Friedreich ataxia (FA) is a progressive neurodegenerative disease caused by expansion of a trinucleotide repeat within the first intron of the gene that encodes frataxin. In our study, we investigated the regulation of frataxin expression by iron and demonstrated that frataxin mRNA levels decrease significantly in multiple human cell lines treated with the iron chelator, desferal (DFO). In addition, frataxin mRNA and protein levels decrease in fibroblast and lymphoblast cells derived from both normal controls and from patients with FA when treated with DFO. Lymphoblasts and fibroblasts of FA patients have evidence of cytosolic iron depletion, as indicated by increased levels of iron regulatory protein 2 (IRP2) and/or increased IRE-binding activity of IRP1. We postulate that this inferred cytosolic iron depletion occurs as frataxin-deficient cells overload their mitochondria with iron, a downstream regulatory effect that has been observed previously when mitochondrial iron–sulfur cluster assembly is disrupted. The mitochondrial iron overload and presumed cytosolic iron depletion potentially further compromise function in frataxin-deficient cells by decreasing frataxin expression. Thus, our results imply that therapeutic efforts should focus on an approach that combines iron removal from mitochondria with a treatment that increases cytosolic iron levels to maximize residual frataxin expression in FA patients.
Friedreich ataxia (FA) is a progressive disease caused by expansion of a GAA trinucleotide repeat in the gene that encodes frataxin. Patients develop ataxia, sensory loss and cardiomyopathy, and symptoms are limited mainly to the central nervous system and heart (reviewed in 1,2). Frataxin is important in iron–sulfur cluster assembly where it likely functions as a chaperone that provides iron in a bioavailable form in the early steps of iron–sulfur cluster biosynthesis (reviewed in 3). Patients manifest mitochondrial iron–sulfur cluster assembly deficiency, with decreased activity of iron–sulfur proteins such as mitochondrial aconitase and succinate dehydrogenase. As disease progresses, mitochondrial iron overload develops in the heart and central nervous system (4–6). Because of the mitochondrial iron overload, it has been considered that iron chelation therapy should ameliorate disease by protecting mitochondria from the adverse effects of iron overload. Studies in tissue culture performed with the primarily cytosolic iron chelator, desferal (DFO) (7), or with lipophilic iron chelators that specifically target mitochondrial iron, including pyridoxal isonicotinoyl hydrazone and analogues of 2-pyridylcarboxaldehyde isonicotinoyl hydrazone (8,9), as well as studies in patients using deferiprone, a membrane-permeable chelator thought to be capable of shuttling chelated iron from cells to transferrin (10) have shown positive results.
Although many chelator studies have been performed on FA cells and models, none has evaluated the effects of iron chelation on levels of residual frataxin expression in patients. Low expression of FXN causes iron accumulation in mitochondria, a common phenotype observed when proteins involved in iron–sulfur cluster biogenesis are deficient (11). Previous studies have shown that yeast which lacks the mitochondrial ABC transporter, Atm1 (a homologue of human ABCB7), display features of both mitochondrial iron overload and cytosolic iron starvation (12), which is similar to the phenotype of frataxin-deficient yeast (13). Interestingly, levels of yfh1 (frataxin) mRNA level decreased 4-fold in Atm1-deficient yeast, whereas atm1 mRNA levels were unchanged in the yfh1 mutant (12). Although this study suggested that yfh1 levels were regulated by iron, the authors were unable to confirm iron-dependent regulation of frataxin when they manipulated iron levels in yeast.
Since we have previously observed that expression of the iron–sulfur scaffold protein, ISCU, decreases with iron chelation (14), we decided to evaluate the effects of iron depletion on frataxin expression. We treated human cell lines with the iron chelator, DFO, or the iron salt, ferric ammonium citrate (FAC), to determine whether frataxin expression changes in normal controls or in cell lines derived from FA patients. We discovered that frataxin mRNA levels decrease significantly in multiple human cell lines treated with DFO. In addition, frataxin levels decrease in fibroblasts and lymphoblast cells derived from both normal controls and from patients with FA when treated with DFO. Using activity of iron regulatory proteins as an indicator of cytosolic iron status, we find evidence to support that there is cytosolic iron depletion in fibroblast and lymphoblast cell lines from FA patients compared with normal control cell lines. Our results suggest that cytosolic iron depletion may worsen FA by further diminishing expression of frataxin.
To study whether frataxin expression was regulated by iron, we studied human cell lines, including HEK293. After adding either FAC or the iron chelator DFO (Sigma) to cells for 16 h, we observed that frataxin mRNA levels decreased in cells treated with DFO, as detected by northern blot (Fig. 1A) and confirmed by quantitative real-time PCR (qRT–PCR) (Fig. 1B). In addition to finding the major frataxin transcript of 1.24 kb as previously reported (15), we observed some similarly regulated bands (e.g. ∼1.7 and 2.3 kb), whereas some others did not appear to be regulated (e.g 3.2, 4.5 and 5.8 kb bands). All of these bands were detected using an antisense RNA probe to the sense strand of FXN, and we do not know whether the larger weak bands are non-specific, or whether they represent different isoforms or incompletely spliced intermediate forms. We also tested fibroblast cell lines (GM08399, Coriell Cell Repositories), and found increased mRNA levels of frataxin in cells treated with FAC versus DFO (Fig. 1C), which was confirmed by qRT–PCR (Fig. 1D). This decrease in frataxin expression with DFO treatment was also observed in other cell lines we tested, including the rhabdomyosarcoma cell line, RD4 (data not shown).
To evaluate the mechanisms of regulation of FXN mRNA levels by iron status, we cloned the previously defined human frataxin promoter region (16,17), and generated constructs that incorporated 1.2 (luc 1), 1.8 (luc 2) or 3.2 kb (luc 3) upstream of the first AUG of exon 1 into a luciferase reporter plasmid, as described previously (17). In another construct (pGL-GAA-luc), we fused a DNA fragment including the promoter region (1.3 kb upstream of the first AUG in exon 1), the entire first exon, and 1.2 kb of the 5′ end of the first intron extending to the position of the GAA triple repeat expansion with luciferase. Each of the four constructs showed that DFO treatment significantly reduced the expression of luciferase compared with untreated controls (Fig. 1E), although the pGL4-GAA-luc construct showed lower promoter activity.
To study whether frataxin protein levels were regulated by iron, as expected based on mRNA results, we analyzed the half-life of frataxin in metabolic labeling experiments to define the optimal length of iron or iron chelator treatments. After immunoprecipitation (IP) of frataxin radiolabeled with 35S-methionine, we observed the frataxin precursor and the frataxin intermediate band generated during mitochondrial import (Fig. 2A), but we were unable to observe the mature fully processed product (Fig. 2D), as the mature form does not contain methionine (18). There are two methionines in the unprocessed frataxin precursor, both of which are removed during mitochondrial import. The inferred cleavage sites are indicated in Figure 2D with inverted arrowheads (18,19), and methionines are underlined. In order to measure the half-life of mature frataxin in cells, we replaced two leucines (L) (amino acids 200th and 203rd) of frataxin, with methionine (M) (described in Materials and Methods). Metabolic labeling results showed that the half-life of the initial precursor was 10 min, whereas that of the intermediate was 2 h 10 min, and that of the mature form was 50 h (Fig. 2B and C).
We then analyzed regulation of frataxin mRNA and protein levels in fibroblast cell lines of controls and FA patients after treatment with DFO or FAC for 48 h. We found that frataxin mRNA levels decreased in fibroblasts (GM03665) of patients treated with DFO (Fig. 3A) and we verified this result with qRT–PCR (Fig. 3B). We then verified that protein levels changed in RD4 (top), HEK293 (middle), fibroblasts (GM04078 and GM03816, not shown) and lymphoblasts (GM15850) from affected patients (bottom, Fig. 3C). The increased levels of transferrin receptor and decreased levels of ferritin H and L chains after DFO treatment in lymphoblasts confirmed that the iron treatment was efficacious. Instantaneous biosynthetic rate measurements showed markedly diminished biosynthesis of endogenous frataxin after treatment with DFO (Fig. 3D), further confirming regulation of frataxin expression by iron.
To more fully characterize cellular iron status in cell lines from FA patients, we compared fibroblast and lymphoblast cell lines derived from either normal controls or patients that are either age and sex-matched, or proband and parent-paired, for levels of frataxin, iron regulatory proteins 1 (IRP1) and 2 (IRP2), IRE-binding activity and aconitase activities. Comparing lymphoblasts from patients (P) to control (C) cell lines, we found that frataxin expression levels varied from 8.5% [e.g. the pair of GM16200 (C5) and GM16197 (P5) in Fig. 4A, right column] to 25% [e.g. the pairs of GM16215 (C6) and GM16214 (P6), AG15799 (C7) and GM16209 (P7), Fig. 4A, right column] of control levels, whereas in fibroblasts from three different patients, we found that frataxin levels varied from 10% [e.g. the pair of GM08399 (C3) and GM03665 (P3), Fig. 4A, left column] to 66% [e.g. the pair of AG09429 (C2) and GM03816 (P2), Fig. 4A, left column] of normal control levels, quantified as described in Materials and Methods. We found that without any iron manipulations, lymphoblast cell lines of FA patients showed increased levels of IRP2 in western blots (Fig. 4A) and RNA gel-shift studies (Fig. 4B) compared with normal controls. In the presence of high iron and oxygen, IRP2 undergoes iron-dependent degradation (20,21). Although low oxygen levels and increased cobalt levels can stabilize IRP2, high IRP2 levels are in general an excellent indicator of cytosolic iron status in mammalian cells (reviewed in 22). IRP2 was not detectable in fibroblasts, but gel-shift studies indicated that the IRE-binding ability of IRP1 was markedly increased, as has been previously observed in FA patient cells (23) (Fig. 4B), as well as in RNAi knockdown studies in HeLa cells (24) and in mouse models (25). In correlation, both mitochondrial and cytosolic aconitase activities were decreased in patient fibroblasts, and cytosolic aconitase activity was significantly decreased in lymphoblasts (Fig. 4C). Thus, IRP1 converted from the cytosolic aconitase form to the IRE-binding form without change of IRP1 protein levels in cell lines of patients with FA.
To further verify the effects of iron on FXN expression, we challenged the FA patient cells with different concentrations of DFO or FAC. We observed regulation of FXN, as well as sensitivity of FA fibroblasts to iron depletion, where as little as 1 or 2 µM DFO abolished cytosolic aconitase activity of IRP1, and 10 µM DFO was sufficient to dramatically decrease mitochondrial aconitase activity (Fig. 5). In addition, the IRE-binding activity of IRP1 and IRP2 were maximal at 5 µM of DFO concentration, suggesting that the cells of the FA patients were relatively iron-starved prior to treatment.
In summary, our results demonstrated that cytosolic iron depletion may contribute to decreased expression of frataxin in FA patients and may thereby worsen disease (modeled in Fig. 6).
In this study, we have demonstrated that frataxin expression levels are regulated by iron in various human tissue culture cell lines, and in lymphoblasts and fibroblasts from controls and FA patients. Through use of promoter-luciferase constructs, we have further observed that transcriptional regulation likely accounts for much of the iron-dependent regulation of frataxin in controls. Importantly, significant iron-dependent regulation was also observed in the lymphoblast and fibroblast cell lines derived from FA patients. To further evaluate the importance of iron-dependent regulation of frataxin in patients, we analyzed cytosolic iron status by assaying activity of IRP1 and IRP2. IRP1 and IRP2 are mammalian proteins that register cytosolic iron concentrations and post-transcriptionally regulate expression of numerous iron metabolism genes to optimize cellular iron availability. IRP1 registers cytosolic iron status mainly through an iron–sulfur cluster at its active site; when the iron–sulfur cluster is present, IRP1 is an active cytosolic aconitase, whereas when the cluster is absent, IRP1 becomes an apoprotein that undergoes a significant conformational change, enabling it to bind to IREs, RNA stem-loop structures found within transcripts (reviewed in 22,26,27). Although IRP2 is homologous to IRP1, IRP2 activity is regulated primarily by iron-dependent degradation through the ubiquitin-proteasomal system in iron-replete cells (22,26,27). Targeted deletions of IRP1 and IRP2 in animals have demonstrated that IRP2 is the chief physiologic iron sensor in animals (28,29). Gel-shifts in this study showed that the IRE-binding activities of both IRP1 and IRP2 were increased in FA patient lymphoblasts, and IRP2 levels were increased relative to controls (Fig. 4), suggesting that there is cytosolic iron depletion in FA cells. Our measurements were not performed on the patient tissues that are very adversely affected in FA, including the heart (6) and cerebellum (5). Thus, even though frataxin levels were markedly reduced, mitochondrial aconitase activity was normal in lymphoblasts from a patient with only 10% of normal frataxin levels, although cytosolic aconitase levels were significantly reduced (Fig. 4C, lanes C4 and P4, C5 and P5). Levels of both cytosolic aconitase and mitochondrial aconitase were significantly reduced in FA fibroblast cells in which frataxin levels were ∼35% of normal (Fig. 4C, 30% for lane C1 and P1, 66% for lane C2 and P2, 10% for lane C3 and P3), but not as significantly as has been reported in heart samples of patients (30), consistent with the observation that disease affects primarily heart and central nervous system, whereas other tissues are less adversely affected.
Since we found evidence for cytosolic iron depletion in cell types that do not develop phenotypes in FA, we postulate that cytosolic iron depletion may be even more pronounced in areas of the brain and heart that are functionally compromised in FA patients. Interestingly, expression of ISCU was decreased in the heart of frataxin-deficient mice (31), and cytosolic iron deficiency perhaps could explain the decrease in ISCU expression (14). In the brain and heart of FA patients, mitochondrial iron overload is prominent and is readily detected with tissue iron stains. In numerous model systems, mutations or deficiencies of iron–sulfur cluster biogenesis proteins result not only in diminished activities of iron–sulfur enzymes such as aconitase and succinate dehydrogenase, but also in mitochondrial iron overload, which appears to be a secondary effect that results from the compromise of iron–sulfur cluster biogenesis (32) (reviewed in 11). Thus, the cytosolic iron depletion that we observe in FA cells conforms to a pattern of compartmental iron misregulation observed when iron–sulfur cluster biogenesis is impaired, characterized by mitochondrial iron overload and cytosolic iron depletion.
Therapeutic approaches to treatment of FA presently include three broad areas: methods to increase frataxin expression by decreasing heterochromatin formation at the GAA repeat (33–35), methods to chelate iron and thereby reduce associated oxidative stress (9) (10) and methods to directly reduce oxidative stress (36), which have recently been shown to be effective in a fly model for FA (37).
Since evidence in favor of cytosolic iron depletion has not been previously reported in FA patients of cell lines, there have been no efforts to increase cytosolic iron stores in patients. Interestingly, several studies have reported that adverse phenotypes can be ameliorated by treatment with hemin, although the mechanism is unknown (38,39). Perhaps hemin treatment increases frataxin expression by increasing cytosolic iron levels, as we have observed in this study when cells were treated with the iron salt, FAC (Figs 3C and 5).
Our results imply that the proposed cytosolic iron depletion that develops with frataxin deficiency may exacerbate disease by further reducing frataxin transcription. Since we observed that frataxin levels varied over at least a 2.5-fold range with manipulations of iron, we suggest that frataxin expression could potentially be increased significantly in patients by judicious management of iron status. We conclude that addition of an iron supplementation regime that would replenish cytosolic iron might be of value in conjunction with other therapeutic approaches such as HDAC inhibitors (35), iron chelators that would reduce mitochondrial iron stores (9,40), and antioxidants that would mitigate the effects of mitochondrial iron overload (36).
MATERIALS AND METHODS
Chemicals and cell cultures
FAC and deferoxamine (DFO, commercial name ‘desferal’) were purchase from Sigma (St Louis, MO). HEK293 cell line was obtained from ATCC (Rockville, MD), lymphoblasts GM15851 and AG15799 (healthy), GM15849, GM16215 and GM16200 (unaffected carriers), GM15850, GM16209, GM16197, GM16214 and GM04079 (FA patients), apparently healthy fibroblasts, GM08399, AG09429 and GM08402, and FA patient fibroblasts GM03665, GM03816 and GM04078 from the Coriell Cell Repostories (Camden, NJ). All the medium and antibiotics (when necessary) were bought from Invitrogen (Carlsbad, CA). The condition for cultivation of the cells was followed according to the instruction of the manufactories. Cells were incubated in a humidified atmosphere of 5% CO2 at 37°C.
Northern blot and qRT–PCR
Total RNA was isolated from 80% confluence of cells after treatment as needs with TRizol (Invitrogen). polyA RNA for northern blot was isolated with PolyATract mRNA isolation kit (cat# Z5210, Promega, Madison, WI). Northern hybridization was carried out and the blot was exposed to PhosphoImage screen and detected with Typhoon 9200 (GE Healthcare). For qRT–PCR, 1 µg total RNA was used in the reverse transcription reaction with random primers (Applied Biosystems, cat# 4368814). Quantitative PCR was carried out with a kit (Invitrogen, cat#11744). Primers for FXN gene are: 5′-CCTTGCAGACAAGCCATACA-3′ and 5′-CCACTGGATGGAGAAGATAG-3′, for control GAPDH gene are: 5′-TGCACCACCAACTGCTTAGC-3′ and 5′-GGCATGGACTGTGGTCATGAG-3′.
Plasmid construction and transfection
Wild-type full length of human frataxin gene coding for 210 amino acids was amplified with primers 5′-AATCTCGAGATGGAAACTCTCGGGCGCCGCGCAGTAG-3′ and 5′-GGGGGAAGCTTTCAAGCATCTTTTCCGGAATAGGCC-3′, then cloned into plasmid pcDNA3.1(−) with restriction sites (XhoI and HindIII) to generate plasmid pcDNA3.1-mito-hFRDA. Site-directed mutagenesis was carried out with QuikChange® II Site-Directed Mutagenesis Kits (Stratagene, La Jolla, CA) to mutate 200th and 203rd amino acid leucine (L) into methionine (M) to produce plasmid pcDNA3.1-mito-hFRDA L200/203M with primers 5′-CCAAACTGGACATGTCTTCCATGGCCTATTCCGG-3′ and 5′-CCGGAATAGGCCATGGAAGACATGTCCAGTTTGG-3′. Plasmid transfections were performed using Fugene6 (Roche Applied Science, Indianapolis, IN) according to the manufactur′s instructions. Stable cell line formation was confirmed by western and metabolic labeling.
Reporter gene assays
A template DNA construct FXN81658 (16) or a BAC construct RP11-265B8 (BACPAC Resources Center, Oakland, CA) was used to generate different promoter constructs for cloning by PCR. Promoter fragments that extended 5′ of the initiation methionine of FXN exon 1 were cloned by adding KpnI and NcoI sites to the PCR primers, and were then inserted into pGL4.10[luc2] (Promega) (17). Cloned fragments contained the 1.2 kb promoter region (16) for construct FXN-luc1, a larger 1.8 kb fragment for FXN-luc2, or a 3.2 kb fragment in FXN-luc3. In addition, a fragment that contained the promoter (1.3 kb), the entire first exon, and the first 1241 bases of intron 1 of human FXN was included in the FXN-GAA-luc plasmid. For the FXN-GAA-luc plasmid, a reverse primer containing the splice acceptor site from the 3′ end of intron 1 and an NcoI site was designed so that the FXN coding sequence in exon 1 would be translated in frame with the luciferase open reading frame from pGL4.10[luc2] (16). The four plasmids, generated as FXN-luc1, FXN-luc2, FXN-luc3 and FXN-GAA-luc, were, respectively, co-transfected with pGL4.75 into HEK293, HeLa and RD4 cells. The luciferase assays were conducted with Veritas microplate luminometer (Turner Biosystems, Sunnyville, CA) according to the manufacturer's manual (Promega, cat#TM040). To control for transfection efficiency in each well, firefly luciferase activity was normalized to Rellina luciferase activity.
HEK293 cells and stably transfected HEK293 with pcDNA3.1-mito-hFRDA L200/203M were cultivated to reach confluence 80%, then labeled with [35S]-methionine (Perkin Elmer, Waltham, MA) for 1 h. After radiolabeling, the cells were chased. Harvested cells were lysed, followed by IP with antibody against human frataxin (MitoScience, Eugene, OR). After the first IP, a second IP (recapture) was performed as previously described (41).
Western blot analysis was performed as previously described (42), using the above-described frataxin antibody. IRP1 and IRP2 antibodies were polyclonal, raised from rabbit (29,14). H- and L-ferritin antibodies against human spleen ferritin were purchased from Santa Cruz Biotech Inc. (Santa Cruz, CA), transferrin receptor antibody from Zymed (San Francisco, CA), α-tubulin and actin antibodies from Sigma. Quantification of the density of the western bands was done with program ImageJ (http://rsb.info.nih.gov/ij/).
RNA mobility shift and in-gel aconitase activity assays
Gel retardation assays that separate human IRP1 and IRP2 and in-gel aconitase activity assays were performed as described previously (14).
All statistical analysis in this study was done with the Student's t-test (http://www.danielsoper.com/statcalc/).
This work was supported by the intramural program of the NICHD, National Institutes of Health, and in part by Friedreich ataxia research association.
We thank Dr Manik Ghosh for preparation of the IRE probe, and Dr Karen Usdin for the generous gift of plasmid FXN81658. Funding to pay the Open Access publication charges for this article was provided by the intramural program of NICHD.
Conflict of Interest statement. None declared.