Nature Neuroscience

Human ALS/FTD brain organoid slice cultures display distinct early astrocyte and targetable neuronal pathology

hiPSC culture

hiPSC lines were obtained from three suppliers (Supplementary Table 1). Cell lines were adapted and cultured in StemFlex medium (Thermo Fisher Scientific, A3349401) on plates coated with Geltrex (Thermo Fisher Scientific, A1413302). Cells were fed daily and passaged using 0.5 mM EDTA (Thermo Fisher Scientific, 15575020) when confluency reached 60–70%. For banking, cells were frozen in StemFlex medium containing 10% dimethylsulfoxide (Sigma Aldrich, D2650).

Generation of cerebral organoids and ALI-COs

For the generation of hiPSC-derived COs cultured at the ALI, we modified a previously published method5 utilizing a STEMdiff Cerebral Organoid kit (StemCell Technologies, 08570). Briefly, 18,000 cells were plated in the presence of poly(lactide-co-glycolide) copolymer microfilaments to achieve improved cortical development58. Initially, media changes were performed at days 3, 5 and 7, then every 3–4 days. From 35 DIV onwards, Matrigel (Corning, 354234) was added in 1:50 dilution to achieve a polarized cortical plate formation. Between 50 and 80 DIV, 300-μm thick ALI-CO cultures were prepared for subsequent long-term cultures on Millicell-CM (Merck Millipore, PICM0RG50) inserts and were fed daily with slice medium containing neurobasal medium (ThermoFisher Scientific, 21103049) supplemented with 1× B27 supplement (ThermoFisher Scientific, 17504044), 0.45% (w/v) glucose (Sigma-Aldrich, G8769), 1× Glutamax (ThermoFisher Scientific, 35050038) and 1% antibiotic–antimycotic (ThermoFisher Scientific, 15240062).

Organoid cell dissociation

For organoid cell dissociation, slices were transferred into a 10-cm2 dish containing 1× dPBS (Sigma Aldrich, D8537). Washed slices were then placed into a gentleMACS C tube (Miltenyi, 130-093-237) containing 2 ml of papain solution (20 units per ml; Worthington, PAP2) and ran on a gentleMACS Octo dissociator (Miltenyi) using the default ABDK program. After dissociation, the cell suspension was triturated, diluted with dPBS containing 0.5 mg ml−1 DNAse (Sigma Aldrich, 11284932001) and then spun down at 300 × g for 5 min. The cell pellet was resuspended and filtered through a 70-μm strainer (Miltenyi, 130-098-462) to remove any remaining aggregates before centrifugation again under the same conditions. For scRNA-seq samples, the cell pellet was resuspended in dPBS (316 cells per μl) containing 0.04% BSA (Sigma, A9418), and the suspension was kept on ice for 30 min until being processed. For the cell culture experiments, cells were resuspended in N2B27 medium containing 10 μm Y-27632 (Tocris, 1254/10) and plated on coverslips pre-coated with polyethylenimine (Sigma Aldrich, P3143) and Geltrex.

ER stress and rescue assays

For ER stress induction, ALI-COs at 220 DIV were treated with 50 μM SA (Sigma-Aldrich, 1062771000) or vehicle (equal volume of slice medium). For short-term rescue experiments, slices were either pretreated with 5 μM GSK (R&D Systems, 5107/10) or vehicle overnight in inserts, followed by their immersion in slice medium in 24-well plates for an additional 4 h of treatment with GSK or vehicle before SA or vehicle administration. For the long-term rescue experiments, control and C9 CO slices at 200–240 DIV were kept on inserts but were immersed in medium and fed daily with either vehicle only or supplemented with 5 μM GSK for 14 days. Each untreated–treated pair of CO slices consisted of two adjacent slices derived from the same whole organoid, and at least three independent CO slice pairs (derived from different whole organoids) were used for assessing GSK-mediated effects in each experiment.

Real-time cell vulnerability assay

Six-well plates (Appleton Woods, Corning, CC010) were coated with 1% Geltrex (Thermo Fisher Scientific, A1413302) in DMEM/F-12 (Fisher Scientific, Gibco, 11514436). hiPSCs were seeded in StemMACS iPS-Brew XF (Miltenyi Biotec, 130-104-368) medium at a density of 1 × 105 cells per well. After 24 h, the survival assay was performed in IncuCyte chambers at 37 °C and 5% CO2, and cells were either left untreated or treated daily with 1.25 nM, 2.5 nM, 5 nM or 10 nM topotecan (Apexbio, B2296). Images were acquired every 6 h over a duration of 5 days using a ×10 magnification objective of the Incucyte S3 Live-Cell Analysis system (Sartorius, 4647). For assessing cell viability, cell confluency was determined as a cell body cluster area by phase microscopy using the IncuCyte NeuroTrack software59.

Alkaline comet assay

The alkaline comet assay was performed to compare DNA accumulation and repair in control and C9 ALS/FTD hiPSC lines60. Cells were seeded overnight before topotecan (10 μM for 1 h) treatment in the presence or absence of an ATMi (Selleck Chemicals, AZD0156, S8375; 30 nM for 1 h). Half of the plates were trypsinized (0.25%, Sigma-Aldrich, T4049) for the assessment of DNA damage load (damage; D). The rest were washed in PBS and further incubated for 6 h in StemMACS iPS-Brew XF (Miltenyi Biotec, 130-104-368) without topotecan or ATMi before trypsinization to allow for DNA repair (recovery; R). The cells were then resuspended in 1× PBS (Mg/Ca-free) (Sigma-Aldrich, D8537) at 2 × 105 cells per ml concentration. Cell suspension (75 μl) was mixed in 500 μl LMAgarose (Trevigen, 4250-050-02; melted for 5 min at 100 °C and kept at 37 °C), then 70 μl of the mixture was pipetted on preheated (37 °C) 1% agarose-coated glass slides. The cell-containing droplets were covered with a coverslip and kept in the dark for 30 min at 4 °C before the application of the CometAssay lysis solution (Trevigen, 4250-050-01) followed by the alkaline unwinding solution (200 mM NaOH, 1 mM EDTA, pH > 13) for 1 h at each step at 4 °C in the dark. The slides were subjected to electrophoresis at 35 V for 10 min in the alkaline electrophoresis solution (200 mM NaOH, 1 mM EDTA, pH > 13). Following fixation with 70% ethanol and drying at 37 °C, slides were stained with SYBR green I (Invitrogen, in 10 mM Tris-HCl, pH 7.5, 1 mM EDTA, pH 8.0). The tail percentage was measured using the OpenComet plugin for ImageJ (SRC V1 version). For each condition, the comet tail percentage was measured (>50 cells).

Western blotting

Cell lysates for protein samples were obtained from ALI-CO slices at 150, 200, 220 and 240 DIV using RIPA lysis buffer (Sigma-Aldrich, R0278) plus protein and phosphatase inhibitors (Thermo Fisher Scientific, 31462, A32957), and standard immunoblotting protocols were used. Briefly, 18 μg of protein was resolved by SDS–PAGE then transferred onto polyvinylidenedifluoride membranes before overnight incubation with primary antibodies (Supplementary Table 2), except for the directly conjugated β-actin antibody, for which 1-h incubation was applied. Species-specific horseradish-peroxidase-conjugated secondary antibodies were applied for 1 h at room temperature at 1:10,000 dilution (Supplementary Table 2) before signal detection with the enhanced chemiluminescence system (GE Healthcare, RPN2232). Standard quantitative WB analysis was performed in ImageJ, and band density levels were expressed as fold-changes to controls after normalization to β-actin and controls within the same blot.


ALI-COs and whole organoids at 30 and 75 DIV were fixed in 4% paraformaldehyde (PFA) for 45 min, 2 or 4 h, respectively, at room temperature. Fixed samples were prepared as frozen blocks for cryostat sectioning, and 12-μm thick frozen sections were immunostained with antibodies (Supplementary Table 2) using our published protocols61. For immunofluorescence-based detection of DDR in hiPSCs, cells were seeded on 1% Geltrex-coated coverslips 24 h before treatment (Geltrex, Thermo Fisher Scientific, A1413302). After treatment, cells were washed in PBS followed by fixation in 2% PFA (Alfa Aesar) for 15 min at room temperature, then washed again three times in PBS. hiPSCs and ALI-CO cryostat sections were then permeabilized with 0.5% Triton X-100 (Sigma-Aldrich, T8787) for 10 min, before blocking at room temperature for 1 h in 5% BSA (w/v; Sigma, A9647), followed by primary antibody (in 5% BSA PBST) incubation overnight at 4 °C. Cells were incubated with secondary antibody (in 5% BSA PBST) for 1 h at room temperature and were counterstained with 4,6-diamidino-2-phenylindole (DAPI; 0.2 mg ml−1) and then were mounted using Fluoromount-G (Affymetrix eBioscience, 15586276). For immunolabeling-enabled 3D imaging, organoids were solvent-cleared using the published iDISCO method62.

Image acquisition and processing

Images were acquired using a confocal microscope (Leica TCS SPE, z-stack step: 0.5–1 μm, ×20–63 objective; 1,024 × 1,024 or 2,024 × 2,024 pixel images), an automated confocal slide scanner (Pannoramic Confocal, 3DHISTECH, 9,216 × 11,520–15,616 × 15,872 pixel ranges for images), a confocal spinning disk microscope (Andor Dragonfly 302, ×20 objective, 1,177 × 1,177 pixel images) or a fluorescence microscope (Leica DM6000, ×10 objective, 1,392 × 1,040 pixel images). Camera exposure and gain were kept the same while collecting images for each experiment. Unmodified images were used for manual analyses. For unbiased semi-automated analyses of the cell nucleus and immunoreactive SYT1+, HOMER+, P62+, LC3+ and 53BP1+ particle counts, automated batch normalization with background fluorescence signal subtraction was uniformly performed on images using established image analyzer software plugins with settings specified in the relevant method sections. Automation was verified by manual cell and particle counts. For illustration purposes, the recommended guidelines were followed. Representative images were only minimally and uniformly processed in ImageJ (v.2.0.0 Fiji) or in Adobe Photoshop without affecting data presentation. This included changes in exposure and/or contrast parameters (= 0.3) when clear views were obscured in merged images by interference between strong DAPI/cytoplasmic GFAP staining and other immunoreactive objects (SOX2, SOX9, SATB2, CTIP2, P62 and LC3) on coverslips and histological slides. Cyan or magenta pseudocolors were rendered to images in ImageJ for immunolabelling of SOX9+, CTIP2+ cell nuclei (magenta or gray) for multicolor visualization. For WB densitometry, chemiluminescence on WB membranes was detected by the Alliance 4.7 CCD image system (UVITEC), and the original membrane images were used. For focused illustration, the images were cropped, leaving a minimum of six band width in all lanes with corresponding β-actin loading controls. For figure assembly, images were embedded in Adobe Illustrator. For schematic illustrations, parts of drawings from the Motifolio drawing toolkit were utilized ( for figure preparation.

Synapse density analysis

For synapse analysis of ALI-COs, z-stacks of images were taken from 3 cortical plate areas in 11–17 sections per each ALI-CO by confocal microscopy, using the same parameters set for the control samples (×63 lens, ×1.5 digital zoom, 1,024 × 1,024 image resolution, phase correction x value = −33.4). This included four to six ALI-COs for each hiPSC line. For z-stack images, standard batch image-normalization was performed in ImageJ by applying a 0–255 range histogram stretch (no pixel value alteration) and the “subtract background” function63 using the same parameters for the entire dataset. Synapse quantification was carried out using the open-source CellProfiler software64 (v.3.1.9; with optimization of a publicly available pipeline ( Using the MaskObjects function, masks for presynaptic and postsynaptic particle recognition reflected close proximity between SYT1 and HOMER1, respectively (synaptic puncta size threshold = 6–15 pixels). From each z-stack image, 3–4 confocal planes at 1 μm apart were analyzed to prevent double counts for synapses (96–178 images per group). Synapse densities were expressed as the number of colocalizing masks over areas defined by MAP2+ dendrites identified by the “tubeness” function in CellProfiler.

P62 and autophagy marker detection

For the analysis of P62+ and LC3+ particles in astroglia, ALI-CO slices at 190–220 DIV were dissociated as described above. Cells (20,000) were plated on nitric-acid-treated coverslips, coated overnight with polyethyleneimine and subsequently with Geltrex for 4 h at 37 °C. After 20 days, cultures were treated with the autophagy inhibitor chloroquine (Autophagy Assay kit, Abcam, 139484) or vehicle for 18 h at 30 μM concentration. Following immunostaining of dissociated cells, z-stack images were taken from 3–5 different areas per coverslip by confocal microscopy (×63 lens, ×1.5 digital zoom, 1,024 × 1,024 image resolution, phase correction x value = −33.4) using the same settings. Principles of z-stack image sampling and the settings for batch image normalization in ImageJ were the same as described for synapse analysis, and immunoreactive particles were quantified using CellProfiler. Using the MaskObjects function, particle recognition was carried out within GFAP+ areas. The counts and sizes of individual and overlapping P62+ and LC3+ puncta were measured for comparison between control and C9 ALI-CO-derived GFAP+ astroglia. For the analysis of cell-type-specific distribution of P62+ particles in whole ALI-CO sections, immunolabeled cryostat sections were imaged using a confocal microscope or a confocal microscopy slide scanner as described above. The colocalization pipeline was used to elucidate differences in the overlap of P62+ areas with GFAP+ astroglial/MAP2+ neuronal territories in control versus C9 ALI-COs. Data are expressed as the proportion of overlapping areas or as a correlation coefficient generated by CellProfiler.

DNA damage detection

DNA damage accumulation was assessed by immunolabeling for γ-H2AX or 53BP1 (refs. 43,65). hiPSCs were seeded overnight on coverslips (Academy, NPC16/13) coated with 1% Geltrex (ThermoFisher Scientific, A1413302; diluted in DMEM/F-12, Fisher Scientific, Gibco, 11514436). After 1 h of treatment with 10 μM topotecan, a TOP1i, alone or in combination with 30 nM of an ATMi, AZD0156, coverslips were washed with PBS containing 0.1 % Tween-20 (PBST) and were fixed with 2% PFA (w/v; Alfa Aesar, 43368) in PBS for 10 min. ALI-CO slices were first incubated in a 24-well plate, floating in 1 ml slice medium for 2 h at 37 °C, then in 1 ml of medium alone or supplemented with 10 μM TOP1i alone or in combination with 30 nM AZD0156 for 1 h. Coverslips and ALI-CO cryostat sections were processed as described in the immunohistochemistry section. To verify the specificity of DNA damage accumulation, ALI-CO slices were exposed to γ-radiation (Xhtrahl, RS225M; 2GY), and subsequently DNA repair was allowed for 30 min up to 6 h. This also served for calibrations of the DNA damage analysis pipeline for C9 hiPSCs or in ALI-COs (Extended Data Fig. 5b). Confocal microscopy (Zeiss, LSM880 100) imaging parameters were set to the intensity of γ-H2AX immunoreactivity in a control ALI-CO or CO section, which were kept uniform, and stack images were analyzed using the CellProfiler software66.

scRNA-seq pipeline and analysis

The initial scRNA-seq data-analysis pipeline was generated using the CellRanger 3.1 software package according to our published protocols5. Reads were aligned to the GRCh38 human genome. CellRanger detected 85% fraction reads on average per cell and 1,600–2,100 median genes per cell. Bias from false cell discovery due to potential ambient RNA contamination/low UMI counts was eliminated in two samples using the “force cell” function in CellRanger. Data were processed in Seurat 3.1/4.0.1 and filtered based on a minimum of three cells expressing each gene. As a standard method, only those cells were included in the analysis that expressed 200–5,000 genes and in which the proportion of mitochondrial genes were below 25% out of all genes. Cells displaying a greater fraction in mitochondrial transcripts are conventionally regarded as nonviable and were discarded to avoid bias in the transcriptomic analysis, resulting in a dataset representing 148,223 cells. Clustering was performed in Seurat over 14 dimensions with a resolution of 0.4, and cell type or state identities were determined by the expression of previously defined markers. For merged representation of all samples, a subset of 75,497 cells was taken for batch correction using canonical correlation analysis67. Cell maturity analysis was achieved by independently projecting data from ALI-COs onto two different scRNA-seq datasets for fetal brains16,17 using the scmap 1.12.0 software18 (‘scmapCluster()’ threshold = 0)). For each cell type, the proportion of cells projecting to a particular actual age of the fetal brain was plotted. For trajectory reconstruction, Monocle 3 (v. was used on the entire merged dataset. Cells with a transcriptomic signature of cell stress11 were removed and 75,497 cells were processed. The top 3,000 variable features calculated by Seurat over 100 dimensions were used for the ‘preprocesscds()’ function. iRG cells marked by expression of TOP2A, SOX2, VIM and NUSAP1 were used as the root for ‘order_cells()’. DEGs were obtained for each batch independently (batch 1: C9-L1 versus H-L1/H-L2; batch 2: C9-L2 versus ISO-L2) and were only retained if significantly different (adjusted P value < 0.05) for both individual and combined comparisons. Overlap was then identified between the batch-specific list of DEGs. WGCNA was used to reveal potentially affected pathways in each cell type in the two separate batches of ALI-CO datasets using the top 3,000 variable features (WGCNA v.1.70-3). A minimum module size of 15 and a deep split of 4 were retained. The resulting modules were then projected onto the merged dataset using ‘moduleEigengenes()’. Significant differences between module eigengene values for C9 versus controls are presented in box plots. Overlap between genes from each module and DEGs calculated for the cell type was then plotted using Python. In addition, STRING ( was utilized for network interaction analysis (‘medium confidence’), and the disconnected nodes were eliminated. Finally, enrichment analysis was performed for gene sets included in the networks of highly correlated genes by GO analysis via the EnrichR platform. The –log() of adjusted P values or the false discovery rate was taken to generate clustermap plots. To determine whether C9 ALI-COs recapitulate gene expression changes seen for C9 ALS/FTD samples, the expression of cell-type-related DEGs were compared to upregulated and downregulated genes detected in human samples from patients with ALS/FTD using various publicly available databases20,21,22,23,24,25. To infer a potential functional relevance for differentially affected genes in C9 ALI-COs, TF activity analysis was performed using SCENIC v.0.9.6 (python 3.6)29, for which a standard pipeline was followed (GRNboost2, ‘ctx’, AUCell). The resulting matrix of cells and TFs were binarized using the recommended threshold calculation in R (v.4.0.3). For each cell type, the proportion of cells with ‘active’ TFs was calculated in C9 versus control datasets.

Southern blotting

DNA was extracted from the hiPSC cultures using a Qiagen DNeasy Blood and Tissue kit (69506). DNA (5 µg) was restriction digested with Bsu36I (New England Biolabs, R0524S) and ran on 0.8% agarose gels accompanied by molecular weight markers II (Roche, 11218590910) and III (Roche, 11218603910). Southern blotting was performed according to published protocols68 using a 1-kb probe (Prepared by IDT; sequence: GGGGCC) in salmon sperm DNA (ThermoFisher Scientific, 15632011) for hybridization and an anti-DIG AP antibody (Roche, 11093274910) for visualization.

Repeat-primed PCR

DNA was extracted from samples using a QIAamp DNA Mini kit (Qiagen, 51306), and the C9ORF72 locus amplified by PCR with the 6-FAM-labeled forward and reverse primers (see below) as previously described69. The PCR product was denatured and analyzed by capillary electrophoresis on an Applied Bioscience 3730XL DNA Analyzer (Thermo), and chromatographs were aligned in GeneMapper v.6. software (Thermo).




MEA recordings

ALI-COs were secured with a platinum harp onto 3D MEAs (Multi Channel Systems, MEA2100, 60–3DMEA200/12iR-Ti-gr, 60 electrodes, 12 μm in diameter, 200-μm spacing). Six-minute-long recordings of spontaneous activity at 37 °C (n = 35 ALI-COs) were acquired and exported to Matlab (MathWorks) for analysis5. The raw signal was bandpass-filtered (third-order Butterworth, 600–8,000 Hz) and spikes detected using a threshold of 3 standard deviations (s.d.) above background noise using a 1.5-ms refractory period after each spike. Correlated activity between electrodes was analyzed using the spike-time tiling coefficient (STTC)70 with a synchronicity window of 175 ms. Using graph theory, the functional connectivity is shown as the edge weights and node degree of each electrode for STTC > 0.6. The node degree distribution and binary connection matrices were compared against surrogate graphs of synthetic spike matrices for temporally randomized spike trains with equivalent spike rate distribution.

Statistics and reproducibility

The subject identifiers were blinded for the observers. Details of statistical tests and exact sample sizes are listed in Supplementary Table 3. Briefly, the sample sizes using hiPSC lines and organoids were estimated from previously performed experiments5,61. In total, 233 whole COs were used for this study, and 587 ALI-CO slices were grown deriving from 99 independent whole COs that were generated from three control and two disease lines harboring the C9ORF72 mutation. Sample allocations into groups included independent organoids, ALI-COs or immersed CO slices grown from different cell lines and/or as separate batches (independent biological replicates). ALI-COs or immersed CO slices derived from identical whole organoids were only used in separate studies or as adequate control–treatment slice pairs for each independent biological replicate per group. Studies carried out on cultures of non-differentiated hiPSCs included two disease and two control lines (one of which is a genetically corrected isogenic line) in at least three independent experiments. Experiments were repeated three times (or two times for GSK treatments), which included at least three independent biological replicates, and all had similar results. At least three independent biological replicates were used per group for all statistical analyses in biological experiments. For the anti-γ-H2Ax antibody validation studies (using irradiation; Extended Data Fig. 5b), three sections sampled from different positions within one organoid were subjected to analysis for each time point. The GraphPad software (GraphPad Prism v.7.0/8.0) was used for distribution analysis, statistical analysis and for generating graphs. When normality was not assumed or defined, nonparametric tests were used. Area under the curve graphs were generated using the integrated Prism formula without modification (baseline is considered y = 0). The specific type of statistical tests with exact n values and P values are indicated in the figures and legends, and further details are included in Supplementary Table 3. Unless stated otherwise, statistical significance was accepted at P < 0.05, and the exact P values are included in the graphs. In cases where no statistical difference was found between more than two groups, the overall analysis of variance (ANOVA) P value is presented.

Reporting Summary

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

Source link

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button