C9ORF72-derived poly-GA DPRs undergo endocytic uptake in iAstrocytes and spread to motor neurons

Cell culture

1321N1 astrocytoma cells were cultured in DMEM (Sigma-Aldrich) supplemented with 10% FBS (Gibco) and 5 U ml⁻¹ Penstrep (Lonza). Hb9-GFP mouse stem cells were cultured as described (Wichterle et al, 2002) and differentiated into motor neurons with 2 μM retinoic acid (Sigma-Aldrich) and 1 μM Smoothened Agonist (SAG) (Millipore) for 5 d. Embryoid bodies were then dissociated with papain. All cells were maintained in a 37°C incubator with 5% CO₂.

Conversion of skin fibroblasts to iNPC

Skin fibroblasts from one healthy control (see Table 1) were reprogrammed as previously described (Meyer et al, 2014). Briefly, 10⁴ fibroblasts were grown in one well of a six-well plate. Day 1 post-seeding the cells were transduced with retroviral vectors containing Oct 3/4, Sox 2, Klf 4, and c-Myc. Following 1 d of recovery in fibroblast medium, DMEM (Gibco), and 10% FBS (Life Science Production), the cells were washed 1× with PBS, and the culture medium was changed to Neural Progenitor Cell (NPC) conversion medium comprising of DMEM/F12 (1:1) GlutaMax (Gibco), 1% N2 (Gibco), 1% B27 (Gibco), 20 ng/ml FGF2 (PeproTech), 20 ng/ml EGF (PeproTech), and 5 ng/ml heparin (Sigma-Aldrich). As the cell morphology changes and cells develop a sphere-like form, they can be expanded into individual wells of a six-well plate. Once an iNPC culture is established, the medium is switched to NPC proliferation medium consisting of DMEM/F12 (1:1) GlutaMax, 1% N2, 1% B27, and 40 ng/ml FGF2.

iAstrocyte differentiation and co-culture system

iAstrocytes were yielded as previously described (Meyer et al, 2014; Hautbergue et al, 2017). Briefly, iNPCs were switched to astrocyte proliferation medium, which includes DMEM (Thermo Fisher Scientific), 10% FBS (Life science production) and 0.2% N2 (Gibco). Cells were grown in 10 cm dishes coated with fibronectin for 7 d unless otherwise stated. For the co-culture system, we treated iAstrocytes with 1 µM DPRs for 24 h, then plated Hb9-GFP mouse motor neurons on top of the astrocyte layer and kept this co-culture system for 48 h before fixation and confocal imaging.

Primary mouse cortical neurons

Primary cortical neurons were produced from E15.5 embryos of wild-type C57BL/6 mice. Brains were harvested and hemispheres were divided. In HBSS−/− medium, meninges and midbrain were removed to isolate the cortical tissue, which was then incubated with trypsin (Gibco) for cell dissociation. Single-cell suspension was obtained by mechanical pipetting in appropriate trituration solution (HBSS+/+ with 1% Albumax, 25 mg trypsin inhibitor, 10 mg/ml DNAse stock). Finally, cortical neurons were resuspended in Neurobasal medium (Thermo Fisher Scientific) with B27 (Gibco), 1% Pen/Strep (Thermo Fisher Scientific), and 1% glutamine (Lonza) and seeded on Xona silicon device (#RD450) coupled with a 35-mm dish previously coated with poly-D-lysine (Sigma-Aldrich). Cells were maintained in a 37°C incubator with 5% CO₂, changing the medium every 2 d. Neurotrophic factors (2 ng/ml Brain-derived neurotrophic factor [BDNF], 2 ng/ml Glial cell line-derived neurotrophic factor [GDNF]) were added into the medium to favour the correct directionality of axonal growth through the microgrooves. After 12 d in culture, the cells were stained with LysoTracker Green (Thermo Fisher Scientific) and live-imaging was performed with Airyscan microscopy (LSM 880; Zeiss) at 1.5 Hz for 2 min (561 and 488 channels; 63× 1.4 NA oil immersion lens).

Dipeptide repeat proteins cloning, purification, and labelling

The V5-tagged 34-GA repeat was obtained using an expandable cloning strategy with Age1 and Mre1 as compatible enzymes (Mcintyre et al, 2008). A “start acceptor” pCi-Neo vector (Promega) was first constructed by cloning a V5-3xGly/Ala insert into the Xho1/Not1 sites (ctc gag gcc acc atg ggc aaa ccg att ccg aac ccg ctg ctg ggc ctg ctg gat agc acc ggt gca ggt gct ggc gcc ggc gga tcc gaa ttc tag ccg cgg ccg c) and a “start donor” vector with a 14xGly/Ala insert (ctc gag acc ggt gca ggt gct gga gct ggt gca ggt gct gga gca ggt gca ggt gct gga gct ggt gca ggt gct gga gca ggt gct ggc gcc ggc gga tcc gaa ttc ccg cgg ccg c) in the Xho1/Not1 sites. These vectors (“start acceptor” and “start donor”) were then used to propagate the GA repeats to construct 34-GA repeats. The plasmid pAG416-Gal, which encodes for the PAx50 dipeptide repeats, was a gift from Aaron Gitler (plasmid #84902; Addgene; http://n2t.net/addgene:84902) (Jovičić et al, 2015). DNA sequences encoding V5-tag followed by 34 repeats of GA or encoding FLAG-tag followed by 50 repeats of PA were subcloned in a bacterial expression vector containing an N-terminal 6xHis-Tag and a TEV protease cleavage site (pETM-11 vector, EMBL). The pETM-11 vector was a gift from Frank Schulz (plasmid #108943; Addgene; http://n2t.net/addgene:108943) (Dirkmann et al, 2018). In particular, the V5-tagged 34-GA construct was subcloned in the pETM-11 vector using NcoI/NotI restriction sites; whereas the FLAG-tagged 50-PA repeat construct was subcloned in the pETM-11 vector using NcoI/XhoI sites. An additional pCI-Neo vector with DNA sequences encoding V5-tag followed by 24 repeats of GP was subcloned in the pETM-11 vector using NcoI/NotI restriction sites. Transformation of bacterial cells (E. coli BL21) with plasmids containing these repeat constructs (V5-tagged 34-GA; FLAG-tagged 50-PA; V5-tagged 24-GP) was performed. Thus, proteins were expressed in E. coli BL21 and purified on a 5-ml Talon column (Clontech) loaded with cobalt. The proteins were eluted with a linear gradient of 12 ml from buffer A (20 mM Tris, pH 7.5, 250 mN NaCl, 5 mM imidazole, 1 mM β-mercaptoethanol, and glycerol 10% PMSF 0.1 mM) to buffer B (20 mM Tris, pH 7.5, 250 mM NaCl, 250 mM imidazole, 1 mM β-mercaptoethanol, and glycerol 10% PMSF 0.1 mM). Eluted fractions were analysed by SDS–PAGE, and proteins were quantified spectrophotometrically using a molar extinction coefficient of ε_GA = 2,980 M⁻¹ cm⁻¹ and ε_PA = 4,470 M⁻¹ cm⁻¹. Proteins were assembled into fibrils at 4°C without shaking for ≤14 d. Proteins were labelled with Atto-550 (Atto:DPR 5:1) or Atto-647N (Atto:DPR 2:1) dyes (#AD550-35, #AD647-35; Atto-Tec). Unreacted NHS-dye and non-fibrillar polypeptide was removed by centrifugation (100,000g, 30 min, 4°C). Labelled assemblies were fragmented by sonication for 5 min in 2-ml Eppendorf tubes in a Vial Tweeter powered by an ultrasonic processor UIS250v (250 W, 2.4 kHz; Hielscher Ultrasonic) to generate fibrillar particles that are suitable for endocytosis (with an average size of 45–55 nm), flash-frozen in liquid nitrogen and stored at −80°C. Immunoreactivity and fluorescent labelling of the generated GA/PA-repeat recombinant proteins were confirmed by resolving the samples with dot-blotting or protein gel electrophoresis (Fig S2).

The morphology of DPR assemblies was assessed by Transmission Electron Microscopy in a Jeol 1400 microscope before and after fragmentation following adsorption onto carbon-coated 200 mesh grids and negative staining with 1% uranyl acetate. The images were recorded with a Gatan Orius CCD camera (Gatan). The β-sheet amyloid component of fibrillar poly-GA assemblies was assessed and confirmed by Fourier-transform infrared spectroscopy as described (Brasseur et al, 2020). Importantly, before the addition to the medium, fibrillar DPRs were sonicated for 5 min at 80% amplitude with a pulse cycle of 5 s on and 2 s off (MSE Soniprep 150); this procedure is required to disperse the aggregated β-sheet assemblies.

Coalescence measurements

For phase-separation experiments, soluble GA and PA DPRs (stock concentration = 100 μM) were diluted in ddH₂O to either 20, 10 or 1 μM. After vortexing, the mixture was pipetted onto glass-bottom slides (Ibidi), and assembly formation was monitored over time. Images were acquired immediately, as well as after 2 and 24 h using a Leica SP5 confocal microscope with a 63× 1.4 NA oil immersion objective, 561 channel. To capture fine details for 3D rendering, assemblies were imaged with a z-stack following Nyquist sampling for optimized pixel density. Huygens Professional version 19.10 (Scientific Volume Imaging, http://svi.nl) was used to deconvolve z-stack data using the CMLE algorithm (with SNR:10 and 40 iterations) and subsequently for 3D-volume and surface rendering thus generating Videos 1 and 2. The 3D rendering for Videos 3 and 4 was performed using the software Imaris v7.7.2 (Bitplane); no deconvolution was applied here. Number, circularity and size of DPR assemblies were quantified with FIJI (Schindelin et al, 2012) and plotted using GraphPad Prism 8. The FIJI plug-in Trainable Weka Segmentation (Arganda-Carreras et al, 2017) was used to finely measure the circularity of liquid droplets in heterogeneous PA assemblies (20 μM, 24 h) (Fig S3D and E).

MSD(Δt) analysis

Human iAstrocytes were exposed to 1 µM ATTO550-labelled Poly-GA DPR fibrils for 24 h. Cells were then washed several times with PBS and, while keeping them in 5% CO₂ and 37°C, z-stack live-imaging was performed by taking 60 frames at ∼1 frame/s rate (1 frame = 1 full z-stack) at ∼190 nm lateral resolution (Zeiss LSM880, airyscan mode). Single DPR aggregates were then detected as single particles and analysed by the open-source FIJI-plug-in TrackMate (Tinevez et al, 2017) using difference of Gaussians (DoG) detection (estimated blob diameter = 0.8 µm; threshold = 100) and Simple LAP tracker (linking max distance = 1 µm; gap-closing = 1 µm; max frame gap = 5). Video 5 was produced in 3D-volume rendering mode using Arivis Vision 4D software (Arivis AG) after complete image stack deconvolution was performed for each frame with Huygens Professional version 19.10 (CMLE algorithm, SNR:20, 40 iterations). The xy data for each tracked object were analysed to determine mean-squared displacements (MSD) as a function of time-step, Δt. For untethered non-interacting objects moving freely within the medium, MSD(Δt) is expected to evolve linearly with a gradient equal to 4D, where D is the Brownian translational diffusion coefficient. If the object also experiences ballistic motion (i.e., constant translational velocity magnitude and direction), which could approximately describe microtubule transport (van den Heuvel et al, 2007), the MSD(Δt) becomes quadratic and is represented by the equation: MSD = 4DΔt + v²Δt², where v is the average ballistic velocity (Dunderdale et al, 2012). Consequently, MSD(Δt) data were fitted to this expression to identify any trajectories that showed non-zero values for v, and so possibly reflect microtubule directed motion. For this analysis, the time range fitted was 20 s, and only trajectories greater than 30 s in length were analysed (380 separate trajectories met this criterion).

For the MSD(Δt) analysis on DPR trajectories in primary mouse cortical neurons (Fig S6A and B), we first used TrackMate for producing DPR tracks and then we implemented the MATLAB class @msdanalyzer written by Jean-Yves Tinevez (https://github.com/tinevez/msdanalyzer GitHub), already used in a previous study (Tarantino et al, 2014) and explained in its details (Miura & Sladoje, 2020). MSD plots and Log–Log fit plots were produced with MATLAB R2018b using the aforementioned class. A detailed explanation of how the analysis was performed (from the generation of TrackMate DPR tracks to the production of graphs after MSD analysis) can be found on GitHub at the following link: https://github.com/paoloM1990/Guide-to-TrackMate-Matlab-MSD.

Flow cytometry

1321N1 human astrocytoma cells (1.2 × 10⁶/sample) were washed six times in PBS (to eliminate any remaining DPRs in the media), trypsinized, and then resuspended in 500 μl of PBS. Suspended cells were then analysed by using an LSRII flow cytometer (BD Bioscience) with excitation at 488 nm and BD FACSDiva software (version 8.0.1; BD Bioscience) to excite the ATTO550 fluorophore. Detection of ATTO550+ signal was set at 610/20 voltage. Cells not treated with DPRs acted as control producing the gating to discriminate between the ATTO550+ and the ATTO550− cells.

Immunocytochemistry

After the addition of the DPRs, all the single-cell cultures were washed five to six times with PBS and fixed with 4% PFA for 30 min at room temperature. After fixation, cells were washed two times with PBS, permeabilized for 10 min with 0.1% Triton X-100:PBS and additionally washed twice with PBS. Subsequently, cells were incubated with the blocking agent 3% BSA for 30 min and then incubated overnight at 4°C with primary antibodies (in 3% BSA). Cells were then washed 3× with PBS and incubated for 1 h with the corresponding Alexa Fluor secondary antibodies (Thermo Fisher Scientific) at 1:1,000 dilution (in 3% BSA) and with Hoechst when needed. Cells were washed 3× with PBS, and coverslips were mounted onto glass slides using Fluoromount aqueous mounting medium (Sigma-Aldrich). In the case of imaging with Opera Phenix high-throughput system (PerkinElmer), fixed cells were imaged directly in PBS on a 96-well optical-bottom plate (#165305; Thermo Fisher Scientific), thus coverslipping was not required.

Primary antibodies used were: chicken anti-imentin (1:4,000, #AB5733; Millipore), mouse anti–α-tubulin (1:1,000, #T9026; Sigma-Aldrich), rabbit anti-CD44 (1:1,000, #ab157107; Abcam), mouse anti-Cathepsin B (1:500, #sc-365558; Santa Cruz Biotechnology), mouse anti-Cathepsin D (1:1,000, #sc-377299; Santa Cruz Biotechnology), mouse anti-Cathepsin L (1:1,000, #sc-390367; Santa Cruz Biotechnology), mouse anti-Galectin 3 (1:100, #sc-374253; Santa Cruz Biotechnology) and mouse anti-LAMP1 (1:25, #ab25630; Abcam).

Immunoblotting

To collect protein extracts, iAstrocytes were washed six times in PBS (to eliminate any remaining DPRs in the medium) and lysed in RIPA buffer (20 mM Tris–HCl, pH 7.5, 137 mM NaCl, 10% glycerol, 1% Triton X-100, 0.5% sodium deoxycholate, 2 mM EDTA, and 0.1% SDS, supplemented with protease inhibitor cocktail; Sigma-Aldrich) on ice for 20 min. The protein extracts were then collected in the supernatant and the concentration of each protein extract was estimated using a BCA assay (Pierce). Equal quantities of protein were mixed with 4× loading buffer (0.4 M sodium phosphate, pH 7.5, 8% SDS, 40% glycerol, 10% 2-mercaptoethanol, and 0.05% bromophenol blue), heated to 95°C for 5 min, and processed for either dot-blot or Western blot. Nitrocellulose membranes (0.22 µm pores) were blocked in 1× TBS with 0.05% Tween (1× TBST) with 5% wt/vol non-fat dry milk for 1 h, and then incubated with primary antibodies in 1× TBST 5% wt/vol non-fat dry milk at either room temperature for 2 h or 4°C overnight. Primary antibodies used were: anti-V5 (1:1,000, #13202S; Cell Signaling Technology), anti-α-tubulin (1:3,000, #T9026; Sigma-Aldrich), anti-GA repeat (1:1,000, #24492-1-AP; ProteinTech), anti-AP repeat (1:1,000, #24493-1-AP; ProteinTech). Membranes were then washed three times for 5 min with 1× TBST and incubated with either an anti-mouse IgG-HRP–conjugate (1:5,000, Cat. no. 172-1011; Bio-Rad) or an anti-rabbit IgG-HRP-conjugate (1:5,000, Cat. no. 12-348; Millipore). ECL substrate was then added to the membrane to enable detection, and nonsaturated images were acquired using a G:BOX EF machine (Syngene) and GeneSys software (Syngene).

Perturbation of endocytosis

Exposure of cells to low-temperature conditions is a commonly used method for nonspecific inhibition of endocytosis. Healthy control iAstrocytes were primed with 30 min of exposure to 4°C, and then the DPR assemblies (diluted in ice-cold DMEM) were delivered to the cells and incubated at 4°C for an additional time of 1 h. The same procedure was applied for Alexa647-labelled transferrin, an established marker of clathrin-coated pit endocytosis (Ehrlich et al, 2004). In parallel conditions, cells exposed to DPR assemblies or transferrin were incubated with DMEM at 37°C for 1 h. For the dynasore experiment, following the established protocol (Kirchhausen et al, 2008), cells were primed for 30 min with Dynasore (or 0.2% DMSO) and incubated at 37°C in serum-free medium before addition of transferrin or poly-GA fibrils for 1 h. Cells were subsequently fixed in Glyoxal solution (pH = 5) for 30 min (Richter et al, 2018) and imaged with confocal microscopy (LSM 880; Zeiss, Airyscan mode) from the plane of sharp focus. From the images, using FIJI and creating a macro (https://github.com/paoloM1990/Quantification-of-cell-fluorescent-intensity), the Mean grey values of Atto550 or Alexa647 whole-cell signals were calculated. In the case of the 37–4°C endocytosis experiment, mean grey values were log-transformed (log₁₀) only for better graph visualization.

Lysosomal assays

Lysosomal in situ enzyme activity was measured by using lysosome-specific self-quenched substrate (Cat. no. ab234622; Abcam) at manufacturers recommended dosage. In brief, primary mouse cortical neurons were exposed to poly-GA DPRs for 24 h (or were left untreated), and lysosome-specific self-quenched substrate was added during the final 1 h of the 24-h period. Cells were then fixed with 4% PFA at room temperature for 15 min before being imaged with a Zeiss LSM 880 confocal microscope on glass-bottom slides (Ibidi). FIJI was used to analyse the images, and the mean fluorescence intensity of the substrate was quantified per neuronal cell.

Lysosomal staining with the fluorescent acidotropic probe, LysoSensor Green DND-189, was performed according to the manufacturer’s recommendations (#L7535; Thermo Fisher Scientific). Briefly, HeLa cells on a 96-Well Optical-Bottom Plate (#165305; Thermo Fisher Scientific) were exposed to poly-GA DPRs for 24 h (or were left untreated), and LysoSensor Green was added during the final 1 h of the 24-h period (1 μM final concentration). After incubation, live cells were transferred to the PerkinElmer Opera Phenix high-throughput system for imaging (40× 1.1 NA lens). Using the Columbus Image Analysis System, the mean fluorescence intensity of LysoSensor Green was quantified per cell (PerkinElmer).

mRNA isolation and quantitative real time PCR

Primary cortical neurons were produced from E15.5 embryos of wild-type C57BL/6 mice. Total RNA from primary mouse cortical neurons (after 10 d in culture) exposed to poly-GA oligomers or poly-GA fibrils (or untreated) for 24 h was isolated using RNeasy Mini Kit (Cat. no. 74104; QIAGEN), according to the manufacturer’s manual. During RNA extraction, treatment with DNase I was applied to get rid of contaminating DNA. RT-qPCR was carried out using the QuantiFast SYBR Green RT-PCR Kit (QIAGEN). Briefly, a 10-μl volume reaction was set up by using 2 μl total RNA (diluted to a concentration of 10 ng/μl in nuclease-free H₂O), 5 μl 2× QuantiFast SYBR Green RT-PCR Master Mix, 1 µM forward primer, 1 µM reverse primer, 0.1 μl QuantiFast RT mix and nuclease-free H₂O. Following an initial reverse transcription step at 50°C for 10 min and a 5 min denaturation step at 95°C, the cDNA was amplified by 39 cycles of 95°C for 10 s followed by a combined annealing/extension step at 60°C for 30 s, and subsequent melt curve analysis (to ensure primer specificity), with data collected over a temperature range of 65–95°C in 0.5°C increments. RT-qPCR was performed on a Bio-Rad C1000 Touch Thermal Cycler. Bio-Rad CFX Manager software was used to analyse signal intensity and relative gene expression values were determined using the ΔΔCt method, with GAPDH RNA used as a reference gene. The RT-qPCR product was then visualized on a 2% agarose gel, after loading 10 μl of product along with 2 μl of 6× gel loading dye. After electrophoresis (at 120 V for 40 min), the gel visualization showed RT-qPCR products matching the expected bp size (data not shown). For the primer sequences used in this study, see Table 2.

Colour deconvolution for differentiating two lysosomal populations

We developed custom scripts in MATLAB to produce colour deconvolution algorithms that separated “non-colocalized lysosomes” (NCLs) and “colocalized lysosomes” (CLs). NCLs corresponded to the signal of LysoTracker Green devoid of any overlapping with ATTO550-DPRs signal. CLs, instead, corresponded to the merged signal generated by LysoTracker Green–ATTO550-DPR colocalization. Briefly, eight-bit RGB time-lapse images were split into individual frames and a fuzzy colour detection algorithm was applied to identify and isolate regions of interest. Resulting images were binarised by applying a global Otsu threshold, and then de-noised using a median filter.

Release of DPRs in the CM

To detect the cell release of ATTO550 DPRs into the CM, we have initially added 1 µM of DPRs to the culture medium for 24 h. After DPR uptake, cells were washed at least five times with PBS to remove remaining assemblies in the medium and then incubated for 24 h with PhenolRed-free FBS-free DMEM. This CM was harvested in tubes which were subsequently centrifuged at 200g for 4 min to remove any remaining debris and dead cells. Finally, the HTS microplate reader PHERAstar FSX (BMG LABTECH) was used to measure ATTO550 fluorescence intensity (thus relative DPR concentration) in the CM on a 96-well plate. The optical module used for the fluorescence intensity measurement of each well was set to 540–20 nm (excitation light) and 590–20 nm (emission light), covering the whole area of the well. Importantly, we tested for linear dependence of fluorescence on concentration to evaluate potential inner filter effect (Fonin et al, 2014) by using nine known dilutions of purified DPRs in PhenolRed-free FBS-free DMEM; the plotted fluorescence values originated a standard curve with R² > 0.95.

ELISA

APOJ (also known as clusterin) was measured in CM from either untreated or treated iAstrocytes for 24 h with various recombinant DPRs (poly-GA₃₄ fibrils, poly-GA₃₄ oligomers, poly-PA₅₀ oligomers, and poly-GP₂₄ oligomers). The CM was collected in tubes and centrifuged at 500g for 4 min to remove any remaining debris and dead cells. After that, samples were diluted 1:10 and analysed within the range of the standard curve. APOJ protein levels were measured using a human APOJ ELISA kit (#KE00110; ProteinTech) following the manufacturer’s instructions. The HTS microplate reader PHERAstar FSX (BMG LABTECH) was used to measure the absorbance of APOJ standards at 450 nm with the correction wavelength set to 630 nm. Regression analysis using the Four-parameter logistic curve-fit (4-PL) method was used to determine the best-fit standard curve (Fig S8A).

dSTORM imaging

Quantification and statistical analysis

All data are presented as means ± SEM or means ± SD, where indicated. On normally distributed data, statistical differences were analysed using unpaired two-tailed t test (with Welch’s correction, when SDs were not equal) for pairwise comparisons or one-way ANOVA (with Tukey’s correction) for comparing groups of more than two. On non-normally distributed data, the non-parametric Kolmogorov–Smirnov test or Kruskal–Wallis test (with Dunn’s multiple comparisons) were used for pairwise comparisons or for comparing multiple groups, respectively. Normal distribution was tested with Shapiro–Wilk test and Q–Q plot. P < 0.05 was considered statistically significant. All graphs and tests were generated using GraphPad Prism 8.

Ethics statement

Ethical approval to use iPSCs or iNPCs is in place for this project (REC approval 12/YH/0330). This experimental work involves studies on genetically modified vectors already approved by the Health and Safety Executive (Azzouz_GMO_2006-07).