C9orf72 Hexanucleotide Expansions Are Associated with Altered Endoplasmic Reticulum Calcium Homeostasis and Stress Granule Formation in Induced Pluripotent Stem Cell‐Derived Neurons from Patients with Amyotrophic Lateral Sclerosis and Frontotemporal Dementia

Abstract An expanded hexanucleotide repeat in a noncoding region of the C9orf72 gene is a major cause of amyotrophic lateral sclerosis (ALS), accounting for up to 40% of familial cases and 7% of sporadic ALS in European populations. We have generated induced pluripotent stem cells (iPSCs) from fibroblasts of patients carrying C9orf72 hexanucleotide expansions, differentiated these to functional motor and cortical neurons, and performed an extensive phenotypic characterization. In C9orf72 iPSC‐derived motor neurons, decreased cell survival is correlated with dysfunction in Ca2+ homeostasis, reduced levels of the antiapoptotic protein Bcl‐2, increased endoplasmic reticulum (ER) stress, and reduced mitochondrial membrane potential. Furthermore, C9orf72 motor neurons, and also cortical neurons, show evidence of abnormal protein aggregation and stress granule formation. This study is an extensive characterization of iPSC‐derived motor neurons as cellular models of ALS carrying C9orf72 hexanucleotide repeats, which describes a novel pathogenic link between C9orf72 mutations, dysregulation of calcium signaling, and altered proteostasis and provides a potential pharmacological target for the treatment of ALS and the related neurodegenerative disease frontotemporal dementia. Stem Cells 2016;34:2063–2078


INTRODUCTION
A hexanucleotide (GGGGCC) expansion in the first intron of the C9orf72 gene accounts for approximately 40% of cases of familial amyotrophic lateral sclerosis (ALS), up to 7% of sporadic ALS, and approximately 20% of familial frontotemporal lobar degeneration, establishing a firm genetic link between ALS and frontotemporal dementia (FTD) [1][2][3]. The expansion is located in an intronic or promoter region upstream of the C9orf72 coding sequence, and the number of (GGGGCC) n hexanucleotide repeats ranges between 100 and 4,000 repeats in patients [1,2,4]. C9orf72-associated cases exhibit distinctive neuropathological changes characterized by abundant cytoplasmic and intranuclear inclusions immunopositive for ubiquitin and sequestosome-1/p62, suggesting disrupted protein degradation [5][6][7].
While the function of the C9orf72 gene and the pathogenic mechanisms of the hexanucleotide expansion are currently unknown, several hypotheses have been proposed. A toxic gain of function mechanism, mediated by the accumulation of (GGGGCC) n -rich RNA transcripts within nuclear foci observed in the frontal cortex and spinal cord of patients, might lead to sequestration of RNA binding proteins and disruption of the translation of diverse mRNAs [1] or increased nucleolar stress [8]. Alternatively, reduced transcription of C9orf72 could lead to neurodegeneration by interfering with the constitutive function of the protein [1,9]. Lastly, repeatassociated non-ATG (RAN) translation, occurring in the absence of an initiating ATG codon, over the GGGGCC repeat expansion has been shown to produce homopolymeric proteins prone to aggregation [10,11].
The generation of human motor neurons (MNs) in culture from induced pluripotent stem cells (iPSC), reprogrammed from skin fibroblasts of patients with neurodegenerative diseases such as ALS, offers a potentially powerful tool with which to study the key pathological processes in MN degeneration and for screening drugs of potential therapeutic benefit. Previous studies have demonstrated that RNA foci and RANtranslation products can be detected in iPSC-derived MNs from ALS/FTD patients with C9orf72 hexanucleotide expansions [12][13][14]. Evidence of defects in autophagy, sequestration of RNA-binding proteins by the expanded repeat, changes in gene transcription, and altered neuronal excitability suggest that these models can display disease-relevant phenotypes which can be corrected by targeting the expanded RNA with antisense oligonucleotides [13,14].
Calcium (Ca 21 ) dysregulation is believed to play an important role in the pathophysiology of ALS [15], and Ca 21 overload in the cytoplasm of neurons is a potential mechanism that may link excitotoxicity to neuronal death [16]. The endoplasmic reticulum (ER) is the largest intracellular Ca 21 store and high ER Ca 21 concentration plays an essential role in the activity of protein synthesis and processing. Disturbances in ER Ca 21 homeostasis have been linked to chronic activation of the ER stress response and downstream compensatory mechanisms to protect the cell, such as the unfolded protein response (UPR) and autophagy [17]. ER-stress induced cell death can proceed in both Ca 21 -dependent and Ca 21 -independent ways. The antiapoptotic proteins Bcl-2 and Bcl-XL are essential modulators of Ca 21 signaling in the cell, mediating the decrease and release of Ca 21 in the ER thereby protecting mitochondria from Ca 21 overload [18,19].
In order to explore whether altered Ca 21 homeostasis and ER stress are cellular phenotypes associated with C9orf72related ALS/FTD, we generated iPSC lines from skin fibroblasts from patients carrying C9orf72 hexanucleotide expansions and differentiated them to MNs and cortical neurons (CNs). Here we show that in C9orf72 iPSC-derived MNs loss of Ca 21 homeostasis is associated with a decrease in mitochondrial potential and increase in ER stress. Increased susceptibility to cell death was accompanied by reduced levels of the antia-poptotic Bcl-2 and Bcl-X L proteins and upregulation of the proapoptotic protein BAK. Furthermore, both C9orf72 MNs and CNs showed elevated levels of SQST-1/p62 and the formation of poly-A-binding protein (PABP) positive stress granules. Our results indicate disease-specific alterations in MNs and CNs of ALS/FTD patients carrying C9orf72 repeat expansions and reveal a novel pathogenic link between C9orf72, Ca 21 dysregulation, ER stress, and loss of proteostasis.

Ethics Statement
The human iPSC lines used for this study were derived from human skin biopsy fibroblasts, following signed informed consent, with approval from the South East Wales Research Ethics Committee (REC ref no. 12/WA/0186) and the South Central Berkshire Research Ethics Committee, U.K. (REC 10/H0505/71).

Generation and Culture of iPSC Lines
In this study, we used six iPSC-derived lines from four healthy controls: NHDF-1; OX3-9; OX1 clones 19 and 61; AH017 (clones 3 and 13) and six lines from 3 C9orf72 patients: C9-T2 (clones 6 and 7); C9-7245 (clones 1 and 3); C9-02 (clones 2 and 10). All iPSC lines were derived from skin biopsy fibroblasts in the James Martin Stem Cell Facility, University of Oxford, under standardized protocols to minimize any potential variation attributable to laboratory differences and/or handling. Fibroblasts tested negative for mycoplasma using MycoAlert (Lonza, United kingdom, www.lonza.com/) and the derived iPSC lines also tested negative using the same test.  [22]) were derived using the SeVdp(KOSM)302L Sendai virus system [23,24]. This contains the genes for all four reprogramming transcription factors (Klf4, Oct4, Sox2, c-Myc) expressed from a single transcript, packaged into a single Sendai virus, ensuring consistency of gene dosage ratio. Sendai virus is a negative strand RNA virus that is retained in the cytoplasm. As such, the likelihood of insertion of genetic material into the genome is negligible. This version of the system also contains a target for mir302; mir302 is expressed in pluripotent cells, but not in the originating fibroblasts, so the viral RNA will be detected and degraded once the cells have reprogrammed to pluripotency, ensuring complete removal of exogenous genetic material within a few passages. AH017- 13

Assessment of Genome Integrity and Tracking
Genome integrity was assessed by Illumina Human CytoSNP-12v2.1 beadchip array (300,000 markers) or Illumina HumanOmniExpress-24 (700,000 markers) and analyzed using Kar-yoStudio and GenomeStudio software (Illumina). SNP deviations in the iPSC lines were compared to the original pool of fibroblasts. This also confirmed the identity of the iPSC to the original fibroblasts. SNP datasets have been deposited in Gene Expression Omnibus (GEO) under the accession number GSE64582.

Sendai Clearance Assay
Clearance of Sendai virus from reprogrammed cell lines was confirmed by polymerase chain reaction (PCR). RNA was extracted using an RNeasy kit (Qiagen, Hilden, Germany, http://www1.qiagen.com) from iPSC, from fibroblasts (negative control), and from fibroblasts infected with Sendai reprogramming viruses 5 days previously (positive control). Reverse transcription was carried out using a RetroScript kit (Ambion, Austin, TX, http://www.ambion.com), using 2 mg template RNA in 20 mL reaction volume. 2 mL of 1:10 dilution of cDNA product was used in a 25 mL PCR reaction. For lines reprogrammed with SeVdp(KOSM)302L, quantitative reverse transcriptase PCR (qRT-PCR) was carried out on an Applied Biosystems StepOne Plus Real Time PCR machine, with Ste-pOne software, using Applied Biosystems 23SYBR green PCR mix1ROX and 60 degree Celsius anneal. Target gene transcript levels were compared to actin b control (actin b primers, Eurogentec) and subsequently to the positive control [28]. Sendai-specific primers were 5 0 -AGACCCTAAGAGGACGAAG ACAGA-3 0 and 5 0 -ACTCCCATGGCGTAACTCCATAG-3 0 . For lines reprogrammed with the Cytotune Sendai reprogramming system, PCR was carried out according to the manufacturers' instructions, with a primer pair for each of the four reprogramming viruses and also for the backbone, using actin b as a positive control for each line (actin b primers, Eurogentec), and products were visualized by running on a 1.5% agarose gel with ethidium bromide, and 2-log ladder or 100 bp ladder PluriTest RNA was extracted from iPSC using an RNeasy kit (Qiagen) for Illumina HT12v4 transcriptome array analysis. The data files were then uploaded to www.pluritest.org and scored for pluripotency, as previously described. The Illumina HT12v4 transcriptome array results have been deposited in the GEO under accession number GSE64583.

Differentiation of iPSCs to MNs
MNs were differentiated from iPSCs using three different previously published protocols, with several modifications [29,30]. In one of the differentiation protocols, the differentiation process was initiated by neural induction of embryoid bodies (EBs) and was followed by neural progenitor stages and neural maturation. After 4 days in neural induction medium, the EBs were plated onto Geltrex-coated plates, and caudalization of the neural progenitors was obtained by addition of retinoic acid (RA) to the neural differentiation medium at a concentration of 0.1 lM. After 10 days, the colonies displaying neural rosettes structures were isolated and expanded in suspension in the form of neurospheres in the presence of the ventralizing growth factor, sonic hedgehog at a concentration of 100 ng/mL. The protocol was modified at this step to enhance cell survival by addition of Rock inhibitor Y27632, which prevents  [31,32]. After 2 weeks of culture as neurospheres, the cells were plated on Geltrex-coated coverslips, and terminal differentiation was induced by addition of brain derived neurotrophic factor (BDNF 10 ng/mL), glial-derived neurotrophic factor (GDNF, 10 ng/mL), insulin-like growth factor (IGF-1, 10 ng/mL), cyclic AMP (cAMP), and ascorbic acid 0.5lM, Sigma ascorbic acid. The differentiating MNs were cultured for another 3-4 weeks before functional assays were performed. Using the second differentiation protocol [30], the iPSCs were cultured as monolayers, and neural induction was started by addition of compound c (1 lM) Tocris Bioscience, Bristol UK, https://www.tocris.com/ in DMEM/F12 and Neurobasal medium, supplemented with N2, B27, Heparin 0.2%. After 3 days, the medium was supplemented with RA (1 mM) and after another 3 days, smoothened agoinst (SAG) and purmorphamine (1 mM) were added and compound c removed. After 9 days in culture, the neural precursors were replated at a low density for differentiation and the medium was supplemented with the growth factors BDNF, GDNF and IGF-1 (10 ng/mL), ascorbic acid and cAMP. After 20 days, the MNs were replated for experiments and allowed to mature for another 3-4 weeks before functional experiments.

Differentiation of iPSCs to CNs
The differentiation of iPSCs to CNs was performed following a previously established protocol [34]. Briefly, the iPSCs were grown on Matrigel as monolayers, and neural differentiation was induced by changing the medium to basic neural medium DMEM/F12: Neurobasal 1:1, 13 N2, 13 B27, insulin 5 mg/mL (Sigma), nonessential amino acids 100 mM (Life Technologies), 2-mercaptoethanol 100 mM, antibiotic-antimycotic (Life Technologies) and supplementing it with 500 ng/mL Noggin (PeproTech, Rocky Hill, NJ, http://www.peprotech.com), and 10 mM SB431542 (Enzo Life Sciences). Medium was changed every 24 hours until neural rosettes could be observed (12-16 days). The neuroepithelial layer was lifted using Dispase (Life Technologies), and the medium was supplemented with 20 ng/mL Fibroblast growth factor 2 (FGF2) (PeproTech) for 4-6 days. The neural precursors were split using Accutase (Life Technologies) at a low density. The medium was then supplemented with 10 ng/mL BDNF and 10 ng/mL GDNF, while FGF2 was withdrawn. The neurons were allowed to extend processes and to mature for 60-70 days before functional assays.

Immunoblotting
Cells were washed with PBS and then scraped in PBS. Cells were lysed in RIPA buffer (0.5% sodium deoxycholate, 0.1% SDS, 1% Nonidet P40 and 1% protease inhibitor cocktail (Sigma)), sonicated, and incubated on ice for 30 minutes. After centrifugation at 5,000g for 10 minutes, the supernatant was retained, and protein concentration was quantified using the BCA assay according to the manufacturer's instructions (Sigma). Protein was boiled in Laemmli buffer for 5 minutes at 958C, and 10 lg of protein were loaded on gels. For Western blot analysis, lysates were resolved on a SDS-PAGE (10% or 12% Tris-glycine gel) using a Novex X-Cell electrophoresis system (Life Technologies) under constant current (125 mA for conjugated anti-rabbit secondary antibodies were added for 1 hour at room temperature. The membrane was washed three times for 10 minutes in 0.1% Tween/PBS and the signal was detected by ECL (Amersham Pharmacia Biotech).

Propidium Iodide Staining
Propidium iodide (PI) staining was performed according to the manufacturer's instructions (Sigma). Briefly, the neurons were incubated with 1 mg/mL PI stock solution and RNase A for 10 minutes at 378C. The solution was then washed with PBS, and cells were fixed in 4% PFA. Immunocytochemistry was then performed as described above.

Mitochondrial Staining
Cultures of differentiated MNs were incubated with 100 nM of the potentiometric dye MitoTracker Red CMXRos (Thermo Fisher) for 45 minutes at 378C. The neurons were then washed with PBS, fixed in 4% PFA, and immunostained for Islet1 and SMI-32 according to the immunocytochemistry protocol above.

ER Calcium Imaging
Mature MNs were loaded with 10 lM Fluo-4 AM dissolved in 0.2% dimethyl sulfoxide (Sigma-Aldrich, St. Louis, http://www. sigmaaldrich.com) with 0.04% Pluronic acid in Hank's Balanced Salts Solution (HBSS) supplemented with CaCl 2 and MgCl 2 . To image Ca 21 signals, we used wide-field fluorescence microscopy using a Nikon TE 2000 inverted microscope equipped with a 340 oil-immersion objective and EMCCD camera (Evolve 128, Photometrics) for imaging fluorescence emission (510-600 nm) at a frame rate of 300 s 21 . Fluorescence was imaged at 258C from a 12 3 12 mm (128 3 128 pixel) region within 18 mm coverslips, on which MNs were cultured. ER Ca 21 concentration levels were assessed by fluorescence measurements upon application of 1 mM thapsigargin (Life Technologies) in HBSS depleted of CaCl 2 and MgCl 2 . Fluorescence measurements are expressed as a ratio (DF/F 0 ) of the mean change in fluorescence (DF) at a pixel relative to the resting fluorescence at that pixel before stimulation (F 0 ). Image analysis was performed using MetaMorph and ImageJ software.

RNA Fluorescence In Situ Hybridization (FISH)
Cells grown on 12 mm RNase-free coverslips in 24-well plates were fixed using 4% PFA for 20 minutes, washed in PBS, and permeabilized using 0.2% Triton X-100 (Sigma) in PBS for 30 minutes. Cells were then dehydrated in a graded series of alcohols, air dried, and rehydrated in PBS, briefly washed in 23 standard sodium citrate (SSC) and then incubated in pre-hybridization solution and hybridization solution as previously described using Cy3-conjugated 2 0 O-methyl RNA probes (Integrated DNA Technologies) [35]. Probe sequences were (CCCCGG) 4 or (CAGG) 6 . Optional treatments using RNase (0.1 mg/mL; Life Technologies) or DNase (150 U/mL, Qiagen) were performed at 378C for 1 hour following permeabilization. Following hybridization washes, staining with Tuj1 (1:2000, Covance) and Islet1 (1:200, DSHB) were performed as described above.
Southern Blotting 5 mg of DNA were digested with 2 mL each of FastDigest DdeI and FastDigest AluI (Thermo Scientific) for 2 hours at 378C. Digested DNA underwent electrophoresis in 0.8% agarose gel in 31 Tris-acetate-EDTA (TAE) and was transferred to a positively charged nylon membrane (Roche, Basel, Switzerland, http://www.roche-applied-science.com) by capillary blotting. Roche digoxigenin (DIG)-labeled DNA molecular weight marker II was used as a ladder. DNA was crosslinked to the membrane by UV irradiation. Prehybridization was carried out in 20 mL DIG Easy Hyb solution (Roche) at 538C for 2 hours. 10 ng DIG-labeled (GGGGCC) 5 probe (custom synthesized by Integrated DNA Technologies) was incubated at 958C for 5 minutes, snap cooled on ice for 30 seconds, and immediately added to 7 mL DIG Easy Hyb solution prewarmed to 608C. Hybridization took place overnight at 538C. Stringency washes were performed at 648C as follows: 23 SSC, 0.1% sodium dodecyl sulphate (SDS) for 5 minutes, 23 SSC, 0.1% SDS for 10 minutes, 13 SSC, 0.1% SDS for 15 minutes three times, 0.53 SSC, 0.1% SDS for 15 minutes twice.
The membrane was prepared for detection using the DIG Wash and Block buffer set (Roche). The membrane was washed with wash buffer for 2 minutes then incubated with blocking solution for 1-2 hours. Anti-DIG-AP Fab fragments (Roche) were incubated with the membrane at room temperature for 20-25 minutes at 1 in 10,000 dilution in blocking solution. Three washes were carried out for 5, 15, and 15 minutes followed by a 5-minute equilibration step using the supplied Detection buffer. CDP-Star (Roche) was added to the membrane for at least 5 minutes, and signal was detected by exposure to x-ray film. Repeat number was estimated by interpolating a plot of log 10 of band size in base pairs versus distance travelled on the blot, with subtraction of 199 bp from the estimated base-pair number to account for the wildtype allele as performed by Beck et al. [36].

Statistical Analyses
All quantitative data were analyzed using GraphPad Prism. Groups of three or more samples were subject to one-way

Dataset Deposit
The Illumina HT12v4 transcriptome array results have been deposited in the GEO under accession number GSE64583. SNP datasets have been deposited in GEO under the accession number GSE64582. These are both contained in the Superseries GSE64584.

Reprogramming of Patient Fibroblasts to Generate iPSCs
Two lines from each of three patients (C9-T2, C9-7245, C9-02) were generated, along with lines from control donors. Figure 1 shows the representative characterization data for one of the patient iPSC lines generated for this study, iPSC C9-T2-7 (data for the other previously unpublished iPSC lines, including control lines, is given in Supporting Information Fig. S1). SNP analysis confirmed the identity of the iPSC in relation to their parental fibroblasts and that they had not acquired large-scale copy number variations, although a small number of small-scale indels (well below the level that would be detected by G-banding) were noted, particularly in the C9-7245 lines ( Fig. 1A; Supporting Information Fig. S1A). SNP datasets for these lines have been deposited in GEO under the accession number GSE64582.iPSC lines were positive for pluripotency markers (TRA-1-60 and Nanog, Fig. 1B; also Supporting Information Fig. S1B) and showed human pluripotent stem cell-like morphology (cell-cellcontact-dependent clusters, with high nucleus to cytoplasm ratio, Fig. 1C; Supporting Information Fig. S1C). PluriTest is a quantitative assessment of pluripotency based on transcriptome data from iPSCs in comparison to a large reference set of genomewide profiles from multiple cell and tissue types. It measures a "pluripotency score" (expression of pluripotency genes, with high scores correlating with high pluripotency, y-axis) and a "novelty" score (expression of nonpluripotent stem cell genes, therefore this score inversely correlates with pluripotency, xaxis). All the patient iPSC lines were deemed to be fully reprogrammed and pluripotent by this analysis (Fig. 1D). The Illumina HT12v4 transcriptome array results have been deposited in the GEO under accession number GSE64583. Finally, all banked iPSC line stocks had undetectable levels of residual Sendai virus RNA by qRT-PCR ( Fig. 1E; Supporting Information Fig. S1D).

C9orf72 iPSCs Differentiate to Functional MNs
In vitro cell differentiation protocols recapitulate signaling events that occur during embryogenesis. The differentiation of iPSCs to spinal MNs is a three-stage process that involves neuralization, caudalization, and ventralization. The iPSC lines were differentiated to MNs according to three previously published protocols, with some modifications [29,30,33] (Supporting Information Fig. S2A-S2C). The fully differentiated MNs were identified by immunostaining with the MN marker HB9 and neuronal marker b-III tubulin (Tuj1), which indicated successful differentiation of iPSCs to MNs (Supporting Information Fig S2D). The iPSCderived MNs showed expression of synaptophysin, a synaptic vesicle protein essential for the formation of synapses, as well as the expression of choline acetyltransferase, which demonstrates an advanced degree of cholinergic neuron maturation (Supporting Information Fig S2D, S2E).
To assess the functional maturation of the iPSC-derived MNs, we performed whole-cell patch clamp recordings to determine their electrophysiological properties. Trains of action potentials were elicited in both control and patient MNs in response to depolarizing current injection in currentclamp recordings (Fig. 2D, 2E). Inward and outward ionic currents were recorded by voltage clamp in both controls and patients (Fig. 2F, 2G). Ca 21 dynamics were investigated using the calcium-sensitive dye Fluo-4 AM. In the absence of external stimulation, spontaneous calcium transients were detectable in the cell bodies and processes of the neurons (Fig. 2H). The neurons were then stimulated with kainate (KA), to activate ionotropic receptors, and with KCl to depolarize the membrane and activate voltage-gated Ca 21 channels. Exposure to KCl induced Ca 21 transients in Islet11 MNs (Fig. 2I). Similarly, exposure to KA induced characteristic increases in intracellular Ca 21 levels in 80% of the cells (Fig. 2J).

RNA Foci and RAN Dipeptides are Present in C9orf72 MNs
In the MNs derived from C9orf72 patients, we identified a significant increase in intranuclear GGGGCC-repeat containing RNA foci in all lines (Fig. 3A, 3B). The specificity of the probe was confirmed by treatment with RNase, which reduced the number of foci substantially (Supporting Information Fig. S4A), and with DNase I, after which the number of foci was maintained. To investigate evidence for the RAN translation hypothesis, we performed a dot blot analysis using anti-GA, GP, GR, and PR antibodies, and we could detect traces of all the dipeptides in the MNs from C9orf72 patients (Fig. 3C). To assess the total content of ER Ca 21 stores, 10 lM thapsigargin (TG) was applied to the MNs, and the peak increase in fluorescence intensity resulting from ER Ca 21 depletion into the cytoplasm was measured (Fig. 4A, 4B). TG acts as a potent inhibitor of the SERCA pump on the ER membrane, blocking the reuptake of Ca 21 ions from the cytoplasm. In our cultures, we detected significantly higher levels of ER Ca 21 content in the C9orf72 iPSC-derived MNs, which showed up to twofold increase in cytoplasmic fluorescence intensity compared to the control MNs (p < .05, one-way ANOVA) (Fig. 4C). Abnormal elevations in ER Ca 21 levels can cause ER stress and can activate the UPR. We investigated levels of proteins associated with the UPR and ER stress responses, and we found significantly elevated levels of GRP78/BiP in C9-7245-1 and C9-7245-3 iPSC-derived MNs compared to AH017-13 and OX1-61 MNs (p < .05, one-way ANOVA) (Fig. 4D, 4E). We did not detect differences in the splicing of XBP-1 between patients and controls (Supporting Information Fig. S5A).

Elevated ER Ca
The Bcl-2 family of proteins are key components involved in the modulation of calcium homeostasis in the ER, controlling the release of Ca 21 ions through the IP 3 R and through channels formed by its own dimerization [19]. The balance between proapoptotic and antiapoptotic members of the Bcl-2 family is essential for controlling ER stress-induced apoptosis and mitochondrial membrane permeability [17]. In C9orf72 MNs, we detected an imbalance between proapoptotic and antiapoptotic members of the Bcl-2 family, which showed up to 50% downregulation of antiapoptotic Bcl-2 in C9-7245-1, C9-7245-3, and  (Fig. 4F, 4G) and 75% reduction of Bcl-X L (p < .05, one-way ANOVA) (Fig. 4H). Significant upregulation of proapoptotic BAK was detected in the MNs with large expansions, which showed up to 2.5-fold increase in expression (p < .05, one-way ANOVA) (Fig. 4I). A significant upregulation of the proapoptotic BIM was also detected in some of the MN lines (Supporting Information Fig. S5B).
The Bcl-2 family of proteins is distributed throughout the ER and mitochondrial membranes, where they play critical roles in regulating intracellular Ca 21 homeostasis and exchange of Ca 21 between the two organelles. Following dysfunction of these proteins and Ca 21 overload, the mitochondria undergo typical morphological changes, such as swelling and perturbation of the outer membrane, accompanied by the release of apoptotic factors [37]. To assess these alterations, we used the potentiometric dye MitoTracker Red CMXRos, accumulation of which is dependent on mitochondrial potential (Fig. 4J). We found up to 30% reduction in mitochondrial membrane potential in C9 MNs compared to controls (p < .01, one-way ANOVA) (Fig. 4K).
In line with these findings, analysis of morphological alterations of the mitochondria using EM also revealed swollen mitochondria in the neurons from C9orf72 patients compared to the healthy control (Fig. 4L). Alterations of the mitochondrial morphology and abnormalities in the cristae structures were observed in up to 88% of C9orf72 neurons compared to only 2% normal healthy neurons (Fig. 4M).
These mitochondrial alterations are consistent with the release of apoptotic factors, such as cytochrome c, which was found to be elevated in all patient lines, specifically in the C9-7245-1 compared to healthy controls (p < .05, one-way ANOVA) (Fig. 4N, 4O). These results indicate an association between C9orf72 expansions and elevated ER Ca 21 concentrations and stress, as well as mitochondrial alterations which may be the drivers for downstream apoptosis.

C9orf72 iPSC-Derived MNs and CNs have Increased Susceptibility to Apoptosis
To investigate whether MNs and CNs carrying pathogenic hexanucleotide expansions in C9orf72 develop an apoptotic  phenotype under normal culture conditions, we assessed markers of cellular stress and cell death. Caspases are present in the cell as inactive proenzymes which can be activated by proteolytic cleavage, the cleavage of caspase-3 being an essential process that initiates the apoptotic disassembly of the cell. In neurons from C9orf72 patients we found significantly increased immunostaining for cleaved caspase-3 compared to the CNs from OX1-19, OX3-9, and AH017-13 (Fig. 5A). We detected between 8.1% and 10.5% of neurons positive for cleaved caspase-3 in the C9orf72 patient lines, compared to a maximum of 2.9% in OX1-19 and 1.9% in OX3-9 (p < .05, one-way ANOVA) (Fig. 5B). In line with this finding, the ALS/FTD-derived MNs also showed significantly increased staining with PI in up to 32% of C9orf72 MNs compared to healthy MNs, indicating reduced survival compared to control MNs (p < .01, one-way ANOVA) (Fig. 5C).
In CN cultures, we found a significantly high frequency of cleaved caspase-3 in the C902-2 and C902-10 lines, with up to 35% of neurons positive for apoptotic markers compared to controls (p < .05, one-way ANOVA) (Fig. 5D, 5E). These results also correlated with high PI staining, which was found in up to 40% of the neurons analyzed in the C902 patient lines (p < .001, Student's t test) (Fig. 5F). Taken together, these results demonstrate that the presence of the C9orf72 hexanucleotide expansions activates caspase-3 dependent apoptosis in both MNs and CNs of patients.

Stress Granules and Protein Aggregates are Present in
C9orf72 iPSC-Derived MNs and CNs SQST1/p62, a marker of the UPR and autophagy flux is found in neuronal aggregates in autopsy samples from C9orf72 ALS/ FTD cases. We found significantly elevated levels of SQST1/ p62 in MNs derived from each C9orf72 patient, and the patient with the highest number of repeats showed up to twofold higher expression of SQST1/p62 compared to control MNs at baseline (p < .05, one-way ANOVA) (Fig. 6A-6C). No differences were detected in the autophagic marker LC3-II/ LC3-I in MNs (Supporting Information Fig. S5C, S5D). In CNs carrying the C9orf72 expansions we also found approximately 40% neurons positive for p62 aggregates (Fig. 6D-6F). In contrast to MNs, these results correlated with an increase in levels of LC3-II in C902-2 and C902-10 compared to controls OX3-9 and AH017-13 (p < .05, one-way ANOVA) (Fig. 6G).
Stress granules are formed when RNA binding proteins aggregate through their glycine rich domains. These granules function to sequester, silence, or degrade specific RNA transcripts as part of a mechanism that adapts patterns of local RNA translation to facilitate the response to cellular stress [38]. PABP is a major component of stress granules and serves as a robust marker of the activation of the stress response. In our neuronal culture, we found up to fivefold increase in the frequency of neurons with PABP-positive granules under resting conditions in all five C9orf72 patient MN lines compared to controls (p < .05, one-way ANOVA) (Fig. 6H, 6I). The C9orf72 MNs were also immunopositive for the early stress granule marker, TIA-1 (Fig. 6H). PABP positive stress granules were detected in up to 40% of C9orf72 CNs compared to a maximum of 20% control CNs (Fig. 6J, 6K).

DISCUSSION
The inability to directly study neurons in patients suffering from neurodegenerative disorders has been a significant impediment to the understanding of basic mechanisms, particularly those occurring early in the disease process.
Recent developments in stem cell technology have substantially expanded the range of cellular models available in ALS by allowing the direct observation of pathomechanisms in MNs derived from iPSC obtained from fibroblasts harboring the genetic background conferring disease susceptibility.
The mechanisms underlying C9orf72 toxicity in MNs remain to be elucidated, but possibilities include haploinsufficiency and toxic gain-of-function, mediated either by RNA foci or toxic aggregation-prone polypeptides resulting from RAN translation [1,10,39,40]. These potential mechanisms are not mutually exclusive and may each play a role. In this study, we identified several converging pathways that are altered in MNs and CNs from C9orf72 patients.
RNA foci have been consistently detected in the CNs of patients with hexanucleotide repeats [1,14], and we detected discrete RNA foci in our MN cultures from all C9orf72 patient lines. Since the presence and number of RNA foci did not correlate with other phenotypes, the relationship of RNA foci to the pathology we observe in this cellular model requires further investigation. A dot blot analysis using high levels of proteins revealed the presence of the GA, GP, GR, and PR dipeptides in the patient lines, but given their low abundance and difficulty of detection, it is challenging to analyze their contribution to neuronal degeneration in our cellular model. A recent study has shown the formation of stress granules in neuronal cells transfected with poly-GA, poly-GP, and poly-PA, suggesting that RAN translation may be associated with stress granule formation and neurodegeneration [41]. In our cultures of C9orf72 MNs and CNs, we detected the frequent presence of stress granules under baseline conditions. Stress granules are conserved cytoplasmic aggregates of nontranslating messenger ribonucleoprotein complexes, and they are dynamic structures which dissociate when conditions return to normal or they are cleared by autophagy [42]. The accumulation of PABP 1 stress granules suggests impairments in autophagy and proteasomal degradation, and the convergence of these two dysfunctional pathways may partially explain the vulnerability of C9orf72 MNs and CNs to degeneration.
Calcium is known to exert a complex regulatory role in the ER-mitochondrial crosstalk, playing an essential part in cell death signaling [43]. Bcl-2 together with Bcl-X L are located on the ER membrane where they lower the [Ca 21 ] ER by forming Ca 21 permeable leakage pores [44]. Reduced levels of Bcl-2 and Bcl-X L contribute to nonphysiological accumulation of Ca 21 ions into the ER due to reduced leakage, and we confirmed elevated levels of ER Ca 21 stores in the C9orf72 MNs. We have recently shown that TDP-43 pathogenic mutations in mouse MNs and in a human cellular model also interfere with ER Ca 21 , but with the opposite effect, showing reduced ER Ca 21 levels and decreased ER Ca 21 release [45].
The tight connection between the ER Ca 21 stores and mitochondria allows cells to respond promptly and efficiently to various Ca 21 stimuli. However, under pathological conditions, the high levels of Ca 21 from the ER will overload the mitochondria and consequently induce mitochondrial swelling, rupture of the outer membrane, and release of apoptotic factors from the mitochondria [37]. The antiapoptotic protein Bcl-2 is located on the mitochondrial membrane where it prevents proapoptotic BAK/BAX to oligomerize and

STEM CELLS
to form channels that allow release of the apoptotic factor cytochrome c into the cytosol. Low levels of Bcl-2 and increased levels of BAK promote the release of cytochrome c from the mitochondria triggering the activation of caspases that regulate the induction and execution of apoptosis [46]. Consistent with this, we detected high levels of cleaved Disruption of ER Ca 21 homeostasis also plays a direct role in the induction of ER stress by interfering with the function of Ca 21 -sensitive ER chaperones [47]. Constitutive activation of ER stress in C9orf72 MNs, which interestingly appears to be greater in cells with a higher number of hexanucleotide repeats, may contribute to the induction of ER-stress associated apoptosis. This is consistent with a previous study of C9orf72-patient iPSC-derived MNs which found upregulation of transcripts involved in UPR activation and ER stress [48]. While all MNs carrying repeats showed a survival deficit and an active apoptotic pathway, only the patient-derived MNs with large hexanucleotide expansions showed ER stress under baseline conditions. These results suggest that ER stress is activated by high numbers of pathogenic repeats but is not the exclusive pathway in the induction of neuronal death. In accordance with our observations, a similar study using SOD1 1/A4V iPSCderived MNs found elevated ER stress in patient MNs, but treatment with salubrinal or knockdown of XBP-1 did not rescue the survival deficit, suggesting that ER stress is only one of many components that may be contributing to MN death [48]. Selective neuronal vulnerability may involve specific combinations of mutually reinforcing stressors and altered pathways that ultimately converge on neuronal death [49]. Consequently, the gradual increase of stress in an age-dependent fashion in vulnerable MNs may underlie the progression of neurodegeneration in ALS. Taken together, our results highlight the importance of ER Ca 21 regulation in different in vivo and in vitro models and support the hypothesis that MN degeneration in ALS is mediated by cell-type specific impairments in intracellular Ca 21 handling [50][51][52].
MNs and CNs from C9orf72 patients also showed high levels of SQST1/p62, which is an adaptor molecule for selective autophagosomal degradation of ubiquitinated targets and its accumulation reflects the inefficiency of protein aggregate removal by the autophagic machinery [53]. SQST1/p62 is a common component of protein aggregates found in neurodegenerative diseases, such as Lewy bodies in Parkinson's disease, neurofibrillary tangles in Alzheimer's disease, and in huntingtin aggregates [54][55][56]. Elevated p62 levels have previously been reported in CNs derived from iPSCs of FTD patients carrying hexanucleotide expansions in C9orf72, and it is a pathological marker in post-mortem analysis from C9orf72 carriers [5,7]. We detected high SQST1/p62 levels in MNs under basal conditions without changes in LC3-II/LC3-I turnover, indicating a reduced autophagic flux in C9orf72 MNs compared to healthy controls. Interestingly, elevated SQST1/p62 levels were correlated with increased LC3-II in CNs from an ALS/FTD patient, suggesting a potentially distinct accumulation of autophagosomes in these cells and impairment in the autophagic process, which is specific to the CNs and is not recapitulated in MNs.

CONCLUSION
Our study identified several cellular phenotypes that associate with the repeat expansions in C9orf72 in MNs and CNs and molecular changes that may account for their vulnerability in pathological conditions. We have identified an apoptotic pathway that is activated in the C9orf72 neurons, which involves ER calcium dysregulation and stress, followed by mitochondrial alterations, the release of apoptotic factors from the mitochondria, and the formation of stress granules. It is unclear where this chain of phenotypes is initiated, whether the presence of toxic RAN dipeptides induces ER stress and dysregulation of calcium, or whether the presence of the RNA foci leads to the recruitment of essential Ca 21 regulators transcripts. Further work is currently underway to analyze the molecular mechanism linking the hexanucleotide repeats in C9orf72 and the neuronal phenotypes. In this study, a high number of repeat expansions (>1,000) correlated with a more severe phenotype in our patient MN cultures, but a more extensive study is required to establish whether there is a correlation between the number of repeats and severity of the cellular phenotypes. Although our observations suggest that the cellular phenotypes observed are robust enough to be used to as a tool to screen potential therapies, further work is required to establish the exact mechanistic relationship between the pathological pathways activated in C9orf72 MNs, CNs, and neurodegeneration.