Calcitriol increases frataxin levels and restores altered markers in cell models of Friedreich Ataxia

Friedreich Ataxia (FA) is a neurodegenerative disease caused by the deficiency of frataxin, a mitochondrial protein. In primary cultures of dorsal root ganglia neurons, we showed that frataxin depletion resulted in decreased levels of the mitochondrial calcium exchanger NCLX, neurite degeneration and apoptotic cell death. Here we describe that frataxin-deficient dorsal root ganglia neurons display low levels of ferredoxin 1, a mitochondrial Fe/S cluster-containing protein that interacts with frataxin and, interestingly, is essential for the synthesis of calcitriol, the active form of vitamin D. We provide data that calcitriol supplementation, used at nanomolar concentrations, is able to reverse the molecular and cellular markers altered in DRG neurons. Calcitriol is able to recover both ferredoxin 1 and NCLX levels and restores mitochondrial membrane potential. Accordingly, apoptotic markers and neurite degeneration are reduced resulting in cell survival recovery with calcitriol supplementation. All these beneficial effects would be explained by the finding that calcitriol is able to increase the mature frataxin levels in both, frataxin-deficient DRG neurons and cardiomyocytes; remarkably, this increase also occurs in lymphoblastoid cell lines derived from FA patients. In conclusion, these results provide molecular bases to consider calcitriol for an easy and affordable therapeutic approach for FA patients.


INTRODUCTION
Friedreich ataxia (FA) is caused by decreased expression of the mitochondrial protein frataxin due to large expansions of GAA triplet repeats in the first intron of the gene.
Patients with FA suffer progressive limb and gait ataxia, dysarthria, reduced tendon reflex, extensor plantar responses and loss of position and vibration senses. Patients also develop cardiomyocyte hypertrophy causing heart failure, one of the main causes of death in FA patients [1]. Although many efforts are being done, there is no effective cure for the disease [2,3]. The pathologic changes occur first in dorsal root ganglia (DRG) with loss of large sensory neurons, followed by degeneration of the spinocerebellar and corticospinal tracts [4,5]. DRG neurons express the highest levels of frataxin and display a high vulnerability to frataxin down-regulation. The deleterious effects of frataxin depletion have been studied in DRGs using conditional knockout mice [6] and samples from patients with FA [7,8]. Using primary cultures of frataxin-deficient DRG, we observed alterations of several parameters compatible with calcium mishandling such as a decrease in mitochondrial membrane potential (Ψm), increased fodrin cleavage by calpain and caspase (two calcium-activated proteases) and Bax induction. These events leading to apoptotic cell death, can be minimized either by supplementing cultures with BAPTA, a calcium chelator, or by TAT-BH4, the antiapoptotic domain of Bcl-xL fused to TAT peptide [9]. Alteration in mitochondrial membrane potential can affect, among other processes, the opening of mitochondrial permeability transition pore. This event can also be triggered by calcium dyshomeostasis [10,11], a process in which NCLX, the mitochondrial calcium exchanger, plays a relevant role [12,13].
The synthesis of the active form of vitamin D is a mitochondrial process in which the precursor form, 25-OH-vitamin D3 (or calcidiol), is transformed to 1,25-(OH)2-vitamin D3 (calcitriol) by the 1α-hydroxylase, CYP27B1, a mitochondrial heme-containing enzyme, present in kidney as well as in many other tissues including brain [14].
Moreover, there are evidences that calcitriol can be synthesized locally in microglia [15]. The activity of this cytochrome is dependent on its coupling with ferredoxin 1 (FDX1), an Fe/S cluster-containing protein, to transfer electrons from a ferredoxin reductase to CYP27B1 [16][17][18]. The expression of this hydroxylase is controlled by calcitriol which binds to the intracellular vitamin D receptor (VDR), forming a heterodimer with retinoic acid X receptor (RXR) which translocate to the nucleus and contributes to the repression of the CYP27B1 gene transcription [19,20]. VDR has been identified in foetal (E12-E21) and adult DRG neurons [21,22] and its levels increase as a consequence of the rise in calcitriol concentration [23]. Besides self-regulating its synthesis, calcitriol exerts crucial effects in maintaining cellular redox balance and calcium signalling pathways [24]. Of note, calcitriol contributes to induce the expression of Nrf2 [25] and the antiaging protein Klotho [26], transcription factors that are both related to the antioxidant response and calcium homeostasis [27,28]. In terms of neuroprotection, calcitriol exerts anti-apoptotic and protective effects on glutamateinduced excitotoxicity to cultured hippocampal cells [29]. Also, calcitriol and tacalcitol, a structural and functional related compound, they both prevented neuronal damage caused by hydrogen peroxide-induced toxicity in the SH-SY5Y cell line [30]. In endothelial cells submitted to oxidative stress/hypoxic conditions, calcitriol supplementation diminishes cellular damage by increasing the expression of SOD1 and VGEF [31]. Connected to this finding, the positive effects of calcitriol in delaying Amyotrophic Lateral Sclerosis (ALS) progression has been attributed to its ability to induce axonal regeneration [32].
In this paper, we report that frataxin-deficient DRG neurons display reduced amounts of ferredoxin1 (FDX1), a result that prompted us to assay the effects of calcitriol supplementation. Of note, among the beneficial effects, calcitriol significantly increased frataxin amounts in DRG neurons and, consistently, restored altered molecular and cellular markers such as reduced levels of FDX1, mitochondrial calcium exchanger (NCLX) and mitochondrial membrane potential; moreover, fodrin fragmentation, neurite degeneration are clearly reduced and, as a result, cell survival improved.
Additionally, calcitriol increases frataxin amounts also in lymphoblastoid cell lines derived from FA patients and in frataxin-deficient cardiomyocytes. Taken together, these results support the beneficial role of calcitriol on cell physiology and encourage proposing calcitriol supplementation as a potential therapy easily applicable to FA patients.
To prevent growth of non-neuronal cells, culture media were supplemented with the anti-mitotic agent Aphidicolin (Sigma-Aldrich, Cat# A0781) at final concentration 3,4 µg/ml. After 1h of pre-plating in a p60 tissue dish (Corning Incorporated, Cat# 35004) at 37°C/5%CO2, the cells were then plated in a 24 well tissue dish (Corning Plasmids and production of lentiviral particles.-Lentiviral particles are routinely produced in our laboratory as described [9] with some modifications [34]. The shRNA lentiviral plasmids (pLKO.1-puro) for human/mouse/rat frataxin were purchased from Sigma. The RefSeq used was NM-008044, which corresponds to mouse frataxin. The clones used were TRCN0000197534 and TRCN0000006137 (here referred as FXN1, FXN2). The vector SHC002, a non-targeted scrambled sequence, served as a control (SCR). For DRG neurons, we use FXN1 and FXN2 and for cardiomyocytes FXN1. Lentiviral particles were propagated in HEK293T cells using the polyethylenimine (Sigma, Cat# 408727) cell transfection method as described previously (Purroy et al., 2018) using the plasmids pMD2.G (RRID:Addgene_12259) and psPAX2 (RRID:Addgene_12260) and tittered using the Quicktitter Lentivirus ELISA kit (Cell Biolabs, Cat# VPK-108-H). For DRG neurons transduction, the medium containing lentivirus particles (20 ng p24/1.000 cell) was added 2 days after plating and replaced with fresh medium 6h later. For cardiomyocytes transduction, 5.5 ng p24/1000 cells were added 4h after plating and culture medium was replaced 20h later.

Measurement of DRG neuronal survival and neurite degeneration.-Neuronal
survival and neurite degeneration was performed after 5 days of culture as described [33]. Neuronal survival was expressed as the percentage of cells counted at day 5 on the initial value in the same field of a cross-marked well. Four fields per well of three different wells were counted for each condition and the experiments were repeated at least 3 times. Counting was performed by three independent individuals, not blinded.
Neurite degeneration was measured with a ×32 lens and a grid, which was created over each image with NIH Image J with the grid plugin (Image size 680×512 and line area 10,000 pixels 2 ). Healthy and degenerated neurites (displaying neurofilament aggregates) were counted in three grid fields per image and at least three images per well were analysed. For each condition, we used three different wells. Experiments were repeated at least three times.
Mitochondrial membrane potential experiments.-The fluorescent probe JC-1 (5,5',6,6'-tetrachloro-1,1',3,3'-tetraethylbenzimidazolylcarbocyanine iodide) cationic dye (Abcam, Cat# Ab141387) was used to analyse the mitochondrial membrane potential (Ψm). JC-1 exhibits a potential-dependent mitochondrial accumulation and a dual-emission for green fluorescent monomer (λ 530 ± 15 nm) or red fluorescent Jaggregates (λ 590 ± 17.5 nm) depending on depolarized or polarized mitochondria, respectively. Three days after lentivirus transduction DRG neurons were incubated with 5μg/ml JC-1 dye-containing medium for 30-35 minutes at 37°C/5%CO2. Before starting the experiment, neurons were washed once with preheated PBS (GIBCO, Cat# 10010-015) and then maintained with PBS in order to reduce background in fluorescent images taken with an Olympus FluoView IX71 microscope (x16 lens). Representative images of depolarized mitochondria (green) and polarized mitochondria (red/orange fluorescence) obtained with two filters: Olympus U-MWIB2 filter (λ460-490 nm excitation, λ>510nm emission) and Olympus U-MWIG2 filter (λ520-550 nm excitation, λ>580nm emission). For quantitative analysis, the same image was taken with both filters. Images were then imported in NIH Image J software and the entire image was considered as ROI. As this filter overestimated the green fluorescent intensity, for a better analysis channels were split and the green fluorescent intensity, in which was subtracted the overestimated fluorescence, was used for green/red fluorescent intensity ratio analysis. Then, an increase in green/red fluorescent intensity ratio represents mitochondrial depolarization. Ten images per well of at least three different wells were analysed for each condition and the experiments were repeated 6 times.
Mitochondrial calcium measurements.-Mitochondrial calcium measurements were performed with a low affinity calcium indicator Rhod5N-AM (Kd=19µM, λ~551nm excitation, λ ~576 nm emission) (Cayman Chemicals, Cat# 20442) dissolved in DMSO (Sigma-Aldrich, Cat# D2650) and used as previously described [35]. Five days after lentivirus transduction, DRG neurons were incubated with 0,005% Pluronic F127 Values were expressed as mean ± SD for n≤4 or mean ± SEM (error bars) for n≥5. For all experiments, normality of data was assessed by using Kolmogorov-Smirnov and Shapiro-Wilk tests and equal variance by using Bartlett and Brown-Forsythe tests.
Then, one-way ANOVA was used to assess differences between groups for variable treatment. 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. Graphpad Prism 5.0 was used for all above analysis. Grubbs' test (OriginPro 2017) was used to identify outliers, but no outlier, if existent, was eliminated from the data.  S1A and quantification in S1B). The heme-containing 1-α-hydroxylase enzyme, responsible for the last step in the synthesis of the active form of vitamin D, is encoded by CYP27B1 gene. The expression of this gene is dependent on the cellular levels of calcitriol that, in combination with other factors, exerts a repression on CYP27B1 gene [37]. For this reason, measuring the amounts of CYP27B1 after frataxin depletion could also provide further evidences of calcitriol alteration. DRG neurons were processed as mentioned above and, as shown in Fig. 1B and F, the amounts of the enzyme were higher than in control (Scr) cells, thus suggesting a deficiency of calcitriol. In contrast, neurons treated with calcitriol displayed CYP27B1 levels close to those found in control cells. In addition, we tested the levels of CYP24A1, a mitochondrial enzyme induced in response to high levels of calcitriol to convert it into 24,25-hydroxy-vitamin D [38]. Figure 1C shows that in FXN1 and FXN2 neurons, the levels of CYP24A1 are reduced compared to control cultures (Scr) and recovered by calcitriol supplementation (quantification is shown in Fig. 1G). These findings are in agreement with those observed for CYP27B1, again suggesting an alteration in calcitriol biosynthetic pathway. Since those enzymes are all mitochondrial, we tested the amounts of a mitochondrial protein, LONP1, not related to CYP27B1, CYP24A1 or FDX1. As shown in Fig.1D, the levels of LONP1 remain stable in all the conditions tested and suggesting that the alterations described above should be specific events.

Altered amounts of FDX1 and CYP27B1 in frataxin-
Calcitriol increases frataxin amounts.-The beneficial effects of calcitriol supplementation described in previous sections could rely in several reasons, one of them being the restoration of frataxin levels. As shown in Fig. 2, calcitriol supplementation promotes a marked increase in mature form of the protein (about twofold) in frataxin-deficient DRG neurons ( Fig. 2A, quantification in 2B). Also, as a control of the cellular effects of calcitriol, we observed a rise in the VDR levels by calcitriol treatment (Fig.2A), which is a key feature of calcitriol upon entering the cells [39,40]. Furthermore, when calcitriol was tested on lymphoblastoid cells from FA patients and from healthy donors, a significant frataxin increase in FA cells was observed ( Fig. S2A and S2B). Also, frataxin increase was also observed in frataxin-deficient cardiomyocytes (see Supplementary Text and Fig. S3). Based on the relevance of these findings, we analysed the effect of calcitriol on altered cellular parameters caused by frataxin depletion.
Calcitriol supplementation restores normal levels of NCLX.-Mitochondrial calcium levels are mainly balanced by the mitochondrial calcium uniporter (MCU) for calcium entry and NCLX controlling the calcium efflux [41][42][43]. In a previous paper, we found that the amounts of NCLX in frataxin-deficient neurons were decreased and can be recovered by treating these cells with BAPTA, a calcium chelator. Reduced levels of NCLX were also found in frataxin-deficient cardiomyocytes and FA lymphoblastoid cell lines [34]. Maintaining proper levels of this mitochondrial calcium exchanger has been recently revealed of great importance in Parkinson disease [44] and in heart failure [45]. As shown in Fig. 3A, NCLX levels in frataxin-deficient neurons are, around 45% and 35% for FXN1 and FXN2, respectively, of the control values while, in calcitrioltreated neurons, NCLX levels were increased to 75% and 70%, respectively (quantitative analysis is shown in 3B). Since decreased NCLX impacts on the correct balance of mitochondrial calcium, a fluorescent probe (Rhod5N-AM) that predominantly accumulates in mitochondria [46] was used to monitor calcium accumulation in control and frataxin-depleted cells, treated or not with calcitriol. Figure   3C (upper row) shows the increased fluorescence in FXN1 and FXN2 as compared to control (Scr) neurons; in contrast, addition of 20nM calcitriol resulted in a significant reversion of calcium accumulation (Fig. 3C, lower row) which is in agreement to restored NCLX amounts. Quantitative analyses for cell soma and for neurites were performed as described in materials and methods and shown in Figure 3D (whole cells) and 3E (neurites only). As a complement, (Fig. 3F), images obtained by confocal microscopy show with more detail the punctate mitochondrial calcium accumulation (Rhod5N, red) and cytosolic calcium (Fluo8, green) in frataxin-deficient neurons.
Calcitriol restores mitochondrial membrane potential.-We previously showed that frataxin-deficient DRG neurons display altered cellular parameters such as decreased mitochondrial membrane potential (Ψm) [9,47]. Then we used this altered marker to test whether it could be reversed by calcitriol. The mitochondrial membrane potential was tested with the fluorescent JC1 probe as described in Materials and Methods. In this assay normal mitochondria display an orange/yellow fluorescence while mitochondria with altered Ψm, display a green fluorescence. Figure 4A shows that 20 nM calcitriol supplementation is able to reasonably maintain mitochondrial membrane potential in frataxin-deficient neurons (a representative image is shown in 4A. Histograms in Fig.   4B represent the fluorescent intensity ratio showing that a 29% (for FXN1) and 30% (for FXN2) more intensity of green/red fluorescence in frataxin-deficient DRG neurons.
These values are reduced by 15% and 24%, respectively, using calcitriol treatment. In addition, based on the knowledge that frataxin-deficient cardiomyocytes display altered cellular markers such as enlarged mitochondria and lipid droplets accumulation [34,48], the beneficial effects of calcitriol in these cells were also tested (data shown in Fig. S4).

6C). No significant changes were detected in the survival values of control cells (Scr)
with or without calcitriol administration.

DISCUSSION
The clue that frataxin depletion could be connected to vitamin D synthesis came from data [49] revealing that, on a human cell model of granulosa cells, frataxin depletion resulted in a deficient steroid (progesterone) synthesis. In this mitochondrial process, adrenodoxin (ferredoxin) transfer electrons from a reductase to the corresponding cytochrome CYP11A1, for the conversion of cholesterol to pregnenolone. Interestingly, ferredoxin is also the link between ferredoxin reductase (FDXR) and CYP27B1 for the final step of calcitriol synthesis, a reaction that takes place in mitochondria [50]. In addition, it was already described that ferredoxin physically interacts with frataxin [36].
In this context, the possibility that frataxin depletion could impact on enzymes of calcitriol biosynthetic pathway and downstream effects, lead us to analyse the effects of supplementing this compound on frataxin-deficient cells.
The results shown here reveal that FDX1 levels are significantly below of those found in normal DRG neurons or in FA lymphoblastoid cells (Figs, 1 and S1) thus supporting the rationale that calcitriol supplementation should be able to recover cell survival after frataxin depletion. Although we are aware that the explanation on how frataxin deficiency results in decreased FDX1 levels remains to be answered, deficiency in Fe/S cluster-containing proteins has been observed in several models of the disease [51,52].
Nevertheless, other explanations are possible: it would be conceivable that this protein could be the target of mitochondrial calcium-activated proteases, as a result of the calcium imbalance occurring after frataxin deficiency reported here (Fig. 3). Besides, one must bear in mind that frataxin deficiency is known to cause oxidative stress [53] and that Fe/S cluster proteins are prone to oxidative damage and degradation [54,55].
For this reason, we cannot rule out the possibility that calcitriol-induced recovery of FDX1 levels could be promoted by increased antioxidant defences since it is very well known that calcitriol triggers such a response. In fact, both effects can synergistically act on recovering FDX1 amounts.

The beneficial role of calcitriol in neurodegenerative diseases have been reported in
Alzheimer disease by reducing the levels of amyloid beta protein [56,57] or decrease neurotoxic events in animal models of Parkinson disease [58]. Also vitamin D contributes to neuroprotection [59] and prevents cognitive decline in aged brain [60]. It is important mentioning that VDR, the intracellular receptor of calcitriol, has been detected in neurons, astrocytes and oligodendrocytes highlighting the importance that both VDR and calcitriol can play on brain development [61,62]. Also, the presence of CYP27B1 in several cell types in brain including endothelial cells, microglia, neurons and astrocytes [63], suggests that calcitriol synthesis could be an in situ process in central nervous system.
The results also reveal that calcitriol is able to reverse altered cellular markers that we described in previous papers, such as NCLX levels. Restoring the levels of NCLX by calcitriol treatment is a relevant result ( Fig. 3A and B) because this protein is key to maintain the correct sodium/calcium balance in mitochondria [41,64] of several cell types such as neurons, lymphocytes, pancreatic beta cells, cardiomyocytes [43]. In nociceptive neurons, NCLX has a fundamental role due to its coordination with TRPV1 channels [65]; the inhibition of the exchanger and calcium imbalance has been described in a familial form of Parkinson's disease [66]; besides, NCLX has been related to cardiac hypertrophy [45] and involved in insulin secretion [67][68][69]. The reason why this exchanger shows reduced levels in frataxin-deficient cells is not well understood but it could be related to calcium-activated proteases since, as reported in a previous paper [34], BAPTA was able to partially restore NCLX to normal levels. In this context, we have recently shown that calpain inhibitors such as MDL28170 and calpeptin, result in partially restoration of the NCLX levels in frataxin-deficient DRG neurons and the reduction of calpain 1 levels results in neuroprotection [47]. Moreover, as recently published [13], alterations in mitochondrial membrane potential can negatively impact on NCLX function. Since we have observed that DRGs show altered Ψm, it could affect the activity of the remaining amounts of NCLX, thus aggravating the mitochondrial calcium homeostasis.
The main reason why calcitriol could exert these beneficial effects on frataxin-deficient cells is the observed increase in frataxin levels in several cell types. The results shown in DRG neurons (Fig.2) have been corroborated by the increases observed in lymphoblastoid cell lines derived from FA patients (fig, S2). To further complement these results, we have provided additional data showing that frataxin is significantly induced by calcitriol in frataxin-deficient cardiomyocytes (Fig.S3). In this cell model, calcitriol also exerts beneficial effects on altered markers, such as lipid droplets accumulation and enlarged mitochondria (Fig. S4). These restoring effects in cardiac myocytes are attributable to recovering frataxin amounts and improvement of mitochondrial functions, but a direct effect of calcitriol on increasing beta oxidation pathway could be also considered as suggested by the findings described in 3T3-L1 adipocytes [70]. Also, we cannot rule out the possibility that calcitriol contributes to preventing lipid droplet accumulation [71].
Although the mechanism through which calcitriol supplementation results in increased found for genes that are very well known VDR targets: genes related to calcium homeostasis such as parvalbumin, the calcium exchanger NCX1 (SLC8A2-2) or related to antioxidant defence genes such as glutathione peroxidase, or NRF2 (NFE2L2-4) to cite a few [24,28]. It is interesting remembering that NRF2, in turn, activates the transcription of antioxidant and detoxifying genes such as catalase, superoxide dismutases or thioredoxin reductase [72]. In this context, the beneficial effects of calcitriol, as a natural NRF2 inducer, are in agreement with one of the emerging therapies for FA is based on the activation of NRF2 including sulphoraphane [73], dimethyl fumarate [74] or omaveloxolone [75,76]. We also thank Cecabank "Tu Eliges" program. We thank Alba Caparrós for her work on lymphoblastoid cell cultures, Roser Pané for technical assistance, Anaïs Panosa for helping in confocal imaging and Dr. E. Vilaprinyó for assistance in statistics.         with a threshold value of 13. Finally, using the function "analyze particles", particles bigger than 2µm 2 and with a circularity bigger than 0,65 were quantified. The used value was the area of enlarged mitochondria per nuclei. For each condition, 15 images were taken. Lipid droplets were analyzed as previously described [34]. Briefly, cultured cells were washed 3x with PBS and loaded with 5 μM BODIPY 493/503 (ThermoFisher Scientific, Cat# D3922) and 0.05 μg/ml Hoechst33258 (Sigma, Cat# B2261) for nuclei staining during 20 min at 37°C. Cells were washed 1x with PBS and images were taken with a x20 lens using an Olympus IX71 microscope. Lipid droplets of each image were quantified by using ImageJ software: each image was transformed to 16-bit image and, then, to black and white image with a threshold value depending on the background of each culture. The area of lipid droplets was quantified using the function "analyze particles". Finally, the number of nuclei was calculated using the function "cell