A Human iPSC-based model of TDP43 proteinopathy in MNs

(A) Schematic depicting genome editing of healthy iPSCs to knock-in GFP into the C-terminus of endogenous TARDBP, encoding TDP43. Yellow rectangle indicates the TDP43 last exon. Red vertical line indicates the STOP codon. Blue rectangle indicates the TDP43 3’UTR. The resulting TDP43-GFP iPSCs are differentiated into motor neurons (MNs) and transduced with AAVs encoding anti-GFP nanobodies. The “NES” nanobody includes a sequence for a strong nuclear export signal, which transports nuclear TDP43-GFP into the cytoplasm. The control nanobody (lacking NES) has no effect on TDP43-GFP localization.

(B) Representative images depicting TDP43 localization in two homozygous TDP43-GFP lines (E8, E5). MNs were transduced at day 20 of the differentiation process and fixed at day 40 for the immunostaining. TDP43 is expressed in the nucleus in the presence of the control nanobody and relocates to the cytoplasm in the presence of the NES nanobody. TDP43-GFP indicates signal from the anti-GFP antibody while TDP43 indicates signal from the anti-TDP43 antibody. Neurons were also stained with neurofilament-M (red) and Hoechst 33342 (blue nuclear stain). Scale bar = 50 µm.

(C) Quantification of the % of nuclear TDP43 intensity over total TDP43 intensity in the nucleus + soma in individual neurons transduced with control or NES nanobodies.

(D) Quantification of morphological defects and cleaved caspase 3 (CC3) levels in NES nanobody treated MNs versus control. Mislocalized TDP43 (NES) causes a reduction in dendritic complexity (D), soma swelling and elevation of CC3 levels, compared to neurons expressing nuclear TDP43 (control). Low-to-moderate mislocalization indicates neurons with > 40% nuclear TDP43. Severe mislocalization indicates neurons with < 40% nuclear TDP43. Measurements were normalized to data from the control nanobody condition.

Panels C, and D display data for the two homozygous TDP43-GFP lines (E8 and E5), transduced with nanobodies at day 20 and fixed for staining at day 40. N = 3 independent differentiations per clone. At least 100 neurons were included per condition.

(E) Representative images showing cytoplasmic TDP43 puncta at day 40 in homozygous TDP43-GFP MNs transduced with the NES nanobody at day 18. Scale bar = 15 µm. Images were captured with the Zeiss LSM880 Airyscan.

(F) Quantification of the percentage of NES expressing neurons displaying TDP43 puncta. The largest punctum per neuron was used for the analysis. Upper panel shows data for E8 neurons while lower panel displays data for E5 neurons. N = 3 independent differentiations per clone. At least 100 neurons were included per clone for the analysis.

* indicates p < 0.01, ** indicates p < 0.001, *** indicates p < 0.0001.

Transcriptional consequences of TDP43 mislocalization

(A) Principal component analysis of gene counts for E8 and E5 iPSC-MNs.

(B) Volcano plot displaying genes differentially expressed due to TDP43 mislocalization. Red: upregulated genes, Blue: downregulated genes, NC: No Change. Horizontal line denotes a p-value threshold of 0.01.

(C) Gene ontology enrichment analysis on the differentially expressed genes in the TDP43 model. The top five enriched pathways in upregulated and downregulated genes have been displayed.

(D E, F) Venn diagram displaying the overlap between mis-spliced genes identified due to TDP43 mislocalization in the TDP43-GFP iPSC MNs (cyan circles) with publicly available transcriptomic datasets; D: cortical neuronal nuclei displaying TDP43 depletion obtained from ALS/FTD patient tissue (GSE126542), E: TDP43 knockdown using siRNAs in healthy iPSC-MNs (GSE121569), F: TDP43 knockdown using CRISPRi in healthy iPSC-iNs (PRJEB42763). P-values were estimated using a hypergeometric distribution. Only genes with detectable expression in both datasets were used for splicing analysis.

(G) RT-qPCR validating alternative splicing changes resulting from TDP43-GFP mislocalization in iPSC-MNs. MN samples were lysed at day 30, 12 days post-transduction with AAVs. Replicates were three independent differentiations of homozygous TDP43-GFP knock-in lines, (two of E5, one of E8). CE = cryptic exon, TRC = truncated, IR = intron retention. CACNA1E and KCNQ2 displayed alternate exon usage. The exons that were included in the NES samples have been indicated.

** indicates p < 0.01. *** indicates p < 0.001.

An inducible model of TDP43 mislocalization

(A) Schematic depicting the knock-in of the nanobody (control or NES) into the human AAVS1 safe harbour locus of our E8 homozygous TDP43-GFP cell line. Nanobody expression was under the control of a doxycycline (Dox)-inducible promoter. Addition of Dox is expected to induce TDP43 mislocalization in the TDP43-GFP-NES MNs but not in the TDP43-GFP-Control (TDP43-GFP-CTRL) MNs.

(B) Immunofluorescent staining showing TDP43 localization in our TDP43-GFP-CTRL or TDP43-GFP-NES line in response to Dox. Mislocalization only occurs in the TDP43-GFP-NES line with Dox treatment, while all other conditions maintain nuclear TDP43. MNs were fixed and stained at day 35, 15 days post-Dox addition. Scale bar = 50 µm.

(C) TDP43 localization in the TDP43-GFP-NES or –CTRL cell lines at day 40, 20 days post-Dox addition. The box highlights cytoplasmic TDP43 puncta in the NES line. Scale bar = 20 µm.

(D) Western blots of total and phosphorylated TDP43 in TDP43-GFP-NES or –CTRL cell lines at day 40, 20 days post-Dox addition. Alpha-tubulin was used as a loading control.

(E) Quantification of total and phosphorylated TDP43-GFP from figure D. Total TDP43-GFP remains stable, while there is a significant increase in phosphorylated TDP43-GFP in the NES lines. Each sample was normalized to alpha-tubulin, pTDP43 samples were also normalized to total TDP43 levels. Replicates are three independent differentiations of TDP43-GFP-NES or –CTRL.

pTDP43-GFP = phosphorylated TDP43-GFP. * indicates p < 0.05. Error bars indicate SEM.

Splicing defects are one of the earliest changes downstream of TDP43 mislocalization

(A) Immunocytochemical staining of TDP43-GFP-NES MNs at 4, 8, 12, and 24 hours following 1 ug/ml doxycycline addition. TDP43 mislocalisation is observed within 8 hours. Dox was added to MN day 20. Scale bar = 25 um.

(B) RT-qPCR of alternative splicing changes at 4, 8, 12, and 24 hours following 1 ug/ml doxycycline addition. Significant cryptic transcript expression in all genes is detected at 8 hours.

Error bars show SEM. N = 3 independent differentiations of the TDP43-GFP-CTRL/-NES lines.

CE = cryptic exon, TRC = truncated, IR = intron retention. CACNA1E and KCNQ2 displayed alternate exon usage. The exons that were included in the NES samples have been indicated.

* indicates p < 0.05, ** indicates p < 0.01. *** indicates p < 0.001.

Persistent mislocalization of TDP43 in neurons post-Dox withdrawal.

(A) TDP43 localization in the E8 TDP43-GFP MNs expressing dox-inducible V5-tagged nanobodies. Control: control-V5 nanobody. NES: NES-V5 nanobody. Constant: neurons were treated with dox continuously. Removed: Dox was withdrawn five days after addition.

N = 2 for the constant + control condition. N=4 for the NES conditions. Each replicate is indicated by a coloured dot. Scale bar = 20 µm.

(B, C) Representative images of the TDP43-GFP MNs expressing NES-V5 nanobody under constant dox treatment (B) and 21 days post-dox withdrawal (C).

Boxes indicate Pearson correlation coefficient between the TDP43-GFP and V5-nanobody signals across individual pixels within the soma. A higher coefficient indicates higher co-localization. Scale bar = 10 um. Images were captured with the Zeiss LSM880 Airyscan.

(A) Confirmation of three heterozygous and two homozygous TDP43-GFP knock-in clones using PCR-gel electrophoresis. The unedited parent healthy iPSCs served as a control. Homozygous “E5” and “E8” clones were selected.

(B) Representative images of the unedited parent line and untransduced TDP43-GFP homozygous clones E8 and E5. Neurons were stained for MN markers neurofilament-M (NFM) and ISLET1 (ISL1) at day 20. Nuclei were stained with Hoechst 33342 Scale bar = 50 µm.

(C) RT-qPCR of unedited (parent) iPSC-MNs and untransduced homozygous TDP43-GFP knock-in MNs at day 30, suggesting no adverse effects of GFP knock-in on TDP43 expression or alternative splicing. N = 3 independent differentiations for each of the unedited and TDP43-GFP (two of E5, one of E8) lines. P-values were calculated using a one-tailed Student’s T-test. Error bars show SEM. None of the differences were significant. n.s. = not significant.

(D) Representative images of the E5 homozygous TDP43-GFP line showing a cell with and without cytoplasmic puncta. Cells were fixed and stained at D40, 22 days post-transduction. Scale bar = 15 µm.

(A) Related to main figures 1C and 1D. Quantification of morphological defects in NES nanobody treated MNs versus control. Mislocalized TDP43 (NES) causes a reduction in dendritic complexity (D) and soma swelling, compared to neurons expressing nuclear TDP43 (control).

(B) Related to main figure 1D. Quantification of caspase 3 (CC3) levels in NES nanobody treated MNs versus control. Mislocalized TDP43 (NES) causes elevation of CC3 levels, compared to neurons expressing nuclear TDP43 (control).

For A and B, each replicate is indicated by a different colour. Control: control nanobody, NES: NES nanobody, NES-M: Low to moderate levels of mislocalization, defined as neurons with > 60% nuclear TDP43. NES-S: Severe mislocalization defined as neurons with < 60% nuclear TDP43.

(C) Morphological analysis in unedited (parent) iPSC MNs treated with control or NES nanobodies. No significant effect is observed on dendrite morphology, soma swelling or apoptosis (CC3 intensity) at a p-value threshold of 0.01. N = 2 independent differentiations. At least 150 neurons were analysed per condition. CC3 = cleaved caspase-3. Each replicate is indicated by a different colour.

(D) RT-qPCR showing abnormal splicing of UNC13A and STMN2 in day 40 homozygous TDP43-GFP MNs, 20 days post-transduction with NES or control nanobodies. N = 6 replicates of NES and CTRL each (three of E5, three of E8). Error bars show SEM. CE = cryptic exon, TRUNC = truncated.

n.s. = not significant, ** indicates p < 0.01,*** indicates p < 0.001.

UCSC Genome browser screen shots depicting cryptic exon inclusion (UNC13A, STMN2, ELAVL3) and intron retention (CELF5) resulting from TDP43 mislocalization. In all cases, there is at least one TDP43 binding site, identified by SH-SY5Y eCLIP data (GSE122648) shown as a horizontal black bar. Screenshots are from the homozygous E5 TDP43-GFP clone, at MN day 40, 20 days post-transduction with AAVs. Red arrows indicate the alternative splice event.

UCSC Genome browser shots depicting cryptic exon inclusion in CYFIP2, FEZ1, and IGSF21 resulting from TDP43 mislocalization. There are no TDP43 binding sites identified by SH-SY5Y eCLIP data (GSE122648). Screenshots are from the homozygous E5 TDP43-GFP clone, at MN day 40, 20 days post-transduction with AAVs. Red arrows indicate the alternative splice event.

(A) STMN2 transcripts containing the CE in the TDP43-GFP MNs expressing the NES nanobody (isoforms containing the variable length CE are shown in red, reference isoforms are show in dark blue at the top). Each isoform was assigned a unique number. Top five most abundant isoforms are marked by yellow arrows. Isoforms containing the extra 114 bp exon upstream of the CE are marked by purple arrows. (B) Quantification of the top five most abundant STMN2 isoforms in TDP43-GFP MNs expressing control or NES nanobody. Each dot represent data from an independent replicate. N=4.

RT-qPCR showing alternative splicing errors in unedited iPSC MNs, 10 days post-addition of shRNAs targeting TDP43, or controls. The chart showing successful TDP43 knockdown is highlighted in red. Error bars show SEM. CE = cryptic exon, TRC = truncated, IR = intron retention. N = 3 independent differentiations.

* indicates p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001.

TDP43-GFP mislocalization is distinct from stress granule formation.

Representative images of TDP43-GFP-NES or –CRTL lines with and without SA exposure (500 µM for 60 minutes), 3 days in Dox (1 ug/mL). G3BP1+ Stress granules were observed only with SA treatment. Cytoplasmic TDP43-GFP puncta in the NES line do not fully co-localize with stress granules.

Cells were fixed and stained at day 23. Scale bar = 20 µm.

Uncropped western blots for total and phosphorylated TDP43.

Protein from TDP43-GFP-CTRL or –NES MNs was collected day 40 (20 days post-Dox, 1 ug/mL) and stained for total TDP43 and phosphorylated TDP43 (pTDP43-GFP). Alpha-tubulin was used as a loading control. Three replicates from independent differentiations are shown.

Transcriptomic analysis of day 40 MNs using the Dox inducible TDP43 model.

(A) PCA plot using the top 500 most variable genes. Control: MNs expressing the control nanobody, NES: MNs expressing the NES nanobody.

(B) GO enrichment analysis of 494 genes mis-spliced in the NES MNs compared to Control MNs. Splicing changes were detected using leafcutter at an adjusted p-value < 0.01 and delta-Psi > 0.1.

MicroRNA dysregulation in MNs due to TDP43 mislocalization or knockdown.

(A) Principal component analysis of microRNA (miR) counts for the TDP43-GFP-Control and TDP43-GFP-NES MNs.

(B) Volcano plot displaying miRs differentially expressed due to TDP43 mislocalization. Red: upregulated miRs, Blue: downregulated miRs, NC: No Change. Horizontal line denotes a p-value threshold of 0.01.

(C) Principal component analysis of miR counts for unedited healthy iPSC MNs transduced with TDP43 or control shRNAs.

(D) Volcano plot displaying miRs differentially expressed due to TDP43 knockdown. Blue: downregulated miRs, NC: No Change. Horizontal line denotes a p-value threshold of 0.01. Only miR-1249-3p was deemed differentially regulated at the thresholds used.

(E, F) Gene-set enrichment analysis plots displaying overlap of differentially expressed miRs due to TDP43 mislocalization or knockdown. miRs differentially expressed in the TDP43 knockdown samples were ranked from most upregulated to the most downregulated. E: Gene set used for the GSEA include miRs downregulated due to TDP43 mislocalization, F: Gene set used for the GSEA include miRs upregulated due to TDP43 mislocalization.

Table of qPCR primers. AS = Alternative splicing.

Antibodies used for immunofluorescence and western blotting