© 2020. Published by The Company of Biologists Ltd | Disease Models & Mechanisms (2020) 13, dmm045229. doi:10.1242/dmm.045229

RESEARCH ARTICLE Mitochondrial damage and senescence phenotype of cells derived from a novel frataxin G127V point mutation mouse model of Friedreich’s ataxia Daniel Fil1, Balu K. Chacko2,3,4, Robbie Conley1, Xiaosen Ouyang2,3,4,5, Jianhua Zhang2,3,4,5, Victor M. Darley- Usmar2,3,4, Aamir R. Zuberi6, Cathleen M. Lutz6, Marek Napierala1 and Jill S. Napierala1,*

ABSTRACT INTRODUCTION ’ Friedreich’s ataxia (FRDA) is an autosomal recessive Friedreich s ataxia (also called FA or FRDA) is an autosomal neurodegenerative disease caused by reduced expression of the recessive disease with an estimated prevalence of 2-4 per 100,000 mitochondrial protein frataxin (FXN). Most FRDA patients are individuals (Leone et al., 1990; Winter et al., 1981). Patients homozygous for large expansions of GAA repeat sequences in develop progressive ataxia of all four limbs, associated with intron 1 of FXN, whereas a fraction of patients are compound cerebellar dysarthria, absent reflexes in the lower limbs, sensory loss heterozygotes, with a missense or nonsense mutation in one FXN and abnormal pyramidal signs. Additionally, some patients develop allele and expanded GAAs in the other. A prevalent missense optic atrophy, sensorineural hearing loss, diabetes mellitus, foot mutation among FRDA patients changes a glycine at position 130 to deformity, scoliosis and hypertrophic cardiomyopathy (Filla et al., valine (G130V). Herein, we report generation of the first mouse model 1996; Harding, 1981). Currently, there is no effective treatment. harboring an Fxn point mutation. Changing the evolutionarily FRDA is caused in the majority (96%) of cases by conserved glycine 127 in mouse Fxn to valine results in a failure-to- hyperexpansion of trinucleotide GAA repeats located in the first FXN thrive phenotype in homozygous animals and a substantially reduced intron of the frataxin ( ) gene on both alleles. The remaining number of offspring. Like G130V in FRDA, the G127V mutation results (4%) of the patients are compound heterozygotes for GAA FXN in a dramatic decrease of Fxn protein without affecting transcript expansion on one allele and a point mutation on the other synthesis or splicing. FxnG127V mouse embryonic fibroblasts exhibit (Cossee et al., 1999; Durr and Brice, 1996; Filla et al., 1996; Galea significantly reduced proliferation and increased cell senescence. et al., 2016). Although not affecting the coding sequence, GAA These defects are evident in early passage cells and are exacerbated repeats impede transcription, leading to low mRNA and protein at later passages. Furthermore, increased frequency of mitochondrial levels. Regarding point mutations on the protein coding sequence, ≥ DNA lesions and fragmentation are accompanied by marked 30 pathogenic point mutations have been identified (Clark et al., amplification of mitochondrial DNA in FxnG127V cells. Bioenergetics 2019; Cossee et al., 1999; Galea et al., 2016), and only FXN R165C analyses demonstrate higher sensitivity and reduced cellular has been identified in individuals in a homozygous state (Candayan FXN respiration of FxnG127V cells upon alteration of fatty acid availability. et al., 2020). Many of the point mutations affect mRNA Importantly, substitution of FxnWT with FxnG127V is compatible with life, expression or the initiation of translation and result in more severe and cellular proliferation defects can be rescued by mitigation of clinical presentations when compared with classic repeat expansion oxidative stress via hypoxia or induction of the NRF2 pathway. We patients (Bidichandani et al., 1997; Cavadini et al., 2000b; Galea propose FxnG127V cells as a simple and robust model for testing et al., 2016; Santos et al., 2010). Interestingly, a prevalent point therapeutic approaches for FRDA. mutation in FRDA, a single nucleotide change (c.389G>T) resulting in a missense substitution at amino acid 130 of glycine to valine (G130V), is associated with a milder clinical presentation ’ KEY WORDS: Friedreich s ataxia, Senescence, Mitochondria, and slower disease progression than most FRDA cases arising from Frataxin, Point mutation, Oxidative stress homozygous GAA repeat expansions (Bidichandani et al., 1997; Clark et al., 2017; McCabe et al., 2000; Puccio and Koenig, 2000). 1Department of Biochemistry and Molecular Genetics, University of Alabama at Symptoms common for homozygous repeat expansion FRDA, Birmingham, 1825 University Boulevard, Birmingham, AL 35294, USA. including dysarthria, loss of tendon reflexes and ataxia, are not 2Department of Pathology, University of Alabama at Birmingham, 901 19th Street South, Birmingham, AL 35294, USA. 3Center for Free Radical Biology, University of typically observed in G130V individuals, although they do Alabama at Birmingham, Birmingham, AL 35294, USA. 4Mitochondrial Medicine distinctively exhibit an early onset spastic gait and increased Laboratory, University of Alabama at Birmingham, Birmingham, AL 35294, USA. prevalence of optic disk pallor and diabetes (Cossee et al., 1999; 5Department of Veteran Affairs Medical Center, Birmingham, AL 35294, USA. 6The Rare and Orphan Disease Center, JAX Center for Precision Genetics, 600 Main Galea et al., 2016). This suggests that the frataxin G130V Street, Bar Harbor, ME 04609, USA. (FXNG130V) protein is biologically active and contributes to a

*Author for correspondence ( [email protected]) unique pathological mechanism. Frataxin is a nuclear-encoded mitochondrial protein that is highly J.S.N., 0000-0002-5141-0822 conserved throughout evolution, with homologs in eukaryotes

This is an Open Access article distributed under the terms of the Creative Commons Attribution including mammals, invertebrates and yeast (Dhe-Paganon et al., License (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use, 2000). Complete absence of wild-type Fxn (FxnWT) is incompatible distribution and reproduction in any medium provided that the original work is properly attributed. with life, because Fxn null mice die early during embryogenesis

Handling Editor: Steven Clapcote (Cossee et al., 2000), and mouse embryonic fibroblasts (MEFs) WT Received 9 April 2020; Accepted 16 June 2020 lacking Fxn fail to divide (Calmels et al., 2009). After translation Disease Models & Mechanisms

1 RESEARCH ARTICLE Disease Models & Mechanisms (2020) 13, dmm045229. doi:10.1242/dmm.045229 in the cytoplasm, frataxin is transported into the mitochondria using expression of the FxnG127V protein is sufficient for viability, an N-terminal targeting sequence. Upon entry, it undergoes a two- although significant postnatal lethality was observed. Moreover, step proteolytic cleavage by mitochondria processing peptidase cells derived from homozygous mutant embryos (FxnG127V/G127V; (MPP) to generate the mature protein (Koutnikova et al., 1998). It MUT) demonstrate a significant increase in a senescent phenotype, has been reported that the G130V mutation modulates interaction significant mitochondrial damage and reduced proliferative capacity with MPPβ and affects the maturation process (Cavadini et al., that can be rescued after incubation in a hypoxic environment or 2000a). In fact, the ratio of intermediate to mature frataxin protein is treatment with the nuclear factor erythroid 2 (NF-E2)-related increased in fibroblasts from FRDA G130V patients, suggesting an factor 2 (NRF2) activators dimethylfumarate (DMF) or impairment in maturation processing (Clark et al., 2017). The omaveloxolone (RTA 408), conditions and compounds that G130V mutation has been found to destabilize the protein structure, mitigate oxidative stress (La Rosa et al., 2020; Saidu et al., 2019; leading to low levels of mature frataxin in yeast (Cavadini et al., Wong et al., 1999). This new cellular model could be considered as 2000a). a screening platform for testing the efficacy of small molecules and Despite the availability of numerous disease models, no single human FXN or FXN replacement therapies aimed to alleviate the model perfectly mimics the human FRDA state. Several FRDA effects of frataxin deficiency. mouse models have been generated thus far, but all of them aimed to recapitulate a condition of low frataxin expression by engineering RESULTS long tracts of the expanded GAAs (Miranda et al., 2002; Pook et al., Generation of frataxin G127V knock-in mice 2001), conditional deletion of Fxn (Puccio et al., 2001) or transcript Human and mouse mature frataxin proteins share 71.5% amino acid depletion (short hairpin RNA and small interfering RNA) sequence identity (CLUSTALW; 73% by BLASTP), indicating that (Chandran et al., 2017). Thus far, no animal models have been the outcome of the amino acid change is comparable between the reported that specifically analyze pathological consequences of two (Fig. 1A). To study the effects of the disease-causing point frataxin point mutations. mutation on the endogenous Fxn protein in vivo, we used CRISPR/ To determine the impact of the G130V mutation in vivo,we Cas9 to introduce the (c.401G>T) mutation in exon 4 of murine Fxn generated mice harboring the homologous mutation in murine (Fig. 1B). The resulting G127V substitution is equivalent to the frataxin (G127V). Our studies indicate that in the absence of FxnWT, human pathogenic G130V missense mutation. Additionally, a silent

Fig. 1. Generation of FxnG127V mice via CRISPR/Cas9-mediated knock-in strategy. (A) Alignment of human and mouse frataxin amino acid sequences, with the position of glycine 130/127 in bold type and surrounded by a box. (B) Sequence of the edited region of the Fxn gene: scissors illustrate location of DNA break in exon 4 of Fxn; gray box indicates position of single guide RNA (sgRNA); arrow indicates location of missense mutation (red box) and silent mutation (green box). The result of the modification is G→V substitution and creation of an AatII restriction site. F and R illustrate locations of genotyping primers. The abbreviation ssODN refers to the single-stranded oligodeoxynucleotide donor template. PAM, protospacer adjacent motif. (C) Representative agarose gel electrophoresis image showing restriction fragment length polymorphism genotyping products for WT, HET and MUT cells. (D) Observed frequency of genotype distribution at weaning is plotted, with the number of animals per group indicated on each pie chart; n=276 biological replicates. (E) The observed frequency of the genotype distribution of embryos at 20 dpc is plotted, with the number of animals per group indicated on each pie chart; n=67 biological replicates. Disease Models & Mechanisms

2 RESEARCH ARTICLE Disease Models & Mechanisms (2020) 13, dmm045229. doi:10.1242/dmm.045229 mutation (c.399T>C) was introduced to serve as a Cas9 blocking embryos were significantly underweight compared with their WT mutation and to create a new restriction site (AatII/ZraI) for and HET counterparts, suggesting a disadvantage for competitive genotyping (Fig. 1C). survival after birth (Fig. S1). Furthermore, mating of HET to WT After CRISPR/Cas9 editing, the integrity of the Fxn coding (C57BL/6J) did not result in a significant difference in mutant allele sequence and the incorporation of the G127V point mutation were transmission (P=0.074; Table S1). verified by sequencing complementary DNA generated from Owing to the low number of MUT animals immediately available cerebella of two mice heterozygous for the FxnG127V allele for study, we conducted molecular analyses on effects of the G127V (FxnG127V/+). The results from both animals indicated the mutation on Fxn expression and function using MEFs, because they presence of only the two engineered point mutations, with the could readily be isolated and cultured ex vivo. This homogeneous remaining Fxn coding sequence being free of errors. Inter-crossing cellular model allows, for the first time, in-depth analyses of the of FxnG127V/+ mice resulted in progeny of all possible genotypes: endogenous FxnG127V protein in the absence of FxnWT. WT, mice homozygous for wild-type frataxin (Fxn+/+; n=77; 28%); HET, mice heterozygous for Fxn+ and FxnG127V alleles (FxnG127V/+; Endogenous Fxn protein levels are markedly reduced by the n=175; 63%); and MUT, mice homozygous for the FxnG127V allele G127V mutation (FxnG127V/G127V; n=24; 9%). MUT mice were viable but were To determine how the mutation affected Fxn expression, we first under-represented postweaning (9% instead of the expected 25%; measured the levels of Fxn mRNA and Fxn protein in MEF cell Fig. 1D). However, reduced representation of MUT animals was not lines established from WT, HET and MUT embryos. Analyses by detected at the late embryonic stage, 20 days postcoitum (dpc) quantitative reverse transcription PCR (RT-qPCR) demonstrated (Fig. 1E; Fig. S1), suggesting that MUT mice develop normally in that the G127V mutation did not decrease Fxn mRNA expression utero but fail to thrive shortly after birth. Notably, late-stage MUT (Fig. 2A). In addition, because glycine 127 is the second amino acid

Fig. 2. FxnG127V mRNA expression is not correlated with immunodetectable protein levels. (A) Frataxin mRNA expression levels measured by real-time RT- qPCR using RNA extracted from WT, HET and MUT MEFs. Fxn transcripts were normalized to Gapdh and plotted relative to WT samples. Data are shown for three different primer pairs used for analyses: Ex2_Ex3 (spanning exons 2 and 3; upstream of the mutation); Ex3_Ex4 (spanning exons 3 and 4; encompassing the mutation site); and Ex4_Ex5 (spanning exons 4 and 5; downstream of the mutation). Bars show the mean±s.d. and represent data from two independent MEF lines per genotype (n=2 biological replicates), combined from two independent experiments; total measurements per bar=4. Student’s unpaired t-test (*P<0.05). (B) Frataxin protein expression levels in WT and MUT MEFs was determined by western blot. FxnG127V protein can be detected only with enhanced sensitivity western blotting techniques. To avoid oversaturation of the signal, 10 μg of WT lysate and 100 μg of MUT lysate were loaded per lane. The blot shown is representative of three experiments performed with two independent MEF lines per genotype (n=2 biological replicates); total measurements=6. (C) Western blot analysis of protein samples fractionated to soluble and insoluble fractions is shown, with Hprt and Ponceau S staining serving as loading controls. The blot shown is representative of two experiments performed with two independent MEF lines per genotype (n=2 biological replicates); total measurements=4. (D) A representative western blot showing mitochondria-enriched fractions prepared from WT, HET and MUT MEFs. Fifty micrograms of each fraction was loaded per lane. Hprt serves as a positive control for the cytosolic fraction, whereas Nfs1 and Iscu are positive controls for the mitochondrial fraction. The blot shown is representative of two experiments performed with two independent MEF lines per genotype (n=2 biological replicates). Quantitative values are plotted as the mean±s.d. (total measurements per bar=4), with significant differences determined using ordinary one-way ANOVA (**P<0.01, ***P<0.001). IB, immunoblot. Disease Models & Mechanisms

3 RESEARCH ARTICLE Disease Models & Mechanisms (2020) 13, dmm045229. doi:10.1242/dmm.045229 of exon 4, we tested whether the nucleotide substitutions introduced below the levels of FxnWT protein measured in human FRDA cases near the 5′ end of this exon affected splicing of the Fxn transcript. caused by homozygous repeat expansions (10-30% of levels in RT-qPCR results using primer pairs designed upstream, spanning healthy people) (Deutsch et al., 2010; Long et al., 2017; Nachbauer and downstream of the targeted site revealed that introduction of the et al., 2011; Saccà et al., 2013) and other FRDA mouse models [e.g. mutations did not decrease steady-state levels of the Fxn transcript knock-in/knockout (KIKO) mice (25-36% of WT) (Miranda et al., (Fig. 2A). Moreover, RT-qPCR products spanning the Fxn exon 3-4 2002; Perdomini et al., 2013)]. Nevertheless, this low amount of junction were amplified from MUT mRNA, cloned and sequenced, FxnG127V protein was sufficient for MEF survival and mouse and the results (5/5 clones) demonstrated no abnormally spliced development. transcripts. Taken together, analyses of Fxn mRNA from WT, HET and MUT MEFs indicated that introduction of the G127V mutation Cells expressing only FxnG127V exhibit reduced proliferation did not decrease production or negatively affect splicing of Fxn We next turned our attention to overt cellular phenotypes observed transcripts. while establishing and culturing independent batches of MUT, HET Next, we turned our attention to how the G127V mutation and WT MEF cell lines. To measure growth rates of lines impacted Fxn protein levels, because we have previously reported representing the three genotypes, equal numbers of cells were greatly reduced total Fxn levels, especially of the mature (M) seeded, and the cells were counted every 24 h over a 6-day period. isoform, in fibroblasts derived from FRDA G130V patients (Clark We consistently noted slower proliferation of FxnG127V et al., 2017). We first tested three distinct, commercially available homozygous MUT MEFs when compared with WT and HET cell antibodies using lysates prepared from HEK-293T cells transiently lines that were established in parallel (Fig. 3A,B). Calculated expressing exogenous murine FxnWT or FxnG127V proteins to population doubling times (PDTs) of WT, HET and MUT MEFs address the possibility of epitope masking by the point mutation from three independent experiments were 31.8, 30.7 and 62.3 h, (Fig. S2A). Albeit at much lower levels than FxnWT, the FxnG127V respectively (Fig. 3C), which is a 2-fold slower growth rate of MUT protein was detected by both FXN polyclonal antibodies (first and MEFs when compared with WT or HET. second panels). However, as previously reported (Koutnikova et al., To test whether the diminished growth rate of MUT MEFs was 1998), the FXN monoclonal antibody (clone 1G2) failed to attributable to slower progression through the cell cycle, we recognize the mutant protein (third panel). Western blotting of analyzed their distribution between G1, S and G2 phases using whole cell lysates revealed that FxnG127V protein levels were flow cytometry. In synchronized cultures, a smaller percentage of profoundly decreased in MUT compared with HET or WT MEFs to MUT cells were actively dividing compared with WT or HET MEFs such an extent that western blot techniques with enhanced (Fig. 4A,B). The WT and HET MEFs entered S phase ∼18 h after sensitivity were necessary to detect the FxnG127V protein in whole serum supplementation, whereas MUT MEFs mostly remained cell lysates (Fig. 2B,C). Similar results were obtained with the other quiescent beyond 26 h (the last measurement taken). These results FXN polyclonal antibody tested (Fig. S2B; GTX54036). aligned with the calculated PDTs (Fig. 3C) and demonstrated a In light of the low western blot signal from MUT samples, we profound proliferation defect in cells expressing a low level of examined the possibility of FxnG127V protein aggregation. As FxnG127V protein in the absence of FxnWT. evident from analysis of insoluble fractions of whole cell lysates, neither FxnWT nor FxnG127V was found to aggregate (Fig. 2C). FxnG127V MUT MEFs exhibit increased senescence Finally, despite lower immunodetectable levels, the FxnG127V In addition to suppression of the cell cycle, we considered additional protein was exclusively found in mitochondria-enriched fractions, processes that could negatively impact MUT cell numbers in culture in a similar manner to FxnWT protein (Fig. 2D). Western blot (Fig. 3). Given that most of the MUT cell population appeared to be quantitation revealed that FxnG127V levels in MUT mitochondrial arrested in the G1 phase (Fig. 4B), we turned our attention first to extracts were only 5% of those observed for FxnWT protein in WT cellular senescence. It was previously shown that acute and severe mitochondrial extracts (Fig. 2D). These results indicated that depletion of FXN (∼80% reduction) in immortalized cells resulted steady-state levels of immunodetectable FxnG127V protein were in inhibited proliferation, accompanied by increased G1 phase arrest

Fig. 3. FxnG127V MUT MEFs exhibit slow growth in culture. (A) Representative phase-contrast images of WT, HET and MUT MEFs at days 1 and 3 in culture after equal plating. Three independent growth curve experiments were performed using two independent MEF lines per genotype (n=2 biological replicates). Scale bars: 500 μm. (B) Growth analysis of WT, HET and MUT MEFs over 6 days in culture. Cells (at passage 3) were seeded at equal densities in duplicate wells at day 0, then counted every 24 h. The experiment was repeated three times using two independent MEF lines per genotype (n=2 biological replicates), and a representative curve is shown as the mean±s.d. (C) WT, HET and MUT MEF population doubling times as calculated from the growth curves (B). Bars show the mean±s.d. calculated from three independent growth curve experiments; total measurements per bar=3. Significant differences were determined using ordinary one-way ANOVA (*P<0.05). Disease Models & Mechanisms

4 RESEARCH ARTICLE Disease Models & Mechanisms (2020) 13, dmm045229. doi:10.1242/dmm.045229

Fig. 4. FxnG127V MUT MEFs proliferate more slowly after cell cycle synchronization. (A) A schematic representation of the cell cycle synchronization protocol is shown, whereby quiescence is induced by serum deprivation for 72 h, followed by serum restoration and progression into the cell cycle. (B) Flow cytometry analysis of cell cycle distribution of WT, HET and MUT MEFs (at passage 4) monitored from 16 to 26 h after serum restoration. Each time point is an average of two replicates. and cellular senescence (Bolinches-Amoros et al., 2014). To test the proportion of apoptotic cells was higher in MUT than in WT whether MUT MEFs, expressing low levels of FxnG127V, were and HET MEF populations, the difference did not reach statistical prone to a similar fate, we examined induction of senescence by significance (MUT, 7.7±1.1%; WT, 5.5±2.7%; and HET, 3.5 detection of senescence-associated β-galactosidase (SA-β-gal) ±0.4%). Taken together, these data demonstrated that in the activity, a known characteristic of senescent cells (Dimri et al., absence of FxnWT, expression of FxnG127V was sufficient for 1995). As expected, WT and HET MEF cultures were nearly devoid survival, but that murine cells expressing only the mutant protein of senescent cells, with only 2.4±1.6% and 1.7±0.5% of the cultures had reduced proliferative capacity and augmented cell death and staining positive for SA-β-gal activity, respectively (Fig. 5A,B). In senescence. contrast, cells positive for SA-β-gal activity represented 10.3±1.1% in MUT MEF cultures, indicating a ∼4.3-fold increase in Mitochondrial integrity in FxnG127V MUT MEFs decreases senescence induction in MUT MEFs compared with WT over time (Fig. 5B). Moreover, we found that MUT MEFs senesced at a Considering that consequences of frataxin deficiency are acutely significantly faster rate than WT cells when compared at each observed in mitochondria (Stepanova and Magrané, 2020), we passage (Fig. S3). sought to evaluate effects of FxnG127V expression on functions of In addition to senescence, acute frataxin depletion has been this organelle. Increased mitochondrial fragmentation has been shown to induce cell death via apoptosis (Loría and Diáz-Nido, reported for some neuronal subtypes derived from FRDA mouse 2015; Mincheva-Tasheva et al., 2014; Palomo et al., 2011). To models (Lin et al., 2017b; Mollá et al., 2017), and fragmentation in determine whether MUT MEFs were predisposed to apoptosis, we FRDA patient cells becomes apparent after an external oxidative quantified the percentage of viable, apoptotic and dead cells in insult (Lefevre et al., 2012). To assess mitochondrial morphology actively dividing cultures using flow cytometry. WT and HET and network integrity in WT and MUT MEFs, cells were labeled MEF populations were composed mostly of viable cells (89±4% with MitoTracker Deep Red and imaged. As shown in Fig. 6A, and 92±0.7%, respectively), whereas the percentage of viable cells mitochondrial networks were not qualitatively different between the in MUT MEF populations was significantly lower (75±0.8%; cell lines at early passages, but fragmented networks were observed P<0.0001) (Fig. 5C,D). By contrast, the percentage of dead cells more frequently in MUT MEFs at late passages compared with [propidium iodide (PI) and annexin V positive] was significantly those in WT cells. Measurement of the length of mitochondria with higher for MUT populations (12.8±1.7%) compared with WT (4.5 IMARIS Filament Tracer software revealed shorter filament lengths

±0.8%; P<0.001) or HET (3.3±0.5%; P<0.001). Finally, although in late passage MUT cells (Fig. 6B). Disease Models & Mechanisms

5 RESEARCH ARTICLE Disease Models & Mechanisms (2020) 13, dmm045229. doi:10.1242/dmm.045229

Fig. 5. FxnG127V MUT MEFs are prone to senescence and cell death. (A) Representative phase-contrast images of WT, HET and MUT MEFs stained for detection of SA-β-galactosidase (blue cells) taken at ×200 total magnification. Scale bars: 200 μm. (B) Quantification of senescent cells in WT, HET and MUT MEF cultures is plotted as a percentage of total cells counted. Each bar represents the mean±s.d. of six independent fields, in which ≥125 cells were counted per field; n=2 biological replicates, total measurements per bar=6. A significant difference between the MUT group and the WT and HET groups was determined by ordinary one-way ANOVA (****P<0.0001). (C) Representative scatter plots are shown for flow cytometry analyses after annexin V and propidium iodide (PI) staining of WT, HET and MUT MEFs. The histograms illustrate the number of cells stained with PI (y-axis) and/or annexin V (x-axis), and the populations are divided into quadrants (Q1-Q4). Ten thousand events were collected for each measurement, and measurements were repeated in two independent experiments with two technical replicates per experiment; n=2 biological replicates, total measurements=4. (D) The percentages of live (Q4), apoptotic (Q3) and dead (Q2) cells within WT, HET and MUT MEF cultures were averaged from two independent annexin V/PI flow cytometry experiments; n=2 biological replicates; total measurements per bar=4. Significant differences were determined between MUT and WT, HET groups by two-way ANOVA using Tukey’s method for multiple comparisons (***P<0.001, ****P<0.0001). Cells used for all staining experiments were early passage (passage 3 or 4).

Frataxin deficiency is also linked with mitochondrial DNA Mitochondria of FxnG127V MUT MEFs exhibit altered fatty acid (mtDNA) damage (Haugen et al., 2010; Karthikeyan et al., 2003). utilization Indeed, our previous study in FRDA fibroblasts found that low FXN FXNWT knockdown (KD) and overexpression both negatively levels led to increased mtDNA damage and decreased mtDNA impact mitochondrial energy production (Bolinches-Amoros et al., repair capacity (Bhalla et al., 2016). To determine whether the 2014; Vannocci et al., 2018). However, the effect of FXN point integrity of mtDNA was compromised in cells expressing only mutations on cellular bioenergetics has not been studied. To assess FxnG127V, we analyzed mtDNA isolated from WT and MUT MEFs the effect of FxnG127V expression on bioenergetics, we measured using a quantitative PCR assay based on the principle that DNA cellular oxygen consumption rate (OCR) directly in living cells. lesions can slow down or block the progression of DNA polymerase Initially, we performed mitochondrial stress tests for WT and MUT (Bhalla et al., 2016; Ponti et al., 1991; Redmann et al., 2018). This MEFs in basal conditions (unchallenged; Fig. 7A,B) (Dranka et al., type of assay detects various forms of DNA damage, from single- or 2011; Hill et al., 2012). Interestingly, basal and maximal respiration double-strand DNA breaks and gaps to specific chemical lesions rates were not different between WT and MUT MEFs at early or late (Lehle et al., 2014). Quantitation of long (10,085 bp) and short passages in these conditions. However, an increase in proton leak (117 bp) mtDNA fragments amplified from WT and MUT MEFs was detected in the MUT cells (compared with WT) at both early revealed an increase in lesion frequency in cells expressing only and late passages (Fig. 7B), which is consistent with a decrease in FxnG127V (Fig. 6C), consistent with a higher level of mtDNA bioenergetic efficiency, possibly related to mitochondrial damage. membrane damage. Finally, we calculated the mtDNA copy number by comparing Next, we assessed the status of fatty acid oxidation in FxnG127V the amount of mtDNA with an intergenic region of nuclear DNA MUT MEFs. Depletion of endogenous fatty acids followed by [fragment of the hexokinase 2 (Hk2) gene]. These results revealed palmitate supplementation revealed similar profiles for early that the mtDNA content was not changed in early passage MUT passage WT and MUT MEFs, suggesting that both had the MEFs but that it increased significantly in later passage cells capacity to utilize exogenous fatty acids as an energy source in these compared with WT MEFs (Fig. 6D). Taken together, these effects test conditions (Fig. 7C,D). To determine the dependence of the on mtDNA integrity and the number of mitochondria suggested that cells on oxidation of endogenous fatty acids, we next treated the FxnG127V MUT MEFs accumulated defective mtDNA over time MEFs with etomoxir, which, at low concentrations (10 µM), blocks which might then contribute to the emergence of the senescence ∼80% of mitochondrial import of fatty acids via irreversible phenotype. inhibition of carnitine palmitoyltransferase 1 (CPT1) (McGarry Disease Models & Mechanisms

6 RESEARCH ARTICLE Disease Models & Mechanisms (2020) 13, dmm045229. doi:10.1242/dmm.045229

Fig. 6. Increased mitochondrial damage in FxnG127V MUT MEFs. (A) Representative confocal images of WT and MUT passage (p) 4 and 6 MEFs stained with MitoTracker DeepRed FM are shown (z-stack maximum intensity projections). Cells were imaged using an oil immersion ×63 objective. Three independent experiments were conducted (staining and imaging) using two independent MEF lines per genotype (n=2 biological replicates). (B) Fields from ten images per group (A) were analyzed, and mitochondrial network filament lengths were calculated from an average of 415 measurements per field using IMARIS Filament Tracer software. Each bar represents the mean±s.d. of the averaged measurements per field; total averaged measurements per bar=6-9. The significant difference between MUT and WT filament lengths was determined by Student’s unpaired t-test (*P<0.05). (C) Relative quantitation of mitochondrial DNA lesions in WT and MUT MEFs calculated from the ratio of long PCR (10 kb) product normalized to the short PCR (117 bp) product is plotted as the mean±s.d; n=2 biological replicates, total measurements per bar=4. Significant differences were determined by Student’s unpaired t-tests (*P<0.05, **P<0.01). (D) Relative quantitation of mitochondrial DNA copy number normalized to genomic DNA is plotted as the mean±s.d; n=2 biological replicates, total measurements per bar=4. Significant differences were determined by Student’s unpaired t-tests (ns, not significant; *P<0.05, **P<0.01). et al., 1991; Yao et al., 2018). After etomoxir treatment, we (Fig. 8D) to levels that no longer differed significantly from observed a significant reduction of maximal respiration in both WT untreated WT cells, whereas idebenone treatment had no effect and MUT MEFs, with a greater inhibition shown by MUT cells for (Fig. 8E). Taken together, these results implicated oxidative stress as both ATP-linked and maximal respiration, in addition to their a driver of reduced survival and proliferation of FxnG127V MUT reserve capacity (Fig. 7E,F). In a similar manner, we analyzed fatty MEFs and suggested that restoring the redox balance via NRF2 acid utilization of late passage WT and MUT MEFs and observed activation significantly improved the viability of these cells. significant decreases in OCR for both types of cells after both treatments, with exacerbated effects observed for MUT MEFs DISCUSSION (Fig. 7G-J). Herein, we report generation of mice carrying a G127V mutation in the endogenous Fxn locus, analogous to the pathogenic G130V Mitigation of oxidative stress increases viability of FxnG127V mutation observed in individuals diagnosed with FRDA MUT MEFs (Bidichandani et al., 1997). Previous studies of FXNG130V in It was recently demonstrated that hypoxic conditions rescue frataxin MEF cultures demonstrate that transgenic expression of the mutant null organisms from lethality (Ast et al., 2019). Indeed, we observed human protein can rescue the lethal phenotype of Fxn knockout a significant increase in proliferation of MUT MEFs, rescued to (Calmels et al., 2009). The viability of our homozygous FxnG127V rates similar to those observed for WT and HET MEFs, when the mice also demonstrates that the FxnG127V mutant protein retains a cells were cultured in a hypoxic environment (Fig. 8A). The level of function that, on its own, sustains life in a whole organism. calculated PDTs for WT, HET and MUT cells grown in hypoxic Individuals carrying homozygous FXN G130V mutations have not conditions were 29.4, 28.4 and 31.1 h, respectively, with no been identified. The possibility exists that homozygous FXN significant differences determined between the genotypes (Fig. 8B). G130V mutations could result in embryonic or perinatal lethality, We also used XTT assays to test the effects of several compounds an idea supported, in part, by our results demonstrating reduced shown to improve biochemical and cellular phenotypes of FRDA representation of FxnG127V MUT mice after birth. Other cell line models, including idebenone (IDB) and omaveloxolone contributory factors could be under- or misdiagnosis owing to (RTA 408; Omav), which are molecules that have been or are underutilization of high-throughput sequencing methods to support currently included in clinical trials for FRDA (Zhang et al., 2019), molecular diagnoses and atypical clinical presentations. and dimethyl fumarate (Tecfidera; DMF), which is a molecule The normal Mendelian genotype distribution of mouse embryos approved for treatment of multiple sclerosis (http://www.fda.gov) at 20 dpc and reduced representation of MUT neonates is intriguing. (Gold et al., 2012; Saidu et al., 2019). Viability of MUT MEFs was MUT mice seem to develop normally in utero, but the majority fail improved after 24 h of treatment with Omav (Fig. 8C) and DMF to thrive shortly after birth. It is plausible that deficiencies resulting Disease Models & Mechanisms

7 RESEARCH ARTICLE Disease Models & Mechanisms (2020) 13, dmm045229. doi:10.1242/dmm.045229

Fig. 7. Bioenergetic characteristics of FxnWT and FxnG127V MUT MEFs. (A-J) Shown are OCRs and mitochondrial function indices recorded during mitochondrial stress tests conducted on the following: (A,B) early (p4) and late (p8) passage WT and MUT MEFs; (C,D) early (p4) passage WT and MUT MEFs with or without palmitate-BSA (25 µg/ml) treatment; (E,F) early (p4) passage WT and MUT MEFs with or without etomoxir (10 µM) treatment; (G,H) late passage (p8) WT and MUT MEFs with or without palmitate-BSA (25 µg/ml) treatment; or (I,J) late passage (p8) WT and MUT MEFs with or without etomoxir (10 µM) treatment. Data for unchallenged WT and MUT MEFs (A,B) are replotted on all graphs for side-by-side comparison of all treatments/conditions. Comparisons were made with Student’s unpaired t-tests between WT and MUT MEFs for each condition. Three independent experiments were performed for panels A-J, each with at least four technical replicates per sample. Representative plots are shown for each as the mean±s.e.m.; total measurements per bar=12 (unchallenged), 4 (etomoxir) and 4 (palmitate). AntiA, antimycin A; Eto, etomoxir; FCCP, carbonyl cyanide-4 (trifluoromethoxy)phenylhydrazone; Oligo, oligomycin; Palm, palmitate. *P<0.05, **P<0.01, ***P<0.0.001, ****P<0.0001. from low levels of FxnG127V protein hinder adequate metabolic Molecular analysis of MUT MEFs revealed that the G127V adaptation at birth, when the neonate must make a transition from a mutation does not reduce Fxn mRNA levels but has a profound continuous transplacental supply of glucose to a variable fat-based negative impact on immunodetectable Fxn protein levels. The food supply via a series of coordinated metabolic and hormonal G130V mutation was shown to reduce the solubility of purified changes coupled with a dramatic increase in oxygen availability recombinant human FXN proteins in vitro (Correia et al., 2008) or (Ward Platt and Deshpande, 2005). Extremely low levels of Fxn when exogenously expressed at high levels (Clark et al., 2017). (WT or G127V) could potentially exaggerate iron sulfur cluster However, our data indicate that the FxnG127V protein expressed from (ISC) deficiency, affect multiple ISC-dependent processes, the endogenous locus is detectable exclusively in soluble fractions. including respiration (Brzóska et al., 2006), DNA replication and Moreover, endogenous FxnG127V is localized to mitochondria, as is repair (Fuss et al., 2015) or ribosome maturation (Kispal et al., expected and observed for the FxnWT protein. 2005), and curtail the survival of MUT mice in the crucial period Depending on the FRDA model, different effects on the integrity after birth. Comprehensive studies will be necessary to assess (e.g. ultrastructure), number and function of mitochondria have developmental milestones and to determine the precise mechanism been reported. For instance, reduced mitochondrial biogenesis and, and timing of premature lethality in MUT mice. Owing to reduced subsequently, the number of mitochondria was reported in FRDA representation of postnatal mutant animals, we focused our initial patient-derived cells, in FXNWT KD control cells and in brain and analyses on MEFs isolated from FxnG127V MUT mice, and full skeletal muscle from the KIKO mouse model (Jasoliya et al., 2017; molecular and behavioral phenotyping of FxnG127V mice will be Lin et al., 2017a,b), whereas accumulation of damaged described elsewhere. mitochondria was observed in FRDA induced pluripotent stem Disease Models & Mechanisms

8 RESEARCH ARTICLE Disease Models & Mechanisms (2020) 13, dmm045229. doi:10.1242/dmm.045229

Fig. 8. Viability of FxnG127V MUT MEFs can be rescued by mitigating oxidative stress. (A) Growth analysis of WT, HET and MUT MEFs grown in hypoxic conditions over 6 days in culture. Cells were seeded at equal densities in duplicate wells at day 0, then counted every 24 h. The experiment was repeated twice using two independent MEF lines per genotype (n=2 biological replicates), and a representative curve is plotted as the mean±s.d. (B) WT, HET and MUT MEF population doubling times as calculated from the growth curves shown in A. Bars represent the mean±s.d.; total measurements per bar=4. (C-E) XTT assays were performed after 24 h treatments of WT and MUT MEFs with respective compounds. Absorbances were recorded (specific A465 nm and background A660 nm), and data for each treatment are expressed relative to untreated WT cells. Each bar represents the mean±s.d. of at least three independent experiments (black dots) performed using two independent MEF lines per genotype (n=2 biological replicates); total measurements per bar=5 (Omav), 4 (DMF) and 3 (IDB). Significant differences were determined by ordinary one-way ANOVA comparing each treatment with untreated WT (*P<0.05, **P<0.01, ****P<0.0001). cell-differentiated cardiomyocytes and cardiac Fxn conditional and late passage MUT MEFs. Extracellular flux analyses have so far knockout (cKO) mice (Hick et al., 2013; Perdomini et al., 2014; been reported only recently in FRDA cellular models, namely in Vyas et al., 2012). Our results demonstrate a significantly increased sensory neurons differentiated from induced pluripotent stem cells frequency of mtDNA lesions in FxnG127V MUT MEFs, even at early derived from patients harboring homozygous repeat expansions (i.e. passages, whereas an increased number of mitochondria is chronically low FXNWT expression) (Igoillo-Esteve et al., 2020) correlated with time in culture (i.e. with increasing passage). and an inducible FXN overexpression HEK293 cell line (i.e. short- Elevated mitochondrial biogenesis could be a compensatory term high FXNWT expression) (Vannocci et al., 2018). Basal and mechanism initiated by FxnG127V MUT cells to overcome the maximal respiration and ATP production were significantly reduced underlying load of damaged mitochondria. Indeed, this type of in FXNWT-deficient sensory neurons (Igoillo-Esteve et al., 2020), mechanism was previously proposed to explain similar results and no beneficial effects were observed for these parameters upon obtained using FRDA fibroblasts harboring homozygous repeat FXNWT overexpression (Vannocci et al., 2018). Expression of only expansions (García-Giménez et al., 2011). In addition to total FxnG127V protein was sufficient to maintain basal respiration at content, analyses of the dynamic state of mitochondria also levels observed for WT MEFs at early and late passages. provided insight into the networks formed over time in FxnG127V Impaired glucose utilization and increased β-oxidation in FRDA MUT MEFs. Although the mitochondrial networks formed in early patients was revealed by metabolic labeling of freshly isolated passage MUT cells appear indistinguishable from those in WT cells, platelets (Worth et al., 2015), and lipid metabolism defects have been filament lengths in later passage MUT MEFs are significantly reported in various FRDA cellular and animal models (Tamarit et al., shorter than those of WT cells at matched passage. Increased 2016). Our study on FxnG127V MUT MEFs also reveals altered fragmentation could serve as an independent indicator of excessive endogenous fatty acid utilization, and our results could indicate that mitochondrial damage in late passage FxnG127V MUT MEFs. MUT MEFs depend on oxidation of endogenous FAs for energy Elevated mitochondrial content is one feature of senescent cells production more than WT MEFs do, or that their available fatty acid and is often accompanied by increased production of reactive pool is lower. The increased dependence of MUT mitochondria on oxygen species, reduced membrane potential and reduced fatty acid oxidation could also be a compensatory mechanism for respiratory coupling (Korolchuk et al., 2017), additional impairments in glycolysis or other metabolic pathways and might be phenotypes that were revealed by our bioenergetic studies of early related to increased susceptibility to oxidative stress. Disease Models & Mechanisms

9 RESEARCH ARTICLE Disease Models & Mechanisms (2020) 13, dmm045229. doi:10.1242/dmm.045229

Some growth phenotypes typified by FXNWT-deficient cellular the Care and Use of Laboratory Animals published by the US National models, such as reduced proliferation and viability and increased Institutes of Health (NIH publication no. 85-23, revised 1996) and were sensitivity to oxidative stress (Carletti et al., 2014; Cotticelli et al., approved by the Institutional Animal Care and Use Committee at the 2019; La Rosa et al., 2019; Loría and Diáz-Nido, 2015; Palomo University of Alabama at Birmingham. et al., 2011), are observed in FxnG127V MUT MEFs. However, most of these studies report significantly increased apoptosis upon Generation of Fxn G127V mice depletion of FxnWT, whereas FxnG127V MUT MEFs instead The G127V knock-in mutation (GGC→GTC codon change) was introduced undergo growth arrest and senescence, suggesting a residual using CRISPR/Cas9 endonuclease-mediated genome editing via function of the mutant protein that sustains cell viability. The homologous-directed repair. Sequences for the guide RNA and ability of certain antioxidant agents to restore proliferation of MUT oligonucleotide donor used are provided in Table S2. A silent D126D (GAT→GAC) was co-introduced with the G127V mutation as a PAM MEFs supports this hypothesis. blocker. C57BL/6J zygotes with well-recognized pronuclei were Activation of the NRF2 pathway as a therapeutic target in FRDA microinjected with guide RNA, the mutation-bearing oligonucleotide has gained attention because positive results of a phase 2 study for donor and Cas9 nuclease. After injection, the embryos were transferred to Omav treatment of FRDA patients were recently reported pseudopregnant females. Two founders in 31 progeny were identified that (NCT02255435) (La Rosa et al., 2020; Lynch et al., 2019). contained the desired D126D, G127V mutations. Both were mated to Increased NRF2 levels (mRNA and protein), in addition to the C57BL/6J, and a single founder demonstrated transmission of the targeted induction of NRF2 target gene expression, are observed after 24 h of D126D, G127V mutation through the germline. This mouse strain was treatment of FRDA fibroblasts with DMF, idebenone or Omav designated JR 30822 (C57BL/6J-Fxnem8Lutzy/J). The population of (Petrillo et al., 2019). Moreover, treatment of fibroblasts from FRDA JR 30822 mice was expanded with an additional backcross to C57BL/6J patients with DMF was shown to increase FXN mRNA levels mice before intercrossing to score for the effect on homozygosity. The colony is maintained by backcrossing of heterozygous mice to wild type (Jasoliya et al., 2019), but treatment with Omav or idebenone had no (C57BL/6J or wild-type littermates). effect on FXN gene expression (Petrillo et al., 2019). Finally, Omav treatment resolved several mitochondrial defects attributed to low WT WT Genotyping FXN /Fxn levels in FRDA fibroblasts and was protective against Mouse genomic DNA was isolated from ∼3 mm tail biopsies with oxidative stress-induced cell death (Abeti et al., 2018). Likewise, our QuickExtract™ DNA Extraction Solution (Lucigen # QE09050) and used study indicates that activation of NRF2 signaling in cells expressing as a template for PCRs. A 300 bp fragment encompassing the Fxn point only the FxnG127V protein is protective against cell death, at least via mutation was amplified with JumpStart™ Taq DNA Polymerase (Sigma- Omav- and DMF-mediated actions. Omav and DMF both work to Aldrich, #D9307) using Fxn G127V restriction fragment length increase NRF2 levels and transcriptional activity by regulating its polymorphism primers found in Table S2. After amplification, the PCR repressor, Kelch-like ECH-associated protein 1 (KEAP1), thereby product was digested overnight at 37°C with 0.5 μl of the restriction enzyme preventing its degradation and augmenting its nuclear translocation AatII (NEB, #R0117) added directly to the PCR tube. Subsequently, the (Probst et al., 2015; Takaya et al., 2012). We can speculate that the reaction products were resolved by 1% agarose gel electrophoresis and visualized with ethidium bromide. typical NRF2 antioxidant transcriptional response is hampered, but we cannot rule out the involvement and regulation of cell cycle Derivation and culture of MEFs protein expression, and therefore additional studies will be necessary MEFs were derived according to (Jozefczuk et al., 2012). Briefly, pregnant to determine the mechanism(s) of action of DMF and Omav in female mice at 13 or 14 dpc were sacrificed and embryos extracted into Petri G127V enhancing the viability of Fxn MUT MEFs. dishes. After removal of heads and red organs, the remaining tissues were Our results demonstrate that cells expressing only the FxnG127V minced, trypsinized with 0.05% trypsin/EDTA and incubated for 15 min at mutant protein exhibit many shared pathophysiological 37°C. Disrupted tissues were centrifuged at 300 g for 5 min, the supernatant characteristics of cells expressing low levels of FxnWT protein, was removed, and pelleted cells were suspended in Dulbecco’s modified especially reduced growth potential and sensitivity to oxidative stress. Eagle’s medium (DMEM) high-glucose medium with pyruvate (Life However, some phenotypes, especially those pertaining to lipid Technologies, 11995) supplemented with 10% fetal bovine serum (HyClone, handling, appear to be attributed uniquely to expression of the SH30396.03), 1% GlutaMAX (Life Technologies, 35051), and 1% penicillin- streptomycin (Life Technologies, 15140). Cells from each embryo were plated FxnG127V mutant protein. These results support a hypothesis of G127V G130V separately on gelatin-coated 100 mm dishes and propagated in DMEM (as Fxn /FXN proteins assuming alternative functions rather above) supplemented with 10% fetal bovine serum and 1% GlutaMAX. than loss of function. Changes in Fxn/FXN function induced by the Experiments denoted as ‘early passage’ were performed with cells between G127V/G130V mutation, whether it be differential association with passage numbers 3 and 5, whereas ‘late passage’ denotes cells of passages 6-9. ISC assembly proteins or by eliciting new interactions, could help to MEF cell lines derived from at least two different embryos per genotype were shed light on unique molecular events underlying the atypical clinical used for each experiment, and considered as independent biological replicates. presentation of FRDA G130V patients. It will be crucial to define functional relationships of the FxnG127V protein in neurons involved Treatments and XTT assays in prominent FRDA G130V symptoms, such as spastic paraparesis. WT and MUT MEFs were seeded at 20,000 cells per well in 96-well plates 24 h Examining the effects of FxnG127V expression on disease-relevant before treatments. Compounds were added and cells incubated for an additional tissues, such as those of the central nervous system and the heart, in 24 h, followed by incubation with XTT [(sodium 2,3,-bis(2-methoxy-4-nitro- the mouse model, and accounting for temporal aspects of 5-sulfophenyl)-5-[(phenylamino)-carbonyl]-2H-tetrazolium)] reagent for 3 h. The XTT assays were performed as recommended by the manufacturer development and aging processes are of immediate interest. (ATCC). Commercially available compounds were obtained as follows: Omav (RTA 408; Cayman Chemical), DMF (Tocris) and idebenone (Sigma- MATERIALS AND METHODS Aldrich). Animals Mice were housed in the animal facility at the University of Alabama at Quantitation of mRNA Birmingham under 12 h/12 h light/dark conditions and fed ad libitum. All RNA was isolated from MEFs using the RNeasy Mini Kit (Qiagen, #74104). experimental procedures were conducted in accordance with the Guide for Genomic DNA contamination was removed with the TURBO DNA-free™ Disease Models & Mechanisms

10 RESEARCH ARTICLE Disease Models & Mechanisms (2020) 13, dmm045229. doi:10.1242/dmm.045229

Kit (Invitrogen, #AM1907). The level of specific mRNA was quantitated by FITC Annexin V Apoptosis Detection Kit I (BD BioSciences, #556547) Power SYBR™ Green RNA-to-CT™ 1-Step Kit (Applied Biosystems, according to the manufacturer’s instructions. For cell cycle analysis, MEFs #4389986). Transcript (mRNA) levels relative to that of the housekeeping were fixed with cold 70% ethanol overnight, then stained with PBS gene glyceraldehyde phosphate dehydrogenase (Gapdh) were calculated supplemented with Triton X-100 (final concentration 0.1%), RNase A (final using the ΔΔCt method. Primer sequences can be found in Table S2. concentration 200 µg/ml) and PI (final concentration 20 µg/ml) for 30 min at RT. Stained MEFs were counted with an LSRII Analyzer, BD Diva Western blot v.8.0.1 (BD BioSciences) and data analyzed with the FlowJo software For preparation of total protein lysates, MEFs were homogenized in NP-40 package using a Watson pragmatic model. lysis buffer containing 1% NP-40, 20 mM Tris-HCl pH 7.4, 150 mM NaCl, 5 mM EDTA, 10% glycerol, 1 mM dithiothreitol, 10 mM sodium fluoride, Senescence-associated β-galactosidase staining 1 mM sodium orthovanadate and 5 mM sodium pyrophosphate, with Cell staining was performed with a Senescence β-galactosidase Staining Kit protease inhibitor cocktail (Millipore, #539134). The lysates were rotated (Cell Signaling Technology, #9860), following the manufacturer’s for 30 min at 4°C, followed by centrifugation at 15,800 g for 20 min. The instructions. At least ten images were taken per group at ×200 total supernatant was removed as the soluble fraction. To reduce carryover, the magnification, and the fraction of blue-stained cells counted by a blinded pellets were washed with lysis buffer and resuspended in urea-sodium investigator. To avoid staining attributable to cell confluence rather than to dodecyl sulfate buffer (NP-40 lysis buffer with 8 M urea/3% sodium proliferative senescence (Severino et al., 2000), the assay was performed in dodecyl sulfate) followed by sonication, with three 20 s pulses at 20% subconfluent cultures displaying comparable cell densities. amplitude on a Fisher Scientific™ Model 120 Sonic Dismembrator. The lysates were then centrifuged again at 15,800 g for 20 min at 4°C and the Staining of mitochondria supernatant was collected as the insoluble fraction. Protein concentration of MEFs were plated on glass coverslips, cultured until ∼60-80% confluent soluble fractions were estimated by Bradford assay using Protein Assay Dye and stained with OptiMEM supplemented with 100 nM (final Reagent (Bio-Rad, #5000006). Samples were heated with NuPAGE™ LDS ™ ® concentration) MitoTracker Deep Red FM (Thermo Fisher Scientific, Sample Buffer (4×) (Invitrogen, #NP0008) and NuPAGE Reducing Agent #M22426) for 30 min at 37°C followed by 30 min incubation in OptiMEM (10×) (Invitrogen, #NP0004) for 10 min at 70°C. Protein samples were without the dye. Stained MEFs were fixed with 3.7% formaldehyde for resolved on a 4-12% NuPAGE gel and transferred to nitrocellulose 15 min at 37°C, mounted on microscope slides with ProLong™ Gold ™ membranes with an iBlot 2 Gel Transfer Device (Invitrogen, #IB21001). Antifade Mountant with 4′,6-diamidino-2-phenylindole (Invitrogen, ™ Membranes were blocked with 5% ECL Advance Blocking Reagent (GE #P36935) and imaged with a Nikon A1R confocal microscope using ×63 Healthcare, #RPN418) followed by incubation with primary and secondary objective magnification (University of Alabama at Birmingham High (anti-rabbit horseradish peroxidase linked, GE Healthcare, #LNA934V) Resolution Imaging Facility). Maximum-intensity projections of confocal antibodies. Antibody information is provided in Table S3. For enhanced image z-stacks were rendered as three-dimensional reconstructions using ™ sensitivity, membranes were incubated with SuperSignal Western Blot Imaris analytical software (Bitplane AG). Imaris Filament Tracer tool with Enhancer (Thermo Fisher Scientific, #46640) for 10 min at room subjective thresholding was used to detect mitochondrial filament lengths temperature (RT) before blocking, and high-intensity horseradish automatically. peroxidase substrate SuperSignal™ West Femto (Thermo Fisher Scientific, #34094) was used to produce a signal. All images were captured on a ChemiDoc MP Imaging System and analyzed using Mitochodrial DNA damage and copy number in MEFs ImageLab v.6.0.1 software (Bio-Rad). Mitochondrial DNA damage was determined by quantitative PCR (qPCR) Mitochondrially enriched protein lysates were prepared as described (Lin as described by Furda et al. (2014), with some modifications. Briefly, total DNA was isolated with the QIAamp DNA Mini Kit (Qiagen, #51304) et al., 2017b), with modifications. Cells were resuspended in hypotonic ’ buffer [225 mM mannitol, 75 mM sucrose, 5 mM Hepes, 1 mM EGTA, according to the manufacturer s instructions. Long mtDNA fragments were amplified with high-fidelity AccuPrime™ Taq DNA polymerase (Thermo 0.1 mM EDTA (pH 8), 0.1% bovine serum albumin (BSA), 1% protease ™ inhibitor cocktail (PIC; Sigma-Aldrich, P8340)], kept on ice for 10 min, Fisher Scientific, #12346086) and quantified with the Quant-iT PicoGreen™ dsDNA Assay Kit (Thermo Fisher Scientific, #P7589). A then homogenized. The suspensions were centrifuged at 1500 g for 4 min at ™ 4°C, after which supernatants were transferred to fresh tubes, and the short mtDNA fragment was quantified with Power SYBR Green PCR suspensions were centrifuged again at 20,000 g for 15 min at 4°C. (Thermo Fisher Scientific, #4367659) relative to genomic DNA. Supernatants were collected as ‘cytoplasmic’ fractions, and pellets were Quantitation of the average lesion frequency in mtDNA was performed washed with buffer [50 mM Tris (pH 7.5), 0.25 M sucrose, 0.2 mM EDTA assuming a random distribution of lesions in accordance with the Poisson (pH 8), 0.1% BSA and 1% PIC], then resuspended in lysis buffer [0.1% NP- equation (Westbrook et al., 2010). The mtDNA copy number was estimated 40, 0.25 M NaCl, 5 mM EDTA, 50 mM Hepes (pH 7.5) and 1% PIC] and by determination of the mtDNA to nuclear DNA ratio with a qPCR assay as kept on ice for 10 min for protein extraction. Lysates were centrifuged at described by Quiros et al. (2017). Primer sequences can be found in 20,000 g for 10 min at 4°C, and clarified supernatants were transferred to Table S2. fresh tubes as ‘mitochondrially enriched’ fractions. Measurement of mitochondrial respiration Growth curves and population doubling time of MEFs The mitochondrial function of MEFs was determined using a Seahorse Early passage WT, HET and MUT MEFs were seeded in duplicate in six- Extracellular Flux Analyzer (XF96 Analyzer; Agilent Technologies, Santa well plates at 1×105 cells per well and cultured in either a cell culture Clara, CA, USA), which measures the OCR in live cells. Cells were plated at incubator with ‘normoxic’ gas composition (air supplemented with 5% 40,000 cells per well in a XF96 assay plate and allowed to attach for 4 h in CO2)or‘hypoxic’ gas composition (90% N2,5%CO2 and 5% O2). Cells culture medium before the assay, washed using XF-DMEM assay medium and were detached with trypsin and counted daily after seeding (days 1-6). The allowed to equilibrate in a non-CO2 incubator for 1 h at 37°C. Mitochondrial PDT was calculated during the logarithmic phase of the growth using the stress tests were performed by sequentially injecting oligomycin (Oligo, 1 µg/ formula: PDT=T×ln2/ln(A/A0), where T corresponds to the time duration of ml), carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP, 1.5 µM) culture (in hours), A corresponds to the number of cells in the well at the time and antimycin A (Anti A, 10 µM) (Chacko et al., 2013). To record fatty acid- of measurement and A0 is the initial number of cells. mediated OCR measurements, BSA-palmitate substrate (25 µg/ml) was incubated with MEFs for 30 min before the assay in XF-DMEM medium Measurement of cell death and cell cycle (Ravi et al., 2015). Etomoxir (10 µM) was added to determine fatty acid- MEFs were plated at 1×105 WT or 2×105 MUT cells per 100 mm dish and sensitive OCR (Ravi et al., 2015). The data were normalized to the total protein cultured until ∼80% confluent. Cells were detached with accutase and content in each well, measured using the Lowry HS protein assay (Bio-Rad) washed with PBS. For cell death analysis, MEFs were processed with the and expressed as the mean. Disease Models & Mechanisms

11 RESEARCH ARTICLE Disease Models & Mechanisms (2020) 13, dmm045229. doi:10.1242/dmm.045229

Statistical analyses Cavadini, P., Adamec, J., Taroni, F., Gakh, O. and Isaya, G. (2000a). Two-step Statistical analyses were conducted using GraphPad Prism v.6. Statistical processing of human frataxin by mitochondrial processing peptidase. Precursor significance was determined by performing Student’s unpaired two-tailed and intermediate forms are cleaved at different rates. J. Biol. Chem. 275, t t – 41469-41475. doi:10.1074/jbc.M006539200 -tests, multiple -tests (with Holm Sidak correction for multiple Cavadini, P., Gellera, C., Patel, P. I. and Isaya, G. (2000b). Human frataxin comparisons), ordinary one-way ANOVA or two-way ANOVA (with maintains mitochondrial iron homeostasis in Saccharomyces cerevisiae. Hum. Tukey correction for multiple comparisons). Significant differences were Mol. Genet. 9, 2523-2530. doi:10.1093/hmg/9.17.2523 considered as P<0.05. Chacko, B. K., Kramer, P. A., Ravi, S., Johnson, M. S., Hardy, R. W., Ballinger, S. W. and Darley-Usmar, V. M. (2013). Methods for defining distinct bioenergetic profiles in platelets, lymphocytes, monocytes, and neutrophils, and the oxidative Acknowledgements burst from human blood. Lab. Invest. 93, 690-700. doi:10.1038/labinvest.2013.53 The authors wish to acknowledge the technical support of the University of Alabama Chandran, V., Gao, K., Swarup, V., Versano, R., Dong, H., Jordan, M. C. and at Birmingham High Resolution Imaging Facility. We also thank Ms Yu-Yun Chen for Geschwind, D. H. (2017). Inducible and reversible phenotypes in a novel mouse assistance with antibody validation experiments. model of Friedreich’s Ataxia. eLife 6, e30054. doi:10.7554/eLife.30054 Clark, E., Butler, J. S., Isaacs, C. J., Napierala, M. and Lynch, D. R. (2017). Competing interests Selected missense mutations impair frataxin processing in Friedreich ataxia. Ann. The authors declare no competing or financial interests. Clin. Transl. Neurol. 4, 575-584. doi:10.1002/acn3.433 Clark, E., Strawser, C., Schadt, K. and Lynch, D. R. (2019). Identification of a ’ Author contributions novel missense mutation in Friedreich s ataxia -FXN(W) (168R). Ann. Clin. Transl. Conceptualization: A.R.Z., C.M.L., M.N., J.S.N.; Methodology: A.R.Z., C.M.L., M.N., Neurol. 6, 812-816. doi:10.1002/acn3.728 Correia, A. R., Pastore, C., Adinolfi, S., Pastore, A. and Gomes, C. M. (2008). J.S.N.; Formal analysis: D.F., B.K.C., J.Z., V.M.D.-U., A.R.Z., C.M.L., M.N., J.S.N.; Dynamics, stability and iron-binding activity of frataxin clinical mutants. FEBS J. Investigation: D.F., B.K.C., R.C., X.O., A.R.Z., J.S.N.; Writing - original draft: D.F., 275, 3680-3690. doi:10.1111/j.1742-4658.2008.06512.x J.S.N.; Writing - review & editing: D.F., B.K.C., J.Z., V.M.D.-U., A.R.Z., C.M.L., M.N., Cossee, M., Durr, A., Schmitt, M., Dahl, N., Trouillas, P., Allinson, P., Kostrzewa, J.S.N.; Supervision: J.S.N.; Project administration: J.S.N.; Funding acquisition: M., Nivelon-Chevallier, A., Gustavson, K.-H., Kohlschutter, A. et al. (1999). M.N., J.S.N. Friedreich’s ataxia: point mutations and clinical presentation of compound heterozygotes. Ann. Neurol. 45, 200-206. doi:10.1002/1531- Funding 8249(199902)45:2<200::AID-ANA10>3.0.CO;2-U This work was supported by the Office of Extramural Research, National Institutes of Cossee, M., Puccio, H., Gansmuller, A., Koutnikova, H., Dierich, A., LeMeur, M., Health, National Institute of Neurological Disorders and Stroke (R03 NS099953 to Fischbeck, K., Dolle, P. and Koenig, M. (2000). Inactivation of the Friedreich J.S.N., R21 NS101145 and R01 NS081366 to M.N.) and Friedreich’s Ataxia ataxia mouse gene leads to early embryonic lethality without iron accumulation. Research Alliance (to J.S.N.). The Nathan Shock Center (P30 AG050886) grant Hum. Mol. Genet. 9, 1219-1226. doi:10.1093/hmg/9.8.1219 supported the bioenergetics measurements. The Jackson Laboratory’s Genetic Cotticelli, M. G., Xia, S., Lin, D., Lee, T., Terrab, L., Wipf, P., Huryn, D. M. and ’ Engineering Technologies Scientific Service partially supported the development of Wilson, R. B. (2019). Ferroptosis as a novel therapeutic target for Friedreich s the Fxn mutant mouse model described in this publication. Mouse strain JR083822 ataxia. J. Pharmacol. Exp. Ther. 369, 47-54. doi:10.1124/jpet.118.252759 is available from the Jackson Laboratory mutant mouse collection (www.jax.org). Deutsch, E. C., Santani, A. B., Perlman, S. L., Farmer, J. M., Stolle, C. A., Marusich, M. F. and Lynch, D. R. (2010). A rapid, noninvasive immunoassay for frataxin: utility in assessment of Friedreich ataxia. Mol. Genet. Metab. 101, Supplementary information 238-245. doi:10.1016/j.ymgme.2010.07.001 Supplementary information available online at Dhe-Paganon, S., Shigeta, R., Chi, Y.-I., Ristow, M. and Shoelson, S. E. (2000). https://dmm.biologists.org/lookup/doi/10.1242/dmm.045229.supplemental Crystal structure of human frataxin. J. Biol. Chem. 275, 30753-30756. doi:10. 1074/jbc.C000407200 References Dimri, G. P., Lee, X., Basile, G., Acosta, M., Scott, G., Roskelley, C., Medrano, Abeti, R., Baccaro, A., Esteras, N. and Giunti, P. (2018). Novel Nrf2-inducer E. E., Linskens, M., Rubelj, I., Pereira-Smith, O. et al. (1995). A biomarker that prevents mitochondrial defects and oxidative stress in Friedreich’s ataxia models. identifies senescent human cells in culture and in aging skin in vivo. Proc. Natl. Front. Cell Neurosci. 12, 188. doi:10.3389/fncel.2018.00188 Acad. Sci. USA 92, 9363-9367. doi:10.1073/pnas.92.20.9363 Ast, T., Meisel, J. D., Patra, S., Wang, H., Grange, R. M. H., Kim, S. H., Calvo, Dranka, B. P., Benavides, G. A., Diers, A. R., Giordano, S., Zelickson, B. R., S. E., Orefice, L. L., Nagashima, F., Ichinose, F. et al. (2019). Hypoxia rescues Reily, C., Zou, L., Chatham, J. C., Hill, B. G., Zhang, J. et al. (2011). Assessing frataxin loss by restoring iron sulfur cluster biogenesis. Cell 177, 1507-1521; e16. bioenergetic function in response to oxidative stress by metabolic profiling. Free Radic. Biol. Med. 51, 1621-1635. doi:10.1016/j.freeradbiomed.2011.08.005 doi:10.1016/j.cell.2019.03.045 Durr, A. and Brice, A. (1996). Genetics of movement disorders. Curr. Opin. Neurol. Bhalla, A. D., Khodadadi-Jamayran, A., Li, Y., Lynch, D. R. and Napierala, M. 9, 290-297. doi:10.1097/00019052-199608000-00009 (2016). Deep sequencing of mitochondrial genomes reveals increased mutation Filla, A., De Michele, G., Cavalcanti, F., Pianese, L., Monticelli, A., Campanella, ’ load in Friedreich s ataxia. Ann. Clin. Transl. Neurol. 3, 523-536. doi:10.1002/ G. and Cocozza, S. (1996). The relationship between trinucleotide (GAA) repeat acn3.322 length and clinical features in Friedreich ataxia. Am. J. Hum. Genet. 59, 554-560. Bidichandani, S. I., Ashizawa, T. and Patel, P. I. (1997). Atypical Friedreich ataxia Furda, A., Santos, J. H., Meyer, J. N. and Van Houten, B. (2014). Quantitative caused by compound heterozygosity for a novel missense mutation and the GAA PCR-based measurement of nuclear and mitochondrial DNA damage and repair triplet-repeat expansion [letter]. Am. J. Hum. Genet. 60, 1251-1256. in mammalian cells. Methods Mol. Biol. 1105, 419-437. doi:10.1007/978-1- Bolinches-Amoros, A., Molla, B., Pla-Martin, D., Palau, F. and Gonzalez-Cabo, 62703-739-6_31 P. (2014). Mitochondrial dysfunction induced by frataxin deficiency is associated Fuss, J. O., Tsai, C.-L., Ishida, J. P. and Tainer, J. A. (2015). Emerging critical with cellular senescence and abnormal calcium metabolism. Front. Cell Neurosci. roles of Fe-S clusters in DNA replication and repair. Biochim. Biophys. Acta 1853, 8, 124. doi:10.3389/fncel.2014.00124 1253-1271. doi:10.1016/j.bbamcr.2015.01.018 Brzóska, K., Meczyńska, S. and Kruszewski, M. (2006). Iron-sulfur cluster Galea, C. A., Huq, A., Lockhart, P. J., Tai, G., Corben, L. A., Yiu, E. M., Gurrin, proteins: electron transfer and beyond. Acta Biochim. Pol. 53, 685-691. doi:10. L. C., Lynch, D. R., Gelbard, S., Durr, A. et al. (2016). Compound heterozygous 18388/abp.2006_3296 FXN mutations and clinical outcome in friedreich ataxia. Ann. Neurol. 79, 485-495. Calmels, N., Schmucker, S., Wattenhofer-Donzé, M., Martelli, A., Vaucamps, N., doi:10.1002/ana.24595 ı́ ́ ı́ Reutenauer, L., Messaddeq, N., Bouton, C., Koenig, M. and Puccio, H. (2009). Garc a-Gimenez, J. L., Gimeno, A., Gonzalez-Cabo, P., Das , F., Bolinches- ́ ́ ́ The first cellular models based on frataxin missense mutations that reproduce Amoros, A., Molla, B., Palau, F. and Pallardo,F.V.(2011). Differential α spontaneously the defects associated with Friedreich ataxia. PLoS ONE 4, expression of PGC-1 and metabolic sensors suggest age-dependent induction of mitochondrial biogenesis in Friedreich ataxia fibroblasts. PLoS ONE 6, e20666. e6379. doi:10.1371/journal.pone.0006379 doi:10.1371/journal.pone.0020666 Candayan, A., Yunisova, G., Çakar, A., Durmuş, H., Başak, A. N., Parman, Y. Gold, R., Kappos, L., Arnold, D. L., Bar-Or, A., Giovannoni, G., Selmaj, K., and Battaloğlu, E. (2020). The first biallelic missense mutation in the FXN gene in Tornatore, C., Sweetser, M. T., Yang, M., Sheikh, S. I. et al. (2012). Placebo- a consanguineous Turkish family with Charcot-Marie-Tooth-like phenotype. controlled phase 3 study of oral BG-12 for relapsing multiple sclerosis. Neurogenetics 21, 73-78. doi:10.1007/s10048-019-00594-1 N. Engl. J. Med. 367, 1098-1107. doi:10.1056/NEJMoa1114287 Carletti, B., Piermarini, E., Tozzi, G., Travaglini, L., Torraco, A., Pastore, A., Harding, A. E. (1981). Friedreich’s ataxia: a clinical and genetic study of 90 families Sparaco, M., Petrillo, S., Carrozzo, R., Bertini, E. et al. (2014). Frataxin with an analysis of early diagnostic criteria and intrafamilial clustering of clinical silencing inactivates mitochondrial Complex I in NSC34 motoneuronal cells and features. Brain 104, 589-620. doi:10.1093/brain/104.3.589 alters glutathione homeostasis. Int. J. Mol. Sci. 15, 5789-5806. doi:10.3390/ Haugen, A. C., Di Prospero, N. A., Parker, J. S., Fannin, R. D., Chou, J., Meyer,

ijms15045789 J. N., Halweg, C., Collins, J. B., Durr, A., Fischbeck, K. et al. (2010). Altered Disease Models & Mechanisms

12 RESEARCH ARTICLE Disease Models & Mechanisms (2020) 13, dmm045229. doi:10.1242/dmm.045229

gene expression and DNA damage in peripheral blood cells from Friedreich’s McCabe, D. J. H., Ryan, F., Moore, D. P., McQuaid, S., King, M. D., Kelly, A., Daly, ataxia patients: cellular model of pathology. PLoS Genet. 6, e1000812. doi:10. K., Barton, D. E. and Murphy, R. P. (2000). Typical Friedreich’s ataxia without 1371/journal.pgen.1000812 GAA expansions and GAA expansion without typical Friedreich’s ataxia. Hick, A., Wattenhofer-Donze, M., Chintawar, S., Tropel, P., Simard, J. P., J. Neurol. 247, 346-355. doi:10.1007/s004150050601 Vaucamps, N., Gall, D., Lambot, L., Andre, C., Reutenauer, L. et al. (2013). McGarry, J. D., Sen, A., Esser, V., Woeltje, K. F., Weis, B. and Foster, D. W. Neurons and cardiomyocytes derived from induced pluripotent stem cells as a (1991). New insights into the mitochondrial carnitine palmitoyltransferase enzyme model for mitochondrial defects in Friedreich’s ataxia. Dis. Model. Mech. 6, system. Biochimie 73, 77-84. doi:10.1016/0300-9084(91)90078-F 608-621. doi:10.1242/dmm.010900 Mincheva-Tasheva, S., Obis, E., Tamarit, J. and Ros, J. (2014). Apoptotic cell Hill, B. G., Benavides, G. A., Lancaster, J. R., Jr, Ballinger, S., Dell’Italia, L., death and altered calcium homeostasis caused by frataxin depletion in dorsal root Jianhua, Z. and Darley-Usmar, V. M. (2012). Integration of cellular bioenergetics ganglia neurons can be prevented by BH4 domain of Bcl-xL protein. Hum. Mol. with mitochondrial quality control and autophagy. Biol. Chem. 393, 1485-1512. Genet. 23, 1829-1841. doi:10.1093/hmg/ddt576 doi:10.1515/hsz-2012-0198 Miranda, C. J., Santos, M. M., Ohshima, K., Smith, J., Li, L., Bunting, M., Igoillo-Esteve, M., Oliveira, A. F., Cosentino, C., Fantuzzi, F., Demarez, C., Cossée, M., Koenig, M., Sequeiros, J., Kaplan, J. et al. (2002). Frataxin knockin Toivonen, S., Hu, A., Chintawar, S., Lopes, M., Pachera, N. et al. (2020). mouse. FEBS Lett. 512, 291-297. doi:10.1016/S0014-5793(02)02251-2 Exenatide induces frataxin expression and improves mitochondrial function in Mollá, B., Muñoz-Lasso, D. C., Riveiro, F., Bolinches-Amorós, A., Pallardó, Friedreich ataxia. JCI Insight 5, e134221. doi:10.1172/jci.insight.134221 F. V., Fernandez-Vilata, A., de la Iglesia-Vaya, M., Palau, F. and Gonzalez- Jasoliya, M. J., McMackin, M. Z., Henderson, C. K., Perlman, S. L. and Cabo, P. (2017). Reversible axonal dystrophy by calcium modulation in frataxin- Cortopassi, G. A. (2017). Frataxin deficiency impairs mitochondrial biogenesis in deficient sensory neurons of YG8R mice. Front. Mol. Neurosci. 10, 264. doi:10. cells, mice and humans. Hum. Mol. Genet. 26, 2627-2633. doi:10.1093/hmg/ 3389/fnmol.2017.00264 ddx141 Nachbauer, W., Wanschitz, J., Steinkellner, H., Eigentler, A., Sturm, B., Hufler, Jasoliya, M., Sacca, F., Sahdeo, S., Chedin, F., Pane, C., Brescia Morra, V., Filla, K., Scheiber-Mojdehkar, B., Poewe, W., Reindl, M. and Boesch, S. (2011). A., Pook, M. and Cortopassi, G. (2019). Dimethyl fumarate dosing in humans Correlation of frataxin content in blood and skeletal muscle endorses frataxin as a ’ increases frataxin expression: a potential therapy for Friedreich s Ataxia. PLoS biomarker in Friedreich ataxia. Mov. Disord. 26, 1935-1938. doi:10.1002/mds. ONE 14, e0217776. doi:10.1371/journal.pone.0217776 23789 Jozefczuk, J., Drews, K. and Adjaye, J. (2012). Preparation of mouse embryonic Palomo, G. M., Cerrato, T., Gargini, R. and Diaz-Nido, J. (2011). Silencing of fibroblast cells suitable for culturing human embryonic and induced pluripotent frataxin gene expression triggers p53-dependent apoptosis in human neuron-like stem cells. J. Vis. Exp. 3854. doi:10.3791/3854 cells. Hum. Mol. Genet. 20, 2807-2822. doi:10.1093/hmg/ddr187 Karthikeyan, G., Santos, J. H., Graziewicz, M. A., Copeland, W. C., Isaya, G., Van Perdomini, M., Hick, A., Puccio, H. and Pook, M. A. (2013). Animal and cellular Houten, B. and Resnick, M. A. (2003). Reduction in frataxin causes progressive models of Friedreich ataxia. J. Neurochem. 126 Suppl. 1, 65-79. doi:10.1111/jnc. accumulation of mitochondrial damage. Hum. Mol. Genet. 12, 3331-3342. doi:10. 12219 1093/hmg/ddg349 Perdomini, M., Belbellaa, B., Monassier, L., Reutenauer, L., Messaddeq, N., Kispal, G., Sipos, K., Lange, H., Fekete, Z., Bedekovics, T., Janáky, T., Bassler, Cartier, N., Crystal, R. G., Aubourg, P. and Puccio, H. (2014). Prevention and J., Aguilar Netz, D. J., Balk, J., Rotte, C. et al. (2005). Biogenesis of cytosolic reversal of severe mitochondrial cardiomyopathy by gene therapy in a mouse ribosomes requires the essential iron-sulphur protein Rli1p and mitochondria. model of Friedreich’s ataxia. Nat. Med. 20, 542-547. doi:10.1038/nm.3510 EMBO J. 24, 589-598. doi:10.1038/sj.emboj.7600541 Petrillo, S., D’Amico, J., La Rosa, P., Bertini, E. S. and Piemonte, F. (2019). Korolchuk, V. I., Miwa, S., Carroll, B. and von Zglinicki, T. (2017). Mitochondria in Targeting NRF2 for the treatment of Friedreich’s ataxia: a comparison among cell senescence: is mitophagy the weakest link? EBioMedicine 21, 7-13. doi:10. drugs. Int. J. Mol. Sci. 20, 5211. doi:10.3390/ijms20205211 1016/j.ebiom.2017.03.020 Ponti, M., Forrow, S. M., Souhami, R. L., D’Incalci, M. and Hartley, J. A. (1991). Koutnikova, H., Campuzano, V. and Koenig, M. (1998). Maturation of wild-type Measurement of the sequence specificity of covalent DNA modification by and mutated frataxin by the mitochondrial processing peptidase. Hum. Mol. antineoplastic agents using Taq DNA polymerase. Nucleic Acids Res. 19, Genet. 7, 1485-1489. doi:10.1093/hmg/7.9.1485 2929-2933. doi:10.1093/nar/19.11.2929 La Rosa, P., Russo, M., D’Amico, J., Petrillo, S., Aquilano, K., Lettieri-Barbato, Pook, M. A., Al-Mahdawi, S., Carroll, C. J., Cossée, M., Puccio, H., Lawrence, L., D., Turchi, R., Bertini, E. S. and Piemonte, F. (2019). Nrf2 induction re- Clark, P., Lowrie, M. B., Bradley, J. L., Cooper, J. M. et al. (2001). Rescue of the establishes a proper neuronal differentiation program in Friedreich’s ataxia neural Friedreich’s ataxia knockout mouse by human YAC transgenesis. Neurogenetics stem cells. Front. Cell Neurosci. 13, 356. doi:10.3389/fncel.2019.00356 La Rosa, P., Bertini, E. S. and Piemonte, F. (2020). The NRF2 signaling network 3, 185-193. doi:10.1007/s100480100118 Probst, B. L., Trevino, I., McCauley, L., Bumeister, R., Dulubova, I., Wigley, W. C. defines clinical biomarkers and therapeutic opportunity in Friedreich’s ataxia. and Ferguson, D. A. (2015). RTA 408, a novel synthetic triterpenoid with broad Int. J. Mol. Sci. 21, 916. doi:10.3390/ijms21030916 Lefevre, S., Sliwa, D., Rustin, P., Camadro, J.-M. and Santos, R. (2012). anticancer and anti-inflammatory activity. PLoS ONE 10, e0122942. doi:10.1371/ Oxidative stress induces mitochondrial fragmentation in frataxin-deficient cells. journal.pone.0122942 Biochem. Biophys. Res. Commun. 418, 336-341. doi:10.1016/j.bbrc.2012.01.022 Puccio, H. and Koenig, M. (2000). Recent advances in the molecular pathogenesis Lehle, S., Hildebrand, D. G., Merz, B., Malak, P. N., Becker, M. S., Schmezer, P., of Friedreich ataxia. Hum. Mol. Genet. 9, 887-892. doi:10.1093/hmg/9.6.887 ́ Essmann, F., Schulze-Osthoff, K. and Rothfuss, O. (2014). LORD-Q: a long- Puccio, H., Simon, D., Cossee, M., Criqui-Filipe, P., Tiziano, F., Melki, J., run real-time PCR-based DNA-damage quantification method for nuclear and Hindelang, C., Matyas, R., Rustin, P. and Koenig, M. (2001). Mouse models for mitochondrial genome analysis. Nucleic Acids Res. 42, e41. doi:10.1093/nar/ Friedreich ataxia exhibit cardiomyopathy, sensory nerve defect and Fe-S enzyme gkt1349 deficiency followed by intramitochondrial iron deposits. Nat. Genet. 27, 181-186. Leone, M., Brignolio, F., Rosso, M. G., Curtoni, E. S., Moroni, A., Tribolo, A. and doi:10.1038/84818 Schiffer, D. (1990). Friedreich’s ataxia: a descriptive epidemiological study in an Quiros, P. M., Goyal, A., Jha, P. and Auwerx, J. (2017). Analysis of mtDNA/nDNA Italian population. Clin. Genet. 38, 161-169. doi:10.1111/j.1399-0004.1990. Ratio in Mice. Curr. Protoc. Mouse Biol. 7, 47-54. doi:10.1002/cpmo.21 tb03566.x Ravi, S., Chacko, B., Kramer, P. A., Sawada, H., Johnson, M. S., Zhi, D., Lin, H., Magrane, J., Clark, E. M., Halawani, S. M., Warren, N., Rattelle, A. and Marques, M. B. and Darley-Usmar, V. M. (2015). Defining the effects of storage Lynch, D. R. (2017a). Early VGLUT1-specific parallel fiber synaptic deficits and on platelet bioenergetics: The role of increased proton leak. Biochim. Biophys. dysregulated cerebellar circuit in the KIKO mouse model of Friedreich ataxia. Dis. Acta 1852, 2525-2534. doi:10.1016/j.bbadis.2015.08.026 Model. Mech. 10, 1529-1538. doi:10.1242/dmm.030049 Redmann, M., Benavides, G. A., Wani, W. Y., Berryhill, T. F., Ouyang, X., Lin, H., Magrane, J., Rattelle, A., Stepanova, A., Galkin, A., Clark, E. M., Dong, Johnson, M. S., Ravi, S., Mitra, K., Barnes, S., Darley-Usmar, V. M. et al. Y. N., Halawani, S. M. and Lynch, D. R. (2017b). Early cerebellar deficits in (2018). Methods for assessing mitochondrial quality control mechanisms and mitochondrial biogenesis and respiratory chain complexes in the KIKO mouse cellular consequences in cell culture. Redox. Biol. 17, 59-69. doi:10.1016/j.redox. model of Friedreich ataxia. Dis. Model. Mech. 10, 1343-1352. doi:10.1242/dmm. 2018.04.005 030502 Sacca,̀ F., Marsili, A., Puorro, G., Antenora, A., Pane, C., Tessa, A., Long, A., Napierala, J. S., Polak, U., Hauser, L., Koeppen, A. H., Lynch, D. R. Scoppettuolo, P., Nesti, C., Brescia Morra, V., De Michele, G. et al. (2013). and Napierala, M. (2017). Somatic instability of the expanded GAA repeats in Clinical use of frataxin measurement in a patient with a novel deletion in the FXN Friedreich’s ataxia. PLoS ONE 12, e0189990. doi:10.1371/journal.pone.0189990 gene. J. Neurol. 260, 1116-1121. doi:10.1007/s00415-012-6770-5 Lorıa,́ F. and Dıaz-Nido,́ J. (2015). Frataxin knockdown in human astrocytes Saidu, N. E. B., Kavian, N., Leroy, K., Jacob, C., Nicco, C., Batteux, F. and triggers cell death and the release of factors that cause neuronal toxicity. Alexandre, J. (2019). Dimethyl fumarate, a two-edged drug: Current status and Neurobiol. Dis. 76, 1-12. doi:10.1016/j.nbd.2014.12.017 future directions. Med. Res. Rev. 39, 1923-1952. doi:10.1002/med.21567 Lynch, D. R., Farmer, J., Hauser, L., Blair, I. A., Wang, Q. Q., Mesaros, C., Santos, R., Lefevre, S., Sliwa, D., Seguin, A., Camadro, J.-M. and Lesuisse, E. Snyder, N., Boesch, S., Chin, M., Delatycki, M. B. et al. (2019). Safety, (2010). Friedreich ataxia: molecular mechanisms, redox considerations, and pharmacodynamics, and potential benefit of omaveloxolone in Friedreich ataxia. therapeutic opportunities. Antioxid Redox Signal. 13, 651-690. doi:10.1089/ars.

Ann. Clin. Transl. Neurol. 6, 15-26. doi:10.1002/acn3.660 2009.3015 Disease Models & Mechanisms

13 RESEARCH ARTICLE Disease Models & Mechanisms (2020) 13, dmm045229. doi:10.1242/dmm.045229

Severino, J., Allen, R. G., Balin, S., Balin, A. and Cristofalo, V. J. (2000). Is β- Westbrook, D. G., Anderson, P. G., Pinkerton, K. E. and Ballinger, S. W. (2010). galactosidase staining a marker of senescence in vitro and in vivo? Exp. Cell Res. Perinatal tobacco smoke exposure increases vascular oxidative stress and 257, 162-171. doi:10.1006/excr.2000.4875 mitochondrial damage in non-human primates. Cardiovasc. Toxicol. 10, 216-226. ́ Stepanova, A. and Magrane,J.(2020). Mitochondrial dysfunction in neurons in doi:10.1007/s12012-010-9085-8 ’ Friedreich s ataxia. Mol. Cell. Neurosci. 102, 103419. doi:10.1016/j.mcn.2019. Winter, R. M., Harding, A. E., Baraitser, M. and Bravery, M. B. (1981). Intrafamilial 103419 correlation in Friedreich’s ataxia. Clin. Genet. 20, 419-427. doi:10.1111/j.1399- Takaya, K., Suzuki, T., Motohashi, H., Onodera, K., Satomi, S., Kensler, T. W. and Yamamoto, M. (2012). Validation of the multiple sensor mechanism of the 0004.1981.tb01052.x Keap1-Nrf2 system. Free Radic. Biol. Med. 53, 817-827. doi:10.1016/j. Wong, A., Yang, J., Cavadini, P., Gellera, C., Lonnerdal, B., Taroni, F. and ’ freeradbiomed.2012.06.023 Cortopassi, G. (1999). The Friedreich s ataxia mutation confers cellular sensitivity Tamarit, J., Obis, E. and Ros, J. (2016). Oxidative stress and altered lipid to oxidant stress which is rescued by chelators of iron and calcium and inhibitors of metabolism in Friedreich ataxia. Free Radic. Biol. Med. 100, 138-146. doi:10. apoptosis. Hum. Mol. Genet. 8, 425-430. doi:10.1093/hmg/8.3.425 1016/j.freeradbiomed.2016.06.007 Worth, A. J., Basu, S. S., Deutsch, E. C., Hwang, W.-T., Snyder, N. W., Lynch, Vannocci, T., Notario Manzano, R., Beccalli, O., Bettegazzi, B., Grohovaz, F., D. R. and Blair, I. A. (2015). Stable isotopes and LC-MS for monitoring metabolic Cinque, G., de Riso, A., Quaroni, L., Codazzi, F. and Pastore, A. (2018). Adding disturbances in Friedreich’s ataxia platelets. Bioanalysis 7, 1843-1855. doi:10. ’ a temporal dimension to the study of Friedreich s ataxia: the effect of frataxin 4155/bio.15.118 overexpression in a human cell model. Dis. Model. Mech. 11, dmm032706. doi:10. Yao, C.-H., Liu, G.-Y., Wang, R., Moon, S. H., Gross, R. W. and Patti, G. J. (2018). 1242/dmm.032706 Identifying off-target effects of etomoxir reveals that carnitine palmitoyltransferase Vyas, P. M., Tomamichel, W. J., Pride, P. M., Babbey, C. M., Wang, Q., Mercier, β J., Martin, E. M. and Payne, R. M. (2012). A TAT-Frataxin fusion protein increases I is essential for cancer cell proliferation independent of -oxidation. PLoS Biol. 16, lifespan and cardiac function in a conditional Friedreich’s ataxia mouse model. e2003782. doi:10.1371/journal.pbio.2003782 Hum. Mol. Genet. 21, 1230-1247. doi:10.1093/hmg/ddr554 Zhang, S., Napierala, M. and Napierala, J. S. (2019). Therapeutic Prospects for Ward Platt, M. and Deshpande, S. (2005). Metabolic adaptation at birth. Semin. Friedreich’s Ataxia. Trends Pharmacol. Sci. 40, 229-233. doi:10.1016/j.tips.2019. Fetal. Neonatal. Med. 10, 341-350. doi:10.1016/j.siny.2005.04.001 02.001 Disease Models & Mechanisms

14