Frataxin‐deficient neurons and mice models of Friedreich ataxia are improved by TAT‐MTScs‐FXN treatment

Abstract Friedreich ataxia (FA) is a rare disease caused by deficiency of frataxin, a mitochondrial protein. As there is no cure available for this disease, many strategies have been developed to reduce the deleterious effects of such deficiency. One of these approaches is based on delivering frataxin to the tissues by coupling the protein to trans‐activator of transcription (TAT) peptides, which enables cell membranes crossing. In this study, we tested the efficiency of TAT‐MTScs‐FXN fusion protein to decrease neurodegeneration markers on frataxin‐depleted neurons obtained from dorsal root ganglia (DRG), one of the most affected tissues. In mice models of the disease, we tested the ability of TAT‐MTScs‐FXN to penetrate the mitochondria and its effect on lifespan. In DRG neurons, treatment with TAT‐MTScs‐FXN increased cell survival, decreased neurite degeneration and reduced apoptotic markers, such as α‐fodrin cleavage and caspase 9 activation. Also, we show that heat‐shock protein 60 (HSP60), a molecular chaperone targeted to mitochondria, suffered an impaired processing in frataxin‐deficient neurons that was relieved by TAT‐MTScs‐FXN addition. In mice models of the disease, administration of TAT‐MTScs‐FXN was able to reach muscle mitochondria, restore the activity of the succinate dehydrogenase and produce a significant lifespan increase. These results support the use of TAT‐MTScs‐FXN as a treatment for Friedreich ataxia.


Introduction
Friedreich ataxia (FA) is the most common autosomal recessive ataxia known to date, with a population frequency of 1-2/50,000 [1] and early life mortality caused mainly by cardiomyopathy [2]. FA is attributed to partial depletion of the mitochondrial protein Frataxin, caused by a trinucleotide GAA expansion in intron 1 of the Frataxin gene. When both Frataxin alleles contain the intron 1 expansion, frataxin level is reduced below the normal threshold and FA is overt [3]. Additionally, increased number of GAA repeats (100-1700 in FA patients) is associated with more severe clinical outcome [2]. Frataxin is a nuclear encoded protein that is translocated to the mitochondria due to a mitochondrial targeting sequence (MTS) that is expressed on its N-terminus [4,5]. Frataxin biological function is not completely elucidated; however, it was found to regulate the biogenesis of mitochondrial Fe-S clusters [6,7] Fe-S clusters serve as prosthetic groups for a series of enzymes including, among others, energy metabolism enzymes such as aconitase [8] or proteins of the respiratory chain including succinate dehydrogenase [9]. These enzymes were found to be impaired in FA patients [10]. Nevertheless, frataxin may have additional roles as Fe-S deficiency is not a universal consequence of frataxin deficiency [11,12].
Leading drug candidates for FA are now in phase 2 clinical trials, but currently, no cure is available. Drug candidates include, among others, mitochondrial function enhancers, antioxidant and neuroprotection agents, enhancers of frataxin gene expression and frataxin replacement. The latter consists in delivering frataxin to cells using the HIV-trans-activator of transcription (TAT) as carrier molecule [13]. Using a conditional-KO mouse model in which Frataxin was depleted in the brain [14], intraperitoneal (IP) TAT-FXN injections showed evidence of mitochondrial localization and rescued the fatal phenotype. TAT-FXN has also been shown to cross the blood-brain barrier and protect dopaminergic neurons in a mouse Parkinson disease model after IP injections [15].
Most of the mitochondrial proteins are synthesized with mitochondrial targeting sequence (MTS), a positively charged amphipathic a-helix structure [16] typically 15-30 amino acid long, which allows their import from the cytoplasm into the mitochondria through the translocase of the outer and inner mitochondrial membranes (TOM/TIM). Upon translocation through TOM complex, the inner mitochondrial membrane potential (Dw m , negative on matrix side) [17] and the ATP-dependent action of mitochondrial heat-shock protein 70 (mtHSP70) drive the protein import into matrix space [18] where MTS is recognized by mitochondrial processing proteases (MPP), retaining the mature protein within the mitochondria. In this context, one of the mitochondrial proteins imported into the mitochondrial matrix is HSP60, a molecular chaperone which is highly conserved from bacteria to humans. When the importing machinery is altered, HSP60 precursor form (MTS-HSP60) accumulates and can be detected by Western blot [19,20]. Frataxin precursor -FXN 1-210 -contains an unusual 80-amino acid-long MTS with several cleavage sites, which undergoes two-step processing upon entry into the mitochondria, leading to at least a transient processing intermediate -FXN  -before of an alternative processing -FXN 56-210 - [21] and/or the mature form -FXN 81-210 - [22]. Recently, a construct was designed (TAT-MTScs-FXN) consisting in human Frataxin mature protein -FXN 81-210 -sequence coupled with TAT sequence, that promotes cellular internalization and the mitochondrial targeting sequence of citrate synthase (MTScs), allowing efficient mitochondrial cleavage. This fusion protein was used to deliver frataxin into mitochondria of human Bjab lymphoma cell line as well as into cell cultures obtained from FA patients [23]. The results showed that frataxin was delivered with higher efficiency compared to TAT-MTScs-FXN, a construction carrying the native MTS of human frataxin.
In this study, we have used our model of frataxin-deficient DRG neurons [24] to show that TAT-MTScs-FXN is able to reach mitochondria of DRG neurons, partially rescuing cell viability, reducing neurite degeneration and decreasing a-fodrin cleavage, a marker of apoptotic process. These neurodegenerative markers were observed in our model as a consequence of altered calcium homoeostasis [24]. Moreover, when TAT-MTScs-FXN was injected in two mice models of FA [14,25], with either 4 or 10 mg/kg, functional frataxin was observed in muscle tissue, where it was able to recover the activity of succinate dehydrogenase (SDH) to the levels of heterozygous healthy mice. Reduced SDH activity impairs mitochondrial function and has been demonstrated to occur in Friedreich ataxia heart autopsies [10]. TAT-MTScs-FXN is also able to promote an increase in the mean survival of mice treated with 10 mg/kg. The results obtained with both cell and animal models indicate that this approach can be a promising strategy for FA treatment.

Materials and methods
Plasmids and production of lentiviral particles Reduction in frataxin in cultured DRG neurons was achieved by short-hairpin RNA-interfering sequences. Lentiviral vectors are routinely produced as described in our previous publication [24]. The shRNA lentiviral plasmids (pLKO.1-puro) for human/mouse/rat frataxin were purchased from Sigma-Aldrich (Merck. Madrid. Spain). The RefSeq used was NM_008044, which corresponds to mouse frataxin. The clones used were TRCN0000197534, TRCN0000006137 and referred in this paper as Fxn1 and Fxn2, respectively). A non-targeted scrambled sequence (the vector SHC002) served as a control (here referred as Scr).

Isolation and culture of DRG sensory neurons
DRGs were extracted and purified as previously described in Mincheva-Tasheva et al., 2014 [24]. Briefly, DRGs were extracted from neonatal rats (P3-P4) and dissociated with 0.05% trypsin (Sigma-Aldrich). Ganglia were mechanically disrupted with a pipette tip until obtaining a single cell suspension in culture media supplemented with DNAse I grade II at final concentration 3 mg/ml (Roche Diagnostics, Madrid, Spain), which was centrifuged at 8009g r.p.m. (302 rcf) through 7,5% BSA solution (Sigma-Aldrich) for 5 min., followed by resuspension in enriched neurobasal culture media (Thermo Fisher Scientific-TFS, Madrid, Spain) (NBMc) consisting of 2% horse serum (TFS), 2% B27 supplement (TFS), 0.5 mM L-glutamine (TFS), 100 U/ml penicillin plus 100 ng/ml streptomycin (TFS) and supplemented with murine b-nerve growth factor at 50 ng/ml (PeproTech, New Jersey, USA). To prevent growth of non-neuronal cells, culture media were supplemented with the anti-mitotic agent Aphidicolin (Sigma-Aldrich) at final concentration 3,4 lg/ml. After 1 hr of pre-plating in a p60 tissue dish (Corning Inc., Fisher Scientific. Madrid. Spain) at 37°C/5%CO2, the cells were then plated in a 24-well tissue dish (Corning Incorporated) pre-treated with 0,1 mg/ml of collagen (Sigma-Aldrich) at a cell density of 30,000 cells/ well. After 1-2 days, lentiviral transduction was performed, and lentivirus particles (20 ng/1000 cells) were added. After transduction, allowed to proceed for 6 hrs, media containing lentivirus were substituted by fresh culture media. TAT-MTScs-frataxin was added to cultures at 12 hrs or 48 hrs after ending lentivirus transduction protocol and used at the concentrations indicated in each figure. Usually, concentrations of 1, 3 or 7 lg/ml were used. The culture medium was not changed after the addition of TAT-MTScs-frataxin.

Immunocytochemistry of DRG neurons
Cell cultures were fixed in 4% paraformaldehyde and incubated overnight at 4°C with specific antibodies at indicated dilutions: rabbit anti-frataxin (1:50) from Santa Cruz Biotechnology (Dallas, USA) and mouse anti-OxPhos (1:50) from Novex (Invitrogen, Madrid, Spain)). Anti-rabbit or antimouse secondary antibodies conjugated with Alexa Fluor 555 or Alexa Fluor 488 (Invitrogen, Madrid, Spain), respectively, were used at 1:300 dilutions and incubated for 2 hrs at room temperature, protected from light. For cell nucleus staining, the fixed cells were incubated for 5 min. with DAPI. Micrographs were taken using an Olympus FluoView 500 (609).

Immnunohistochemistry of heart tissue
Hearts were obtained from wild-type mice (C57BL/6J) and MCK mice treated with vehicle or 10 mg/Kg TAT-MTScs-FXN. Unstained slides were deparaffinized and rehydrated. Heat-induced epitope retrieval was performed with Diva antigen retrieval buffer (Biocare Medical Pacheco, CA, USA). Following a rinse, a protein blocking step was performed with Sniper protein block (Biocare Medical). Slides were again rinsed and anti-frataxin antibody, ab175402 from Abcam (Cambridge, UK), was applied at 1:50 dilution for 60 min. Slides were rinsed. Endogenous peroxidase was blocked using Peroxidase 1 [Biocare Medical]. After a rinse, the secondary antibody, biotinylated goat anti-rabbit IgG (1:500 dilution) from Jackson Immunoresearch (ref 111-065-144) was applied for 30 min. The slides were rinsed, and ABC Elite was added for 30 min., followed by another rinse. Colour development was achieved by adding Sigma Fast DAB solution for 5 min. Slides were rinsed, counterstained with haematoxylin and coverslipped. All rinse steps were performed with TBS-Autowash (Biocare Medical).

Survival and neurite degeneration analyses
Neuronal survival was measured with a 916 lens and cross-marked wells as described [24]. Neurons at zero and 5 days were counted at four fields per cross-marked well, using three wells per condition tested. Experiments were repeated at least three times. Neurite degeneration was measured with a 932 lens and a grid, which was created over each image with NIH Image J with the grid plugin (Image size 680 9 512 and line area 10,000 pixels2). Healthy and degenerated neurites (displaying neurofilament aggregates) were counted at three grid fields per image and three images per well were analysed. For each condition, we used three different wells. Experiments were repeated at least three times.

Animal care
All experiments were conducted in accordance with the protocols described by the National Institutes of Health's Guide for the Care and Use of Animals and were approved by JAX institutional animal care and use committees. Mice were ear notched for identification and housed in individually and positively ventilated polycarbonate cages with HEPA-filtered air at a density no greater than of four to five mice per cage. Pine shaving-based corn cob bedding was used, and cages were changed every 2 weeks. The animal room was lighted entirely with artificial fluorescent lighting, with a controlled 12-hrs light/dark cycle (7 a.m. to 7 p.m. light). The normal temperature and relative humidity ranges in the animal rooms were 22 AE 4°C and 50 AE 15%, respectively. The animal rooms were set to have 15 air exchanges per hour. Filtered tap water, acidified to a pH of 2.8-3.2 and LabDiet 5LL4 were provided ad libitum.  [25].

Administration of TAT-MTScs-frataxin to mice and survival monitoring
Treatment was initiated at time of weaning (at 3-4 weeks of age). Mice were intravenously (IV) injected twice weekly (at least 3 days apart) with either vehicle (containing 50 mM Tris-HCl, 300 mM NaCl, 10% glycerol pH 8), or with TAT-MTScs-FXN at doses of 4 mg/kg or 10 mg/ kg.
For survival experiments, treatment was initiated at time of weaning (3-4 weeks of age). MCK mice were intraperitoneally (IP) injected twice weekly following the protocol described above. MCK mice survival was monitored daily. Additionally, mice that lost >15% weekly weight were killed and registered as death incidents. Significance was defined using log-rank (Mantel-Cox) test.

Isolation of mitochondria from muscle tissue
Mice tissue was extracted, and the mitochondrial fraction was isolated using mitochondria isolation kit (Abcam, ab110168) according to the manufacturer's protocol. Mitochondrial pellets were preserved at À80°C.

Measurement of frataxin amounts in muscle mitochondrial fractions
Human frataxin level was analysed in isolated mitochondrial fractions using Abcam ab176112 ELISA kit, according to the manufacturer's instructions.

Mitochondrial complex II activity assay
Complex II activity (succinate dehydrogenase, or SDH) in mitochondrial fractions was determined using a commercially available kit (BioVision, K660-100, Milpitas, California, USA) according to manufacturer's instructions. Mitochondrial pellets were resuspended with 90-150 ll succinate dehydrogenase (SDH) buffer, according to mitochondrial pellet amount (0.2V of resuspended volume) was used in duplicates. Protein was determined by the Bradford assay. Enzyme activity was followed at 600 nm and measured every 2 min. up to 20 min. Specific activity was given as DOD per mg of protein.

Western blot analysis
Western blot analysis were performed using equal volumes of mitochondrial extracts, diluted in sample buffer, separated using 4-12% gradient SDS-PAGE gels. External human Frataxin protein was used as a positive control (5 lg/well). COX-IV mitochondrial marker was detected using specific antibody from R&D Systems Abingdon, UK (MAB6980) diluted 1:1000 in blocking buffer at room temperature for 1.5 hrs. Following analysis, nitrocellulose membranes were washed, striped and subjected to Frataxin detection, using Frataxin antibody from Abnova, Walnut, CA, USA (H00002395-M02). Both Frataxin and COX-IV signals were quantified using Image J software, and Frataxin signal was normalized to COX-IV signal. For DRGs, cells were rinsed three times in ice-cold PBS (pH 7.4) and lysed with 2% SDS, 125 mM Tris and protease inhibitors (Roche) to obtain crude extracts used for SDS-polyacrylamide gel electrophoresis. In Western blot analysis, PVDF membranes were used for HSP60, fodrin, procaspase 9 and actin detection and nitrocellulose membranes for Frataxin detection. Monoclonal antibody against HSP60 was from Acris-Origene Technology Rockville, MD, USA (TA326374) used at 1/6000 dilution; a-fodrin cleavage products antibody was from ENZO Life Sciences (Farmingdale, NY, USA) used at 1:1000 dilution, and anti-caspase 9 monoclonal antibody was from Cell Signaling Technology (9508) used at 1:1000 dilution. Frataxin was detected using anti-Frataxin antibody from Santa Cruz Biotechnology 1:250 dilution. Fodrin, HSP60, procaspase 9 and frataxin signals were normalized to actin signal, using antib-actin antibody from Chemicon-EMD Millipore, Madrid, Spain at 1:50,000 dilution.

Data analysis
The data obtained from at least three independent experiments were used for statistical analyses. Values were expressed as mean AE S.E.M. (error bars). One-way ANOVA was used to assess survival, neurite degeneration between groups. If the ANOVA test was statistically significant, we performed post hoc pairwise comparisons using the Bonferroni test. The P-values lower than 0.05( * ), 0.01( * * ) or 0.001 ( * * * ) were considered significant. For in vivo studies, statistical analyses were performed using Statistical Package for the Social Sciences (SPSS software version 22. licence number f11340ae0 8e3b46186e4).

TAT-MTScs-FXN penetrates DRG neurons
TAT-fusion proteins have been used to successfully deliver proteins and peptides into cells as treatment for various diseases. Although the mechanism is unknown, transduction mechanism requires an initial ionic interaction between TAT peptide and the cell surface, followed by TAT-fusion protein internalization through lipid-raft-dependent micropinocytosis [27]. Using our model of rat primary culture, we tested the ability of the TAT-MTScs-FXN to penetrate frataxin-deficient DRG neurons. DRG neurons were transduced with lentiviruses carrying either Fxn1 or Fxn2 shRNAs to reduce the endogenous expression level of frataxin or lentivirus carrying Scr shRNA to use it as control condition for frataxin level. Efficient reduction in endogenous levels of frataxin was achieved by Fxn1 and Fxn2 interference RNAs (shown in Fig. S1). After 48 hrs of lentiviral transduction, TAT-MTScs-FXN (7 lg/ml) or vehicle solution was added to cultures. To evaluate the presence and mitochondrial location of frataxin in these cells, cultures were co-stained by immunofluorescence with anti-frataxin and anti-OxPhos antibodies. The latter being a cocktail of antibodies that recognize different subunits of the mitochondrial respiratory chain. Frataxin is almost undetectable in frataxin-depleted DRG neurons (data not shown). In contrast, frataxin can be detected after TAT-MTScs-FXN treatment, indicating that the fusion protein is able to penetrate frataxin-deficient cells (Fig. 1). Moreover, the punctate pattern of frataxin staining in the cell soma and along the neurites mostly co-localize with mitochondrial marker, which shows that TAT-MTScs-FXN is targeted to the mitochondria.

TAT-MTScs-FXN processing in DRG neurons
The results in Figure 1 indicate that frataxin reach mitochondria. To ensure that TAT-MTScs-FXN is correctly processed by DRG neurons, cells were taken at day 5, lysed and protein samples analysed by Western blot to evaluate frataxin levels. The frataxin signal in TAT-MTScs-FXN-treated DRG cultures is shown in Figure 2 where the full length, intermediate and mature forms can be detected. It is worth noting that in Fxn1 and Fxn2 cultures the levels of unprocessed construct were higher than those exhibited Scr cultures, suggesting a reduced ability to process frataxin. This could be a consequence of the fact that, as we previously reported [24], frataxin-depleted neurons display decreased mitochondrial membrane potential (Dw m ) that could potentially affect  TAT-MTScs-FXN penetration into the mitochondria. Indeed, Fxn1 and Fxn2 cultures displayed higher amounts of whole fusion protein and intermediate forms compared to control cultures (Scr) for each dose used ( Fig. S2A and B). Interestingly, the level of mature FXN form in TAT-MTScs-FXN-treated Fxn1 and Fxn2 cultures is not inferior to the level in Scr culture as one could expect (Fig. 2B).

HSP60-deficient processing is reversed by TAT-MTScs-FXN
Impairment of the frataxin construct processing described above prompted us to analyse whether this was only affecting TAT-MTScs-FXN fusion protein or could also affect other mitochondrial proteins. For this reason, we analyse the HSP60, a mitochondrial molecular chaperone involved in protein folding, acting after the mitochondrial import machinery [28,29]. As HSP60 is targeted to mitochondria, it should also be processed to reach the mitochondrial matrix [30]. Figure 3 (lanes 1-3) shows that in Fxn1 and Fxn2 cells, the precursor form (p) is increased with respect to control cultures (Scr), and accordingly, mature (m) form is decreased. As this could be a direct consequence of low frataxin levels displayed in Fxn1 an Fxn2 cultures, we tested the effect of adding TAT-MTScs-FXN to these cultures. As shown in Figure 3 (lanes 4-6), the levels of precursor and mature forms displayed in Fxn1 and Fxn2 are as those present in control Scr cultures, indicating that the mature frataxin form generated in Fxn1 and Fxn2 by TAT-MTScs-frataxin was sufficient to revert the anomalous processing of the molecular chaperone.

Fodrin cleavage is reduced by TAT-MTScs-FXN
The ability of TAT-MTScs-FXN to restore mitochondrial fitness to improve HSP60 processing leads us to analyse the reduction in apoptotic markers. In a previous study, we reported that apoptotic markers such as caspase 3 activation or a-fodrin cleavage were clearly observed [24]. In response to intracellular free Ca 2+ increase, calpain and caspase proteases are activated, and among others, they cleave a-fodrin, a cytoskeletal protein of 280 kD, rendering protease-specific breakdown products (SBDPs) [31]. Both proteases cleave the a-fodrin to produce a non-specific 150-kD breakdown fragment (SBDP150), and then, a second cleavage results in a calpain or caspase 3-specific product of 145 (SBDP145) or 120 kD (SBDP120), respectively. The presence of SBDP120 is considered a marker of apoptotic process [32]. Based on these results, we evaluated the therapeutic effect of TAT-MTScs-FXN on frataxin-deficient DRG neurons. Treating Scr and Fxn cultures with different doses of TAT-MTScs-FXN (1, 3 and 7 lg/ml) or vehicle solution at different time-points (48 and 12 hrs after lentiviral transduction), we observed that TAT-MTScs-FXN concentrations reduced a-fodrin cleavage in a dose-dependent manner (Fig. 4A). A histogram showing such decrease in each culture is presented in Figure 4B. Also, addition of TAT-MTScs-FXN has a higher impact on reduction in fodrin cleavage when added shortly after lentivirus transduction. As shown in Figure 4C and D, addition of TAT-MTScs-FXN 7 lg/ml at 12 hrs after lentiviral transduction promotes optimal decrease in SBDPs as compared with results obtained by adding TAT-MTScs-FXN at 48 hrs. These results are also in agreement with those obtained by analysing caspase 9 activation, a marker of intrinsic or mitochondrial apoptotic pathway (for a review see [33]). In Figure 4E and F, reduced levels of the precursor form are observed in Fxn1 and Fxn2 cells, indicative of cleavage (activation) of this caspase. When treated with TAT-MTScs-FXN, these cells display caspase 9 amounts similar to those observed in Scr cultures. These results show that apoptotic traits are reduced by TAT-MTScs-FXN and, consequently, degeneration of frataxin-deficient DRG neurons should be prevented.

Neurite degeneration and neuronal death are reduced by TAT-MTScs-FXN
Frataxin depletion in DRG neurons promotes apoptosis and neurite degeneration, indicated by loss of mitochondrial membrane potential and formation of neurofilament aggregates [24]. To further evaluate the positive effects of TAT-MTScs-FXN on frataxin-deficient DRG neurons, neurite degeneration was analysed in our cultures. As shown in Figure 5A, addition of TAT-MTScs-FXN (7 lg/ml, 12 hrs after lentiviral transduction) to Fxn1 and Fxn2 cultures reduced neurofilament aggregates in neurites by 50% in Fxn1 and 35% in Fxn2. Images are shown in Figure 5B where aggregates are marked with arrows. In Figure 5C, we show that neurite degeneration is reduced in a concentration-dependent manner, with 1, 3 and 7 lg/ml treatment after 48 hrs of transduction. Control Scr cultures were not affected by TAT-MTScs-FXN treatment (Fig. 5B, panels on the left). Besides neurite degeneration, Fxn1 and Fxn2 cultures showed a drastic decreased survival in accordance with increased apoptosis [24]. For this reason, we analysed the ability of TAT-MTScs-FXN to recover survival in frataxin-deficient DRG neurons. As shown in Figure 6A and B, addition of TAT-MTScs-FXN to Fxn1 and Fxn2 cultures after 12 hrs of lentiviral transduction resulted in a significant decrease in the percentage of mortality. At day 5, these cultures showed an increased survival rate of 16.6% and 14.6% per cent, respectively. When the compound was added 48 hrs after lentivirus transduction (Fig. 6C), the compound was still able to rescue the cultures from cell death although, compared to values achieved with TAT-MTScs-FXN added at 12 hrs, survival increased to a lesser extent (14.9% and 11.4% for 7 lg/ml and 9.2% and 6.8% for 3 lg/ml, in Fxn1 and Fxn2, respectively). Treatment with 1 lg/ml did not show any significant effect on cell survival. From these results, it can be concluded that in cultured frataxindeficient DRG neurons, TAT-MTScs-FXN is able to reduce neurite degeneration and, consequently, increase survival.

Organ penetration, mitochondrial localization of TAT-MTScs-Frataxin in FXN-KIKO mouse model
To test whether TAT-MTScs-FXN was also able to reach mitochondria in vivo, we used two mice models developed to study FA. For organ penetration and mitochondrial localization studies, mice model, described by Sarsero et al. [25], was IV injected with TAT-MTScs-FXN twice a week at 10 mg/kg dose, followed by 2 weeks of washout period (Fig. 7A). Mitochondrial fractions were obtained from muscles harvested at each time-point. As can be seen (Fig. 7B), muscle Frataxin levels in mitochondria increased one week after study initiation, reaching highly significant level after three weeks of drug administration in comparison with untreated mice. The values show a significant increase of more than threefold compared to untreated mice that show values around 550 pg/ mg. It is worth mentioning that the maximum values achieved of 1500 pg/mg of total protein at 21 days, represents around 15% of the normal values (i.e. concentration is typically around 10,300 pg/mg of total protein in muscles of healthy animals-data  not shown). Interestingly, during the washout period, 2 weeks after ceasing administration, the frataxin levels were reduced but still remained significantly above (around twofold) compared to those of untreated mice. Besides to evaluate the increase of frataxin amounts in tissues other than muscle, brains from treated and control animals were used and shown in Figure S3.

Mitochondrial functions restored with TAT-MTScs-Frataxin
The KIKO-FA model described above, provided that frataxin is reaching mitochondria, should display an improved phenotype. Hence, to examine frataxin function in the mitochondria, we used the MCK-conditional mouse model [14] because striated muscle tissue is depleted of frataxin. As shown in Figure 8A, processed frataxin is clearly observed in muscle mitochondrial fractions obtained from mice treated with TAT-MTScs-FXN. For comparison, a mitochondrial protein, CoxIV, was used as internal standard. The presence of frataxin in muscle was dose-dependent as, as shown in Figure 8B, animals treated with 10 mg/kg showed higher amounts (around threefold increase) compared to those treated with 4 mg/kg. To examine the ability of this frataxin to restore mitochondrial functions, the activity of succinate dehydrogenase (complex II) was tested. This is of special interest because deficiencies in succinate dehydrogenase have been associated with FA [10] and late-onset neurodegeneration characterized by progressive optic atrophy, ataxia and myopathy [34]. As shown in Fig. 8C, the enzymatic activity of SDH was increased, using either 4 mg/kg or 10 mg/kg, when compared to vehicle-treated mice and restored the enzymatic activity values to those exhibited by heterozygous healthy

Survival of MCK mice treated with TAT-MTScs-FXN
To test whether TAT-MTScs-FXN was able to rescue MCK phenotype, mouse survival was analysed as described in Materials and methods. Using intravenous route (IV) of injection, we did not observe a significant increased survival (Fig. S4). The reason for this could rely on the fact that when IV was used, amounts of frataxin in heart mitochondria of treated animals did not vary substantially when compared to controls (Fig. S5). MCK phenotype is mainly related to cardiac disfunctions. Firstly, we tested different ways of administration to deliver TAT-MTScs-FXN to heart tissue. Intraperitoneal injections resulted in an improvement compared to SC, IM and IV (Fig. S5). To test the presence of frataxin in cardiomyocytes, we performed immunohistochemistry on heart tissue treated with vehicle or TAT-MTScs-FXN. As shown in Figure 9, frataxin was undetectable in vehicle-treated MCK mice while wild-type mice and

Scr Fxn1 Fxn2
Vehicle TAT-MTScs-FXN-treated MCK mice show accumulation of frataxin near intercalated discs of cardiomyocytes. Indeed, mitochondria have been shown to be associated with intercalated discs in cardiomyocytes [35]. Moreover, the disorganization of the muscle fibres of MCK mice, a key trait of the model [36], was partially restored with 10 mg/kg TAT-MTScs-FXN treatment.

TAT-MTScs-FXN
Then, for survival assays, MCK mice were treated with IP injections of 4 mg/kg or 10 mg/kg. Although at 4 mg/kg they did not show significant differences with respect to vehicle (not shown), treatment with 10 mg/kg (Fig. 10A) shows an increase in total lifespan. The increased lifespan become evident in the interval from 65 to 75 days, where the maximum differences were observed. The mean survival was significantly higher for treated mice compared to control mice (67 versus 71 days, respectively). Moreover, in Figure 10B, age of death distribution is shown and the beneficial effects of the treatment are evident for the great majority of mice.

Discussion
Frataxin depletion is the main cause of Friedreich ataxia. To date, there are no cures available and many efforts are currently made to develop a treatment for this disease (for an updated review of the different therapeutic approaches see curefa.org/pipeline). One of the strategies consists in replacing frataxin by an exogenous source as TAT-based carrier molecule: TAT-FXN. This technology allows cell penetration and mitochondria localization of exogenous FXN in FA mice model in vivo and in cell derived from FA patients [13,15]. Recently, TAT-MTScs-FXN was shown to be more efficient than TAT-MTSfxn-FXN in mitochondria localization [23]. Pioneering studies conducted at Payne's laboratory showed that TAT-frataxin fusion protein can deliver frataxin to mitochondria and increase lifespan of animal models of the disease by 53% [11]. In this study, we used TAT-FXN-based construct in a neuronal model of the disease and show that TAT-MTScs-FXN was able to penetrate frataxin-deficient sensory neurons and display a mitochondrial staining pattern. The amount of mature frataxin in TAT-MTScs-FXN-treated Fxn1 and Fxn2 neurons was greatly increased compared to vehicle. Using mouse models, we also showed that mitochondria isolated from treated mice display detectable amounts of the protein. The study demonstrated that CCCP treatment, which reduce Dw m, of 293T cells, increased precursor form but not mature form which remains unaltered [37]). Our findings would be in agreement with these results. In muscle samples obtained from mice models, we did not observe any accumulation of TAT-MTScs-FXN, suggesting that it is mostly processed. One can hypothesize that impairment in protein processing is due to altered mitochondrial functions induced by frataxin deficiency, including the decreases in mitochondrial membrane potential (Dw m ) and considering that mitochondrial alterations are probably stronger in frataxin-deficient neurons than in frataxin-deficient muscle cell. As Dw m is a critical hub for the efficiency of mitochondrial protein import [17] and matrix protein translocation additionally requires ATP for mtHSP70 activity [18], major TAT-MTScs-FXN accumulation in frataxin-depleted DRG neurons than in Scr neurons could be explained by the shift of mitochondrial energetic state and the quality of mitochondria, as protein import activity can act for sensing mitochondrial fitness [38]. For this reason, we explore whether it was a condition specifically related to the nature of the construct or other mitochondrial proteins can suffer such impairment. For this reason, we selected HSP60, a molecular chaperone that must be imported and processed to reach the mitochondrial matrix. As shown in Figure 3, HSP60 processing is clearly impaired in Fxn1 and more clearly in Fxn2. The difference between both cultures can be attributable to lower levels of frataxin in Fxn2 cultures (as shown in Fig. S1). HSP60 impaired processing was not reported yet in frataxin-deficient cells, and it could be relevant for the disease. In this context, it is worth mentioning that, beyond its function assisting polypeptides to fold properly once imported into the mitochondria, defects affecting its molecular chaperone activity have been related to neurodegenerative diseases such as spastic paraplegia [39,40,41] or the MitCHAP 60 disease [42,43]. The impact that such deficiency could have in frataxin-depleted cells remains to be explored.
Concerning other neurodegenerative markers, we previously described [24] that frataxin-deficient cells DRG neurons showed decreased Dw m and apoptotic traits such as increased neurite degeneration and a-fodrin cleavage. The fragmentation of this structural protein has been observed in normal ageing [44] and neurodegenerative diseases such as Alzheimer disease [45]. The results presented here show that TAT-MTScs-FXN was able to reduce a-fodrin cleavage, neurite degeneration and cell death in a dose-dependent manner. Also, when administered at 12 hrs after transduction, a-fodrin cleavage is more efficiently reduced than when added at 48 hrs ( Fig. 4C and D). Rescuing from apoptotic process is reasonably more difficult at 48 hrs than after 12 hrs. Indeed, a-fodrin cleavage is produced by caspase 3 and calpain proteases, which become activated by altered calcium homoeostasis through activation of Bax, a pro-apoptotic protein [46,47]. Fodrin cleavage summarizes the activation of pro-apoptotic pathway involving proteins such as P-CREB, BAX, and others [48] with the participation of altered levels of calcium. Reduction in this fragmentation implies a reduction in pro-apoptotic factors and restoring calcium homoeostasis. Then, it is reasonable that reduction in a-fodrin cleavage-as shown in Figure 4A and B-and increase in procaspase 9 ( Fig. 4E and F) are both due to restoring frataxin levels. These results agree with decreased neurite degeneration (Fig. 5) and with the increased survival observed by supplementing the cultures with TAT-MTScs-FXN at several doses (Fig. 6).

845
Conditional mice models of the disease used in this study showed alterations in both aconitase activity and mitochondrial electron transport chain [14], thus compromising mitochondrial energy production. We showed here that mice injected with TAT-MTScs-FXN reaches the muscle mitochondria achieving 15% of the values displayed by healthy animals. Although the increase in these values is moderate, they showed to be able to restore succinate dehydrogenase activity (Fig. 8C). Concerning animal survival, although the differences in mean survival treated with 4 mg/kg were not significant, treatment with 10 mg/kg proved to be partially efficient as the mean survival showed a slight but significant increase, mainly in the interval of 60-70 days. Moreover, the age of death was significantly increased with 10 mg/kg treatment.
FA is characterized by gradual neurodegeneration that precedes fatal cardiac condition. Unfortunately, there is no suitable neuronal model that mimics neurodegeneration and allows screening for drugs efficiency. In this paper, we show the ability of TAT-MTScs-FXN to protect neurodegeneration of primary neurons that are frataxindepleted. These "in vitro" results combined with the data provided by tests performed "in vivo" using mice models of Friedreich ataxia suggest that TAT-MTScs-frataxin can be considered a candidate for the disease therapy. Figure S1 Frataxin levels in Scr, Fxn1 and Fxn2 after lentivirus transduction. Levels of frataxin were analysed by Western blotting in control (Scr) and frataxin-depleted neurons (Fxn1 and Fxn2). Actin was used as a control of protein loading.

Figure S2
Relative amounts of full length, processed (intermediate and mature) TAT-MTScs-frataxin forms. (A) The amounts of the frataxin forms were evaluated in Scr, Fxn1 and Fxn2 cultures after addition of TAT-MTScs-frataxin at 1, 3 or 7 lg/ml (as indicated) to each culture, 48 hrs after lentivirus transduction. Error bars represent mean AE S.E.M., n = 4 (B) Endogenous frataxin levels compared to treated cells. Cells from Scr, Fxn1 and Fxn2 untreated and treated with 1 and 7 lg/ml TAT-MTScs-FXN 48 hrs after lentiviral transduction were lysed after 5 days of culture and blotted to detect frataxin levels (b-actin was used for normalization). Note the increase in mature frataxin levels in treated cells compared to untreated cells.

Figure S3
Increased frataxin in brain samples (Sarsero model). Brain tissue samples were assayed for frataxin amounts using ELISA test. Treatment was initiated at 6-8 weeks of age. Mice were intravenously injected twice a week with vehicle or TAT-MTScs-frataxin at 4 mg/kg. Values are expressed as pg/mg of whole tissue extract. N = 3 for each condition.  Figure S4 Per cent of survival by administrating TAT-MTScs-frataxin intravenously. Survival of mice treated with vehicle or with TAT-MTScs-frataxin at 10 mg/Kg was followed (see Materials and methods). For controls, 5 animals were used and 10 animals for treatment. Note that no significant differences were observed in neither in mean nor in total lifespan. N = 5 for vehicle and N = 10 for TAT-MTScs-FXN.

Figure S5
Frataxin amounts in heart mitochondria of MCK mice as function of injection route. Hearts from MCK mice were assayed for frataxin amounts. Note that intraperitoneally (IP) Injected mice showed the higher amounts of frataxin compared to subcutaneously (SC), intramuscular (IM) or intravenous (IV). Values were expressed as picograms/total mitochondrial protein. N = 3 for each condition.