Assembly of targeting constructs

eGFP sequence and 100 codon-optimized polyGR or polyPR repeats were synthesized (Thermo Fisher Scientific). A pMC cloning vector was synthesized to contain 600 base pairs (bp) of the 5′ homology arm, a double HA-tag, a V5 epitope tag, an SV40 polyA tail and 250 bp of the 3′ homology arm. eGFP and 100 codon-optimized DPRs were cloned into this vector within the BbsI and BsmBI sites to generate pMC-eGFP, pMC-(GR)100 and pMC-(PR)100. Then, 400 codon-optimized DPRs were assembled with two consecutive rounds of recursive directional ligation taking advantage of the restriction enzymes BbsI and BsmBI to generate pMC-(GR)400 and pMC-(PR)400.

A selection cassette (FRT-PGK-gb2-neo-FRT, Gene Bridges) was inserted in the pMC-eGFP, pMC-(GR)400 and pMC-(PR)400 in the NheI site. A BAC subcloning kit (Red/ET recombination, Gene Bridges) was used to clone full-length homology arms (2.7 kilobases (kb) in 5′ and 3.2 kb in 3′) from the BAC clone (RP23-434N2), containing the C57BL/6J sequence of the mouse C9orf72 gene, into the targeting vector pBlueScript II SK (+). Knock-in constructs were obtained by inserting sequences from pMC-eGFP, pMC-(GR)400 and pMC-(PR)400 into targeting vectors within the BstXI and XcmI sites.

Animals

All procedures involving mice were conducted in accordance with the Animal (Scientific Procedures) Act 1986 and the Animal Research: Reporting of In Vivo Experiments guidelines and were performed at University College London (UCL) under an approved UK Home Office project license reviewed by the Institute of Prion Diseases Animal Welfare and Ethical Review Body. Mice were maintained in a 12-h light/dark cycle at a temperature of 20–24 °C and relative humidity of 45–55% with food and water supplied ad libitum. The knock-in mice generated are available from the European Mutant Mouse Archive, strain numbers (GR)400: EM:14658, (PR)400: EM:14659, eGFP: EM:15241.

To generate the DPR knock-in mouse strains, we performed CRISPR-assisted gene targeting in JM8F6 embryonic stem cells (C57BL/6N) using our targeting vector(s) and a CRISPR–Cas9 designed against the insertion site. The CRISPR construct (pX330-Puro-C9orf72) expressed Cas9 and a U6 promoter-driven single-guide RNA designed against the following sequence: AGTCGACATCCCTGCATCCC. This was generated by annealing two oligos (5′-CACCgAGTCGACATCCCTGCATCCC-3′; 5′-AAACGGGATGCAGGGATGTCGACTc-3′) and cloning this into the unique BbsI sites of pX330-U6-Chimeric_BB-CBh-hSpCas9 (Addgene no. 42230), modified by the addition of a PGK-Puro cassette. Then, 1 × 106 embryonic stem cells were electroporated with 2.5 µg of the cloned pX330-Puro-C9orf72 plasmid and 2.5 µg of targeting vector using the Neon Transfection System (Thermo Fisher Scientific) (3 × 1,400 V, 10 ms) and plated on puromycin-resistant fibroblast feeder layers. After approximately 24 h, 600 ng ml−1 puromycin was applied for a further 48 h to allow transient selection. After a further 5 d in culture without selection, individual colonies were isolated, expanded and screened for the desired targeting at both the 5′ end (5′-TCGGGGATTATGCCTGCTGC’3′ and 5′-GCATCCCAGGTCTCACTGCA-3′) and the 3′ end (5′-TCGAAAGGCCCGGAGATGAGGAAG-3′ and 5′-GGGTTCAGACAGGTACAGCAT-3′). Embryonic stem cells from correctly targeted clones were injected into albino C57BL/6J blastocysts and the resulting chimeras were mated with albino C57BL/6J females. The presence of the targeted allele in the F1 generation was confirmed at the DNA level by the above PCR and Sanger sequencing. Germline-transmitting founders were obtained and backcrossed to WT C57BL/6J mice to maintain hemizygous lines.

Mouse genotype was determined by PCR for knock-in sequence with the following set of primers: forward 5′-TAAGCACAGCAGTCATTGGA-3′ and reverse 5′-AAGCGTAATCTGGAACATCG-3′. Repeat length was determined by PCR with the following set of primers: forward 5′-CCCATACGATGTTCCAGATTACGCTTACCC-3′ and reverse 5′-GCAATAAACAATTAGGTGCTATCCAGGCCCAG-3′.

Males were used for all experiments in the main text, except in vivo two-photon calcium imaging and Neuropixels recording where females were used. Phenotyping was also performed on female mice, with similar results to males, and these data are included in Extended data figures.

Homozygous TAR4/4 mice overexpressing WT human TARDBP (TDP-43)59 were used as a positive control for detecting phosphorylated TDP-43.

Human tissues

Human C9orf72 ALS/FTD samples for polyGR MSD immunoassay were described previously42 and protein extracted as described in the biochemical analysis section.

Biochemical analysis

Brains and spinal cords were homogenized in lysis buffer (RIPA buffer (Pierce), 2% SDS, protease (Roche) and phosphatase (Thermo Fisher Scientific) inhibitors). Lysates were sonicated and microcentrifuged for 20 min at 13,000g at room temperature and soluble fractions collected. Proteins were separated on NuPAGE 4% to 12% bis-tris gels (Invitrogen) and transferred to nitrocellulose membranes (Bio-Rad Laboratories). Membranes were blocked in 5% milk in PBS-T (PBS, 0.1% Tween-20) for 1 h at room temperature. The membranes were incubated overnight at 4 °C with the following primary antibodies: C9orf72 (12E7, kindly donated by Prof. Dr. Manuela Neumann; 1:4 dilution), COL6 (ab182744, Abcam; 1:1,000), β-Actin (A2228, Sigma-Aldrich; 1:5,000 dilution), Phospho-TDP-43 (Ser409/410) (22309-1-AP, Proteintech; 1:1,000 dilution), Calnexin (sc-6465, Santa Cruz Biotechnology; 1:1,000 dilution). After three washes in PBS-T, membranes were incubated with secondary HRP-conjugated antibodies for 1 h at room temperature. After three washes in PBS-T, signals were visualized by chemiluminescence (Amersham imager 680) and quantifications performed using ImageJ software.

MSD immunoassays were performed as previously described60,61, using our custom rabbit anti-(GR)7 antibody60,61, and PR32B3 (Helmholtz Zentrum 2 µg ml−1) for capture and detection. GFP levels were measured by the GFP ELISA Kit (ab171581, Abcam) according to manufacturer’s instructions.

qPCR with reverse transcription

Tissues were dissected and flash-frozen. Total RNA was extracted with miRNeasy Micro Kit (Qiagen) and reverse-transcribed using SuperScript IV Reverse Transcriptase (Invitrogen) with random hexamers and Oligo(dT)20 primers. Gene expression was determined by quantitative real-time PCR using a LightCycler and SYBR green (Roche). Relative gene expression was determined using the ΔΔCT (cycle threshold) method. Primers for mouse C9orf72 are: 5′-TGAGCTTCTACCTCCCACTT-3′ and 5′-CTCTGTGCCTTCCAAGACAAT-3′. Primers to amplify the knock-in sequence are: 5′-GCGGCGAGTGGCTATTG-3′ (primer located within mouse C9orf72 gene at exon boundary 1-2) and 5′-GGGTAAGCGTAATCTGGAACATC-3′ (sequence within the HA-tag sequence). Primers for mouse Actin are: 5′-CTGGCTCCTAGCACCATGAAGAT-3′ and 5′-GGTGGACAGTGAGGCCAGGAT-3′.

Total RNA from cells was extracted with ReliaPrep RNA Cell Miniprep System (Promega) and reverse-transcribed using SuperScript IV VILO Master Mix (Invitrogen). Gene expression was determined as described above. Primers for human TGFB1 are: 5′-GGCTACCATGCCAACTTCT-3′ and 5′-CCGGGTTATGCTGGTTGT-3′. Primers for human COL6A1 are: 5′-ACTTCGTCGTCAAGGTCATC-3′ and 5′-CATCTGGCTGTGGCTGTA-3′. Primers for human GAPDH are: 5′-ACTAGGCGCTCACTGTTCT-3′ and 5′-CCAATACGACCAAATCCGTTG-3′.

Drosophila were flash-frozen in liquid nitrogen. Total RNA was isolated using TRIzol reagent (Thermo Fisher Scientific). RNA samples were treated with TURBO DNase (Thermo Fisher Scientific), and converted to complementary DNA using oligod(T) primers and Superscript II reverse transcriptase (Invitrogen). Gene expression was determined by quantitative real-time PCR using QuantStudio 6 Flex Real-Time PCR System (Applied Biosystems). Relative gene expression was determined using the ΔΔCT method. For qPCR, primers for Drosophila daw are: 5′-GGATCAGCAGAAGGACTCCAA-3′ and 5′-CAGTGTTTGATGGGCCACTC-3′. Primers for Drosophila Mp are: 5′-CTGGGCACCTTCAAGGCATT-3′ and 5′-ATCGCCACGAGTGTTCACC-3′. Primers for Drosophila Tubulin are: 5′-TGGGCCCGTCTGGACCACAA-3′ and 5′-TCGCCGTCACCGGAGTCCAT-3′.

Immunohistochemistry

Mice were perfused with prechilled PBS and then 4% paraformaldehyde (PFA). Brains and spinal cords were dissected and postfixed in 4% PFA at 4 °C for 2 h. After fixation, brains and spinal cords were washed with PBS, allowed to sink in 30% sucrose solution at 4 °C, then stored in 0.02% sodium azide at 4 °C until further processing. Brains and spinal cords were embedded in optimal cutting temperature (OCT) compound (Tissue Tek, Sakura), and 10-μm sections were cut with a cryostat (CM1860 UV, Leica Microsystem). For immunofluorescence, cryosections were washed three times in PBS and blocked in 5% BSA, 1% normal goat serum and 0.2% Triton-X in PBS for 1 h at room temperature. Sections were then incubated with primary antibodies in blocking solution overnight at 4 °C. After three washes with PBS, sections were incubated for 1 h at room temperature in blocking solution with secondary antibodies conjugated with Alexa 488, 546, 594 and 633 (Invitrogen). After three washes in PBS, sections were mounted with ProLong Gold Antifade Mountant with DAPI (Invitrogen).

Alternatively, after fixation, brains were washed with PBS, processed overnight using an automated tissue processor (Leica ASP300) and embedded in paraffin (Leica EG1150H). For immunofluorescence, 5-µm sections mounted on glass slides were incubated for 2 h at 60 °C. Sections were deparaffinized in xylene and rehydrated in decreasing grades of alcohol. Slides were incubated in methanol/hydrogen peroxide (0.3%) solution for 10 min at room temperature to block endogenous peroxidase activity. For antigen retrieval, slides were then transferred to a boiling solution of 0.1 M citrate buffer (pH 6.0) and pressure cooked at maximum pressure for 10 min. For immunofluorescence, slides were then blocked in 10% milk for 1 h at room temperature and incubated with primary and secondary antibodies as described above. For 3,3ʹ-diaminobenzidine (DAB) staining, slides were incubated in methanol/hydrogen peroxide (0.3%) solution for 10 min at room temperature and then blocked in 10% milk for 1 h at room temperature and incubated with primary antibody in PBS overnight at 4 °C. After three washes with PBS, sections were incubated for 30 min at room temperature in biotinylated secondary antibody (Vector Laboratories) in PBS. Slides were then washed in PBS and incubated in VECTASTAIN Elite ABC-HRP Kit, Peroxidase (Vector Laboratories) for 30 min at room temperature. Sections were washed three times with PBS and incubated in DAB chromogen (Abcam). Slides were then dehydrated in increasing grades of alcohol (70%, 95% and 100% ethanol), cleared in xylene and mounted with DPX mounting medium (Sigma-Aldrich).

The primary antibodies used were: HA clone 3F10 (11867423001, Roche; 1:100 dilution), NEUN (ABN91, Millipore; 1:500 dilution), IBA1 (019-19741, FUJIFILM Wako Pure Chemical; 1:500 dilution), GFAP (AB5804, Abcam; 1:500 dilution), GFAP (2.2B10, Invitrogen; 1:500 dilution), S100β (ab41548, Abcam; 1:300 dilution), CD68 (MCA1957, Bio-Rad Antibodies; 1:200 dilution), TDP-43 (12892-1-AP, Proteintech; 1:400 dilution), COL6 (ab182744, Abcam; 1:200), CTIP2 (ab18465, Abcam; 1:500), polyPR (PR32B3, Helmholtz Zentrum; 1:100).

Images were taken using a Zeiss LSM 880 confocal microscope or ZEISS Axio Scan.Z1 slide scanner. Image analyses were performed using ImageJ, Imaris or QuPath-0.3.2 software.

Surgical procedures for in vivo recordings

Surgical and experimental procedures were conducted in accordance with the Animal (Scientific Procedures) Act 1986, approved by the Animal Welfare and Ethical Review Body at UCL and performed under an approved UK Home Office project license at UCL. Before surgical procedures, WT, PR(400) and GR(400) mice were anesthetized with isoflurane (3–4% induction, 1.5–2% for maintenance) and given subcutaneous carprofen for pain relief. The animal’s head was shaved to remove fur and the animal placed in a stereotaxic frame (WPI). Exposed skin was disinfected using diluted chlorhexidine and cleaned with ethanol. Anesthetic depth was confirmed by monitoring pedal reflex and breathing rate. Following a small incision, the skin overlying the skull was retracted and connective tissue over the skull cleared. A small craniotomy overlying motor cortex was then performed using a hand-held microdrill (WPI). In some animals, a silver chloride wire was then attached to a small indentation in the skull overlying cerebellum using cyanoacrylate glue (to act as a ground/reference electrode) and dental cement (Jet) was used to secure the wire and build a well encircling the craniotomy. Animals were then transferred directly to the electrophysiology station. In remaining animals, a glass coverslip (5-mm diameter) was placed over the craniotomy, and dental cement (Jet) applied to secure the cranial window and cover remaining exposed skull. On completion of these procedures, a subcutaneous injection of buprenorphine (0.1 mg kg−1) was administered for immediate postsurgical pain relief. After recovery, animals were returned to the holding room in single-housed conditions. Recordings were performed at least 2 weeks following recovery.

In vivo two-photon Ca2+ imaging and analysis

In vivo two-photon Ca2+ imaging of motor cortex was performed under light isoflurane anesthesia (~1%) using a custom-built resonant-scanning two-photon microscope (Independent NeuroScience Services) controlled by ScanImage (MBF Bioscience), and equipped with a Coherent Chameleon Discovery NX tunable laser and a ×16, 0.8 numerical aperture, Nikon water immersion objective. Imaging of neuronal activity was performed at a wavelength of 1,070 nm and fluorescence detected with a GaAsP photomultiplier tube (Hamamatsu). Images (512 × 512 pixels) were acquired at 30-Hz frame rate, and each field-of-view was recorded for at least 5 min. Image analysis was performed with Suite2p (ref. 62) and custom MATLAB scripts. The recorded image stacks were loaded into Suite2p for motion correction, region of interest (ROI) detection and Ca2+ signal extraction. For each detected ROI (putative cell somata), the neuropil-corrected signal was extracted by subtracting the neuropil fluorescence signal surrounding the ROI (Fn) from the raw fluorescence signal within the ROI (F): Fcorr(t) = F(t) − 0.7 × Fn(t). The baseline fluorescence (F0) was estimated by using robust mean estimation and relative fluorescence change (ΔF/F = (Fcorr(t) − F0)/F0) over time (t) was generated for each ROI. Ca2+ transients were identified as relative changes in ΔF/F that were two times larger than the standard deviation of the noise band. Following automatic peak detection, the peaks were inspected using custom-written MATLAB scripts and manually curated to exclude false positives and include false negatives. Silent and hyperactive neurons were defined as those with individual activity rates of 0 and >3 transients per min, respectively.

In vivo Neuropixels recordings and analysis

For recordings following surgical recovery, a small aperture overlying the motor cortex was made in the glass coverslip using a diamond-tipped drill bit and microdrill, and a silver chloride ground/reference electrode affixed to the edge of the cranial window. In all animals, a Neuropixels probe63 (IMEC) was connected to the ground/reference electrode and slowly implanted into motor cortex using stereotactic procedures at a rate ~5–10 µm s−1 under remote micromanipulator control (QUAD, Sutter Instruments) and visualized through a microscope (GT Vision). Following successful implantation, the cranial well was filled with warm sterile saline and the brain allowed to rest for at least 45 min before recordings. A 10-min recording of spontaneous neuronal activity was then performed in each animal under light isoflurane anesthesia (~1%). Neuropixels recordings (30-kHz sampling rate) were processed and automatically spike sorted using Kilosort3 using default parameters64. Processed data were then imported into PHY software (https://github.com/cortex-lab/phy) for interactive visualization of putative cortical clusters, and data manually inspected and curated to exclude false positives and include false negatives, and improve clustering through merging where appropriate. Following curation, spike-sorted data were analyzed using custom-written MATLAB scripts and subjected to an additional quality control where only units in which less than 1% of associated spikes violated the physiological refractory period of 2 ms were included for further analysis. Mean firing rates of all units (as log10Hz) across the entire 10-min recording session were calculated over 1-s time-bins for superficial cortex (0–400 µm below surface) and layer 5 (550–800 µm below surface). LFP data for the entire 10-min recording session were low pass filtered and decimated from 2.5 kHz to 500 Hz using an eighth order Chebyshev Type 1 IIR filter and common average referenced. LFP data were averaged across cortical channels and mean power in the slow (0.1–1 Hz), mid gamma (40–90 Hz) and high gamma (90–120 Hz) frequency bands estimated using Welch’s technique (40-s windows with 50% overlap, MATLAB function ‘pwelch’). Data were tested for normal distributions using Kolmogorov–Smirnov or Anderson–Darling tests.

Locomotor, grip strength and body weight assessment

Behavioral tests were performed monthly from 3 to 9 months of age, and in 12-month-old mice. Motor coordination was measured by rotarod analysis (Ugo Basile). A power calculation using GPower predicted that for an effect size of 10% deviation from the group mean, with a power of 0.85 and an alpha of 0.05, groups sizes of 28 were needed, so we used 14 females and 14 males per group. Mice were trained the week before starting the test. Then, mice received a session which included three trials of accelerated rotarod for a maximum of 300 s. Trials started at 4-r.p.m. speed and accelerated up to 40 r.p.m. in 4 min; the final minute of the test was performed at 40 r.p.m. The average of the three trials was used. A grip strength meter (Bioseb) was used to measure forelimb and hindlimb grip strength. The highest muscle force score of three independent trials was used. Body weight was measured weekly from 3 months of age. Mice were randomized into different experimental groups and the operator was blind to genotype.

Motor neuron counts

The 10-μm-thick OCT-embedded spinal cord sections were stained with Cresyl Violet and motor neurons located within the sciatic motor pool were counted in each ventral horn on 35 sections, collected every 60-μm of tissue, covering L3 to L5 levels of the spinal cord.

In vivo isometric muscle tension physiology

Isometric muscle tension physiology was performed as previously described65,66. Briefly, under deep anesthesia (isoflurane inhalation via nose cone), hindlimbs were immobilized and the distal tendons of the tibialis anterior and EDL muscles of both hindlimbs were exposed and consecutively attached to force transducers in parallel. Sciatic nerves were exposed bilaterally, at mid-thigh level, severed and the distal stumps placed in contact with stimulating electrodes. EDL muscle MUNEs were determined by gradually increasing the amplitude of repeated square wave stimuli, thereby inducing stochastic changes in contractile force. The total number of motor units recruited over the full range of amplitudes was counted for individual muscles and averaged for each genotype.

Mouse hindlimb lumbrical muscles preparation and NMJ imaging and analysis

Muscles were dissected and NMJs were stained as previously described67. NMJs were analyzed with the NMJ‐morph workflow. The following antibodies were used: mouse anti‐neurofilament (2H3, Developmental Studies Hybridoma Bank (DSHB), supernatant; 1:250 dilution), mouse pan anti‐synaptic vesicle 2 (SV2, DSHB, supernatant; 1:25 dilution) and Alexa Fluor 555 α‐bungarotoxin (α‐BTX; Life Technologies, B35451; 1:1,000 dilution).

Proteomic analysis

Mouse lumbar spinal cords and cortices were solubilized in SDS lysis buffer (2% (wt/vol) SDS in 100 mM triethylammonium bicarbonate supplemented with Roche protease mini and Phos-STOP cocktail tablets). Automated homogenization was performed using the Precellys evolution homogenizer and Bioruptor-assisted sonication. Protein estimation was by BCA assay and protein quality was confirmed by SDS–PAGE. Lysate (200 µg) was aliquoted and processed using S-Trap-assisted On-column tryptic digestion as described previously (https://doi.org/10.17504/protocols.io.bs3tngnn). Peptide eluates were then subjected to TMT labeling and high-pH fractionation for liquid chromatography with tandem mass spectrometry (LC–MS/MS); a total of 96 fractions were collected and pooled to 48 fractions, vacuum dried and stored at −80 °C until LC–MS/MS analysis.

LC–MS/MS analysis: a total of 48 basic reverse-phase liquid chromatography fractions were prepared for mass spectrometry analysis using an Orbitrap Tribrid Lumos mass spectrometer in-line with an Ultimate 3000 RSLC nano-liquid chromatography system. The mass spectrometer was operated in a data-dependent SPS-MS3 mode at a top speed for 2 s. Full scan was acquired at 120,000 resolution at 200 m/z measured using an Orbitrap mass analyzer in the scan range of 350–1,500 m/z. The data-dependent MS2 scans were isolated using a quadrupole mass filter with 0.7-Th isolation width and fragmented using normalized 32% higher-energy collisional dissociation and detected using an ion trap mass analyzer which was operated in a rapid mode.

Data analysis for mouse tissue: spinal cord raw mass spectrometry data were searched with MaxQuant software suite (v.2.0.1.0)68 against the Uniprot Mouse database appended with (GR)400 and (PR)400 sequences for C9orf72, and a common contaminant list exists within MaxQuant. FDR was set at 1% for both protein and peptide-spectrum match levels. The protein group output files were further processed using Perseus software suite (v.1.6.15.0)69 for downstream statistical analysis. Two-sided Welch’s t-test with 5% FDR multiple-correction was performed to identify differentially regulated proteins. GO analysis was performed on differentially regulated proteins using enrichR software70.

Reanalysis of C9orf72 iPS cell-derived motor neurons: we downloaded the SWATH mass spectrometry raw data from the CHORUS repository containing ten control and seven ALS samples. The .wiff and .wiff.scan files were then converted to mzML using mass spectrometry msConvert with the peak picking filter added71. Converted files were then searched using Spectronaut software suite version Rubin: 15.7.220308.50606 (ref. 72) using a direct-data-independent acquisition strategy. A FASTA file from the Human UniProt database was used to generate a predicted library within Spectronaut and a search was performed using the Pulsar search algorithm with default search parameters. The output protein group file was then processed using the Perseus software suite as described above to perform two-sided Welch’s t-test with 5% FDR multiple-correction to identify differentially regulated proteins between ALS and control samples.

Microarray analysis

We analyzed the transcriptional signatures in laser-captured spinal motor neurons from postmortem C9ALS patients32. Raw microarray data are available from the Gene Expression Omnibus (GEO) with accession number GSE56504. We performed Gene Set Enrichment Analysis (GSEA) using the gseGO function from the clusterProfiler R package73. Differentially expressed genes were ranked and then subjected to GSEA. The top enriched gene sets included all had normalized enrichment score > 0.

IPA

Data were analyzed with the use of QIAGEN IPA (QIAGEN, https://digitalinsights.qiagen.com/IPA). We used the user dataset as the reference dataset, and P = 0.05 and log fold change −1.5 to 1.5.

Human frontal cortex RNA-seq data analysis

We used previously published RNA-seq data from the frontal cortex of pathologically diagnosed patients with frontotemporal lobar degeneration (with and without ALS) and control samples. We extracted the summary statistics and residual expression values for our gene(s) or pathways of interest from the differential gene expression and Weighted Gene Co-expression Network Analysis with adjustment for cell-type markers33.

iPS cell differentiation and lentiviral transduction

The WTC11 iPS cell line harboring a doxycycline-inducible NGN2 cassette (kind gift of Dr. Michael E. Ward) was differentiated into cortical neurons (i3Neurons) as previously described35. On DIV 3, neural progenitor cells were dissociated with accutase and replated onto poly-l-ornithine (Merck) and laminin-coated plates in neuronal maintenance media: Neurobasal (Gibco), supplemented with 1 × B27 (Gibco), 10 ng ml−1 BDNF (PeproTech), 10 ng ml−1 NT-3 (PeproTech) and 1 µg ml−1 laminin. Neurons were plated at the desired ratio (typically 6 × 105 cells per well of a 6-well plate). At DIV 3, lentivirus was added to i3Neurons to overexpress (GR)50 or GFP constructs under the control of the neuron-specific human synapsin 1 promoter as previously described36. From DIV 3 to DIV 7, cells were maintained in neuronal maintenance media and imaged on the Incucyte S3 Live-Cell Analysis System using a confluency mask to quantify and track cell survival.

Drosophila stocks and maintenance

Drosophila stocks were maintained on SYA food (15 g l−1 agar, 50 g l−1 sugar, 100 g l−1 brewer’s yeast, 30 ml l−1 nipagin (10% in ethanol) and 3 ml l−1 propionic acid) at 25 °C in a 12-h light/dark cycle with 60% humidity. The UAS-(GR)36 flies have been previously described7. The elav-GS stock was obtained from a kind gift of H. Tricoire74. The fly lines GMR-Gal4 (Bloomington no. 9146), UAS-daw RNAi (Bloomington no. 50911 and no. 34974) and UAS-Mp RNAi (Bloomington no. 52981 and no. 28299) were obtained from the Bloomington Drosophila Stock Center.

Drosophila ortholog prediction

The Drosophila RNAi Screening Center Integrative Ortholog Prediction Tool (DIOPT; http://www.flyrnai.org/diopt) was used to search for orthologues of TGF-β1, COL6A1, COL6A2 and COL6A3. DIOPT predicted daw as the Drosophila orthologue of TGF-β1, and Mp as the Drosophila orthologue of human COL6A1, COL6A2 and COL6A3.

Drosophila eye phenotype analysis

Flies carrying the UAS-daw RNAi or UAS-Mp RNAi construct were crossed to the GMR-GAL4, UAS-(GR)36 line at 25 °C. Two-day-old adult F1 flies were imaged using a stereomicroscope, with female eyes used for analyses. All eye images were obtained under the same magnification; eye area was calculated from each image using ImageJ (ref. 75).

For toxicity scoring, 2-day-old adult F1 flies were examined with a dissecting microscope. Adult flies were separated into four groups based on the severity of the rough eye phenotype: WT, low, medium and high. The percentage of flies in each category was calculated. The results were analyzed by chi-squared test.

Drosophila lifespan assays

Lifespan assays were carried out as previously described76. The parental generation of the genotype used in each lifespan assay was allowed to lay for 24 h on grape agar plates supplemented with yeast. Eggs were placed at a standard density into bottles containing SYA medium. Adult experimental flies were allowed to emerge and mate for 48 h before being lightly anesthetized with CO2, and females randomly allocated onto SYA containing RU486 (200 μM) or ethanol vehicle at a standard density per vial (n = 15). Flies were transferred to fresh vials three times per week, with deaths scored at least three times per week. Escaping flies were censored from the data. Log-rank test was performed using Microsoft Excel (template available at http://piperlab.org/resources/).

iPS cell-derived neuronal differentiation and glutamate toxicity

Non-neurological control and C9orf72 iPS cells were obtained from the Answer ALS repository at Cedars Sinai (see Supplementary Table 1 for demographics) and maintained in mTeSR Plus medium at 37 °C with 5% CO2. iPSNs were differentiated in accordance with the diMNs protocol previously described28,38,77 and maintained at 37 °C with 5% CO2. On day 32 of differentiation, iPSNs were dissociated with accutase to facilitate the generation of a single-cell suspension. A total of 5 × 106 iPSNs were nucleofected with 4 µg of plasmid DNA (Origene) in suspension as previously described28,38. Following nucleofection, 100 µl of cell suspension was plated in each well (total of six wells per cuvette) of a glass-bottom or plastic 24-well plate for propidium iodide (PI) and Alamar Blue toxicity and viability experiments, respectively. Medium was exchanged daily for a total of 7 d to facilitate the removal of iPSNs that failed to recover postnucleofection. On the day of the experiment (day 39 of differentiation), iPSN medium was replaced with artificial cerebrospinal fluid solution containing 10 µM glutamate. For those iPSNs undergoing Alamar Blue viability assays (plastic dishes), Alamar Blue reagent was additionally added to each well according to the manufacturer’s protocol at this time. Following incubation, iPSNs for PI cell death assays were incubated with PI and NucBlue live ready probes for 30 min and subjected to confocal imaging as previously described28. The numbers of PI spots and nuclei were automatically counted in FIJI. Alamar Blue cell viability plates were processed according to the manufacturer’s protocol. As a positive control, 10% Triton-X-100 was added to respective wells 1 h before processing.

Statistical analysis

All data are presented as mean ± s.e.m. Statistical differences of continuous data from two experimental groups were calculated using unpaired two-sample Student’s t-test. Comparisons of data from more than two groups were performed using a one-way analysis of variance (ANOVA), followed by Bonferroni correction for multiple comparisons, or Tukey’s multiple comparisons, or Fisher’s least significant difference procedure. When two independent variables were available, comparisons of data from more than two groups were performed using a two-way ANOVA followed by Bonferroni correction for multiple comparisons, or Tukey’s multiple comparisons. Data distribution was tested for normality using the Kolmogorov–Smirnov test; when normality could not be tested, we assumed data distribution to be normal. The statistical significance threshold was set at P < 0.05, unless otherwise indicated. Statistical methods were used to predetermine sample sizes.

Data collection

Data collection and analysis were performed blind to the conditions of the experiments.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.