in the simple ⌬YFH1 strain, it was decreased to 1.1 nmol/mg of protein in the quadruple deletion strain. This strongly suggests that in the latter strain less iron penetrates the cell. Deletion of the FET3 gene in a ⌬YFH1 strain substantially improved cellular growth on 100 M ferroxiamine B (Fig. 4C, first and third rows) and decreased mitochondrial free iron concentration to 0.4 nmol/mg FIG. 1. Expression of representative genes in wild-type and ⌬YFH1 strains grown in various media. Northern blots were carried out as reported under "Experimental Procedures." Wild-type (wt) and ⌬YFH1 (⌬) strains were grown in glucose-rich or glycerol-rich media. Wild-type strain was also grown in raffinose minimum medium in the absence (raffinose) or presence of 100 M BPS (raffinose ϩ BPS). Autoradiography exposure times were 2 h for YOR382w, YOR383c, and YDR534c probes, 15 h for FET3, ISU1, and YLR136c probes, 6 h for SOD1, and 3 days for SOD2 and HSP30 probes. In YDR534c Northern blot, the band of higher mobility results from cross-hybridization with YOR383c.
FIG. 2. Glycerol growth and mitochondrial DNA stability in wild-type and ⌬YFH1 strains with an AFT1 gene deletion. Cell growth was analyzed for the four spores A, B, C, and D of a tetrad (tetratype). a, an iron concentration gradient was formed by dropping 50 l of 1 M FeSO 4 on a filter paper in the center of a Petri dish containing glycerol-rich medium, and for each strain four 10-l drops of confluently growing cells were deposited on the plate. b, the spores were grown for 16 h in liquid glucose-rich medium in the presence of increasing concentrations of FeSO 4 . Cells were spread for single colonies and the small colonies were scored after 4 days at 30°C. For each iron concentration 500 -1000 colonies were counted in ⌬YFH1 and ⌬AFT1⌬YFH1 strains and ϳ100 colonies were counted in wild-type and ⌬AFT1 strains. of protein, indicating that a large part of the iron contained in ferroxiamine B was imported by the high affinity iron uptake system. Introduction of the ⌬YOR382⌬YOR383⌬YDR534 deletion in the ⌬YFH1⌬FET3 strain slightly increased cellular growth on 100 M ferroxiamine B (Fig. 4C, third and fourth rows). Altogether, these data suggest that the GPI-anchor proteins facilitate siderophore-mediated iron uptake both in a FET3-dependent and -independent manner.
YLR136c bears homology to mammalian TIS, an RNA-binding protein that promotes deadenylation and degradation of the tumor necrosis factor-␣ mRNA, thereby, regulating tumor necrosis factor-␣ mRNA half-life (41,42). TIS and tumor necrosis factor-␣ genes are induced by the same agents. By analogy, we reasoned that mRNA half-life of the iron regulon genes might be decreased by YLR136c. As expected for an iron-replete glucose-rich medium the expression of YOR382w was low, yet, it was slightly higher in the ⌬YLR136 mutant than in the wildtype strain (Fig. 5, No BPS). In the presence of BPS which results in iron deprivation and Aft1p activation, the expression of YOR382w was induced but the mRNA levels were significantly higher in the ⌬YLR136 strain (Fig. 5, ϩBPS). Some increase in FET3 mRNA levels was also observed in the ⌬YLR136 mutant (Fig. 5). By analogy with the mammalian TIS protein, these data suggest that the mRNAs of some genes of the iron regulon could be less stable in the wild-type than in the ⌬YLR136 strain. YLR136c could act by binding either directly to iron regulon mRNAs or to another factor regulating the expression of the iron regulon. It must be noted that the expression of AFT1 or YFH1 was not modified in a ⌬YLR136 mutant (data not shown).
Expression of the Iron Regulon in Glycerol Grown Cells-We have previously reported that the phenotypes observed in null yfh1 mutants are strongly dependent on the laboratory strain (7). In contrast with several other S. cerevisiae strains, W303-1B⌬YFH1 does not significantly lose mitochondrial DNA and grows well in a rich medium containing glycerol, an obligatory respiratory carbon source. In the presence of glycerol, W303-1B and W303-1B⌬YFH1 strains had similar high respiration rates, and moreover, the mutant cells did not accumulate free iron in mitochondria (data not shown). In addition, we found that in both strains the expression of typical members of the iron regulon such as FET3, YOR382w, YOR383c, and to a much lesser extent, YLR136c, was high (Fig. 1, lanes 3 and 4, and see  below, Fig. 6).
We analyzed the evolution of FET3 and YOR382w gene expression after a shift from glycerol to fresh glucose and glycerol media. At the time of the shift, cells were in prestationary phase of growth with low levels of FET3 and YOR382w mRNAs both in wild-type and ⌬YFH1 strains (Fig. 6,  time zero). After wild-type cell transfer to glucose medium YOR382w expression decreased slowly (Fig. 6B) while FET3 expression decrease was preceded by a transient increase (Fig.  6C). In contrast, a rapid induction of these genes that was complete within 2 h was observed in the wild-type and ⌬YFH1 strains incubated in the presence of glycerol (Fig. 6, A and D), and in the ⌬YFH1 strain cultivated in the presence of glucose (Fig. 6, A and C). These data show that both in wild-type and ⌬YFH1 strains grown on glycerol Aft1p encounters low iron concentrations, implying that little iron is extruded from mitochondria.

DISCUSSION
Mitochondria are the subcellular compartment which utilizes most of the cellular iron, mainly in the heme of the cytochromes and in iron-sulfur cluster proteins. Mitochondria are also the site of iron-sulfur cluster biosynthesis (for reviews, see Refs. 13 and 14). Using DNA microarray hybridization analysis in a frataxin-deficient mutant we found that all known iron sources were mobilized in an AFT1-dependent manner, implying that cells are starved for iron (Fig. 7). Thus, we have substantially extended previous findings showing that YFH1 deletion elicits increase in the expression of the high affinity iron uptake genes, FET3 and FTR1 (4,8). Our data suggest that iron is not only actively mobilized from the external me-  Fig. 2 were grown for 5 h in glucose-rich medium to an OD of 0.7 and total RNA was extracted and hybridized with appropriate probes. Autoradiography exposure times were 5 h for YOR382w and 1 day for YLR136c and HSP30. dium via the high affinity iron uptake system and siderophore transporters but also from intracellular compartments (Fig. 7). Iron could be mobilized from the vacuole, a presumed iron storage compartment (48), via the Fth1p/Fet5p (28) and Smf3p (37) transporters. YLR205c, a predicted heme oxygenase, could also provide cells with iron, in agreement with the observation that this enzyme is induced in mammalians and plants during iron deprivation (49,50). The altered iron homeostasis resulting from frataxin deficiency might be associated with a more general redistribution of ion fluxes since the expression of several genes involved in copper, manganese, or cobalt homeostasis is significantly increased in a ⌬YFH1 strain. A possible explanation would be that iron starvation increases the expression of those metal transporters with broad specificity that are also able to transport iron (36). Consequently, heavy metal uptake and homeostasis would be altered.
However, the most impressive increase in gene expression concerned YOR382w, YOR383c, and YDR534c, three predicted cell-wall GPI-anchor protein genes of unknown function that are regulated by AFT1. The triple deletion of these genes results in a less efficient utilization of ferroxiamine B, a hydroxamate-type siderophore (23,24). Thus, a ⌬YFH1 strain harboring the triple ⌬YOR382⌬YOR383⌬YDR534 deletion accumulates less mitochondrial iron and grows better in the presence of high ferroxiamine B concentrations than a simple ⌬YFH1 strain. More work will be necessary to find the specific conditions under which these GPI-anchor proteins become important and which function they have.
The iron regulon is regulated at the transcriptional level by AFT1. Our data suggest that it could also be regulated at the post-transcriptional level by YLR136c. YLR136c deletion elicits increased mRNA levels of YOR382w, YOR383c, and to a lesser extent, FET3. By analogy with the mammalian TIS protein which regulates the mRNA half-life of specific genes by promoting RNA deadenylation and degradation (41,42), it is possible that YLR136c is a post-transcriptional regulator of the iron regulon genes, perhaps acting as a safeguard gene ensuring a faster response to new iron conditions. Iron homeostasis has mostly been studied in cells cultivated in the presence of glucose. Under these conditions, Aft1p encounters high iron levels released from mitochondria and the expression of the iron regulon is low. In contrast, we found that in glycerol-grown wild-type cells with high respiratory activity, the iron regulon is highly expressed, implying that Aft1p encounters low iron levels and that little iron is released from mitochondria (8). Moreover, glycerol-grown ⌬YFH1 cells show high respiration and no mitochondrial iron accumulation. These data indicate that frataxin has a less essential role in glycerol-grown cells, when respiratory activity is high. It has In a frataxin-deficient strain mitochondrial free iron accumulates and cannot be extruded from mitochondria, so that Aft1p encounters a low iron signal and stimulates the transcription of genes encoding the plasma membrane iron and siderophore transporters, the heme oxygenase (YLR205c) and the vacuolar iron transporter (FET5/FTH1). The resulting increased iron flux in the cytosol allows more iron to be imported in mitochondria. In cells grown in glucose-rich medium, frataxin deficiency is associated with defects in cytochromes and ironsulfur proteins. The symbol (?) indicates that a direct role for frataxin in heme and iron cluster pathways is still questionable. To simplify the scheme, ferric reductases have not been shown and siderophore transporters have been localized to the plasma membrane even though there is good evidence that Arn1p and ARN3p localize to intracellular vesicles (25,26). recently been reported that yeast frataxin is an iron-binding protein (12). The authors have found that in yeast mitochondria frataxin is present as monomers that contain no iron and multimers of high molecular weight which contain more than 16 iron atoms per frataxin molecule. An interesting hypothesis would be that in glycerol-grown cells with high respiratory needs, iron is mainly used for mitochondrial heme and ironsulfur cluster biosynthesis and does not bind much to frataxin. Conversely, in glucose-grown cells with low respiration and low iron needs, frataxin would be essentially present as iron binding multimers, mediating iron export to the cytosol. This favors the idea that frataxin controls iron partitioning between mitochondria and cytosol to regulate mitochondrial iron supply. A recent report suggests that the cAMP-dependent protein kinase Tpk2 plays a role in this regulation (51) by connecting respiratory and iron metabolism pathways. However, mitochondrial control seems restricted to genes responsible for iron mobilization. Ribonucleotide reductase, ferrochelatase, or Sur2p and Erg25p desaturases, are maintained outside of this regulatory pathway.
Although the strongest signals were obtained for the iron regulon genes, the expression of several other genes was altered in a ⌬YFH1 strain. We can mention the mating pathway, sulfur metabolism, protein recycling in the vacuole and proteasome, and a number of genes of unknown function, such as the seven-transmembrane helix protein Hsp30p, an opsin-related protein (29). These data point out to new unexpected links between frataxin deficiency and remote metabolic pathways. Surprisingly, none of those genes that protect cells against oxidative stress, and in particular the mitochondrial manganese superoxide dismutase SOD2 gene, were significantly induced in our ⌬YFH1 strain. Currently, we have no satisfactory explanation but lack of SOD2 induction might contribute to the dysfunction of iron-overloaded mitochondria.
No AFT1 homologue has been found in humans. However, human genes involved in iron metabolism are post-transcriptionally regulated by cytosolic iron regulatory proteins which are iron sensor proteins (for a review, see Ref. 52). It can be predicted that in humans also frataxin deficiency results in cytosolic iron deprivation and this opens the possibility that Friedreich ataxia causes a profound alteration in ionic homeostasis.