Mutant C9orf72 human iPSC‐derived astrocytes cause non‐cell autonomous motor neuron pathophysiology

Abstract Mutations in C9orf72 are the most common genetic cause of amyotrophic lateral sclerosis (ALS). Accumulating evidence implicates astrocytes as important non‐cell autonomous contributors to ALS pathogenesis, although the potential deleterious effects of astrocytes on the function of motor neurons remains to be determined in a completely humanized model of C9orf72‐mediated ALS. Here, we use a human iPSC‐based model to study the cell autonomous and non‐autonomous consequences of mutant C9orf72 expression by astrocytes. We show that mutant astrocytes both recapitulate key aspects of C9orf72‐related ALS pathology and, upon co‐culture, cause motor neurons to undergo a progressive loss of action potential output due to decreases in the magnitude of voltage‐activated Na+ and K+ currents. Importantly, CRISPR/Cas‐9 mediated excision of the C9orf72 repeat expansion reverses these phenotypes, confirming that the C9orf72 mutation is responsible for both cell‐autonomous astrocyte pathology and non‐cell autonomous motor neuron pathophysiology.

The intronic hexanucleotide repeat expansion GGGGCC (G 4 C 2 ) in the chromosome 9 open reading frame 72 (C9orf72) gene is the most common genetic cause of ALS, accounting for~40% of fALS and~10% of sALS (DeJesus-Hernandez et al., 2011;Renton et al., 2011). In contrast to extensive studies focused on MN biology and pathology, the consequences of G 4 C 2 expansion at physiological levels on astrocytes is understudied (Meyer et al., 2014). Moreover, the functional interaction between MNs and astrocytes is yet to be investigated in a completely humanized model of C9orf72-mediated ALS.
We and others have previously shown that ALS patient-derived iPSC lines recapitulate key aspects of ALS pathology and MN dysfunction (Bilican et al., 2012;Devlin et al., 2015;Donnelly et al., 2013;Serio et al., 2013;Shi et al., 2018). An important advance in human iPSC-based disease modeling is the use of paired isogenic control lines which help establish causality between a given mutation and phenotypes (Sandoe & Eggan, 2013;Selvaraj, Livesey, & Chandran, 2017;Wang et al., 2017). Using CRISPR/Cas9-mediated genome editing to selectively excise G 4 C 2 repeats we have recently shown selective human MN vulnerability to AMPA receptor mediated excitotoxicity that is mutation dependent (Selvaraj et al., 2018). In this study, we have utilized this system to further explore the cell-autonomous and non-cell autonomous consequences of C9orf72 mutation on iPSCderived astrocytes and MNs. We report C9orf72-dependent cell autonomous astrocyte pathology and astrocyte mediated loss of MN function independent of overt effects on MN viability. Furthermore, we suggest possible molecular pathways, highlighted from RNA-Seq data, which may underlie loss of MN function.
C9ORF72 patient iPSCs were transfected with these vectors to induce double strand break in the DNA sequence at a precise locus resulting in the deletion of G 4 C 2 repeats. Individual iPSC clones were screened for deletion of G 4 C 2 using the repeat-primed PCR. One positive clone for each isogenic control line C9-Δ1, C9-Δ2, and C9-Δ3 was selected, and Sanger sequencing for the C9orf72 G 4 C 2 locus in these clone demonstrated complete deletion of G 4 C 2 repeats in the mutant allele and one remaining G 4 C 2 in the wild-type allele (Selvaraj et al., 2018).

| Generation of MNs from iPSCs
Differentiation of iPSCs into a neuronal and MN lineage was performed using minor modifications of previously established and validated protocols (Amoroso et al., 2013;Bilican et al., 2012;Devlin et al., 2015). The iPSCs were neutralized to neuroectoderm using dual SMAD inhibition in Phase I medium for 4-10 days. Neurospheres were patterned to a caudal, spinal cord identity in Phase II medium for 4-10 days. Caudalized neural stem cells (NSCs) were next ventralized in Phase III medium for 4-10 days, and then cultured in Phase III-FGF for another 4-14 days. Caudalized and ventralized NSCs were transitioned to MN maturation medium for 2-6 weeks. These MN spheres were dissociated into single cells which were plated onto monolayers of astrocytes for co-culture as described below. 0.5% N-2 supplement, 2 ng/ml Heparin, 10 ng/ml brain-derived neurotrophic factor (BDNF; R&D systems), 10 ng/ml glial cell line-derived neurotrophic factor (GDNF; R&D systems), 10 μM forskolin (R&D systems), 0.1 μM retinoic acid, and 0.1 μM purmorphamine. At least 4 iPSC differentiations were performed for each experiment.

| Generation of cortical neuron cultures from human stem cells
A complete and systematic description of the derivation of human cortical neurons from human stem cells can be found in Bilican et al. (Bilican et al., 2014). Briefly, human cortical neurons were differentiated from anterior neural precursors, which were derived from the H9 human embryonic stem cell line (WiCell), obtained under ethical/IRB approval of the University of Edinburgh. Experiments were carried out on human cortical neurons that had been differentiated and maintained in culture for at least 30 days in vitro (DIV). At these time points, around 70% of cells were neuronal (β3-tubulin+), with little contamination from neural precursor cells (nestin+), astrocytes (GFAP+) or GABAergic (GAD65/67+) interneurons (Bilican et al., 2014;Livesey, Magnani, Hardingham, Chandran, & Wyllie, 2015). Neurons exhibited markers (VGLUT1+) consistent with an excitatory identity and also exhibited properties of neurones of the upper and lower layers of the cortex (Bilican et al., 2014;Livesey et al., 2015).

| Immunofluorescence
Cells were fixed in 4% (wt/vol) paraformaldehyde for 10 min, permeabilized with 0.2% Triton X-100 for 5 min and blocked in 3% (vol/vol) goat serum (Dako) or donkey serum (Sigma) for 45 min. They were then incubated in primary antibodies for 45 min followed by secondary antibodies for 30 min (Alexa Fluor dyes, 1:1000, Invitrogen). All antibodies were diluted in the blocking buffer. Nuclei were counterstained with DAPI (Sigma) for 5 min and coverslips were mounted on slides with FluorSave (Merck). All procedures were performed at room temperature. Primary antibodies used in this study were Vimentin Fluorescent imaging was performed on fields of view containing uniform DAPI staining using either an Axio Observer.Z1 (Zeiss) epifluorescence microscope or an LSM710 confocal microscope (Carl Zeiss). Images were processed and blindly analyzed by using the ImageJ64 (v 1.47) software.
Amplite™ Fluorimetric Glutamic Acid Assay Kit (ATT Bioquest) was used to determine the residual concentrations of L-glutamic acid in supernatants following the manufacturer's instructions. The glutamate uptake was calculated by subtracting the remaining concentration from 100 μM. The cell number in each well was determined by using a CyQUANT ® NF Cell Proliferation Assay Kit (Life Technologies, C35006) following the manufacturer's instructions. The glutamate uptake was normalized to the cell number and presented as uptake concentration per 1,000 cells.

| qRT-PCR
Total RNA was isolated from 2-week old differentiated astrocytes using an RNeasy Mini Kit (Qiagen) following the manufacturer's instructions. Five hundred nanograms of RNA was reverse transcribed to complementary DNA (cDNA) using a DyNAmocDNA Synthesis Kit (Thermo Scientific) following manufacturers' instructions. RT-PCR reactions were performed in triplicate using a DyNAmo™ ColorFlash SYBR ® Green qPCR Kit (Thermo Scientific, F-416) following the manufacturers' instructions, and a C1000™ Thermal Cycler with a CFX96 Real-time System (Bio-Rad) was used to conduct the cycling.

| Western blot
2-week old differentiated astrocytes were lysed in cold radioimmunoprecipitation assay (RIPA) buffer (50 mM Trizma ® base, 150 mM NaCl, 1% TritonX-100, 0.5% sodium deoxycholate, 0.1% SDS, and 2 mM EDTA, all from Sigma) supplemented with 1× protease inhibitor (Roche) and 1× phosphatase inhibitor (Roche), and incubated for 30 min on ice. Lysate was then centrifuged at 13000 rpm for 30 min at 4 C, and the resulting supernatant was collected as the RIPA-soluble fraction. RIPA-insoluble pellets were further washed with RIPA buffer once and then dissolved in Urea buffer (7 M Urea, 2 M Thiourea, 4% CHAPS, and 30 mM Trizma ® base, all from Sigma) supplemented with 1× protease inhibitor and 1× phosphatase inhibitor of a volume in proportion to the soluble fraction.
Sonication was performed to further dissolve the protein.

| RNA-fish
Astrocytes on glass coverslips were fixed with 4% paraformaldehyde (Agar Scientific) for 15 min at room temperature followed by permeabilization in 70% ethanol at 4 C overnight. Cells were then rehydrated in 50% formamide (Sigma)/2x SSC (Sigma) for 10 min at room temperature and blocked in hybridization buffer (50% Formamide (Sigma), 2×SSC (Sigma), 10% Dextran Sulfate (Millipore), 1 mg/ml Yeast tRNA (Invitrogen) and 1 mg/ml Salmon Sperm DNA (Invitrogen)) for 30 min at 45 C. 50 ng of an Alexa Fluor ® 546-conjugated (GGCCCC) 4 probe (IDT) diluted in the hybridization buffer was applied on cells for 2 hours at 45 C in a humidified chamber. After the hybridization, cells were washed twice with 50% formamide/2× SSC for 30 min at 45 C and then once with 2× SSC for 30 min at room temperature. After another three washes with PBS at room temperature, immunofluorescence imaging was performed as described previously.
As controls, cells were treated with either 3 U/ml DNase (Life Technologies) or 100 μg/ml RNase (Sigma) diluted in 2x SSC for 1 hour at 37 C prior to the hybridization step. In addition, an anti-sense RNA probe against the CCTG repeat expansion was also applied on cells to assess the specificity of the (GGCCCC) 4 probe.

| LDH assay
Two-week old differentiated astrocytes were dissociated into single cells using Accutase and plated on 96-well plates coated with Matrigel (1:80 diluted) at a density of 2.5 × 10 4 cells per well. Cells were washed once with Neurobasal ® Medium prior to replacement with fresh AstroMed CNTF medium. Twenty-four hours later, conditioned medium was collected and the concentration of lactate dehydrogenase (LDH) was measured using a CytoTox-ONE™ Homogeneous Membrane Integrity Assay (Promega) following the manufacturer's instructions.

| Population viability assay
APCs were plated at a density of 1.5x10 5 cells per well in 24-well plates on plastic coverslips (Thermo Scientific) coated with laminin (Sigma), fibronectin (Sigma), and Matrigel (BD Biosciences) in NSCR EF20 medium for 5-7 days followed by differentiation into astrocytes for a further 2 weeks in AstroMED CNTF medium. Control iPSC-

| RNA-Seq
Total RNA from mature, iPSC-derived astrocytes was assessed for quality (Agilent Bionalyzer) and quantity (Invitrogen Qubit) before library preparation. Illumina libraries were prepared from 1 μg of total RNA using TruSeq RNA Sample Prep Kit v2 with a 10 cycle enrichment step as per the manufacturer's recommendations. Final libraries were pooled in equimolar proportions before Illumina sequencing on a HiSeq 2500 platform using 100 base paired-end reads. Reads were mapped to the primary assembly of the human (hg38) reference genome contained in Ensembl release 90 (Zerbino et al., 2018). Alignment was performed with STAR, version 2.5.3a (Dobin et al., 2013).

| Electrophysiology
Whole-cell patch-clamp recordings were used to assess the functionality of iPSC-derived MNs. Voltage-clamp mode was used to investigate intrinsic membrane properties. Current-clamp mode was used to investigate the firing properties of MNs. Experiments were carried out in a recording chamber which was perfused continuously with oxygenated artificial cerebral spinal fluid (aCSF) at room temperature Electrophysiological data were analyzed using Clampfit10 software (Axon Instruments). Data from control iPSC-derived MNs cocultured with C9orf72 iPSC-derived astrocytes lines (3 lines) were pooled for all analyses. Peak Na + currents and peak K + currents (log10 transformed), Cm, RN, and RMP were compared across the three different co-culture groups using one-way ANOVAs.
For the purposes of statistical comparisons, action potential generation was classified as either present or absent. These binary data were fitted with a general linear model and contrasts were made using Wald's tests and p values adjusted using a Bonferroni correction.

| Statistics
At least three independent derivations of astrocytes and MNs were used in each assay. All data are presented as mean ± S.E.M. Difference between means of two groups was analyzed by two-sided Student's t-test, whereas difference between means of three or more groups were analyzed by one-way ANOVA with Bonferroni correction or Turkey's post-hoc test. Two-way ANOVA was performed where two independent factors were involved. For all analyses, the null hypothesis was rejected at 0.05.
Repeat-primed PCR was used to confirm both the presence of G 4 C 2 repeat expansion in all three C9orf72 mutant lines and its absence in control and gene corrected lines ( Figure S1c). We next generated astrocyte progenitor cells (APCs) from iPSC lines and differentiated them into astrocytes using a previously established protocol (Serio et al., 2013). Immunocytochemistry showed high expression of APC markers vimentin and nuclear factor I-A (NFIA), and quantitative immunolabeling at 2 weeks post differentiation revealed >90% of cells positive for astrocyte markers, S100 calcium-binding protein B (S100B) and glial fibrillary acidic protein (GFAP). Comparable differentiation efficiency was observed across all six iPSC lines (Figure 1a 3.2 | Mutant astrocytes manifest RNA foci and dipeptide repeats that are reversed upon gene correction C9orf72 is believed to cause disease by three putative mechanism(s); haploinsufficiency, sequestration of RNA binding proteins by RNA foci and/or di-peptide repeat (DPR) mediated toxicity (Mizielinska & Isaacs, 2014;Shi et al., 2018;Tabet et al., 2018). As intranuclear RNA foci are observed in astrocytes in post-mortem derived material from  Figure S2). In addition, no foci were detected in C9orf72 mutant astrocytes when using a probe against the myotonic dystrophy type 2 (DM2) repeat expansion (CCTG) n , confirming the specificity of the G 4 C 2 anti-sense probe ( Figure S2). Quantification of RNA FISH revealed up to 60% of mutant astrocytes contained nuclear foci with no foci observed in control astrocytes (Figure 2b,c).
Notably, RNA foci were absent in astrocytes derived from the genecorrected C9-Δ line (Figure 2a-c), demonstrating a direct causal link between C9orf72 mutation and the formation of RNA foci in astrocytes.
We next examined the transcript levels of total C9orf72 in iPSCderived control and mutant astrocytes. C9orf72 transcripts detected in astrocytes were almost four-fold less compared to cortical neurons derived from a control human embryonic stem cell line (Figure 2d).
These data are in agreement with the in vivo finding that C9orf72 is more highly expressed in neurons compared to astrocytes (Jiang et al., 2016;Suzuki et al., 2013). Although no difference between control and mutant astrocytes was evident when data from all lines were pooled, a significant reduction was detected when comparing C9-3 astrocytes directly to its isogenic control C9-Δ astrocytes (Figure 2d; C9-3, 0.482 ± 0.039, n = 5; C9-Δ, 0.868 ± 0.129, n = 4; p < .05, Student's t-test). We further performed western blot analysis in iPSC (d) Differentiated human astrocytes had significantly lower C9orf72 transcript levels compared to cortical neurons, and a significant reduction of C9orf72 transcript levels was detected when comparing C9-3 astrocytes to its isogenic control C9-Δ astrocytes. (ns, not significant, between control and C9; #, p < .05, between C9-3 and C9-Δ; Student's t-test). (e) A western blot of urea-soluble protein fraction showed presence of the poly-GP DPR (indicated by red arrow heads) in C9-2 and C9-3 astrocytes, which was absent in the gene edited C9-Δ astrocytes. (f) A population-based LDH release assay revealed no differences in viability under basal culture conditions either between control and mutant astrocytes or between the isogenic pair (ns, not significant; Student's t-test) [Color figure can be viewed at wileyonlinelibrary.com] poly-GA and poly-PA DPR was detected in mutant astrocytes (data not shown).
TDP-43 proteinopathy is a pathological hallmark of ALS with cytoplasmic misaccumulation of TDP-43 found in MNs and glial cells (Arai et al., 2006;Ling, Polymenidou, & Cleveland, 2013;Neumann et al., 2006). However, using immunocytochemical labeling of TDP-43 we found predominantly nuclear localization ( Figure S4a) with no difference in nuclear or cytoplasmic TDP-43 intensity upon densitometric analysis between control and mutant astrocytes, or between C9-3 and C9-Δ astrocytes ( Figure S4b,c). In addition, quantitative immunoblot analysis showed equivalent protein levels of soluble TDP-43 in astrocytes derived from all six iPSC lines ( Figure S4d,e).

| Mutant astrocytes cause control MNs to lose functional output without overtly effecting cell viability
Accumulating evidence from pathological and experimental studies suggests that astrocytes may both undergo degeneration in ALS (Serio et al., 2013;Tong et al., 2013) and exert toxic effects on MNs (Ilieva et al., 2009). To first address whether G 4 C 2 expansion adversely affects the viability of isolated astrocytes, we undertook LDH assays that showed no difference between control and C9orf72 mutant astrocytes (Figure 2f; ns, not significant; student t-test). We next co- In view of the absence of any viability differences in co-cultures and our previous finding of mutant iPSC-derived MNs demonstrating physiological dysfunction prior to any changes in survival (Devlin et al., 2015), we next examined whether mutant astrocytes affect MN function. To facilitate comparisons with our previous study (Devlin et al., 2015), MNs derived from the same control iPSC line (Con-2) that we used in our previous work were co-cultured with astrocytes for up to 10 weeks post-plating. Electrophysiological analyses were used to investigate whether patient iPSC-derived astrocytes had any effect on the function of control MNs. Whole-cell patch-clamp recordings were obtained from the largest neurons visualized via IR-DIC microscopy in the co-cultures from 3 to 10 weeks post MN plating. Selecting the largest neurons ensured recordings were predominantly obtained from MNs (Devlin et al., 2015).
We first compared the passive membrane properties of control MNs co-cultured with astrocytes from a healthy individual (Con-2), three ALS patients carrying the C9orf72 hexanucleotide repeat expansion (C9-1, C9-2, and C9-3) as well as an isogenic control line for C9-3 (C9-Δ). For these and all other electrophysiological analyses, data were pooled for control MNs co-cultured with mutant astrocytes (see Figure S5 for data from individual lines). At weeks 3-4 post-plating, whole-cell capacitance (C m ) values were similar across MNs plated on control, mutant and gene edited-astrocytes (see Table 1 for x ± SEM and sample sizes; one-way ANOVA with Tukey's honest significant difference). From weeks 5-10, MNs plated on mutant astrocytes had smaller C m values compared to those on gene-edited astrocytes and from weeks 7-10 compared to MNs on control  (Table 1). Input resistance (R N ) values were similar in MNs co-cultured with control, mutant and gene-edited astrocytes throughout the time period studied (Table 1). MNs co-cultured with mutant astrocytes had a more depolarized resting membrane potential (RMP) at weeks 3-4 compared to MNs co-cultured with gene-edited astrocytes (Table 1; p < .05, one-way ANOVA with Tukey's honest significant difference). However, resting membrane potential did not differ at other time points in co-culture. These findings indicate that C9orf72 patient iPSC-derived astrocytes cause time-dependent changes to some of the passive membrane properties of control iPSCderived MNs.
As reported previously (Devlin et al., 2015), current injection elicited four output patterns described as repetitive, adaptive, single, and no firing, in MNs cultured on astrocytes (Figure 4a). Repetitive firing was defined as a train of action potentials that lasted for the duration of the square current injection (1 s), while adaptive firing was defined as multiple action potentials that stopped before the end of the current stimuli. Cells defined as having an adaptive output pattern were unable to repetitively fire in response to any of the series of current steps applied. In order to compare the excitability of repetitively firing MNs co-cultured with control, mutant or gene-edited astrocytes, frequency-current (f-I) relationships were generated from responses to a series of injected current steps (0 to 145 pA, in 10 pA increments, 1 s duration). Comparisons were performed on data pooled from recordings of repetitively firing cells at weeks 2-6 post MN plating. Analyses of the slope of the combined f-I relationship

These data demonstrate a clear loss of functional output in control
MNs co-cultured with mutant astrocytes compared to control or gene edited astrocytes, consistent with the idea that mutant astrocytes alone are sufficient to cause physiological toxicity to MNs.

| Mutant astrocytes cause loss of voltageactivated currents in control MNs
To investigate the mechanisms underlying the progressive loss of action potential output in control MNs co-cultured with mutant astrocytes, voltage-clamp recordings were performed to assess voltageactivated currents involved in action potential generation. Fast inactivating Na + currents were first investigated by using a series of voltage steps (−70 to 20 mV, 2.5 mV increments, 10 ms duration) from a holding potential of −60 mV (Figure 5a). We found no differences in the current (I-V) relationships ( Figure S6a) or peak Na + currents between MNs co-cultured with control, mutant or gene-edited astrocytes at 3-4 weeks post MN plating (Figure 5b; peak current: control, x 2,232 ± s.e.m. 211 pA, n = 49; mutant, 2,067 ± 120 pA, n = 153; gene-edited 2,110 ± 209 pA, n = 74). However, from weeks 5-10 post MN plating, there was a progressive decrease in peak Na + currents in MNs co-cultured with mutant astrocytes compared to cocultures with control or gene-edited astrocytes (Figure 5b and We next investigated whether the progressive loss of Na + currents reflected a more general decrease in voltage-activated currents in control MNs co-cultured with mutant astrocytes. Persistent K + currents were measured by using a series of voltage steps (−70 to 40 mV, 10 mV increments, 500 ms duration) from a holding potential of −60 mV (Figure 5c). At weeks 3-4 post MN plating, peak K + currents were comparable in MNs co-cultured with control, mutant or gene-edited astrocytes (Figure 5d and Figure S6c; Peak current: control, x 1,788 ± s.e.m. 148 pA, n = 46; mutant, 1913 ± 98 pA, n = 147; gene-edited, 1,634 ± 127 pA, n = 74). Similar to Na + currents, peak K + currents progressively declined in control MNs co-cultured with mutant astrocytes from weeks 5-10 compared to MNs co-cultured with control or gene-edited astrocytes (Figure 5d and Figure S6d; Weeks 5-6: mutant, x 1,070 ± SEM 165 pA, n = 53; gene-edited, 2,673 ± 233 pA, n = 43; Weeks 7-10: control, 2069 ± 160 pA, n = 41; mutant, 438 ± 99 pA, n = 60; gene-edited, 1816 ± 248 pA, n = 33; p < .0001, one-way ANOVA after log transformation with Tukey's post-hoc test).

| Functional perturbations in C9orf72 mutant MNs are mediated by mutant astrocytes
We, and others, have previously shown that mutant C9orf72 MN cultures demonstrate functional perturbations (Devlin et al., 2015;Naujock et al., 2016;Sareen et al., 2013). However, all these studies, including our own (Devlin et al., 2015), used MN generation protocols that also resulted in the production of a significant fraction of astrocytes (up to approximately 20%). One interpretation of these earlier findings is therefore that the observed pathophysiological phenotype was a consequence of contaminant astrocytes and not cell processes intrinsic to MNs. To address this possibility, we next used a recently published method to generate highly enriched mutant C9orf72 MN cultures with negligible astrocyte contamination (Maury et al., 2015;Selvaraj et al., 2018). In cultures derived from the C9orf72 ALS patient lines C9-1 & C9-3 and their respective gene edited controls (C9-Δ1 & C9-Δ3), we assessed firing output and voltage-gated Na + and K + currents using the same protocols described above for co-culture experiments.

| DISCUSSION
Here, we show that expression of the C9orf72 mutation in astrocytes recapitulates key aspects of C9orf72-related ALS pathology and directly results in physiological dysfunction of control and C9orf72 MNs upon co-culture, thus highlighting both cell-autonomous astrocyte pathology and non-cell autonomous MN pathophysiology. with formation of RNA foci and poly-GP DPR in human astrocytes. It remains to be determined whether other DPRs, in addition to poly-GP, are also produced in patient iPSC-derived astrocytes. In line with previous in vivo findings we observed relatively low levels of C9orf72 transcripts and protein in astrocytes (Jiang et al., 2016). Although TDP-43 proteinopathies are the pathological hallmark of ALS regardless of patients' genotypes, with TDP-43 deposition observed in astrocytes in post-mortem tissues (Yamanaka & Komine, 2018) no post-mortem stud- ies have yet reported TDP-43 proteinopathies in astrocytes in C9orf72related ALS. Our previous study has shown that iPSC-derived astrocytes carrying a TARDBP mutation do not display TDP-43 aggregates or loss of nuclear TDP-43 despite increased cytoplasmic TDP-43 expression (Serio et al., 2013), nor do we detect TDP-43 proteinopathies in astrocytes harboring a C9orf72 mutation in the present study, suggesting that iPSC-derived astrocytes may not manifest all TDP-43 proteinopathies in vitro. Additionally, in the AAV-G4C2-66 mice, only 7-8% of cells in cortex and hippocampus display phosphorylated TDP-43 inclusions (Chew et al., 2015), and no TDP-43 aggregations were observed in BAC-C9 (100-1,000) mice (O'Rourke et al., 2015).
Although there is consistent evidence of non-cell autonomous toxicity mediated by astrocytes harboring SOD1 mutations (Ilieva et al., 2009;Marchetto et al., 2008;Nagai et al., 2007;Papadeas et al., 2011;Tripathi et al., 2017;Tyzack et al., 2017), data are either lacking or conflicting for other ALS-related mutations (Haidet-Phillips et al., 2013;Serio et al., 2013;Tong et al., 2013). For example, human iPSCderived astrocytes from a patient harboring an TARDBP M337V mutation did not affect the survival of control iPSC-derived MNs (Serio et al., 2013). This finding was supported by an independent study where astrocytes lacking TDP-43 or overexpressing mutant TARDBP failed to cause the death of control MNs in co-culture or when implanted into wild-type rat spinal cords (Haidet-Phillips et al., 2013).
Such perturbations in the function of iPSC-derived MNs were previously assumed to reflect cell autonomous disease mechanisms in cultures consisting of approximately 80% neurons, 50% of which were MNs (Devlin et al., 2015). However, in the present study we found that patient iPSC-derived astrocytes caused a reduction in the functional output of control and patient iPSC-derived MNs, supporting non-cell autonomous mechanisms. Our findings therefore implicate "contaminant" astrocytes present in previous studies of enriched motor neuron mixed cultures as key mediators of MN dysfunction.
The toxic effects of the astrocyte secretome have also been demonstrated in the field of C9orf72-mediated ALS, although this area remains grossly unexplored. Madill and colleagues showed that C9-ALS patient iPSC-derived astrocytes modulate the autophagy pathway in a non-cell autonomous manner (Madill et al., 2017). Cells treated with patient conditioned medium demonstrated decreased expression of LC3-II, a key adapter autophagy protein, with a concomitant accumulation of p62 and increased SOD1 expression. Additionally, micro-RNAs secreted through astrocyte-derived extracellular vesicles cause increased neuronal death and deficits in neurite outgrowth in control mouse MNs (Varcianna et al., 2019). Through pathway analysis, they identified that hsa-miR-494-3p regulates axonal maintenance, with its primary target being Semaphorin 3A (SEM3A).
Treatment of the MNs with a miR-494-3p mimic in the presence of C9 iAstrocyte conditioned medium significantly reduced the levels of SEM3A by 25% in the MNs, and increased branching and neurite length, and survival.
An additional hypothesis to explain the toxic effect of ALS astrocytes on MN function is the loss or reduction of normal supportive roles fulfilled by astrocytes, including homeostatic regulation of extracellular glutamate (Foran & Trotti, 2009;Sasabe et al., 2012). This hypothesis is supported by a recent study that showed a reduction in the ability of VCP mutant astrocytes to support MN survival .
RNA-Seq analysis carried out on C9orf72 astrocytes in this study further highlighted alterations in multiple new gene pathways which may be causative towards both cell autonomous and non-cell autonomous pathophysiology. We observed upregulation of many genes involved in ribosome biogenesis and assembly. This is of interest in view of recent interactome studies and yeast genetic modifier screens that show toxic di-peptide repeat proteins play a role in ribosomal processing/biogenesis and reduce overall cell translation (Chai & Gitler, 2018;Hartmann et al., 2018). These studies thus provide indirect support for astrocyte DPRs having a role in the observed dysregulation of ribosomal processing genes. Na + /K + ATPase is a membrane bound pump that exchanges Na + and K + across the plasma membrane to maintain ionic concentration gradients, whilst also modulating neuronal excitability in an activity dependent manner (Picton, Nascimento, Broadhead, Sillar, & Miles, 2017). In our transcriptome analysis we observed upregulation of glial specific Na + /K + ATPase (ATP1B2). ATP1B2 knock-out mice exhibit deficits in motor coordination and develop tremors leading to premature death due to osmotic imbalance (Magyar et al., 1994). Furthermore, ATP1A2 is found to be upregulated in astrocytes expressing mutant SOD1 and contributes to non-cell autonomous toxicity to motor neurons (Gallardo et al., 2014). Furthermore, astrocytic focal adhesion molecules have been implicated in modulating neuronal excitability in seizure paradigms (Cho, Muthukumar, Stork, Coutinho-Budd, & Freeman, 2018) and dysregulation of cell-adhesion by L1CAM deficiency leads to impairment of action potential initiation (Valente et al., 2016). Taken together, the transcriptomic data are consistent with the possibility that impairments in astrocytic cellular processes could lead to a Na + /K + ionic imbalance in the synaptic cleft, leading to pathological changes in neuronal excitability.
Given the complex and dynamic interplay between astrocytes and MNs, it is likely that mechanisms underlying non-cell autonomous dysfunction and neurodegeneration are varied and reflect an imbalance between loss of homeostatic function and gain of toxic effects. It will therefore be important in future studies to comprehensively evaluate these newly discovered gene pathways and determine the consequences of perturbations in the transcriptome and proteome of astrocytes expressing the human C9orf72 mutation in order to fully define the mechanism(s) that underlie the observed pathological consequences of human C9orf72 astrocytes on human MN function.
In summary, our study provides the first report of the direct molecular and cellular impact of the C9orf72 mutation on human astrocytes and their interaction with human MNs. Findings here demonstrate that astrocytes, in addition to MNs, are affected by expression of mutant C9orf72, which leads to the development of pathological changes. In addition, expression of mutant C9orf72 in astrocytes induces progressive dysfunction of MNs due to the loss of voltage-activated currents. These data suggest that non-cell autonomous disease mechanisms are a contributor to C9orf72-mediated ALS. Furthermore, our study demonstrates the value of combining gene-editing with sensitive physiological studies in human iPSC-based neurodegenerative disease modeling.

CONFLICT OF INTEREST
The authors declare no potential conflict of interest. and S.C. and edited and revised the manuscript.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.