Regulation of Mitochondrial Iron Accumulation by Yfh1p, a Putative Homolog of Frataxin

See allHide authors and affiliations

Science  13 Jun 1997:
Vol. 276, Issue 5319, pp. 1709-1712
DOI: 10.1126/science.276.5319.1709


The gene responsible for Friedreich's ataxia, a disease characterized by neurodegeneration and cardiomyopathy, has recently been cloned and its product designated frataxin. A gene inSaccharomyces cerevisiae was characterized whose predicted protein product has high sequence similarity to the human frataxin protein. The yeast gene (yeast frataxin homolog, YFH1) encodes a mitochondrial protein involved in iron homeostasis and respiratory function. Human frataxin also was shown to be a mitochondrial protein. Characterizing the mechanism by whichYFH1 regulates iron homeostasis in yeast may help to define the pathologic process leading to cell damage in Friedreich's ataxia.

Friedreich's ataxia (FRDA) is an autosomal recessive degenerative disease characterized by progressive gait and limb ataxia, signs of axonal sensory neuropathy, pyramidal weakness of the legs, and dysarthria (1). A hypertrophic cardiomyopathy is found in almost all affected individuals (1, 2). Diabetes mellitus accompanied by a loss of pancreatic β cells is seen in about 10% of the cases, carbohydrate intolerance is present in an additional 20%, and all patients show abnormal insulin secretion in response to amino acid stimulation (3).

The molecular defect in FRDA was identified as a deficiency of frataxin, a small protein (210 amino acids) whose function could not be identified by amino acid sequence analysis (4). Frataxin deficiency may occasionally be traced to nonsense, splice site, or missense point mutations, but the primary cause is the hyperexpansion of a polymorphic GAA trinucleotide repeat situated in the first intron of the corresponding gene, which results in a marked reduction in the steady-state level of mature frataxin mRNA (4, 5). RNA analysis of frataxin expression in adult human tissues (4) and in situ hybridization studies in the mouse (6) suggested a direct correlation between the pattern of degeneration observed in the disease and the sites of frataxin transcription, which is highest in heart, spinal cord, and dorsal root ganglia. A recent study proposed that frataxin was encoded in a larger transcriptional unit than originally thought, and that its gene was in fact part of the neighboring STM7 gene, which was transcribed in the same direction and encoded a phosphatidylinositol 4-phosphate 5-kinase (7). New evidence suggests that the inclusion ofSTM7 sequence as part of the FRDA gene was based on polymerase chain reaction (PCR) conditions that amplified rare and illegitimate transcripts not usually detected in vivo (5). All of the FRDA mutations identified to date are located in the original 3′ “frataxin” region of the gene (4, 5).

Genes encoding predicted proteins with high sequence similarity to frataxin have been identified in mouse, rat, Caenorhabditis elegans, and an anonymous open reading frame (YDL120w) inSaccharomyces cerevisiae (6, 8). The predicted yeast frataxin homolog was independently identified as a multicopy suppressor of a mutant (bm-8) that was unable to grow on iron-limited medium. The bm-8 mutant was transformed with a genomic library, and a clone was identified that partially overcame the low-iron growth deficit (9). The region responsible for the complementing activity was identified by subcloning the open reading frame YDL120w on chromosome IV (10). BLAST analysis of this sequence identified two proteins with high sequence similarity, the predicted product of an anonymous open reading frame in the C. elegans genome and human frataxin. We determined thatYFH1 was not allelic to bm-8. First, no complementation was seen when bm-8 was transformed withYFH1 on a centromeric plasmid. Second, allelic segregation demonstrated that a LEU2 marker integrated next to the chromosomal copy of YFH1 segregated away frombm-8 (11). Third, the gene allelic tobm-8 was cloned by genomic complementation and the open reading frame (YMR134w) localized to chromosome XIII. Although these results demonstrate that YFH1 was not responsible for the phenotype of bm-8, the fact that overexpression ofYFH1 alleviated the bm-8 low-iron growth defect suggested that it was involved in iron metabolism.

To define the function of YFH1, we disrupted the chromosomal copy by inserting the HIS3 auxotrophic marker into the open reading frame (12). Disruptants were selected and deletion of YFH1 confirmed by Southern (DNA) analysis. The Δyfh1 (yfh1:: HIS3) strain showed a severe growth deficit on fermentable carbon sources regardless of iron concentration. The deletion strain was unable to grow on rich medium containing glycerol and ethanol (YPGE) as the carbon source, suggesting that Δyfh1 was unable to carry out oxidative phosphorylation. The disruptant was also unable to grow on any media at 37°C. The respiratory incompetence ofΔyfh1 was confirmed by demonstrating a severe reduction in oxygen consumption, even in cells grown in rich medium (13). We hypothesized that the Δyfh1cells were unable to grow on fermentable carbon sources as a result of the generation of rho mutants. Theserho mutations are characterized by defects in or loss of mitochondrial DNA and an inability to carry out oxidative phosphorylation. Lack of growth on respiratory substrates in Δyfh1 was shown to be the result of a loss of mitochondrial DNA. Diploids generated from a cross betweenΔyfh1 and wild-type cells were respiratory competent, whereas diploids from Δyfh1 and arho tester strain (rhoo ) were unable to grow on respiratory substrates. Therefore, once mitochondrial damage has occurred in Δyfh1, the defect is cytoplasmically inherited and is not due to a second nuclear mutation.

These data suggest that loss of the YFH1 gene results in a loss of mitochondrial function. To test this hypothesis, we transformed a yfh1::HIS3/YFH1 diploid with a plasmid that contained the entire YFH1 gene and URA3 as a selectable marker. All of the diploids were capable of growth on YPGE. When the diploids were sporulated, tetrads segregated such thatyfh1::HIS3 cells bearing the plasmid grew well on all media, including YPGE. When the plasmid was cured by growth on 5-fluoroorotic acid, the Δyfh1 haploids lost the ability to use respiratory carbon sources and the slow-growth phenotype was observed on all other media. Transformation of these cells with a wild-type copy of YFH1 restored growth on YPD medium but did not restore the ability to grow on YPGE.

Taken together, these data indicate that the Yfh1 protein (Yfh1p) is involved in iron homeostasis and mitochondrial function. PSORT analysis of the protein sequence of Yfh1p predicts a mitochondrial targeting sequence (14). To localize Yfh1p within the cell, we fused green fluorescent protein (GFP) to the COOH-terminus of Yfh1p expressed under the control of the methionine promoter (15). Plasmids bearing the Yfh1p-GFP fusion construct were transformed intoΔyfh1 cells. The construct was able to complement fully the slow-growth phenotype of the disruptant, indicating that the fusion protein was functional in vivo. Fluorescence microscopy demonstrated that the protein was localized to the mitochondria. Co-staining of cells for porin (a mitochondrial outer membrane protein) and Yfh1p-GFP indicated that Yfh1p-GFP was not localized in the outer membrane (Fig. 1).

Figure 1

Distribution of fluorescence in cells transformed with a Yfh1p–green fluorescent protein construct. (A) Localization of Yfh1p-GFP to the mitochondria. A wild-type strain (DY150) was transformed with the Yfh1p-GFP fusion construct, and cells were grown in YPGE medium lacking methionine, fixed, and examined by fluorescence microscopy. (B) Immunodetection of porin, a mitochondrial outer membrane protein. Mouse antibody to porin was used as a primary antibody (1:500 dilution), and Texas-Red goat antibody to mouse immunoglobulin G was used as a secondary antibody (1:200 dilution). (C) Overlay of the Yfh1p-GFP fluorescence and the porin immunostain, illustrating the different locations of the two proteins within the mitochondria. (D) Wild-type cells transformed with GFP alone as a control (14). Cells were grown on synthetic medium lacking methionine and containing glycerol and ethanol as the carbon source. The fixed cells were attached to poly-l-lysine–treated cover slips and examined by confocal laser fluorescence microscopy on a Bio-Rad MRC 600 microscope with a Nikon Optishot camera. The scale bar in (B) represents 1 μm.

Because overexpression of YFH1 alleviated the low-iron growth phenotype of bm-8, we examined aΔyfh1 strain for perturbations in iron metabolism. The entry of iron through the plasma membrane into S. cerevisiae grown in iron-rich medium is mediated by low-affinity transport systems. Normally, transcription of the high-affinity iron transport system, which consists of a ferroxidase (Fet3p) and permease (Ftr1p), is not detected if cells are iron replete (9, 16). In Δyfh1 cells a marked induction of the high-affinity iron transport system was observed even when cells were grown in iron-rich medium (Fig. 2A). Northern (RNA) analysis showed a 10- to 50-fold induction of FET3 andFTR1 transcripts in Δyfh1 cells grown in YPD compared with wild-type cells, confirming the induction of the high-affinity transport system even in iron- replete conditions (Fig. 2B). As a control, mRNA was isolated from the AFT1-1up strain grown in high- and low-iron media. This strain shows constitutive expression of the components of the high-affinity iron transport system due to a mutation in the transcriptional regulator AFT1 (17). As shown in Fig. 2B, Δyfh1 had the same level of FET3and FTR1 expression in YPD as did an AFT1-1upstrain or a maximally induced wild-type strain.

Figure 2

Constitutive induction of the high-affinity iron transport system inΔyfh1 cells. (A) High-affinity iron uptake into whole cells. Wild-type (DY150) andΔyfh1 cells were grown in YPD or in iron-limited media [BPS(0)] for 6 hours. The cells were washed and assayed for 59Fe transport by incubation for 10 min at 30°C in assay medium that contained 0.5 μM 59Fe in the presence of 1.0 mM ascorbate as described (9). The data (mean ± SEM) are representative of three independent experiments. (B) Northern blot analysis of RNA extracted from DY150 (lanes 1 and 4), Δyfh1 (lanes 2 and 5), and AFT1-1up cells (lanes 3 and 6) showing a constitutive induction of the components of the high-affinity iron transport system. Cells were grown in YPD (lanes 1 to 3) and BPS(0) (lanes 4 to 6), respectively, as described in (A). Northern analysis was performed and the blots probed with FET3,FTR1, and actin as described (29).

An increased rate of iron uptake, coupled with a slower rate of cell division, should be reflected by an increase in cellular iron content. Measurement of cellular iron content by atomic absorption spectroscopy revealed a doubling of iron content inΔyfh1 cells compared with wild-type cells (Fig.3A). Purified mitochondria from Δyfh1cells showed an iron concentration 10 times that in the mitochondria of wild-type cells. No differences were observed in the mitochondrial content of copper or calcium. Cellular and mitochondrial iron were also assayed in AFT1-1up cells. This latter strain shows intracellular iron concentrations comparable with those of theΔyfh1 strain (Fig. 3A). The fact that AFT1-1up cells had high intracellular iron but showed neither an increase in mitochondrial iron (Fig. 3B) nor a respiratory growth defect indicated that the accumulation of mitochondrial iron was specifically associated with the deletion of YFH1 and was not simply a consequence of increased cellular iron. Accumulation of mitochondrial iron in Δyfh1 renders this strain hypersensitive to oxidative stress, as demonstrated by its sensitivity to H202 (Fig. 3C).

Figure 3

Deletion of YFH1causes an increase in cellular and mitochondrial iron accumulation. (A) Analysis of cellular iron content in wild-type (DY150),Δyfh1, and AFT1-1up cells. Cells were grown in galactose medium and washed, and samples were analyzed for metal content. The data (mean ± SEM) are representative of at least three independent experiments. (B) Analysis of purified mitochondria for protein and metal content. Cells were grown in YP medium containing galactose as the carbon source. Mitochondria were isolated by differential centrifugation techniques (30). Mitochondrial fractions were applied to a 15% percoll gradient, spun at 30,000 rpm in a Ti75 rotor for 27 min. Fractions enriched for mitochondria were separated from percoll by centrifugation at 55,000 rpm in a Ti75 rotor for 90 min and atomic absorption analyses were performed on a Perkin-Elmer model 305A instrument. Protein determinations were performed with the BCA procedure (Pierce, Rockford, Illinois) using bovine serum albumin (fraction V) as the standard. Data are normalized to the protein content of the samples. (C) Sensitivity ofΔyfh1 cells to growth on H2O2. Cells (103/10 μl) were spotted on YPD or YPD containing 0.004% or 0.008% H2O2 and allowed to grow for 3 days at 30°C.

Our data indicate that YFH1 encodes a mitochondrial protein that, when deleted, results in the accumulation of mitochondrial iron at the expense of cytosolic iron. This increase in iron compromises mitochondrial function and results in hypersensitivity to oxidative stress, presumably as a result of iron-catalyzed Fenton chemistry. The sensitivity of Δyfh1 cells to oxidative stress may explain why rho mutations are generated at such high frequency in this strain. Inhibition of oxidative phosphorylation may reduce the production of reactive oxygen species, such as superoxide, and attenuate the effects of mitochondrial iron accumulation.

Several lines of evidence are consistent with a mitochondrial defect underlying the etiology of FRDA. Mitochondrial dysfunction is found in several degenerative ataxias (18), and damage to mitochondria is also a strong apoptotic signal in neuronal cells (19). Variable abnormalities of several mitochondrial enzyme activities have been reported in FRDA patients (20), as well as perturbations in heme biosynthesis (21) and hypersensitivity of cultured cells to ionizing radiation (22). Human frataxin was also localized to the mitochondria, as shown by fluorescence microscopy analysis of cultured cells transfected with human frataxin–GFP expression constructs (Fig.4). Mitochondrial dysfunction from iron deposition may account for the biochemical and neurological lesions seen in FRDA. Iron deposits have been found in the myocardium of FRDA patients, even in cells not yet showing signs of degeneration (23). Tissues most affected in FRDA are neurons and cardiac cells, which are postmitotic tissues that largely depend on efficient oxidative energy metabolism (24, 25). However, this cannot fully explain the specificity of pathological involvement in FRDA, because many neuronal types, spinal cord and brainstem motor neurons for instance, are notably spared. Affected tissues appear to be those normally expressing the highest frataxin levels (4, 6). In addition, vitamin E deficiency in humans, either genetic or acquired, leads to a clinicopathological picture very similar to FRDA and sometimes indistinguishable (26). It is possible that frataxin, by stimulating iron transport out of the mitochondria, and the antioxidant vitamin E, which localizes in mitochondrial membranes (27), cooperate to protect certain cell types from mitochondrial oxidative damage. There may be multiple protective antioxidant systems in humans, so other cell types may rely on different mechanisms and be unaffected by frataxin or vitamin E deficiency. Redundancy of these systems and the presence of residual frataxin in patients may in part explain the less dramatic human phenotype compared with the yeast knockout model. So far, no FRDA patients carrying null mutations in both alleles of the frataxin gene, and therefore completely lacking frataxin, have been identified. Genotype-phenotype correlation studies, however, have shown that larger GAA expansions are associated with lower residual frataxin mRNA levels and an increase in the onset and severity of the disease (28).

Figure 4

Mitochondrial localization of human frataxin in live mammalian cells. Expression vectors were constructed containing sequences encoding the entire proposed human frataxin protein fused to a modified GFP tag (Clontech). Cos7 cells transfected separately with these constructs were incubated 15 min with 50 nM mitochondrial-specific MitoTracker Orange stain (Molecular Probes) 24 hours after transfection and then immediately photographed with a Zeiss MC100/Axioskop fluorescent microscope camera system with the appropriate filter sets. (A and C) Construct with GFP at the NH2-terminus. (B andD) Construct with frataxin at the NH2-terminus. The frataxin-GFP fusion protein with frataxin at the NH2-terminus was targeted to organellar structures matching the appearance and distribution of mitochondria as revealed by MitoTracker staining. The frataxin-GFP fusion protein with GFP at the NH2-terminus generated a diffuse cytosolic fluorescence, consistent with the masking by GFP of a frataxin NH2-terminal mitochondrial targeting sequence.

Characterization of Yfh1p, the yeast frataxin homolog, has led to the identification of an in vivo function that may explain the pathophysiology observed in FRDA. Identification of other proteins that interact with Yfh1p will provide a clearer understanding of mitochondrial iron homeostasis and the etiology of FRDA.

Note added in proof: Expression of frataxin-GFP in the Δyfh1 strain resulted in localization of the human protein to yeast mitochondria but was unable to complement the mutant phenotype. This result suggests that frataxin interacts specifically with mitochondrial proteins to effect iron transport and that despite strong similarity of function, proteins of the yeast and human systems are not interchangeable.


View Abstract

Navigate This Article