Cell functions impaired by frataxin deficiency are restored by drug-mediated iron relocation

Humoral factors or mutations of genes that affect iron metabolism often result in pathologic changes linked to systemic or cellular misdistribution of iron. In anemia of chronic disease (ACD),¹  plasma iron deficiency results from retention of iron in the reticuloendothelial system because of hepcidin-mediated inhibition of iron export into plasma.²  Defective delivery and distribution of iron resulting from deficiency in the iron-chaperone protein frataxin are also considered to be key causative factors in Friedreich ataxia (FRDA).^3,4  The disease is expressed in individuals carrying a GAA repeat expansion in the first intron of frataxin that reduces frataxin levels, leading to reduced iron-sulfur cluster (ISC)–protein (ISP) synthesis and a combined deficiency in aconitase and respiratory chain (complex I-III) activity. This leads to a concomitant deficiency in cell respiratory functions⁵  that is further exacerbated by mitochondrial iron accumulation and ensuing oxidative damage.^6,7  FRDA is a neurodegenerative disorder that is often accompanied by hypertrophic cardiomyopathy and increased predisposition for diabetes mellitus.⁸  The precise role of iron in FRDA pathophysiology has hitherto remained controversial.⁹  Support for its direct involvement in the disease is based on histopathologic examination of human specimens and magnetic resonance imaging of patients.¹⁰  Similar conclusions were drawn from biochemical studies with frataxin-deficient yeast and animal cell models.¹¹  The contending view, that cardiac iron accumulation is not a causative factor of the disease, but a late event that follows the onset of damage, is based on models of frataxin knockout mouse embryos¹²  and conditional cardiac-targeted frataxin knockout mice,¹³  as well as frataxin-depleted HeLa cell models¹⁴  and lymphoblastoid cells derived from FRDA patients.¹⁵  Irrespective of the primary or secondary role of iron in the pathology of FRDA, it has been proposed that iron chelation may have potential therapeutic value.^7,16 

In contrast to the treatment of systemic iron overload, the presumed association of oxidative damage with regional iron accumulation, observed in FRDA¹⁰  and in a variety of other diseases, calls for the application of a special form of iron chelation. We have proposed that a suitable agent for this purpose may be deferiprone (DFP), which can relocate iron within and across cells and donate it to putative acceptors¹⁷  as well as cross the blood-brain barrier.¹⁸  The application of DFP in a clinical pilot study involving FRDA adolescents¹⁹  led to a dissipation of cerebellar foci of labile iron without causing systemic iron deficiency or affecting the hematologic status of the patients.¹⁹  However, the initial clinical benefits that resulted from the treatment opened some key questions: specifically, is depletion of cellular iron sufficient or even desirable for treating diseases characterized by iron misdistribution, in which accumulation of iron in one region coexists and often results in depletion of iron in another? We hypothesized that, for correcting this type of condition, exemplified by ACD, X-linked sideroblastic anemia, or FRDA, an agent should act not merely by chelating the accumulated metal but also by rendering it available for physiologic reuse.

Here, we used a cell line engineered to inducibly underexpress frataxin by short-helical RNA (shRNA)²⁰  to assess whether iron relocation by DFP¹⁷  can also lead to the restoration of functions compromised by frataxin deficiency. We report that DFP reduced the levels of the mitochondrial labile iron pool (LIP) that were increased by frataxin deficiency, provided protection against oxidative damage, and concomitantly mediated the restoration of various metabolic parameters, including aconitase activity. We propose that DFP has the unique dual properties of a siderophore, on the one hand capable of chelating labile iron and on the other donating iron to potential physiologic acceptors, thus supporting physiologic reutilization of the chelated metal.

All chemicals were from Sigma-Aldrich (Rehovot, Israel), except calcein-acetomethoxyl ester; carboxy-dichloro-dihydro-fluorescein diacetate acetomethaxyl ester (CDCF-AM), and tetramethylrhodamine ethyl ester (TMRE) from Molecular Probes (Eugene, OR); dihydrorhodamine 123 (DHR) from Biotium (Hayward, CA); annexin V–phycoerythrin (PE)–Cy5 from BioVision (Mountain View, CA); staurosporine from Cayman Chemical (Ann Arbor, MI); salicylaldehyde isonicotinoyl hydrazone (SIH), a gift from Prof P. Ponka (Lady Davis Institute of Medical Research, Montreal, QC); DFP (sold as Ferriprox) from Apopharm (Toronto, ON); and DFR (deferasirox or Exjade) and deferoxamine (DFO) from Novartis-Pharma (Basel, Switzerland). Cells included the following: T-Rex-293 cells (Invitrogen, Carlsbad, CA) stably expressing the tetracycline repressor and tetracycline-inducible shRNA against human frataxin were prepared as described.²⁰  The cells were grown in Dulbecco modified Eagle medium supplemented with 10% tetracycline-free fetal calf serum (FCS; BD Clontech, Mountain View, CA), 20 mM glutamine, 4.5 g/L glucose, and antibiotics and induced with tetracycline (1 μg/mL) to repress frataxin expression. Cells were transfected with CaPi used as recommended.²¹  Antibodies against frataxin were the kind gift of Dr Franco Taroni (Institute Besta, Milan, Italy). Antibodies against mitochondrial aconitase and actin were obtained from Abgen (San Diego, CA) and Sigma-Aldrich, respectively.

Mitochondria-targeted roGFP2 was a kind gift from Dr J. Remington (Institute of Molecular Biology, University of Oregon, Eugene, OR).

Cells were cultured in glass-bottom microscope dishes with 1 mg/mL collagen (Roche, Mannheim, Germany) and analyzed by confocal microscopy using an Olympus Fluoview 1000 (Olympus, Center Valley, PA) with a 60×/1.35 NA PlanApochromat oil-immersion lens and the Image J software for analysis (Scion, Frederick MD). For quantitative image analysis, three-dimensional (3D) image stacks were generated, and the fluorescence intensities for each image channel in the stacks summed. Mitochondrial punctae were quantified using the Image J “Analyze Particles” plugin. To permit comparison of images, we used the same laser power and detector gain and offset. Mitochondrial LIP was measured by exposing cells for 10 minutes to 50 μM DHR, which accumulates in the mitochondria and is oxidized to the highly fluorescent rhodamine 123 (R). This oxidation is iron dependent, as demonstrated by its iron chelator-mediated attenuation²¹  (Figure 1A) and therefore can serve as a means for indirect determination of mitochondrial LIP. Fluorescence (excitation, 488 nm; emission, 520 nm) was analyzed by confocal microscopy. Intramitochondrial redox potential was determined as described by Schwarzer et al.²²  In brief, intramitochondrial redox potentials were obtained in cells expressing ratiometric mitochondrial targeted roGFP2 from in situ calibration curves. The curves were generated by titrating cells with Ringer solutions containing different ratios of trans-4,5-dihydroxy-1,2-dithiane (oxidized dithiothreitol [DTT]) to reduced DTT, which covered a range of redox potentials from −330 mV to −195 mV. Fluorescence ratios of excitation wavelengths 405/488 (emission, 520 nm) were normalized to the range between 0% oxidation (10 mM reduced DTT) and 100% oxidation (10 mM H₂O₂), and the normalized ratios were plotted against the calculated redox potentials and fit (Sigma Plot Statistical Analysis software; Systat, San Jose, CA) by the nonlinear regression E′ = 0.0509 + 0.8305/(1 + exp(−(x + 249.5422)/6.6161)), where E′ is the redox potential and x is the normalized 405/488 excitation ratio. Normalized 405/488 excitation ratios obtained in each experiment were used in combination with the nonlinear regression to determine intramitochondrial redox potential. Reactive oxygen species (ROS) generation and mitochondrial membrane potential (MMP) were determined microscopically with CDCF-AM and TMRE, respectively, using fluorescein (excitation, 488 nm; emission, 520 nm) and Texas Red (excitation, 543 nm; emission, 633 nm) filter sets.

Cells were incubated with glucose oxidase for 24 hours, washed, and supplemented with 0.5 μM calcein-AM in serum-free medium. Calcein fluorescence (excitation, 488 nm; emission, 520 nm) was monitored by a fluorimeter as an indication for intracellular hydrolysis of calcein-AM in viable cells. Rates of fluorescence increases normalized to rates obtained in cells untreated with glucose oxidase served as a measure for cell viability.

All flow cytometry experiments were performed using a FACSCalibur flow cytometer (BD Immunocytometry Systems, San Jose, CA). Cells in 10-cm dishes were stained as indicated and fluorescence was detected in at least 10 000 cells. Apoptotic indices, ROS generation, and MMP were determined by annexin V–PE-Cy5, CDCF, and TMRE fluorescence and measured using 488 nm excitation line and fluorescein isothiocyanate (FITC; FL1), PE (FL2), and PE-Texas Red (FL3) emission detectors, respectively.

Cytosolic LIP was measured by the calcein method as described.²³ 

Adenosine triphosphate (ATP) was determined by a luciferase-based assay (Sigma-Aldrich) following the manufacturer's instructions. In brief, cells were extracted in 0.2% Triton and extracts were diluted 1:100 in water. Diluted ATP assay mixes (1:25) were added to the diluted extracts, and luminescence was read with a 6-second integration time in a Synergy 2 (BioTek, Winooski, VT) luminometer and analyzed by the Gen5 software (BioTek).

Dissolved oxygen was measured using the Pasco Scientific (Roseville, CA) PS-2108 dissolved oxygen sensor (Clark electrode) with its attached DataStudio analysis software; 4.5 × 10⁶ cells were resuspended in 4.5 mL full medium and dissolved oxygen levels were recorded. N_2(g) was pumped to the system to adjust the initial dissolved oxygen levels; 50 μM carbonyl cyanide m-chlorophenylhydrazone (CCCP) were added to determine maximal oxygen consumption rate.

mtDNA content relative to nuclear genomic DNA was determined by real-time polymerase chain reaction (PCR) (7900HT Real-Time PCR system; Applied Biosystems, Foster City, CA), coamplification of the mitochondrial encoded cytochrome b (using the primer pair 5′-TATCCGCCATCCCATACATT-3′ and 5′-GGTGATTCCTAGGGGGTTGT-3′), and the nuclear encoded β-actin (using the primer pair 5′-AGAAAATCTGGCACCACACC-3′ and 5′-AACGGCAGAAGAGAGAACCA-3′) using the SYBR Green (Applied Biosystems) method. The ratio of mtDNA to genomic DNA was calculated as 2^−ΔΔCt, where C_t is the threshold cycle number, ΔC_t is C_t(cytochrome b) − C_t(β-actin) and ΔΔC_t = ΔC_t(treatment) − ΔC_t(no Tet = control).

Aconitase activity was assayed by an adaptation of the indirect method using the reduction of nicotinamide adenine dinucleotide phosphate (NADPH) by isocitrate dehydrogenase.²⁴  Cells grown in 12-well plates were lysed in 0.2 mL Tris-HCl (pH 7.4) containing 0.5% Triton X-100 (Fluka, Buchs, Switzerland) and 2.5 mM sodium citrate, for 5 minutes on ice. Insoluble material was removed by centrifugation (13 000g, 5 minutes), and protein concentration was determined with Bradford reagent (Bio-Rad, Hercules, CA). Formation of NADPH was followed by fluorescence in 96-well plates (excitation 345 nm, emission 550 nm) at 37°C in a mixture containing 50 mM Tris-HCl (pH 7.4), 2.5 mM sodium citrate, 0.6 mM MnCl₂, 0.2 mM NADP, 2 U isocitric dehydrogenase, and 50 μg protein of cell extract. Because of a lag in NADPH formation, aconitase activity was calculated using the slope recorded after 10 minutes from the beginning of the assay.

Aconitase activity was also assayed in nondenaturing gels as described by Tong and Rouault.²⁵ 

To define the cell phenotype that results from repression of frataxin, we used the HEK 293 T-rex cell line, designed to express shRNA against frataxin after tetracycline (Tet) induction.²⁰  As shown in Figure 1A, frataxin levels steadily declined after exposure of T-rex 293 cells to Tet for 2 days or more. The cells were comprehensively characterized in terms of the following biophysical and biochemical parameters: LIPs in cytosol and mitochondria, ROS production, mitochondrial redox potential, MMP, ATP levels, respiration, and susceptibility to ROS and apoptotic stimuli.

The measurement of cellular LIPs relies either on the direct assessment of chelatable iron and/or on the ability of iron chelators to block ROS formation catalyzed by redox active iron.^21,26  We opted for applying online methods of LIP monitoring that minimally perturb the system and model cells that are amenable to manipulation of their frataxin levels.

For determining mitochondrial LIP, we relied on the mitochondria-specific accumulation of DHR, which becomes fluorescent on oxidation by ROS, in a labile iron-dependent reaction, inhibitable by membrane permeant chelators.²¹  This principle is reconfirmed and exemplified in Figure S1 (available on the Blood website; see the Supplemental Materials link at the top of the online article) for frataxin-deficient T-rex cells challenged with H₂O₂, followed (or not) by exposure to the iron chelator DFR (given at 20 minutes). The fact that DFR prevented the H₂O₂-prompted increase in fluorescence between 20 minutes and 30 minutes identifies the original signal as associated with labile iron. Detectable levels of mitochondrial LIP were found in frataxin-deficient T-rex cells (Tet⁺) but not in uninduced cells (Figure 1B). Because the increase in mitochondrial LIP caused by frataxin repression was already apparent at day 2 after induction and was sustained for at least 9 days (Figure 1C), we selected the 4- to 6-day induction window for determining other functional parameters (described in the following sections) affected by repression of frataxin.

For assessing the changes in cytosolic LIP after tetracycline induction, we used the calcein-based method of monitoring intracellular LIP, whereby the intracellular probe undergoes fluorescence quenching on iron binding and dequenching on iron removal by a chelator. As shown in Figure 1D, cytosolic LIP, determined from the extent of dequenching by the high-affinity chelator SIH, was significantly lower in frataxin-deficient cells, relative to controls (no tetracycline induction). The use of SIH here is arbitrary, as SIH and DFR are equally applicable for use in LIP determinations based on calcein and DHR.^21,26  These results are consistent with the reduced levels of cytosolic iron-regulatory proteins reported in frataxin-deficient systems.^27,28  Moreover, they are in accord with data obtained in yeast in which disruptions in ISC assembly and export led to mitochondrial iron accumulation²⁹  and activation of the iron regulon, usually prompted by global iron deprivation.^30,31 

The frataxin-deficiency phenotype has been characterized by increased ROS generation and by changes in mitochondrial functions.^9,11  This has been attributed to (1) increased ROS formation because of mitochondrial accumulation of labile iron, (2) premature electron transfer to oxygen at complexes I and III because of ISC deficiency (caused by inefficient formation and repair), and (3) energy deficit and decreased redox capacity in both mitochondria and cytosol because of lowered NADH and, concomitantly, NADPH levels (via the membrane-potential-dependent nicotinamide nucleotide transhydrogenase). To confirm that the oxidizing potential increases in the present frataxin-deficient cell system, we transfected T-rex 293 cells with a mitochondrial-targeted construct harboring the redox sensor roGFP2, which allows assessment of the mitochondrial redox potential,²²  a parameter responsive to membrane-permeable reductants and oxidants. The results shown in Figure 2A indicate that the mitochondrial redox potential increases (becomes more oxidizing) as a result of frataxin depletion and that its levels remain relatively high during frataxin repression.

Repression of frataxin also caused a marked increase in the rate of cytosolic ROS generation observed with the pro-oxidant–sensitive probe CDCF (Figure 2B). Decreased frataxin was also associated with mitochondrial membrane depolarization (measured with the potentiometric dye TMRE) (Figure 2C), which may result from mitochondrial membrane permeabilization caused by increased ROS formation.³²  The appearance of mitochondrial fragmentation in frataxin-deficient cells (Figure 2C) is also consistent with the dissipation of the mitochondrial membrane potential.³³ 

A recognized characteristic of frataxin-deficient cells is increased susceptibility to proapoptotic stimuli. In the present experimental cell system, frataxin repression in itself evoked an increase in the percentage of dead and dying cells (Figure 3A). This effect was even more pronounced after exposure to proapoptotic agents, such as staurosporine (Figure 3B), or chronic, low-level exposure to H₂O₂ generated by glucose oxidase (Figure 3C,D). Altogether, these data are consistent with previously reported effects of frataxin deficiency,³⁴  which include increased ROS levels, possibly with ensuing mitochondrial membrane damage and cell death.

As DFP has been recently shown to act as a siderophore that can chelate cytosolic and mitochondrial labile iron^21,26  and redistribute the metal between cell compartments,¹⁷  we set out to determine whether it could also support restoration of functions affected by frataxin deficiency. First, we confirmed (Figure 4A) that the increased levels of mitochondrial LIP detected in frataxin-deficient cells with DHR and H₂O₂ (Figure 1) could be prevented by the addition of 50 μM DFP, similar to what was shown for DFR (Figures S1, 4A). Second, we observed that the same DFP treatment shifted the mitochondrial redox potential of −251.6 plus or minus 1.3 mV measured in frataxin-deficient cells to near-control levels of −308.3 plus or minus 15.3 mV. These measurements were obtained by ratiometric fluorescence in cells transfected with the redox probe roGFP2 (Figure 4B). Third, to quantitatively assess changes in MMP and cytosolic ROS levels, control and frataxin-deficient T-rex cells were colabeled with the potentiometric fluorescent probe TMRE and with the ROS probe CDCF-DA–AM. The changes in the fluorescence intensity of both probes were analyzed simultaneously by flow cytometry (Figure 4C). In comparison to control cells, frataxin-deficient cells showed higher CDCF fluorescence intensity, indicating increased ROS generation and lower TMRE fluorescence denoting decreased MMP. Overnight treatment of frataxin-deficient cells with 50 μM DFP shifted their MMP to levels that were statistically indistinguishable from those of control cells and shifted their cytosolic ROS levels to slightly below those obtained in controls (Figure 4C). Fourth, the percentage of cells undergoing spontaneous apoptosis, which rose from approximately 1.5% in controls to 2.8% in frataxin-deficient cells, returned to normal levels (1.3%) after overnight treatment with DFP (Figure 4D). Induction of apoptosis by staurosporine (0.3 μM) generated approximately 52% annexin V⁺ cells in the control and 70% in the frataxin-deficient population (Figure 4E). DFP treatment decreased the percentage of apoptotic cells in the control and frataxin-deficient populations by 5% and 23%, respectively, resulting in 47% apoptosis in both cell types (Figure 4E). We conclude from these results that mitochondrial damage induced by oxidative stress in frataxin-deficient cells predisposes them to apoptosis, and DFP, by providing relief from the stress, reduces the cell death indices observed in basal conditions (Figure 4D) and after staurosporine treatment (Figure 4E).

The restoration of cell functions by DFP in frataxin-deficient cells could be explained by the ability of the chelator to remove labile iron accumulated in mitochondria. However, if chelation is the sole DFP property required for correcting the frataxin-deficient phenotype, then it can be assumed that other high-affinity iron chelators with similar cell-permeating capacities should be equally effective. In agreement with our previous results,^21,26  we found that agents such as SIH or DFR^21,26  can scavenge mitochondrial labile iron and confer protection from oxidative stress comparably to DFP (Figure S2). Moreover, previous in vitro studies indicated that cytosolic iron-sulfur cluster scaffold (ISU) synthetic machinery could be aided by strong iron chelators.³⁵ 

However, SIH and DFR differed from DFP in terms of their ability to restore cell functions impaired by frataxin deficiency, such as ATP levels (Figure 5A), mitochondrial DNA content (Figure 5B), and respiration (Figure 5C,D). Whereas SIH was virtually ineffective and DFR produced only a partial recovery of those parameters (Figure 5), DFP restored them all. We conclude that the beneficial effects of DFP cannot be attributed solely to its mitochondrial iron chelation ability.

Donation of chelated iron to cellular components is a characteristic property of siderophores. Here we compared DFP and 3 other iron chelators in terms of the ability of their iron chelates to restore the activity of the ISP aconitase. Aconitase was selected because (1) the enzyme activity of its mitochondrial form is known to be repressed in frataxin-deficient cells^5,14,20,36 ; (2) iron deprivation causes rapid disassembly of the ISC in both the mitochondrial and cytosolic forms and loss of their activity; and (3) it is readily reactivated by resupplying iron.²⁵  As shown in Figure 6A, overnight incubation of control T-rex cells with up to 100 μM of the chelators DFO, DFR, and SIH caused significant loss of aconitase activity, whereas DFP had virtually no effect. If loss of aconitase activity is indicative of an iron-deficient state, then DFO, DFR, and SIH appear to have an iron-withholding effect, unlike DFP.

We used the paradigm of recovery of aconitase activity after cellular iron depletion to address the question of how DFP compares with other higher affinity chelators in facilitating the rebuilding of ISCs in frataxin-deficient cells (presumably by bypassing frataxin). In contrast to control cells, frataxin-deficient cells failed to show recovery of aconitase activity. As shown in Figure 6B, administration of DFP to frataxin-deficient cells during the 3-hour recovery phase facilitated partial reactivation of total cell aconitase activity, although control levels were not attained. As opposed to its enzymatic activity, the levels of aconitase (its mitochondrial form, which contributes most of the activity detected, as shown by in-gel assays [Figure S3]) were not changed by the different treatments. The other chelators differed in their capacity to donate iron for aconitase reactivation in both frataxin-deficient and control cells in the order DFP > DFR ≈ SIH > DFO. This order corresponds approximately to their relative affinities for iron (respective pM values at pH 7.4 for Fe(III) of DFP, DFR, SIH, and DFO are 19.5, 22.5, 27.7, and 26.6)³⁷  and is consistent with previous results showing that DFP can donate its chelated iron to biologic acceptors, such as transferrin and the hemoglobin synthesis machinery.¹⁷  These effects of DFP were not obtained by chelator-mediated changes in frataxin expression because DFO further lowered, rather than increased, frataxin expression (Figure 6C, in agreement with Li et al²⁷ ), whereas DFP did not affect frataxin levels.

Misdistribution of iron associated with regional iron accumulation is gaining recognition as a central factor in an increasing number of diseases. In neurodegenerative disorders, such as Parkinson disease, neurodegeneration with brain iron accumulation, and FRDA, iron is deposited in specific, affected areas of the brain.³⁸  In ACD, iron accumulates in reticuloendothelial cells leading to systemic iron deprivation, possibly conferring resistance to plasma pathogens.^2,39  In hematopoietic cells of X-linked sideroblastic anemia patients, an impaired synthesis of heme causes iron to accumulate in mitochondria; and in FRDA heart and dentate nuclei, a defective formation of ISC may also lead to depletion of iron from the cytosol.⁴⁰  We reasoned that treatment of diseases involving misdistribution of iron should aim not only at eliminating the regional iron accumulation by chelating the metal but also, and often more importantly, at rendering the metal available for metabolic reuse. For that purpose, therapeutic agents should be endowed with properties analogous to those of siderophores, namely, iron carriers with both metal acceptor and donor capacities. In this work, we assessed the iron chelator DFP in terms of its ability to reinstate compromised functions in a model system of cellular iron misdistribution caused by frataxin deficiency. This agent was chosen because it was recently shown to act both as a cell iron chelator and an iron donor to the physiologic acceptor transferrin and for hemoglobin synthesis in in vitro cell cultures.¹⁷ 

The experimental system based on tetracycline-induced frataxin repression by shRNA²⁰  was adopted because: (1) it allowed the control of frataxin levels over time in a reproducible fashion and (2) it displayed impaired iron-related functions associated with cytosolic and mitochondrial ISPs, oxidant control and stress response, and heme synthesis.

On the basis of these observations, we first carried out a thorough functional characterization of the frataxin-deficient cells and found that (1) the cell LIPs are increased in mitochondria and reduced in the cytosol; (2) the mitochondrial metabolic and redox capacities are markedly reduced; (3) the MMP is dissipated; and (4) cell survival is compromised. These data show that reduction of frataxin below a threshold level (Figure 1A) results in an impairment of mitochondrial functions and increased sensitivity to stress (Figures 2,Figure 3,Figure 4–5). However, no clear dose-response relationship could be established, with the exception of the cytosolic LIP. These results are consistent with the idea that the overall frataxin deficiency is associated with inefficient ISC formation and/or premature ISC destruction and ensuing release of reactive and chelatable iron.¹¹ 

Our results appear to contrast earlier claims that frataxin deficiency results in no detectable iron accumulation in mitochondria, as visualized by electron microscopy,¹³  measured spectrophotometrically,¹⁴  and measured by radioactivity using radiolabeled iron loaded into cells⁴¹  and by in situ fluorimetry in cultured lymphoblasts and fibroblasts derived from FRDA patients.¹⁵  We ascribe the discrepant results to 2 factors: (1) phenotypic differences among the experimental cellular systems expressing frataxin deficiency, particularly in light of the inconsistent pattern of phenotypic changes observed in immortalized lines of lymphoblasts and fibroblasts derived from human samples; and (2) methodologic limitations in the quantitative assessment of mitochondrial labile iron with targeted fluorescent phenanthroline probes, which may affect the cell iron distribution.¹⁵  These drawbacks are largely avoided in the present study by using cells with defined patterns of frataxin repression and analytical methods that allow selective detection of redox-active (as opposed to total) iron and minimal interference with intracellular iron distribution.

The present study demonstrates, for the first time, that a chemical agent such as DFP may act as a “frataxin surrogate” and facilitate the correction of defective energetic parameters observed in frataxin-deficient cells. The effects of DFP were largely dependent on its ability to scavenge as well as to deliver iron, whereas other chelators that only contributed to iron removal (or detoxification) replaced frataxin functions only to a limited extent (Figure 5). We consider it improbable that DFP exerted its effect by enhancing iron delivery to residual frataxin in Tet⁺ cells because the recovery of mitochondrial aconitase in Tet⁺ cells (Figure 6B) occurred in conditions of normal iron supply (complete culture medium) where mitochondria were also shown to be iron replete (Figures 1, 4).

DFP could act as a frataxin surrogate in FRDA either by iron detoxification, thus slowing down oxidative destruction of ISCs, or by iron donation, facilitating ISC repair or synthesis. These iron relocator/donor properties might explain the clinical benefits of DFP treatment in rheumatoid arthritis with ACD⁴²  and, more recently, in FRDA patients.¹⁹  However, because cis-activation of frataxin expression results directly from induction of hypoxia inducible factor-2 (HIF-2)⁴³  and indirectly from HIF-1 via erythropoietin,⁴⁴  it is theoretically possible that DFP, as other chelators, may act also by increasing HIF stability via prolyl hydroxylase inhibition.⁴⁵  Yet, chelation has recently been shown to repress rather than activate frataxin,²⁷  and this was confirmed here (Figure 6C). Because DFP did not markedly affect frataxin levels (Figure 6C), we hypothesize that, at moderate concentrations, its influence on the overall cellular iron status is minor. We used 50 μM of DFP in this study as a representative concentration that would be achieved in the circulation by doses that are considerably lower than those normally administered to iron-overloaded patients.⁴⁶ 

Recently, we showed that DFP-mediated relocation of iron could serve for shuttling iron from cells to external transferrin, thereby subserving hemoglobin synthesis.¹⁷  In this work, we used a cell model of iron misdistribution to demonstrate the modality of DFP as a siderophore that can both sequester cell labile iron and support aconitase formation, thus contributing to ISC maintenance in frataxin deficiency. Some of these properties might underlie the clinical effects recently reported in FRDA patients,¹⁹  neurodegeneration with brain iron accumulation patients,⁴⁷  and in ACD.⁴² 

However, it is clear that any approach involving drug-mediated relocation of iron should neither induce toxicity nor lead to systemic iron deficiency. The intracellular chelation/relocation of labile iron could in principle lead to the generation of redox-active iron chelates. Such tendency prevails among chelators bearing amine and carboxyl-liganding groups that display low redox potentials,⁴⁸  or when substoichiometric chelator/iron complexes are formed, such as shortly after oral intake of a bidentate or tridentate drug. For the hydroxypyridinone-based DFP, the binding affinity favoring 3:1 iron complex involving all 6 coordination sites of iron would prevent redox cycling⁴⁹  and thus protect cells from iron-evoked production of ROS.^21,26  However, another potential side effect of DFP might occur after the fast removal of the chelator, causing a rebound of iron accumulation accompanied by enhanced oxidative stress.⁵⁰  One can speculate that such effects might underlie the appearance of neutropenia/agranulocytosis in a minor percentage of DFP-treated FRDA¹⁹  and thalassemia⁴⁶  patients and that those effects could be reduced by coadministration of DFP with appropriate antioxidants such as idebenone. However, more importantly, DFP, as any other clinically relevant chelator, should⁴⁶  and could^17,19  be used with moderation so as to avoid overchelation that could affect normal cellular iron metabolism and thereby induce iron-deficiency anemia. The importance of exercising moderation in applying chelation relates not only to drug dosage but also to drug exposure time, as dictated by drug pharmacokinetics. DFP has been recently shown experimentally⁵¹  to be cytotoxic and inhibitory to cell aconitase activity when used for extended periods of time at excessive concentrations. However, even at experimentally toxicologic levels, the aconitase activity inhibited by overnight incubation with excessive DFP is recoverable after short incubations in full medium (Figure S4). These experimental conditions recapitulate with more fidelity DFP administration in the clinical settings where it is cleared from the plasma within hours allowing iron to be readily resupplied and correct overchelation in case it occurred.

This study was supported in part by research funding from the Association Francaise contre les Myopathies, the French-Israeli Organization for Research in Neuroscience, the Israel Science Foundation, and the European Economic Community (EEC) Framework 6 (LSHM-CT-2006-037296 Euroiron1; Z.I.C.).

Contribution: O.K., H.M., W.B., and Z.I.C. designed research; O.K., H.M., W.B., and N.M.-B. performed research; C.L., G.C., and A.M. contributed vital new reagents or analytical tools; O.K., H.M., W.B., and N.M.-B. collected data; O.K., H.M., and W.B. analyzed and interpreted data; O.K. performed statistical analysis; and O.K., W.B., G.C., and Z.I.C. wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Or Kakhlon, Alexander Silberman Institute of Life Sciences, Hebrew University of Jerusalem, Safra Campus at Givat Ram, Jerusalem, Israel 91904; e-mail: ork@cc.huji.ac.il.