Co-precipitation of Phosphate and Iron Limits Mitochondrial Phosphate Availability in Saccharomyces cerevisiae Lacking the Yeast Frataxin Homologue (YFH1)*

Saccharomyces cerevisiae cells lacking the yeast frataxin homologue (Δyfh1) accumulate iron in the mitochondria in the form of nanoparticles of ferric phosphate. The phosphate content of Δyfh1 mitochondria was higher than that of wild-type mitochondria, but the proportion of mitochondrial phosphate that was soluble was much lower in Δyfh1 cells. The rates of phosphate and iron uptake in vitro by isolated mitochondria were higher for Δyfh1 than wild-type mitochondria, and a significant proportion of the phosphate and iron rapidly became insoluble in the mitochondrial matrix, suggesting co-precipitation of these species after oxidation of iron by oxygen. Increasing the amount of phosphate in the medium decreased the amount of iron accumulated by Δyfh1 cells and improved their growth in an iron-dependent manner, and this effect was mostly transcriptional. Overexpressing the major mitochondrial phosphate carrier, MIR1, slightly increased the concentration of soluble mitochondrial phosphate and significantly improved various mitochondrial functions (cytochromes, [Fe-S] clusters, and respiration) in Δyfh1 cells. We conclude that in Δyfh1 cells, soluble phosphate is limiting, due to its co-precipitation with iron.

Friedreich ataxia is the most common hereditary recessive ataxia. The gene responsible for this neurodegenerative disease was cloned in 1996 (1). The corresponding protein (frataxin) is a small, highly conserved mitochondrial protein.
It is now generally accepted that frataxin (Yfh1 in yeast) is one of the components of the mitochondrial [Fe-S] cluster machinery, acting either as an iron donor or more generally as a chaperone interacting with the desulfurase/scaffold proteins involved in [Fe-S] assembly (2,3). However, it has proved to be more difficult to determine the precise role of frataxin than of other components of the [Fe-S] cluster machinery, partly because the phenotypes associated with a lack of frataxin are both very diverse and differ according to the cellular model studied (for recent reviews, see Refs. 4 and 5). Indeed, the absence of frataxin results in direct or indirect defects in many pathways other than [Fe-S] biogenesis, including heme synthesis (6,7), oxidative stress defense (8 -10), apoptosis (11), and mitochondrial iron homeostasis (12). One major difficulty in studying the role of frataxin in the yeast model is to distinguish phenotypes that specifically result from the lack of frataxin from phenotypes shared by several mutants affected in [Fe-S] cluster biogenesis (13).
Frataxin-deficient yeast cells, like other mutants defective in [Fe-S] cluster biogenesis, accumulate huge amounts of iron in their mitochondria (14). The mechanisms of iron import, export, and storage in the mitochondria are still very poorly understood. Several proteins in Saccharomyces cerevisiae mitochondria have been proposed as candidate iron transporters. The proteins include the paralogous proteins Mmt1 and Mmt2 (15), and the paralogous mitochondrial carrier proteins Mrs3 and Mrs4 (16). The Mrs proteins (homologous to the mammalian mitoferrin) are the best candidates for the function of mitochondrial iron import (17)(18)(19). However, the mechanism of uptake is still completely unknown and will probably remain so as long as the iron substrate used by these proteins in vivo is not known. In ⌬yfh1 mitochondria, iron accumulates in the form of amorphous nanoparticles of ferric phosphate (6), and there is now evidence that the same is true for other mutants defective in [Fe-S] cluster biogenesis or export (13,20,21). Iron precipitated in this form is unavailable for biological processes, and thus, ⌬yfh1 cells might paradoxically suffer from iron deficiency despite being overloaded with iron (13). The effects of frataxin deficiency on iron metabolism in yeast have been studied extensively (reviewed in Ref. 4). But, as iron precipitates in the form of ferric phosphate in the mitochondria of ⌬yfh1 cells, the physiological consequences of iron accumulation/precipitation in these cells probably involve dysfunctions in both iron and phosphate metabolism. Like iron, the precipitated phosphate is probably unavailable for biological processes. Frataxin-deficient cells are very often studied with regards to disruption in iron homeostasis but were never studied with regards to disruption in phosphate homeostasis. However, the question of ferric phosphate accumulation in the mitochondria of some mu-tants can obviously be studied from both ends: iron and phosphate. In the present study, we decided to study this issue by focusing on phosphate rather than iron. Inorganic phosphate, which is vital for oxidative phosphorylation, is transported to the mitochondrial matrix by two partly redundant carriers, Mir1 and Pic2 (22,23). Interestingly, Mir1 is also involved in protein import into mitochondria (24), and Pic2 can reversibly change its transport mode (antiport-uniport) according to the redox conditions (25). More generally, phosphate is involved in the homeostasis of several cations in different cellular compartments (26). Co-precipitation of phosphate with iron in ⌬yfh1 mitochondria is thus expected to have physiological implications. We therefore investigated the effects of phosphate precipitation in mitochondria of frataxin-deficient yeast cells. As far as we are aware, this is the first report addressing frataxin deficiency from this particular point of view.

EXPERIMENTAL PROCEDURES
Yeast Strains and Growth Conditions-The strains used in this study were YPH499 (wild-type; MATa ura3-52 lys2-801 ade2-101 trp1-⌬63 his3-⌬200 leu2-⌬1 cyh2), YPH499⌬yfh1 (⌬yfh1; ⌬yfh1::TRP1), YPH499⌬mrs3⌬mrs4 (mrs4::kanMX4, mrs3::URA3), YPH499⌬ggc1 (ggc1::kanMX4)), and ERyfh1 (13). To avoid the accumulation of suppressor mutations in frataxin-deficient cells, the yfh1 deletion was covered by a shuffle plasmid bearing a wild-type copy of YFH1 and of CYH2. The covering plasmid was ejected before each experiment by plating the cells on YPD and cycloheximide, as described previously (6). Cells overexpressing MIR1 were transformed with the 2 plasmid Yep352-MIR1. Unless otherwise stated, cells were grown aerobically in either complete medium (YP-raffinose) or defined medium (6.7 g/liter yeast nitrogen base without amino acids, iron, or copper and containing 0.1% glucose, 2% raffinose, 0.8 g/liter amino acids). The defined media were supplemented with 200 mg/liter adenine, 5 M CuSO 4 , and with various amounts of iron, in the form of ferric citrate (1:20) or ferrous ascorbate (1:10). In some experiments, the cells were grown in defined medium without phosphate (yeast nitrogen base without phosphate), to which various amounts of phosphate was added as K 2 HPO 4 / KH 2 PO 4 (pH 6.5). For conditional expression of YFH1 by the ERyfh1 strain, the cells were grown in defined medium containing different concentrations of estradiol (0 -2.5 nM). The steady-state amount of Yfh1 produced by the cells was proportional to the concentration of estradiol in the medium and was estimated by quantitative immunoblotting (13).
Cell Fractionation-Mitochondrial and cytosolic fractions were isolated after treatment of cells with zymolyase, followed by lysis of the protoplasts in 0.6 M sorbitol (buffered with 50 mM Tris, pH 7.8), in the presence of protease inhibitors (protease inhibitor mixture, P8215, Sigma) as described previously (27). The purity of the mitochondrial fraction was estimated by assaying the activity of cytochrome oxidase as described previously (27). Submitochondrial fractionation was carried out as described previously (28).
Iron and Phosphate Uptake by Isolated Mitochondria-Mitochondria were suspended at 1 mg/ml in 0.6 M sorbitol buffered with 50 mM HEPES, pH 7, and preincubated at 30°C for 15 min with 1 mM NADH. For iron uptake experiments, 55 Fe(II)-ascorbate (1:2) was added at a final concentration of 10 M. At various time intervals, 100-l aliquots were withdrawn and added to ice-cold 0.6 M sorbitol buffered with 50 mM HEPES, pH 7, containing 1 mM unlabeled Fe(II)-ascorbate. The aliquots were washed three times by centrifugation before scintillation counting. For phosphate uptake experiments, the mitochondria were preincubated for 15 min at 30°C in the same buffer containing 1 mM NADH and 10 M Fe(II)-ascorbate, and then 0.5 mM [ 32 P]phosphate (as a mix of 1:1 sodium and potassium orthophosphate, pH 7) was added. Aliquots (100 l) were withdrawn at intervals and added to ice-cold buffer (sorbitol-HEPES) containing 50 mM unlabeled phosphate. The aliquots were washed three times by centrifugation before scintillation counting.
Determination of Total, Soluble, and Insoluble Iron and Phosphate in Isolated Mitochondria-The phosphate content of isolated mitochondria (without prior incubation with [ 32 P]phosphate) was determined using a colorimetric molybdate assay (29). Total phosphate was estimated after disrupting whole mitochondria in 10 mM HEPES buffer, pH 7, containing 0.1% SDS. Soluble phosphate was measured as follows; the mitochondria were lysed in hypotonic buffer (10 mM HEPES, pH 7) and sonicated for 30 s; the insoluble fraction was removed by centrifugation at 10,000 ϫ g for 30 min, and the phosphate content of the soluble fraction was measured by the same colorimetric method. Alternatively, the phosphate content and distribution in the mitochondria was determined by scintillation counting after growing the cells in the presence of [ 32 P]orthophosphate. Total, soluble, and insoluble iron ( 55 Fe) contents in whole mitochondria, in the intermembrane space, and in mitochondrial matrices were measured by scintillation counting of the corresponding fractions.
Iron Uptake by Whole Cells as a Function of Phosphate Concentration in Medium-The cells were precultured in defined medium and then transferred to defined medium without phosphate and cultured for a further 15 h. The cells were then harvested, resuspended in the same medium without phosphate at an A 600 nm of 0.01 and distributed into the wells of a microtiter plate (100 l/well). Phosphate (K 2 HPO 4 /KH 2 PO 4 , pH 7) was added to the wells at various concentrations (0.1 mM-0.2 M), and then iron was added as 55 Fe(II)-ascorbate (1:10). The cells were grown at 30°C in the microtiter plate with shaking. After 24 h of growth, the A 600 nm was recorded, and the cells were harvested and washed on the filter of a cell harvester (Brandel) and subjected to scintillation counting.
Miscellaneous-The in-gel aconitase assay was performed as described previously (32). The respiratory activity of isolated mitochondria was evaluated by an oxypolarographic method. The rate of oxygen consumption was measured with a 1-ml thermostatically controlled oxypolarographic cell equipped with a Clark-type electrode. Low temperature absorption spectra (Ϫ191°C) of whole cells were recorded as described previously (33,34).

Increased Iron Uptake by Isolated ⌬yfh1
Mitochondria-Iron is normally stored in the vacuole in yeast (27). Yeast mutants affected in [Fe-S] cluster biogenesis accumulate iron in their mitochondria, and the mechanism by which iron mislocalizes in these mutants is unknown (14). We investigated whether this phenotype of mitochondrial iron accumulation can be detected in vitro; we measured iron uptake (from ferrous ascorbate) by mitochondria isolated from different yeast strains (Fig. 1). Mitochondria isolated from a ⌬yfh1 strain accumulated more iron than wild-type mitochondria. This was also true, although to a lesser extent for mitochondria isolated from another mutant affected in [Fe-S] cluster biogenesis, ⌬ggc1 ( Fig. 1), which lacks the GTP/GDP mitochondrial carrier (35) and that we used as a control throughout this study because it accumulates mitochondrial iron in the same form than the ⌬yfh1 mutant (13). However, mitochondria isolated from an ⌬mrs3⌬mrs4 strain accumulated similar amounts of iron as wild-type mitochondria (Fig. 1). The MRS3/4 gene products are generally considered to be the main mitochondrial iron carriers (18) so it is probable that in vitro iron uptake observed in this experimental system did not reflect the in vivo mechanism of iron acquisition by mitochondria. Moreover, Scatchard experiments did not provide evidence of any saturation for the specific uptake of iron by isolated mitochondria (Fig. 2). Despite this evidence that in vitro iron uptake did not mechanistically reflect in vivo mitochondrial iron uptake, it is striking that a major phenotype of ⌬yfh1 cells (mitochondrial iron accumulation) was observed when measuring iron uptake by mitochondria isolated from this mutant ( Fig. 1). As the mechanism of uptake was not saturable under our experimental conditions (Fig. 2), the greater uptake of iron by ⌬yfh1 mitochondria than by wild-type mitochondria probably did not result from a higher activity of the mitochondrial iron carrier(s) involved in these experiments. Another possibility is that, in ⌬yfh1 but not in wild-type mitochondria, iron loses its mobility after entering the mitochondria, thereby displacing the thermodynamic equilibrium toward increased net intramitochondrial iron accumulation. This would be consistent with the observation that, in Scatchard experiments (Fig. 2), the mean rate of iron uptake by wild-type mitochondria increased about three times when the temperature was increased from 4°C ( Fig. 2A) to 30°C (Fig. 2B). In the same conditions, the rate of iron uptake by ⌬yfh1 only doubled (Fig. 2). This suggests that one component of the iron uptake process that is temperature-independent is greater for ⌬yfh1 than wild-type mitochondria. This component may be iron precipitation inside the mitochondria. We tested this possibility by investigating the fate of iron after its uptake by mitochondria isolated from different strains.

Precipitation of Iron in ⌬yfh1
Mitochondrial Matrix-Mutants affected in [Fe-S] cluster biogenesis accumulate iron in the mitochondria essentially in an insoluble form, as ferric phosphate nanoparticles (6,13,20,21). We tested whether the apparent higher rate of iron uptake by mitochondria isolated from ⌬yfh1 cells was associated with precipitation of iron in vitro in the mitochondrial matrix. Mitochondria were incubated with 55 Fe (as ferrous ascorbate), washed, and then disrupted by sonication in hypotonic buffer. The lysates were then separated into soluble and insoluble fractions by ultracentrifugation. Iron was determined in each of these fractions (Fig. 3A). In ⌬yfh1 mitochondria, but not in wild-type mitochondria, most of the iron precipitated and became insoluble rapidly after its uptake. The same process occurred in ⌬ggc1 mitochondria, although to a lesser extent (Fig. 3A). Analysis of the submitochondrial fractions showed that iron precipitated in the mitochondrial matrix of ⌬yfh1 mitochondria (Fig.  3B). The precipitation of iron in the matrix of ⌬yfh1 mitochondria presumably results from its oxidation to ferric ions in this compartment and binding to an anion with which it forms an insoluble complex. It is plausible that the ferrous iron is oxidized by oxygen, and the resulting ferric ions precipitate with phosphate, which was shown previously to be associated with insoluble mitochondrial iron in ⌬yfh1 cells in vivo (6,13).
Role of Oxygen and Phosphate in Iron Precipitation-Iron did not accumulate in the mitochondria of ⌬yfh1 cells grown anaerobically (2 ppm of O 2 ). Mitochondria isolated from aerobically grown ⌬yfh1 cells and incubated with iron (as ferrous ascorbate) under anaerobiosis also did not precipitate iron in the matrix (data not shown). However, addition of ferrous ascorbate to a suspension of isolated ⌬yfh1 mitochondria under aerobiosis resulted in significant oxygen consumption by the mitochondria (Fig. 4) consistent with the oxidation of iron in the mitochondrial matrix. A good candidate anion for precipitating the resulting ferric iron is phosphate. We used a colorimetric method to assay the phosphate content of ⌬yfh1 and wild-type mitochondria after cell growth in complete medium (YP-raffinose). There was significantly more total phosphate but much less soluble phosphate in ⌬yfh1 mitochondria than in wild-type mitochondria (Fig. 5A). Similar results were obtained by growing the cells in complete medium containing   [ 32 P]orthophosphate and then counting soluble and insoluble phosphate in the mitochondria; the insoluble-to-soluble ratio of phosphate in the mitochondria measured by this method was 2.0 Ϯ 0.3 for the wild-type, 8.7 Ϯ 0.5 for ⌬yfh1 and 4.3 Ϯ 0.9 for ⌬ggc1 mitochondria (mean Ϯ S.E. for three experiments). The rate of orthophosphate uptake was also higher by isolated ⌬yfh1 than isolated wild-type or ⌬ggc1 mitochondria (Fig. 5B). We then checked whether phosphate precipitated in vitro in the matrix of mitochondria isolated from ⌬yfh1 cells. Isolated mitochondria were incubated with ferrous ascorbate and [ 32 P]orthophosphate. Most of the phosphate taken up by wild-type mitochondria remained soluble, whereas a significant part of the phosphate in ⌬yfh1 mitochondria was recovered in the insoluble fraction (Fig. 5C). Much less phosphate precipitated in ⌬ggc1 than in ⌬yfh1 mitochondria (Fig. 5C). These various findings suggest that iron and phosphate coprecipitate in the mitochondria of mutants defective in [Fe-S] cluster biogenesis, both in vivo and in vitro, and that this phenomenon is more extensive in ⌬yfh1 mitochondria than in mitochondria of another [Fe-S] cluster biogenesis mutant (⌬ggc1). We examined the development of iron and phosphate precipitation in vivo, in a strain with conditional expression of YFH1. For this experiment, we used the ERyfh1 in which transcription of the YFH1 gene is controlled by the concentration of estradiol in the growth medium (13). We previously showed that with 2.5 nM estradiol in the medium, the amount of Yfh1 was ϳ100 molecules per cell (i.e. Ͼ10fold less than in wild-type cells), which was enough to prevent mitochondrial iron accumulation (13). We thus analyzed ERyfh1 cells grown with estradiol in the medium at concentrations ranging from 0 to 2.5 nM (0, 0.5, 1, and 2.5 nM), leading to cellular Yfh1 contents ranging from 0 to 100 molecules per cell. At 2.5 nM estradiol, the mitochondrial phosphate and iron contents and distributions in the ERyfh1 strain were similar to the values found in the wild-type strain (data not shown). Fig. 6 shows that the insoluble-to-soluble ratio of mitochondrial iron progressively increased when the estradiol concentration in the medium was decreased from 2.5 to 0 nM. In the same conditions, the insoluble-to-soluble ratio of phosphate in the mitochondria suddenly increased (from about two to eight; Fig. 6) when the estradiol concentration was decreased from 2.5 to 1 nM, a condition in which iron accumulation was still mild, and then increased moderately when estradiol concentration was further decreased (Fig. 6). This result indicates that phosphate precipitation occurred very early, when iron just started to accumulate moderately in the mitochondria.
High Phosphate Concentration in Medium Suppresses Iron Accumulation Phenotype of ⌬yfh1 Cells-Co-precipitation of iron and phosphate in the mitochondria of ⌬yfh1 cells theoretically has several possible implications. One is that excess phosphate in the mitochondria is part of the cause of iron precipitation leading to iron accumulation in the mitochondria. In this case, it should be possible to decrease iron accumulation in the mitochondria by decreasing the cell phosphate content. Another possibility is that co-precipitation with iron reduces the amount of free phosphate available for oxidative phosphorylation in the mitochondrial matrix of ⌬yfh1 cells. Were this the case, increasing the cell phosphate content would be beneficial, by making more phosphate available for biological processes. We studied the effect of phosphate concentration in the growth medium on cell growth and on iron accumulation by the cells (Fig. 7). Increasing the phosphate concentration in the medium resulted in a substantial decrease in iron accumulation by ⌬yfh1 cells, which recovered wild-type levels of iron at a phosphate concentra-   FEBRUARY (Fig. 7A). In parallel with this effect, ⌬yfh1 cell growth was strongly stimulated by high concentrations of phosphate in the medium (Fig. 7C). This effect of phosphate was iron-dependent, being more pronounced when the medium contained excess iron (50 M) (Fig. 7, A and C), and was much larger for ⌬yfh1 cells than for another mutant defective in [Fe-S] cluster biogenesis (⌬ggc1; Fig. 7B). The beneficial effect of phosphate on ⌬yfh1 cell physiology was also evidenced by the observation that in the presence of excess phosphate in the medium (100 mM), ⌬yfh1 cells recovered some ability to grow on a nonfermentable carbon source (YPglycerol plate), indicating that the cells recovered some respiratory activity in this condition (data not shown). The effect of phosphate in the medium on mitochondrial iron accumulation was probably mostly transcriptional because the expression of genes of the iron regulon (FET3 and SIT1, regulated by Aft1) in ⌬yfh1 cells was strongly decreased when the concentration of phosphate in the medium was increased from 0.1 to 200 mM (Table 1). Table 1 also shows that MIR1 and PIC2, encoding the main mitochondrial phosphate carriers, were repressed in ⌬yfh1 cells (compared with wild-type cells) grown on low phosphate medium. Expression of these genes in ⌬yfh1 cells increased when increasing the phosphate content of the medium (Table 1), which, together with the decrease in iron accumulation by the cells, could account for the better fitness of ⌬yfh1 cells grown in phosphate-rich medium.

Iron and Phosphate in Frataxin-deficient Yeast
Overexpression of Mitochondrial Phosphate Carrier Mir1 Improves Growth and Respiratory Functions of ⌬yfh1 Cells-Our results suggest that although total phosphate is more abundant in ⌬yfh1 than wild-type mitochondria, ⌬yfh1 cells suffer from a lack of bioavailable phosphate due to its co-precipitation with iron. In agreement with this hypothesis, we observed that the expression of PHO84, encoding the high affinity phosphate transporter, was strongly increased in ⌬yfh1 cells (compared with wild-type cells) in low phosphate medium (Table 1). Mir1 is the main mitochondrial phosphate carrier in yeast, although Pic2 also contributes to mitochondrial phosphate import under stress conditions (22). We tried to manipulate the mitochondrial phosphate content/availability by overexpressing MIR1 on a multicopy plasmid in wildtype, ⌬yfh1, and ⌬ggc1 strains. When MIR1 was overexpressed, the steady-state level of soluble mitochondrial phosphate increased only moderately (23 Ϯ 4% in wild-type mitochondria and 8 Ϯ 2% in ⌬yfh1 mitochondria; mean Ϯ S.E. for three experiments). However, overexpression of MIR1 in ⌬yfh1 cells, but not that in ⌬ggc1 cells, clearly resulted in an improvement of cell growth and of the mitochondrial functions that are impaired in the frataxin-deficient mutant. Both cytochrome synthesis and [Fe-S] cluster biogenesis (as evidenced by measuring aconitase activity) were improved in ⌬yfh1 cells by overexpression of MIR1 (Fig. 8, A and B). In contrast, when MIR1 was overexpressed in ⌬ggc1 cells, the synthesis of cytochromes was impaired (cytochromes b and c 1 disappeared) (Fig. 8B), indicating that the beneficial effect of MIR1 overexpression was specific to ⌬yfh1 cells. Moreover, growth of the ⌬yfh1 cells but not of ⌬ggc1 cells on a nonfermentable carbon source was partly restored by MIR1 overexpression (Fig. 9, B and C: compare lanes 3 and 4 (⌬yfh1) and  lanes 5 and 6 (⌬ggc1)). The resistance of the cells to an oxida-   tive stress (copper) was also improved when MIR1 was overexpressed in the ⌬yfh1 mutant (Fig. 9D, lanes 3 and 4), but not in the ⌬ggc1 mutant (Fig. 9D, lanes 5 and 6). Suppressor mutations are frequent in ⌬yfh1 cells (6), so we checked that the effects of MIR1 overexpression in ⌬yfh1 cells were reversible by ejecting the plasmid bearing MIR1 (and URA3 as the selection marker) on 5-fluoroorotic acid plates. The phenotype of ⌬yfh1 cells from which the MIR1 plasmid was ejected was the same as the phenotype of untransformed ⌬yfh1 cells (data not shown). This demonstrates that the improvement of respiratory functions observed in ⌬yfh1 cells overexpressing MIR1 was not due to any suppressor mutations.

Physiological Implications of Iron Accumulation and Precipitation in ⌬yfh1
Mitochondria-Frataxin-deficient yeast cells accumulate iron in the mitochondria in the form of amorphous nanoparticles of ferric phosphate, as do other mutants defective in [Fe-S] cluster biogenesis or export (6,13,20,21). This observation has several implications. First, as most of the iron accumulated in the mitochondria is biologically unavailable/inactive, these mutants might paradoxically suffer from a lack of iron, despite being overloaded with iron. Consistent with this, the viability of ⌬yfh1 cells was increased when an excess iron was added to the medium (13). A second implication is that these cells must be affected in their metabolism of phosphate because a large amount of phosphate co-precipitates with iron in the mitochondria and thereby becomes unavailable for vital biological processes, including oxidative phosphorylation. The issue of iron accumulation/precipitation in ⌬yfh1 mitochondria has been extensively studied but that of phosphate co-precipitation in these mitochondria has never been investigated, as far as we are aware. Nevertheless, there should be a link between phosphate and iron homeostasis in yeast because iron and other cations are normally stored in the vacuole of this organism in association with polyphosphates (27). The effect of phosphate on metal ion homeostasis in yeast was recently investigated by Rosenfeld et al. (26), who found that disruptions in phosphate control elicit an Aft1-dependent iron starvation response. In ⌬yfh1 mutants, iron is not stored in the vacuole but accumulates in the mitochondria, and here also, iron is associated with phosphate. Unlike the iron stored in vacuoles of wild-type cells, the iron accumulated in mitochondria of mutant cells cannot be mobilized because it forms an amorphous precipitate of ferric phosphate. Here, we show that a similar process occurs in vitro; both orthophosphate anions and ferrous cations that are taken up by isolated ⌬yfh1 mitochondria rapidly become insoluble in the mitochondrial matrix. The most probable explanation is that the iron is oxidized and then precipitates with phosphate inside the mitochondria.
Cause of Iron and Phosphate Co-precipitation-The reason why this process occurs in frataxin-deficient mitochondria but not in wild-type mitochondria might be related to the presence, in ⌬yfh1 mitochondria, of preformed nanoparticles of ferric phosphate; these nanoparticles may act as nucleation centers for further precipitation of iron and phosphate. It is particularly striking that this phenotype observed in vivo in ⌬yfh1 cells is also found in vitro with isolated ⌬yfh1 mitochondria, although there is no evidence that the same mechanism is at play.
One hypothesis is that iron accumulation/precipitation in vivo in ⌬yfh1 mitochondria is a result of phosphate accumulation, which itself would result from deficiencies in the respiratory chain, with a subsequent decrease in the flux of phosphate utilization for oxidative phosphorylation. Iron entering the mitochondria may precipitate with this excess phosphate, after oxidation by oxygen, the concentration of which is expected to be higher in mutant than in wild-type mitochondria. In this hypothesis, phosphate would accumulate first in ⌬yfh1 mitochondria, followed by iron precipitation due to the excess of phosphate. To address this issue, we tried to determine the time course of iron and phosphate accumulation/ precipitation in the mitochondria when the level of Yfh1 was progressively decreased in a strain expressing YFH1 conditionally. Our data show that phosphate precipitates very early when the level of Yfh1 is decreased, whereas iron accumulation and precipitation is more progressive. Thus, iron probably precipitates in the form of ferric phosphate as soon as it starts to accumulate in the mitochondria, resulting in a rapid decrease in the soluble, bioavailable pool of phosphate. When we progressively decreased the level of Yfh1, we did not observe any transient increase in the soluble phosphate pool preceding the accumulation/precipitation of iron. Thus, our data do not allow us to conclude that iron precipitation in frataxin-deficient cells is initially caused by the presence of excess phosphates in the mitochondria of these cells. Moreover, if this was the case, iron accumulation/precipitation in ⌬yfh1 cells would be expected to be decreased by phosphatelimiting conditions, which is the opposite of what we found. Further investigations are thus needed to identify the cause of iron and phosphate accumulation and co-precipitation. Possibly, iron starts to accumulate because it cannot be incorporated properly in heme and [Fe-S] clusters, and it oxidizes and co-precipitates with phosphate as a result of the low capacity of the mutant mitochondria to reduce oxygen.
Beneficial Effects of Excess Phosphate in Medium and of MIR1 Overexpression-Iron accumulation in ⌬yfh1 cells strongly increased with decreased concentrations of phosphate in the medium, and conversely, growing the cells in the presence of a large excess of phosphate resulted in much less iron accumulation. The relationship between phosphate and iron homeostasis is thus complex because excess phosphate improves the fitness of ⌬yfh1 cells in terms of growth, iron accumulation and the ability of the cells to grow on a nonfermentable carbon source. This effect is probably mostly transcriptional because increasing phosphate concentration in the medium leads to a decrease in the level of transcription of Aft1-regulated genes, which are constitutively induced in mutants affected in [Fe-S] cluster assembly. But here again, the effect of phosphate could be indirect. We showed here that excess phosphate in the medium resulted in increased fitness of ⌬yfh1 cells, which recovered some ability to grow on a nonfermentable carbon source. Excess phosphate in the medium also resulted in increased expression of the genes encoding the mitochondrial phosphate carriers, MIR1 and PIC2. In sep-arate experiments, we showed that overexpressing MIR1 without changing the phosphate content of the medium also improved the respiratory functions of ⌬yfh1 cells. Thus, the most probable explanation is that the mitochondrial functions of frataxin-deficient cells are partially restored by increasing the level of bioavailable phosphate in the mitochondria. Such an improvement of the respiratory functions, involving some restoration of mitochondrial [Fe-S] biogenesis, is itself expected to decrease the level of transcription of the iron regulon genes, leading to decreased iron accumulation.
We suggest that ⌬yfh1 cells suffer from a lack of available phosphate for biological processes, in exactly the same way as they suffer from a lack of bioavailable iron, because iron and phosphate co-precipitate in the mitochondria by a mechanism that remains to be elucidated. Frataxin-deficient cells are impaired in oxidative phosphorylation (36), and there are several possible causes of this defect, including insufficient available inorganic phosphate.
Frataxin-deficient cells suffer from multiple defects, which are probably cumulative and generate vicious spirals. For example, a primary small defect in [Fe-S] cluster assembly could result in iron accumulation/precipitation in the mitochondria, leading to decreased iron availability (including for [Fe-S] biogenesis) thereby continuing the spiral. The same could be true for phosphate; the defect in oxidative phosphorylation resulting from the deficiencies in the respiratory chain may be strengthened by the lack of phosphate resulting from the coprecipitation of iron and phosphate in the matrix. This obviously remains a hypothesis, and further evidence is required.
The beneficial effect of increasing phosphate concentration in the medium was much clearer in the ⌬yfh1 mutant than in the ⌬ggc1 mutant that we used as a control. Both mutants are affected in [Fe-S] cluster assembly and accumulate iron in the mitochondria in the same form of nanoparticles of ferric phosphate (13). Several phenotypes found in frataxin-deficient cells are shared by different mutants affected in the biogenesis of [Fe-S] clusters, and it is thus important to compare different [Fe-S] mutants to determine which phenotypes are specifically related to the lack of frataxin rather than to a general defect in [Fe-S] biogenesis. Here, we observed some coprecipitation of iron and phosphate in both ⌬yfh1 and ⌬ggc1 mitochondria, although this phenotype was much more marked in ⌬yfh1 than in ⌬ggc1 mitochondria. Similarly, increasing the phosphate concentration in the medium resulted in decreased iron accumulation in both mutants, but this effect was again much more spectacular in ⌬yfh1 than in ⌬ggc1 cells. These differences between the mutants might be due to the fact that the phenotypes of iron accumulation and of respiration deficiency are more pronounced in ⌬yfh1 than in ⌬ggc1 cells. However, the beneficial effect of MIR1 overexpression was not observed at all in the ⌬ggc1 mutant, thereby suggesting a specific effect of phosphate in frataxin-deficient cells. Our interpretation of these various findings is that Mir1 overproduction makes phosphate more available for biological processes in the mitochondrial matrix of ⌬yfh1 mitochondria (where most of the phosphate present is precipitated and biologically unavailable), thereby improving mitochondrial functions in a way that remains to be elucidated. Alterna-tively, the mitochondrial functions of ⌬yfh1 cells might be improved by the increased flux of phosphate into the mitochondria because Mir1-mediated phosphate import into mitochondria significantly contributes to the mitochondrial membrane potential (37). We are currently testing these hypotheses.