Generation of Thy1 (GA)149-CFP mice
We inserted a multiple cloning site into the pUC18 based murine Thy1.2 vector using synthetic oligonucleotides . This allowed us to insert a cDNA encoding (GA)149, 31 amino acids corresponding to the 3′ region of the poly-GA reading frame in patients  and cyan fluorescent protein (CFP) (sequence shown in Fig. S1a). Compared to the previous (GA)149-GFP construct  only the fluorescent protein had been changed. Linearized vector was injected into C57BL/6-derived zygotes and transferred into pseudopregnant CD1 females (PolyGene). GA-CFP mice were kept in the C57BL/6N background. Mice were PCR genotyped using the following primers (tccaggagcgtaccatcttc; gtgctcaggtagtggttgtc). We confirmed maintenance of the full length transgene with PCR amplification (Expand Long Template PCR System, Roche, 11681842001; gatccaagcttgccaccatg; tctagctctgccactccaag) and sequencing.
The transgene integration site was determined by whole genome sequencing according to standard protocols using the TruSeq DNA PCR-Free Library Preparation Kit and an Illumina HiSeq 4000 with 150 bp paired-end reads resulting in about 58× coverage from two lanes. Sequences mates mapping to different chromosomes were analyzed using the Integrative Genomics Viewer (IGV) .
Immunohistochemistry of mouse and patient tissue
After killing, 1-, 3-, 6-, and 12-month-old mice were transcardially perfused with 1% sterile PBS and tissue was then formalin fixated for 2 days. Histological stainings were performed on 5–8 µm thick sections from paraffin-embedded tissue. For spinal cord tissue, an additional decalcification step with 5% formic acid for 5 days was performed after formalin fixation. After deparaffinization in xylene and dehydration in graded ethanol, the paraffin sections were treated with citrate buffer (pH 6) for 20 min in the microwave. Mlf2 IHC staining was more prominent when the citrate retreatment was followed by 20 min incubation in 80% formic acid or 5–25 min incubation with 0.1 µg/µl proteinase K in 10 mM Tris/HCl pH 7.6 at 37 °C. Afterwards the slides were incubated with primary antibody overnight at 4 °C. For ChAT, an additional incubation with rabbit anti-goat-IgG was performed the next day for 1 h at room temperature. The slides were detected by the DCS SuperVision 2 Kit (DCS innovative diagnostic-system, Hamburg, Germany) according to the manufacturer’s instructions. Iba1 and GFAP immunohistochemistry was performed with the Ventana BenchMark XT automated staining system (Ventana) using the UltraView Universal DAB Detection Kit (Roche). For Nissl staining, the deparaffinized slides were incubated in 70% ethanol overnight. After 30 min in Cresyl violet and 1 min in 96% ethanol the slides were processed in 100% ethanol with glacial acetic acid. Bright-field images were taken by CellD, Olympus BX50 Soft Imaging System (Olympus, Tokyo, Japan).
For immunofluorescence, after deparaffinization and citrate antigen retrieval, the slides were incubated with primary antibody overnight at 4 °C and the following day incubated for 1 h at room temperature with secondary Alexa Fluor labeled antibodies. For Mlf2 immunofluorescence staining, a 1 min treatment at 37 °C with 0.05 µg/µl proteinase K in 10 mM Tris/HCl pH 7.6 was necessary before citrate antigen retrieval. After the nuclei were counterstained with DAPI, the slices were incubated for 1 min in 0.2% Sudan black B and mounted with Fluoromount Aqueous Mounting Medium (Sigma, F4680). Fluorescent images were taken using a LSM710 confocal laser scanning system (Carl Zeiss, Jena, Germany) with 20x or 40x/63x oil immersion objectives.
α-GFP (1:1000, Clonetech 632592), α-GA clone 5F2  (purified mouse monoclonal, WB unlabeled 1:50; IHC HRP labeled 5F2 1:2500, labeled by AbD Serotec HRP-labeling Kit LNK002P; biotinylated 5F2 7 ng/µl; MSD-labeled 5F2 10 ng/µl, labeled by Meso Scale MSD Sulfo-Tag NHS-Ester R91AN-1), α-GA-CT (C-terminal tail) clone 5C3  (rat monoclonal, 1:50), α-p62/SQSTM1 (IF 1:100, IHC 1:1000, MBL, PM045), α-pTDP-43 (Ser409/Ser410) clone 1D3  (purified rat monoclonal, 1:50), α-TDP-43 (1:1000, Cosmo Bio, TIP-TD-P09), α-RanGAP1 (1:100, Abcam, ab92360), α-nucleolin (1:1000, Abcam, ab50729), α-CD68 (1:1000, Abcam, ab125212), α-Iba1 (1:500, Wako, 091-19741), α-GFAP (1:5000,Dako, Z0334), α-NeuN (1:1000, Abcam, ab177487), α-ChAT (IF 1:300, IHC 1:5000, Millipore, AB144P), α-Calnexin (1:3000, Enzo Life Science, SPA-860F), α-Calbindin (1:300, Abcam, ab49899), α-Calretinin (1:1000, Abcam, ab702) α-Parvalbumin (1:750, Abcam, ab11427), α-Mlf2 #1 (1:1000, Sigma-Aldrich, HPA010811-100UL), α-Mlf2 #2 (1:1000, Santa Cruz, sc-166874), α-Laminin (1:200, Abcam, ab11575), α-goat-IgG (1:400, Dako, E0466), α-mouse Alexa Fluor 488 (1:500, Thermo Fischer Scientific, A11029), α-rabbit Alexa Fluor 488 (1:500, Thermo Fischer Scientific, A11034), α-rat Alexa Fluor 488 (1:500, Thermo Fischer Scientific, A11006), α-mouse Alexa Fluor 555 (1:500, Thermo Fischer Scientific, A21424), α-rabbit Alexa Fluor 555 (1:500, Thermo Fischer Scientific, A21429), α-rat Alexa Fluor 555 (1:500, Thermo Fischer Scientific, A21434), Streptavidin Alexa Fluor 488 (1:500, Thermo Fischer Scientific, S11223), nuclei were stained with DAPI (Roche Applied Science, Penzberg, Germany).
Immunoassay analysis of poly-GA in tissue homogenates
Mouse brainstem and spinal cord samples and C9orf72 patient motor cortex samples were sonicated in 500–700 µl of cold RIPA buffer (137 mM NaCl, 20 mM Tris pH 7.5, 10% Glycin, 1% Triton X 100, 0.5% Na-deoxycholate, 0.1% SDS, 2 mM EDTA, protease and phosphatase inhibitors). 100 µl of this homogenized tissue stock solutions were diluted to 300 µl with RIPA and centrifuged at 100,000×g for 30 min at 4 °C. To avoid cross contamination, the RIPA-insoluble pellets were resuspended in 300 µl RIPA, re-sonicated and re-centrifuged. Afterwards the RIPA-insoluble pellets were sonicated in U-RIPA (RIPA buffer containing 3.5 M Urea) and the protein concentration determined by Bradford assay. Streptavidin Gold multi-array 96-well plates (Mesoscale, L15SA-1) were blocked for 30 min with block solution (1% BSA, 0.05% Tween20 in PBS) and incubated with biotinylated α-GA clone 5F2 overnight at 4 °C. Equal amounts of protein of all samples were added in duplicate wells for 2 h, followed by 2 h incubation with the secondary MSD-labeled α-GA clone 5F2. Serial dilution of recombinant GST-GA15 in blocking buffer was used to prepare a standard curve. The wells intensity of emitted light upon electrochemical stimulation was measured using the MSD Quickplex 520 and the background corrected by the average response obtained from blank wells. Sensitivity and specificity of the immunoassay were confirmed using purified 15-mer DPRs fused to GST (Fig. S3a, b).
Phenotypic analysis of mice
The study was conducted in accordance with European and national guidelines for the use of experimental animals, and the protocols were approved by the governmental committee (Regierungspräsidium Oberbayern, Germany). All experimenters were blind to the genotype.
Barnes maze (Stoelting Europe, Ireland) assay to test spatial, hippocampus-dependent long-term memory in mice was performed on a circular surface (diameter 91 cm) with 20 circular holes (diameter 5 cm) around its circumference . Under one hole was an “escape box” (diameter 4 cm, depth 15 cm). The table surface was brightly lit by overhead lightning (900 lx). For each trial the mice had 3 min to find and hide in the “escape box”. For the statistical analysis failed attempts were set to 3 min.
In the balance beam test, the mice were placed on a wooden beam (round surface, length 58 cm, diameter 8 mm) and had 1 min to cross the beam. The test was finished either when the mice reached the end of the stick, they dropped down or the time ran out. For the statistical analysis failed attempts were set to 1 min. The experimenters were blind to the genotype, and trials were either video documented or recorded by AnyMaze (Stoelting Europe). AnyMaze Software was used to track the mice and to analyze the data.
In the Rotarod test (Ugo Basile), we accelerated the spindle speed from 5 to 50 rpm over 5 min. The test finished either after 5 min or when the mouse dropped down. The average time of two trials with 1 h break in between was used.
Modified SHIRPA analysis and grip strength testing was performed as described .
The beam ladder consists of two Plexiglas screens connected with several metal beams of variable distance. The test is used to evaluate skilled walking of the mice. Mice traverse the ladder and foot slips of fore paws and hind paws are counted separately as well as the time to traverse the beam.
The open field test as an assessment of spontaneous exploratory and anxiety-related behavior in a novel environment was carried out as previously described [12, 14, 39]. It consisted of a transparent and infra-red light permeable acrylic test arena with a smooth floor (internal measurements: 45.5 × 45.5 × 39.5 cm). Illumination levels were set at approximately 150 lx in the corners and 200 lx in the middle of the test arena. Each animal was placed individually into the middle of one side of the arena facing the wall and allowed to explore it freely for 20 min. For data analysis, the arena was divided by the computer in two areas, the periphery defined as a corridor of 8 cm width along the walls and the remaining area representing the center of the arena (42% of the total arena). Data were recorded and analyzed using the ActiMot system (TSE, Bad Homburg, Germany).
Acoustic startle and its prepulse inhibition were assessed using a startle apparatus setup (Med Associates Inc., VT, USA) including four identical sound-attenuating cubicles. The protocol is based on the Eumorphia protocol (http://www.empress.har.mrc.ac.uk), adapted to the specifications of our startle equipment, and constantly used in the primary screen of the GMC . Background noise was 65 dB, and startle pulses were bursts of white noise (40 ms). A session was initiated with a 5-min-acclimation period followed by five presentations of leader startle pulses (110 dB) that were excluded from statistical analysis. Trial types included prepulse alone trials at four different sound pressure levels (67, 69, 73, 81 dB), and trials in which each prepulse preceded the startle pulse (110 dB) by a 50 ms inter-stimulus interval. Each trial type was presented ten times in random order, organized in ten blocks, each trial type occurring once per block. Inter-trial intervals varied from 20 to 30 s.
DNA constructs and lentivirus production
cDNA of rat Mlf2 (NCBI Gene ID: 312709) containing an N-terminal HA-tag was expressed from a lentiviral vector driven by human ubiquitin promoter (FUW2-HA). Previously described (GA)175-GFP cDNA expressed from a synthetic gene lacking repetitive (ggggcc)n sequences with ATG start codon and EGFP was cloned in a lentiviral packing vector (FhSynW2) containing the human synapsin promoter . Lentivirus was produced in HEK293FT cells (Life Technologies) as described previously .
Cell culture, RNA isolation and immunoprecipitation
Primary hippocampal neurons from embryonic day 19 rats were cultured and transduced with lentivirus as described previously . Immunofluorescence staining was performed on 10 min PFA (4% paraformaldehyde and 4% sucrose) fixed primary neurons. The primary and secondary antibodies were diluted in GDB buffer (0.1% gelatin, 0.3% Triton X-100, 450 mM NaCl, 16 mM sodium phosphate pH 7.4) and incubated over night at 4 °C or 1 h at room temperature. Confocal images were taken using a LSM710 confocal laser scanning system (Carl Zeiss, Jena, Germany) with 40× or 63× oil immersion objectives. RNA isolation and qPCR was performed as described previously  using the following primers (CD68 ttctgctgtggaaatgcaag and gagaaacatggcccgaagt; Iba1 acagcaatgatgaggatctgc and ctctaggtgggtcttgggaac; GFAP tttctcggatctggaggttg and agatcgccacctacaggaaa; ACTB atggaggggaatacagccc and ttctttgcagctccttcgtt; GAPDH caacagcaactcccactcttc and ggtccagggtttcttactcctt).
Tissue samples of patient autopsy cases were provided by the Neurobiobank Munich, Ludwig-Maximilians-University (LMU) Munich and collected according to the guidelines of the local ethics committee.
Statistics and analysis
Statistical analysis was performed with GraphPad Prism software (version 7.01). For neuron and motoneuron count, images of the left and right anterior horns of the spinal cord were taken and all positively stained cells were manually counted. The count number represents the neurons/motoneurons averaged on one side. Experiments with two groups were analyzed by t test (unpaired, two-sided, t = size of the difference relative to the variation; df = degrees of freedom). Behavioral data was analyzed by two-way ANOVA with Bonferroni post hoc test (F = equality of variances).
For phosphorylated TDP-43 measurements, sarkosyl-soluble and urea-soluble fractions of mouse spinal cord tissues were prepared as previously described . In brief, 25–60 mg of tissue were subjected to a sequential extraction protocol using Tris–EDTA buffer (50 mM Tris pH 7.4, 50 mM NaCl, 1 mM EDTA), high salt Triton X-100 buffer, Triton X-100 buffer + 30% sucrose, and sarkosyl buffer. Sarkosyl-insoluble material was further extracted in urea buffer. The protein concentrations of sarkosyl-soluble fractions were determined using a bicinchoninic acid assay (Thermo Scientific), whereas a Bradford assay was utilized to measure protein concentrations of urea-soluble fractions. Phosphorylated TDP-43 levels in both these fractions were evaluated using a sandwich immunoassay that utilizes MSD electrochemiluminescence detection technology . A mouse monoclonal antibody that detects TDP-43 phosphorylated at serines 409 and 410 (Cosmo Bio, #CAC-TIP-PTD-M01, 1:500) was used as the capture antibody. The detection antibody was a sulfo-tagged rabbit polyclonal C-terminal TDP-43 antibody (Proteintech, 12892-1-AP, 2 µg/ml). Response values corresponding to the intensity of emitted light upon electrochemical stimulation of the assay plate using the MSD QUICKPLEX SQ120 were acquired and background corrected using the average response from buffer only.