Modeling riboflavin transporter deficiency type 2: from iPSC-derived motoneurons to iPSC-derived astrocytes

Introduction Riboflavin transporter deficiency type 2 (RTD2) is a rare neurodegenerative autosomal recessive disease caused by mutations in the SLC52A2 gene encoding the riboflavin transporters, RFVT2. Riboflavin (Rf) is the precursor of FAD (flavin adenine dinucleotide) and FMN (flavin mononucleotide), which are involved in different redox reactions, including the energetic metabolism processes occurring in mitochondria. To date, human induced pluripotent stem cells (iPSCs) have given the opportunity to characterize RTD2 motoneurons, which reflect the most affected cell type. Previous works have demonstrated mitochondrial and peroxisomal altered energy metabolism as well as cytoskeletal derangement in RTD2 iPSCs and iPSC-derived motoneurons. So far, no attention has been dedicated to astrocytes. Results and discussion Here, we demonstrate that in vitro differentiation of astrocytes, which guarantee trophic and metabolic support to neurons, from RTD2 iPSCs is not compromised. These cells do not exhibit evident morphological differences nor significant changes in the survival rate when compared to astrocytes derived from iPSCs of healthy individuals. These findings indicate that differently from what had previously been documented for neurons, RTD2 does not compromise the morpho-functional features of astrocytes.


Introduction
Riboflavin transporter deficiency (RTD), previously known as Brown-Vialetto-Van Laere syndrome (BVVLS, OMIM 613350, 614,707), is an autosomal recessive rare childhood neurodegenerative disease (<1/1,000,000 prevalence), characterized by sensorineural hearing loss, bulbar dysfunction, diffuse muscle weakness that may result in respiratory insufficiency.In particular, it can be caused by biallelic mutations in the SLC52A3 gene (613350) on chromosome 20p13 (OMIM 211530), encoding the Rf transporter RFVT3, mainly expressed in the testis, small intestine, is kidney, and placenta.
Since 2011, over 180 genetically confirmed RTD2 patients have been identified.Typical clinical features reported include optic atrophy, sensorineural hearing loss, muscle weakness, sensory ataxia, pontobulbar palsy, dysautonomia, and respiratory compromise [5-6, see Table 1].Untreated patients with early onset generally rapidly decline, resulting in severe disability and early death due to respiratory failure (Green et al., 2010;Jaeger and Bosch, 2016).SLC52A1 and SLC52A3 genes, encoding Rf transporter RFVT1 and RFVT3, respectively, have also been linked to riboflavin-responsive disease.RFVT1/2/3 are transmembrane proteins (Fujimura et al., 2010), which mediate the absorption of Rf, as is not synthetized in humans but introduced with the diet (Suwannasom et al., 2020).Rf is the precursor of coenzymes obtained through enzymatic reactions: flavin mononucleotide (FMN) and flavin adenine dinucleotide (FAD).They take part in maintaining cellular redox balance: indeed, they are cofactors of several enzymes involved in both generating and scavenging of reactive oxygen species (ROS) and reactive nitrogen species (RNS) (Fransen et al., 2012).Excess of ROS and RNS can lead to oxidative stress and cause damages to biological molecules (DNA, RNA, proteins, lipids), thus impairing normal cell functions (Espinós et al., 2020).Various cell types are able to counteract ROS/RNS accumulation by activating a specific Nrf2-mediated antioxidant response and, in particular, astrocytes are more resistant than neurons because they can strongly upregulate Nrf2-mediated gene expression (Baxter and Hardingham, 2016).
Astrocytes are the most common glial cell type in the central nervous system (CNS) and are crucial for its proper function and health (Barres, 2008).They are multifunctional cells regulating neuronal synapses by promoting synaptogenesis and display morphological, functional, and molecular differences depending on the area in which they are located (John Lin et al., 2017).Additionally, astrocytes have a role in the uptake and metabolism of glutamate (Santello and Volterra, 2010), maintenance of blood-brain barrier (BBB) and modulation of extracellular space volume.Moreover, they supply energy substrates to neurons, release ROS scavengers and regulate the process of neurite outgrowth and myelination (Verkhratsky and Nedergaard, 2018).
By using an in vitro disease model based on induced pluripotent stem cells (iPSCs) derived from RTD2 patients, we previously characterized the phenotype of RTD2 motoneurons focusing in particular on mitochondria, redox status, and neuronal morphology and activity (Rizzo et al., 2017;Marioli et al., 2020;Colasuonno et al., 2020b;Niceforo et al., 2021).Despite these insights, many aspects of the pathophysiological mechanisms implicated in RTD2 still deserve deep investigation, including those related to the cellular environment, astrocytes' involvement and neuron-glia interactions.With this in mind, we used the patient-derived iPSC model to functionally profile patient-derived astrocytes to explore a possible role of these cells in the pathogenesis of RTD2 (Table 1).

Clinical information
Clinical features of Pt1 had been previously reported (Niceforo et al., 2021).In particular, patient P1 developed macrocytic anemia and dysphagia at 3 months of age.Optic atrophy, axial muscle weakness, sensory ataxia, and respiratory compromise were noted at 1 year, and bilateral sensorineural hearing loss at 2 years.Patient P1 is now 13 years old and has remained neurologically stable since beginning riboflavin (75 mg/kg QID) and antioxidant therapy at 2.5 years of age.Moderate improvements in vision and hearing were measured during the first year after beginning riboflavin treatment, with the patient receiving a successful cochlear implant at age 3.9 years and currently wearing a night-time scoliosis brace since age 10.
Patient 2 is a girl presenting with rapid onset arm weakness and dysphagia at 8 months of age.At the time of diagnosis at 12 months old, severe optic atrophy and bilateral sensorineural deafness, sensory neuropathy, and complete diaphragm paralysis were noted.Genetic analysis performed by whole exome sequencing (WES) showed two mutations in SLC52A2 (c.505C > T, p.Arg169Cys; c.1030_1031del, p.Leu344AlafsX100).Total flavin cerebrospinal fluid levels pre-Rf treatment were ~ 25% of controls, which normalized on 100 mg/kg of a 50/50 mix of riboflavin and riboflavin 5 phosphate QID.At age 6, the child continued to have severe upper limb and axial predominant muscle weakness, has developed severe scoliosis, and no significant improvement in vision or hearing.She continued to receive respiratory support but has regained partial diaphragm movement.This child recently developed cataracts requiring lens replacement, which has not previously been reported in children with RTD Type 2.
Patient 3 is a 4-year-old girl, third child born from healthy and non-consanguineous parents.Poor weight gain was reported at 2-month-old.At 6 months a psycho-motor development delay was evident.The baby showed marked and generalized hypotonia, had not achieved head control, and deep tendon reflexes were absent.Brain MRI showed a thinned appearance of the optic nerves.Visual evoked potential (VEP) was abnormal whereas auditory evoked potentials (AEP) did not detect brainstem abnormalities.ENG/EMG was conclusive for axonal sensorimotor neuropathy.Comprehensive metabolic screening including acyl-carnitines, plasma amino acids and urinary organic acids profiles were normal.Genetic analysis performed by WES showed two compound heterozygous pathogenic variants in SLC52A2, (c.505C > T, p.Arg169Cys; c.593G > A, p.Trp198*).Riboflavin was started at a dosage of 20 mg/kg/day.At the last clinical assessment (4 years), the baby presented poor growth, generalized hypotonia with kyphoscoliosis, nystagmus, possible visual pursuit, good head balance while maintained in a sitting position, independent sitting position with upper limbs support, and slight dysphagia mainly for liquids.She was orally fed by mouth with modified food consistency, and she had often respiratory infections.Polysomnography and video-fluoroscopy studies showed no abnormalities.
The coding sequences of exons 3 and 4 for patient 2, and exon 3 for patient 3 of the SLC52A2 gene (NM_001363118) were amplified by PCR specific primers (available on request).Amplicons were purified using Exo-SAP (GE Healthcare) and directly sequenced using BigDye 3.1 chemistry (Applied Biosystems, Foster City, CA, United States) with an ABI Prism 3,130 xl automatic sequencer (Applied Biosystems).

Alkaline phosphatase assay
Alkaline phosphatase (ALP, SCR004, Sigma Aldrich) staining was performed on cells fixed in 4% PFA.The samples were incubated at RT for at least 30 min with a solution based on naphthol AS-BI and fast red violet LB.The cells were photographed using a Leica DM1000 (Leica Microsystems) equipped with Leica LAS X software (Leica Microsystems).

Differentiation of iPSCs into astrocytes
iPSCs were differentiated into astrocytes (ASTROs) using an adapted protocol from Yan et al. (2013).According to this protocol, when the iPSCs reached the 80% of confluence, they were plated into 6well-plates pre-coated with Matrigel.The cells were split with Stempro Accutase (A1110501, Thermofisher Scientific) and plated at 2,5-3 × 10 5 cells/well with Y-27632 to promote cell adhesion.After 24 h, the culture medium was switched to PSC Neural Induction Medium (A1647801, Gibco).The medium was changed every other day from day 0 to day 4; then, Neural Induction Medium was changed every day until day 7 when the cells reached the confluence.After the induction phase, primitive Neuronal Stem Cells (NSCs) were dissociated with Accutase and plated on Matrigel-coated 6well-plate at a density of 0.5-1 × 10 5 cells per cm 2 in an NSC expansion medium containing 50% Neurobasal medium, 50% DMEM/F12 (10,565,018, Thermofisher Scientific), and Neural Induction Supplement.During the expansion phase, the medium was changed every other day until day 7.When the 90% of confluence was reached, NSCs were split until step 7 and, in each passage, Y-27632 was added to the medium to prevent cell death.The NSCs can be cryopreserved in NSC expansion medium with 10% DMSO.
For the following 7 days, NSCs were cultured in DMEM/F-12 GlutaMAX supplemented with N-2 Supplement 1X, FBS 1% and 10 ng/mL bFGF.In the culture medium, 20 ng/mL recombinant human ciliary neurotrophic Factor (CNTF, 450-13, PeproTech) were added after 7 days from the beginning of ASTRO differentiation.At 90% confluence, ASTROs were split until passage 7 and, in each passage, the cells were cultured in the medium supplemented with Y-27632 to prevent adhesion difficulties.During each phase of differentiation, the cells were cultured at 37°C, 5% CO 2 and 21% O 2 .

SDS-PAGE and western blotting
Preparation of cell extracts and western blot analyses were performed as follows: the cells were lysed in RIPA buffer (R0278, Sigma Aldrich) and supplemented with complete protease inhibitor cocktail (A32959, Sigma Aldrich).Proteins were separated by SDS-PAGE and following electrophoresis, proteins were transferred to nitrocellulose membrane using the Trans-Blot Turbo transfer system (Bio-Rad Laboratories).Membranes were blocked in 5% milk for 1 h, at RT.The primary antibodies used were: anti-GFAP 1:2000 (556,328, BD Pharmingen), anti-EAAT2 1:2000 (SC-365634, Santacruz), anti-GAPDH 1:10000 (G9545, Sigma Aldrich).They were blotted O/N at 4°C.The secondary antibodies were incubated for 1 h, at RT, and membranes stained with Clarity Western ECL Blotting Substrate (170-5,060, Biorad) and acquired using Uvitec.The analyses were performed using ImageJ Software and three biological replicates (n = 3 minimum) were performed for each experiment.

TUNEL assay
TdT-mediated dUTP-X nick end labeling (TUNEL, 11684795910, Roche) was performed to identify double-stranded and singlestranded DNA breaks, labeling the free 3′-OH terminus with modified nucleotides in an enzymatic reaction.The ASTROs were fixed with 4% PFA for 10 min at room temperature (RT); the samples were treated as described in the protocol and nuclei were counterstained using Hoechst 33342 diluted 1:10000 and incubated for 10 min at RT. Coverslips were mounted using PBS:glycerol (1:1).The ratio between the apoptotic and the total cells' number was evaluated and GraphPad Prism was used to calculate the mean values ± SEM.

Neurite length assay
Motoneurons were plated at a density of 5,000 cells/well in a 96-well plate (92,696, TPP) pre-coated with Matrigel and neurites' length was measured at the end of differentiation (30 days) using the live imaging system Incucyte SX5 (Sartorius).Phase-contrast images were acquired and processed using NeuroTrack Incucyte software.The parameters were optimized according to the workflow outlined in the manufacturer's manual.Raw data neurite lengths were exported to Microsoft Excel and GraphPad Prism was used to obtain the graph, calculate mean values ± SEM and perform ad hoc statistical analyses.

RNA isolation and reverse transcriptase-polymerase chain reaction analyses
RNA was extracted from Ctrl and RTD2 iPSCs, iPSCs-derived MNs and ASTROs using Trizol Reagent (15,596,026, Ambion by Life Technologies), according to the manufacturer's instructions.Each RNA sample was quantified using the Nanodrop 2000 Spectrophotometer (Thermo Fisher Scientific) and 1 μg of total RNA was reverse transcribed using LunaScript RT SuperMix Kit (E3010, BioLabs) following the instructions.Three independent experiments were performed for each sample.

Quantitative real-time PCR
Quantitative Real-Time PCR (qRT-PCR) experiments were performed using Fast SYBR Green Master Mix (4,385,612,Applied Biosystems) and the QuantStudio 7 Pro Real-Time PCR System (Applied Biosystems).The analyses were performed following the ΔΔCt method: the data were represented as fold increase versus control sample, calculated by the 2^[−(ΔΔC)] formula.Expression Frontiers in Cellular Neuroscience 06 frontiersin.orglevels of different genes were compared with those of the housekeeping gene GAPDH, and they were normalized to control's expression.The values were expressed in arbitrary units.Three independent Quantitative Real-Time PCR were performed for each sample.Below is a table (Table 2) with the sequence of all primers used.

Statistical analyses
The data obtained were represented using mean ± SEM.For all experiments, multiple technical replicates and biological replicates were utilized and a minimum of three independent experiments were performed for each assay.Quantification and statistical analyses of the control cells were reported as the mean of the data derived from the two cell lines (AG28851 and AG28869).Significance was tested using ordinary one-way ANOVA (parametric tests) for normally-distributed data.GraphPad-Prism software (Prism 8.0.2,GraphPad Software) was used to analyze the data.

Characterization of two new patient-derived iPSCs and RTD2 motoneurons
Two lines of RTD2 patients' fibroblasts with biallelic pathogenic variants in SLC52A2, obtained from skin biopsies, were reprogrammed into iPSCs introducing episomal vectors by nucleoporation.Four clones from each primary fibroblast line were selected based on the ESC-like morphology (rounded shape colonies with defined borders) and picked for following amplification and characterization (Figures 1A,C,E).Sanger sequencing confirmed that the SLC52A2 mutations identified in the patients' fibroblasts were present following reprogramming (Supplementary Figures S1A,B).Pluripotency of clones was confirmed by enzymatic assay using ALP (Figures 1B,D,F) and by Real-time PCR (qRT-PCR) evaluating pluripotency markers, in particular OCT4 (Figure 1I), SOX2 (Figure 1J), and KLF4 (Figure 1K).
The pluripotency of the picked clones was further confirmed with immunofluorescence assays using the pluripotency markers OCT4, SOX2, SSEA4 and TRA1-60 (Figure 1G).In addition, we tested their ability to differentiate into lineages belonging to the three germ layers (endoderm, mesoderm, and ectoderm) by Trilineage differentiation assay.Immunofluorescence assays were used to assess the positivity of the obtained iPSC clones to the following specific markers: SOX17 for the endoderm; TBXT or BRACHIURY for the mesoderm; and NCAM for the ectoderm (Figure 1H).Furthermore, we differentiated Ctrl and RTD2 iPSCs into motoneurons as described in Corti et al. (2012) to analyze the differentiating ability of the iPSCs derived from the two newly obtained patients' lines.Immunofluorescence of TUJ1 confirmed that their neuronal morphology is compromised (Figure 2), as already demonstrated for neuronal cultures derived from patient 1 and other RTD2 patients (Marioli et al., 2020;Colasuonno et al., 2020a;Niceforo et al., 2021).Motoneurons derived from the previously established and newly generated RTD2 iPSC lines showed significantly shorter neurites than Ctrl motoneurons (Figures 2A,C).We evaluated the SLC52A2 mRNA levels in Ctrl and RTD2 iPSCs and Ctrl motoneurons showed increased levels versus Ctrl iPSCs, and interestingly, increased SLC52A2 levels were found in RTD2 motoneurons derived from patient 1 (c.155C > T; c.935 T > C) and patient 2 (c.505C < T; c.1030_1031del) compared to Ctrl motoneurons (Figure 2B).

The generation of ctrl and RTD2 astrocytes
To investigate the role of astrocytes in RTD2 pathology, Ctrl and RTD2 iPSCs (Figure 3A) were differentiated into astrocytes in a 60-day protocol.NSCs were derived from RTD2 and Ctrl iPSCs following 30 days of differentiation (Figure 3B) using an adapted protocol from Yan et al. (2013) and immunofluorescence analyses were performed with NESTIN and PAX6, as markers of neuronal progenitor cells (Figures 3C,D).The cellular morphology changed from a rounded shape to an elongated form and the percentage of OCT4 positive cells (representing the pluripotent cells) correctly
Once the NSCs were obtained, cells were differentiated for other 30 days to obtain mature astrocytes (Figure 3E).To demonstrate that iPSCs successfully differentiated into astrocytes, immunofluorescence and western blotting analyses for the astrocytic marker GFAP were performed (Figures 3F,H).Further western blotting analyses confirmed the protein expression of another astrocytic marker, EAAT2, in the obtained astrocyte cultures (Figure 3H).Statistical differences were not observed between Ctrl and RTD2 astrocytes, both in GFAP (Figure 3J) and EAAT2 (Figure 3I) levels, indicating that RTD2 iPSCs are able to undergo astrocyte differentiation similarly to Ctrl iPSCs.
The expression level of SLC52A2 mRNA was evaluated using qRT-PCR analyses.The results showed that Ctrl astrocytes had significantly higher levels compared to Ctrl iPSCs.Similarly, RTD2  Frontiers in Cellular Neuroscience 09 frontiersin.orgastrocytes had higher levels compared to RTD2 iPSCs (Figure 3G).However, there was no significant statistical difference observed among Ctrl and RTD2 iPSCs (Figure 3G).

Cell viability and death studies of ctrl and RTD2 astrocytes
To evaluate the viability of RTD2 astrocytes compared to Ctrl ones, we measured their cell viability using the MTT assay.The results showed no significant difference between Ctrl and RTD2 astrocytes (Figure 4A).Furthermore, the TUNEL assay was used to analyze cell death in Ctrl and RTD2 astrocytes and the results confirmed no differences between the two groups.The quantification of the dead cells over the total number of cells (Figure 4B) showed that the percentage of TUNELpositive cells was similar in Ctrl and RTD2 astrocytes (Figure 4C).

Analyses of antioxidant response in iPSCs, motoneurons and astrocytes using qRT-PCR
Given the pivotal role of astrocytes in supporting neuronal function and since RTD2 motoneurons and iPSCs were documented to display energy dysmetabolism and redox imbalance (Colasuonno et al., 2020a,b), we investigated these features analyzing the molecular mechanisms underlying the cellular redox homeostasis in Ctrl and RTD2 iPSCs, motoneurons and astrocytes.To this aim, we performed qRT-PCR analyses to evaluate the expression of genes involved in the antioxidant response.Firstly, we analyzed the expression of NRF2, since it is the major orchestrator of the cellular antioxidant defense.The results indicated that in RTD2 iPSCs, NRF2 expression was comparable with that of Ctrl iPSCs (Figure 5A).NRF2 is a transcription factor that regulates several phase II detoxifying and antioxidant enzymes containing the antioxidant responsive element (ARE) in the promoter region, such as SOD1, SOD2, GPX3, HO-1, and System X C − (xCT subunit), thus we investigated the expression level of these ARE genes in RTD2 and Ctrl cells.The data obtained from the qRT-PCR analysis showed that the expression level of SOD1 (Figure 5B), SOD2 (Figure 5C), GPX3 (Figure 5D) and HO-1 (Figure 5E) were comparable between Ctrl and RTD2 iPSCs.
After differentiation from iPSCs, we analyzed the same genes in motoneurons.We observed a reduction in NRF2 expression levels in all RTD2 motoneurons compared to Ctrl motoneurons.Specifically, there was a statistically significant reduction in motoneurons generated from Pt1, Pt2 and Pt3 (Figure 5F).In relation to the expression of ARE genes, it was observed that the SOD1 mRNA levels in RTD2 motoneurons for all patients were similar to those in Ctrl motoneurons (Figure 5G).Notably, an altered expression of SOD2 was observed in RTD2 motoneurons, with a significant increase in all RTD2 motoneurons (Figure 5H) compared to Ctrl motoneurons.Similarly, GPX3 expression levels were increased in all RTD2 motoneurons (Figure 5I).Regarding the mRNA levels of HO-1, a statistically significant reduction was observed in all RTD2 motoneurons compared to Ctrl motoneurons (see Figure 5J).
Concerning the analyses on astrocytes, the expression levels of NRF2 were found to be similar between Ctrl and RTD2 astrocytes (Figure 5K).Similarly, no significant differences in the mRNA levels of SOD1 (Figure 5L), SOD2 (Figure 5M), GPX3 (Figure 5N), HO-1 (Figure 5O) and xCT (Figure 5P) among lines were observed.

Discussion
RTD2 is a rare neurodegenerative disorder caused by mutations in SLC52A2 gene, encoding the Rf transporters RFVT2.Riboflavin, known as vitamin B2, is a water-soluble vitamin that cannot be produced endogenously by humans and is an indispensable part of the diet.Noteworthy, this vitamin is the precursor of the FMN and FAD coenzymes.Since RTD2 is rare and, as other neurodegenerative diseases, the in vivo models fail to recapitulate the human clinical phenotype and disease progression (Colasuonno et al., 2022), hiPSCs represent an informative model system to explore the mechanism of disease and pathophysiology.
It is known that RTD2 motoneurons show morphological/ cytoskeletal abnormalities and activity, and mitochondrial alterations (Colasuonno et al., 2020a,b;Niceforo et al., 2021), which are partially recovered after the treatment with a combination of Rf and antioxidants (Marioli et al., 2020).Despite this, unfortunately, not all patients are responsive to Rf, as some patients do not benefit from Rf supplementation or only for a limited amount of time.With the aim to more deeply understand the RTD2 pathophysiology, we investigated the involvement of astrocytes in RTD2 pathology, based on the important role of these cells in neuronal function (Dermitzakis and Clark, 2002;Odom et al., 2007;Ilieva et al., 2009;Ciccolella et al., 2012).We, therefore, focused our attention on a possible contribution of astrocytes to RTD2 pathogenesis, in particular by analyzing the efficiency of astrocyte differentiation from patient-derived iPSCs and their redox status.
We successfully differentiated RTD2 iPSCs into astrocytes.Firstly, we differentiated Ctrl and RTD2 iPSCs into NSCs expressing the neuronal progenitors' markers PAX and NESTIN (Figure 3).Then, to evaluate the efficiency of astrocyte differentiation, we performed immunostaining for GFAP and western blotting for GFAP and EAAT2, both markers of mature astrocytes.
The positivity of both Ctrl and RTD2 cells (Figure 3) confirms that RTD2 iPSCs can successfully differentiate into astrocytes (Su et al., 2003;Fox et al., 2004).These findings suggest that the difficulties observed in RTD2 iPSCs differentiating into motoneurons (Figure 2A) do not apply when differentiating into astrocytes (Figure 3).
We evaluated the expression level of SLC52A2 mRNA.By qRT-PCR, we observed no differences in SLC52A2 expression between Ctrl and RTD2 iPSCs.Noteworthy, RTD2 astrocytes showed reduced three biological replicates (n = 3).*** p ≤ 0.001, **** p ≤ 0.0001, according to multiple comparisons of Ctrl iPSCs vs. RTD2 iPSCs or Ctrl and RTD2 MNs (ordinary one-way ANOVA).SLC52A2 mRNA levels when compared to Ctrl astrocytes, but further specific studies will be necessary to investigate this peculiar feature of RTD2 astrocytes.Interestingly, SLC52A2 mRNA expression was found to be higher in RTD2 motoneurons compared to Ctrl motoneurons (Figure 2B).This suggests that motoneurons require RFVT2 to perform their functions, and in RTD2 neurons where the RFVT2 protein is partially functional, the cell transcriptional machinery is forced to translate more mRNA copies of the SLC52A2 gene as a compensatory mechanism.
To accomplish a better characterization of RTD2 astrocytes, we evaluated their viability using MTT and TUNEL assays, demonstrating no significant changes in both these features in Ctrl versus RTD2 astrocytes, underling that SLC52A2 mutations did not influence these features in RTD2 astrocytes (Figure 4).
In RTD2 pathology, mutations of the Rf transporters induce most likely alterations in the absorption of Rf, consequently causing defects in the levels of FAD and FMN, and affecting all molecular mechanisms dependent on them, and particularly the redox status regulation (Barile et al., 2016).For this reason, we decided to analyze the mRNA levels of ARE genes involved in the antioxidant response of RTD2 iPSCs, motoneurons and astrocytes.We first evaluated NRF2 mRNA expression, as NRF2 is a key orchestrator of the antioxidant response required to counteract ROS accumulation (Ma, 2013).RT-qPCR analyses indicated that NRF2 levels were the same in both Ctrl and RTD2 iPSCs (Figure 5A).Although the oxidative processes increase following neuronal differentiation, as the energy production in neurons relies on oxidative phosphorylation (Trigo et al., 2022), NRF2 levels in RTD2 motoneurons were reduced with respect to Ctrl motoneurons.The obtained results indicate that this antioxidant defense pathway is altered in RTD2 motoneurons (Figure 5F), and this is consistent with our previously reported data (Niceforo et al., 2021).NRF2-mediated antioxidant response is particularly active in astrocytes.Indeed, several studies demonstrated the role of astrocytic NRF2 signaling in   neuroprotection (Tanaka et al., 1999) as well as in neurotoxicity (D'Ezio et al., 2021).In an in vitro model of Alzheimer's disease, the activation of NRF2-mediated antioxidant response and the consequent upregulation of system Xc − is involved in neuronal death due to increased glutamate release (D'Ezio et al., 2021).Therefore, we hypothesized that the behavior of astrocytes may be relevant in the pathogenesis of RTD2, although the most obvious effects of the disease are on motoneurons.Our findings indicate that NRF2 levels were the same in both RTD2 and Ctrl astrocytes (Figure 5K) and are consistent with an unmodified phenotype of RTD2 astrocytes, and in line with the absence of morphological differences with respect to Ctrl astrocytes.We also evaluated the expression of some NRF2-driven ARE genes as SOD1, SOD2, GPX3, HO-1 and xCT in both Ctrl and RTD2 iPSCs, motoneurons and astrocytes (Figure 5).Interestingly, we found that the expression of all these genes was similar in Ctrl and RTD2 iPSCs, thus confirming the absence of impairment of the antioxidant response in RTD2 iPSCs.
In RTD2 motoneurons, NRF2 levels were reduced in RTD2 compared to Ctrl motoneurons.In RTD2 cells, we also observed higher mRNA levels of some ARE genes when compared to the control cells.In this respect, it should be recalled that in addition to transcriptional control, epigenetic regulation, and posttranscriptional modifications can also play essential roles in controlling the levels of these phase-II enzymes.These processes include, for example, stabilizing SOD mRNAs and regulating their translation (Miao and St Clair, 2009).In all RTD2 motoneurons, we observed that the levels of SOD2 mRNA were increased with respect to Ctrl motoneurons (Figure 5H), suggesting that patients motoneurons try to counteract oxidative stress through different ROS-scavenging systems.Among ROS species, O 2 − is one of the most harmful to cell health.For this reason, enzymes have evolved to metabolize it as SOD1 (Cu,Zn-SOD) in the cytosol and mitochondrial intermembrane space, and SOD2 (Mn-SOD) in the mitochondrial matrix.They catalyze the conversion of superoxide to hydrogen peroxide.Despite having the same function as SOD2, the SOD1 enzyme was not found to be altered in RTD2 motoneurons compared with Ctrl motoneurons.These data suggest that ROS species may be more abundant in mitochondria, which indeed appear damaged, as described in Colasuonno et al. (2020b).SOD2 is also associated with the GPX3 activity, a seleno-protein with antioxidant potential that catalyzes the reduction of hydrogen peroxide, hydroperoxides, and lipid hydroperoxides by reduced glutathione (GSH).GSH plays an important role at the neuronal level as it is a free radical scavenger, redox modulator of ionotropic receptor activity and a neurotransmitter (Bains and Shaw, 1997).In RTD2 motoneurons, also GPX3 mRNA levels were increased when compared to those of Ctrl motoneurons (Figure 5I).Moreover, we discovered out an altered NRF2/HO-1 pathway in RTD2 motoneurons, thus providing a possible explanation for the imbalance of the redox status already observed in these cells (Colasuonno et al., 2020b;Niceforo et al., 2021).Indeed HO-1, whose levels were significantly decreased in all RTD2 motoneurons (Figure 5J), plays a role in attenuating the production of ROS, as well as in reducing secondary endoplasmic reticulum (ER) stress and neuronal apoptosis (Xu et al., 2020).It should be noted that ER is the main intracellular calcium storage (Koch, 1990), and interestingly, RTD2 motoneurons have impaired calcium influx (Niceforo et al., 2021).Hence, the decreased levels of H0-1 mRNA may promote ER stress, thus leading to altered release of calcium, impairment of calcium signaling and neuronal functionality in RTD2 motoneurons.Importantly, the RTD2 cell type that showed the highest significantly altered gene expression levels were the motoneurons, while the analyses of SOD1, SOD2, GPX3, HO-1 and xCT expression levels in Ctrl and RTD2 iPSCs and astrocytes (Figure 5) did not show statistical differences.
Altogether, the data collected on the expression of genes involved in the regulation of antioxidant defenses suggest the presence of increased oxidative stress in RTD2 compared to Ctrl motoneurons.Importantly, while RTD2 motoneurons displayed altered patterns of mRNA levels of genes relevant to respond to oxidative stress, RTD2 astrocytes showed mostly uncompromised mRNA levels of the same genes.

Conclusion
In the present study, we used patient-derived iPSCs to better study and investigate RTD2 pathogenesis focusing on a previously unstudied cell type, the astrocytes.Exploiting this model, we deepened some new aspects of RTD2 pathology.In particular, we confirmed the ability of Ctrl and RTD2 iPSCs to differentiate into astrocytes.The data indicate that RTD2 astrocytes do not show altered features in terms of morphology, cell death nor viability compared to Ctrl astrocytes (Figure 6).Furthermore, by analyzing the mRNA levels of genes that regulate the redox state, we found an alteration in RTD2 iPSCs, and especially in motoneurons when compared to Ctrl cells.We also discovered that RTD2 astrocytes are similar to those derived from healthy individuals, suggesting that they are not affected by alterations of SLC52A2 in terms of antioxidant gene expression.Noteworthy, the most relevant difference we observed in astrocytes concerns the expression of RFVT2.In RTD2 astrocytes, mRNA levels of this transporter are reduced by about half compared to control astrocytes.Additional research is required to clarify this aspect and gain a better understanding of the role played by astrocytes in RTD2, because the current data suggest that the motoneuron is the most affected cell type in RTD2 pathology.

FIGURE 3
FIGURE 3 Characterization of Ctrl and RTD2 NSCs and ASTROs.(A) Brightfield images of Ctrl and RTD2 patient-derived iPSCs.(B) Brightfield and immunofluorescence images of Ctrl and RTD2 NSCs, highlighting the positivity to (C) NESTIN (in magenta), a marker of neuronal progenitors and OCT4 (in green), a marker of pluripotent cells.(D) Immunofluorescence analyses of PAX6 (in green), transient maker of neural progenitor cells.Nuclei are counterstained with Hoechst (in blue) (scale bar = 25 μm).(E) Brightfield images of Ctrl and RTD2 ASTROs (obtained from an optical microscope with 4x magnification).(F) Images of immunofluorescence assays for GFAP staining astrocytes (in magenta).(G) SLC52A2 mRNA levels [F (7, 16) = 4.4, p = 0.0063, ANOVA] evaluated by RT-qPCR, using GAPDH as housekeeping gene.Data are normalized to Ctrl iPSCs and are presented as mean ± SEM of three independent biological replicates (n = 3).* p ≤ 0.05, ** p ≤ 0.01, according to multiple comparisons (ordinary one-way ANOVA).(H) Western blot analyses showing the expression levels of GFAP and EAAT2.GAPDH was used as internal control.The experiments were performed on protein extracts of Ctrl and RTD2 ASTROs after 60 days of differentiation.(I) Bar graphs show the quantification of EAAT2 [F (4, 10) = 1.286, p = 0.3389, ANOVA] relative to GAPDH levels.(J) Bar charts show the quantification of GFAP [F (4, 10) = 1.830, p = 0.1998, ANOVA] relative to GAPDH levels.Data are the mean fold induction with respect to Ctrl ASTROs and are presented as mean ± SEM of three independent biological replicates (n = 3).

FIGURE 4
FIGURE 4 Analyses of viability and cell death in RTD2 and Ctrl ASTROs.(A) Bar graph depicting cell viability using MTT assay [F (4, 34) = 1.764, p = 0.1590, ANOVA].Data are normalized to Ctrl ASTROs and presented as the mean ± SEM of three biological replicates (n = 3).(B) Confocal micrographs of astrocytes cultures following TUNEL assay (in green) and counterstaining of nuclei with Hoechst (in blue) (scale bar = 50 μm).(C) Bright field images of Ctrl and RTD2 iPSC-derived astrocytes (scale bar = 25 μm).(D) Bar graph depicting the quantification of TUNEL positive cells [F (4, 70) = 1.258, p = 0.2948, ANOVA] over the total number of cells.Data are presented as the mean ± SEM of three biological replicates (n = 3).The total number of analyzed cells is ~5,600 cells.

FIGURE 6
FIGURE 6Schematic drawing depicting Ctrl and RTD2 MNs and ASTROs.In particular, RTD2 MNs show aberrant neuronal differentiation: shorter neurites, while RTD2 ASTROs do not show cytoskeletal alterations nor changes in their survival rate compared to Ctrl ASTROs.

TABLE 2 Table
reporting the primer sequences used for the Quantitative Real-Time PCR experiments.