Human iPSC 4R tauopathy model uncovers modifiers of tau propagation

Abstract

Tauopathies are age-associated neurodegenerative diseases whose mechanistic underpinnings remain elusive, partially due to a lack of appropriate human models. Here, we engineered human induced pluripotent stem cell (hiPSC)-derived neuronal lines to express 4R Tau and 4R Tau carrying the P301S MAPT mutation when differentiated into neurons. 4R-P301S neurons display progressive Tau inclusions upon seeding with Tau fibrils and recapitulate features of tauopathy phenotypes including shared transcriptomic signatures, autophagic body accumulation, and reduced neuronal activity. A CRISPRi screen of genes associated with Tau pathobiology identified over 500 genetic modifiers of seeding-induced Tau propagation, including retromer VPS29 and genes in the UFMylation cascade. In progressive supranuclear palsy (PSP) and Alzheimer's Disease (AD) brains, the UFMylation cascade is altered in neurofibrillary-tangle-bearing neurons. Inhibiting the UFMylation cascade in vitro and in vivo suppressed seeding-induced Tau propagation. This model provides a robust platform to identify novel therapeutic strategies for 4R tauopathy.

Keywords: CRISPRi screen; Tau; UFMylation; endolysosome; human neurons; iPSC; neurodegeneration; neuronal activity; retromer; tauopathy.

Figure 1.. Generation and characterization of 4R and 4R-P301S Tau human iPSC-derived neurons

(A) Stepwise strategy for engineering MAPT 4R and 4R-P301S knock-in in human iPSC lines. Stars indicate mutations promoting exon 10 inclusion. (B) Representative immunoblot of lysates from 3R homozygous (i³N), 4R homozygous, and 3R/4R heterozygous 6-week-old neurons (D42) with or without lambda protein phosphatase treatment. n=2 independent experiments. (C) Representative immunofluorescence images of 6-week-old (D44) i³N and 4R neurons stained with MAP2 and ET3 antibodies, and DAPI. Scale bar, 25 μm. (D and E) Representative immunoblots (D) and quantification (E) of phosphorylated Tau (p-tau: AT180, AT270, AT8) and total Tau (t-Tau: HT7) antibodies in 6-week-old (D42) i³N, 4R homozygous, Tau-KO, and 3R/4R heterozygous neurons, normalized to GAPDH. n=4 from two independent experiments performed in duplicate. *p < 0.05, ***p < 0.001, ****p < 0.0001, one-way ANOVA, Tukey’s multiple comparisons test. (F and G) Gene set enrichment analysis pathways identified from upregulated (F) and downregulated (G) DEGs in 4R-P301S vs 4R neurons. Significantly enriched GO terms for biological process, cellular component, and molecular function. (H) Heatmap and hierarchical clustering of DEGs within the Transmembrane Transport and Intracellular Transport pathways in untreated 4R-P301S and 4R neurons. Adjusted p-value < 0.05. GO term for biological process.

Figure 2.. Modeling seeding-induced 4R-P301S Tau inclusions in human neurons

(A) Schematic of two-step differentiation, Tau seeding, and analysis of human iPSC-derived glutamatergic neurons. (B) Representative immunofluorescence images of K18-seeded D25(7+18) 4R and 4R-P301S neurons stained with MC1 and MAP2 antibodies, and DAPI. Scale bar, 25 μm. n=9 from three independent experiments performed in triplicate. (C and D) Representative images (C) and quantification (D) of immunofluorescence time course of MC1/DAPI+ nuclei in 4R-P301S neurons 1–5 weeks post-K18 treatment. n=4 from two independent experiments performed in duplicate. Scale bar, 50 μm. (E) Representative immunofluorescence images of D25(7+18) 4R-P301S neurons seeded with K18 stained with phosphorylated Tau (AT8), oligomeric Tau (TTC18), and MAP2 antibodies and DAPI. Scale bar, 25 μm. (F) Left panel: Representative TEM image of a 3 μg/ml K18-seeded D35(7+28) 4R-P301S neuronal soma with a large Tau inclusion (outlined). Scale bar, 2 μm. Right panel: Enlarged image of Tau inclusion. Scale bar, 200 nm. (G) Percentage of Tau inclusion-bearing neurons from TEM images of 4R, 4R-P301S, 4R+K18, and 4R-P301S+K18 (3 μg/mL) D35 neurons. n=4 averages from somas: 214 (4R), 104 (4R+K18), 187 (4R-P301S), and 175 (4R-P301S+K18) from two independent experiments performed in duplicate. ****p < 0.0001, one-way ANOVA, Tukey post-hoc test. (H and I) Representative immunoblot images (H) of detergent-fractionated lysates from D59(D7+52) 4R and 4R-P301S neurons −/+ 1.5 μg/ml K18 seeding stained with t-Tau (HT7) and p-Tau (AT8). T=Triton soluble, S=SDS soluble. Quantification (I) of lysates from D68(22+46) and D59(D7+52) 4R and 4R-P301S normalized to GAPDH. n=4–6 from two independent experiments performed in either duplicate or triplicate. *p < 0.05, ***p < 0.001, one-way ANOVA, Tukey post-hoc test. (J and K) Representative flow cytometry analysis (J) and quantification (K) of the percentage of MC1+ cells in D42(7+35) 4R and 4R-P301S neurons clone #1 and clone #2 −/+ 1.5 μg/mL K18 groups. n=3–5 biological replicates. ****p < 0.0001, two-way ANOVA, Šídák’s multiple comparisons test. (L) Venn diagram comparing the number and overlap of downregulated DEGs in 4R-P301S+K18 vs 4R+K18 neurons from bulk RNA-seq analysis, and AT8+ vs AT8- neurons from pseudo bulk RNA-seq analysis of AD brains. Fisher’s exact test. (M) ClueGO biological processes pathway enrichment of overlapping downregulated DEGs in 4R-P301S+K18 vs 4R+K18 / AT8+ vs AT8- comparison. Node colors denote functionally grouped networks (kappa connectivity score ≥ 40%). GO Term ≥ 1% of genes. p-value = 0.01–0.04. (N) Venn diagram comparing the number and overlap of upregulated DEGs in 4R-P301S+K18 vs 4R-P301S neurons from pseudo bulk RNA-seq analysis, and AT8+ vs AT8- neurons from pseudo bulk RNA-seq analysis of AD brains. Fisher’s exact test. (O) ClueGO biological processes pathway enrichment of overlapping upregulated DEGs in 4R-P301S+K18 vs 4R-P301S / AT8+ vs AT8- comparison. Node colors denote functionally grouped networks (kappa connectivity score ≥ 40%). GO Term ≥ 4% of genes.

Figure 3.. Tau propagation in 4R-P301S neurons is associated with endolysosomal dysfunction

(A) Representative TEM images from soma D35 4R neurons, D40 4R-P301S neurons, and 1.5 μg/mL K18-seeded D40(7+33) 4R-P301S neurons. Scale bar, 1 μm. (B) Quantification of #MLB/neuron from D35 4R and 4R-P301S, D40 4R and 4R-P301S, and D43(7+36) 4R-P301S+K18 neurons (3 μg/mL) from TEM images. n=16 from two-three independent experiments with eight neurons each. ***p < 0.001, one-way ANOVA, Tukey post-hoc test. (C) Representative TEM images from soma of 3 μg/mL K18-seeded D43(7+36) 4R-P301S neurons at different magnifications. Scale bars left to right: 2 μm, 500 nm, 200 nm. (D) Representative TEM images from neuronal processes of 3 μg/mL K18-seeded D35(7+28) 4R and 4R-P301S neurons. Scale bars left to right: 500 nm, 500 nm, 200 nm, and 100 nm. (E) VAMP7 dominant-negative (VAMP7^DN) functional knockdown mechanism. TM = transmembrane. (F) Strategy for VAMP7 functional knockdown and analysis of 4R-P301S neurons. (G) Quantification of total Tau secreted from 4R and 4R-P301S neurons −/+ 3 μg/mL K18 and overexpressing GFP or GFP-VAMP7^DN incubated with 50 mM KCl. n=7–8 from two independent experiments. **p < 0.01, one-way ANOVA, Tukey’s post-hoc test. (H) Representative immunofluorescence images of D28(14+14) K18-seeded 4R-P301S neurons overexpressing GFP or GFP-VAMP7^DN stained with MC1 antibody and DAPI. Scale bar, 50 μm. (I) Quantification of MC1 area/DAPI+ nuclei from immunofluorescence images of D28(14+14) K18-seeded 4R-P301S neurons overexpressing GFP or GFP-VAMP7^DN. n=81 for GFP and 92 for VAMP7^DN from 10–12 images per replicate, from two independent experiments. ***p < 0.001, unpaired t-test.

Figure 4.. Tau inclusions impair spontaneous and evoked neuronal activity

(A) Strategy for targeting HaloTag at 5′ end of MAPT locus in 4R-P301S iPSC line. (B) Representative fluorescence image of 3 μg/ml K18-seeded 4R-P301S-HaloTag neurons (D24) harboring Tau inclusions incubated with JFX₅₄₉-HaloTag ligand. Scale bar, 50 μm. (C) Representative fluorescence images expressing GCaMP8f, HaloTag, and overlay in D24 K18-seeded 4R-P301S-HaloTag spontaneous activity. Scale bar, 50 μm. (D) Quantification of spike amplitude, firing rate, and synchrony index from calcium imaging of K18-seeded 4R-P301S-HaloTag neurons (D24–30). n=-inclusion/+inclusion: Spike Amplitude 74/121, Firing Rate 76/122, Network 14/16 from 3 biological replicates. *p < 0.05, unpaired t-test. (E) Representative fluorescence images of GCaMP8f, HaloTag, and overlay in D30 K18-seeded 4R-P301S-HaloTag treated with 50 mM KCl. Scale bar, 50 μm. (F) Representative averaged calcium traces from one KCl stimulation experiment in 4R-P301S-HaloTag neurons (D30) with and without Tau inclusions. Recording 400 seconds. n=11 -inclusion and 8 +inclusion. Mean ± SEM. (G) Quantification of peak amplitude from KCl stimulation-induced neuronal responses. n= 33 -inclusion and 20 +inclusion from 3 biological replicates. *p < 0.05, unpaired t-test. (H) Strategy for sustained neuronal activity repression and analysis of 4R-P301S-HaloTag neurons. (I) Representative spontaneous calcium traces (right) of each cell from D36 4R-P301S expressing hM4Di (left) upon 10 μM CNO addition at 60 seconds. Scale bar, 50 μm. (J) Quantification of MC1 area/DAPI+ nuclei from immunofluorescence images of chronically silenced D29 K18-seeded 4R-P301S neurons. n=50 -CNO and 30 +CNO from 10 images, 5 -CNO biological replicates and 3 +CNO biological replicates. **p < 0.01, unpaired t-test.

Figure 5.. Identification of Tau inclusion modifiers by CRISPRi screening

(A) Strategy for CRISPRi screening based on Tau inclusions in 4R-P301S-dCas9 neurons. (B) Volcano plot summarizing knockdown phenotypes and statistical significance (Mann-Whitney U test) for genes targeted in the pooled screen. Dashed lines: cutoff for hit genes (FDR = 0.05). (C) Subcellular classification of gene hits (product above 2.5 and below −2.5, see STAR methods). (D and E) Representative immunofluorescence images (D) and quantification (E) of the percentage of DAPI+/MC1+ cells in D21(7+14) 1.5 μg/ml K18-seeded 4R-P301S;VPS29^−/− neurons clone #1 and #2 compared to the parent line. n=4 replicates from one independent experiment per clone. ****p < 0.0001, one-way ANOVA, Dunnett’s multiple comparisons test. Scale bar, 50 μm. (F and G) Representative flow cytometry analysis (F) and quantification (G) of the percentage of MC1+ cells in D21(7+14) 1.5 μg/ml K18-seeded 4R-P301S;VPS29^−/− neurons clone #1 and clone #2 compared to the parent line. n=6 replicates from one independent experiment per clone. ****p < 0.0001, one-way ANOVA, Dunnett’s multiple comparisons test.

Figure 6.. CRISPRi screen identifies UFMylation pathway as a key modulator of Tau inclusions

(A) UFMylation cascade of a target substrate. VGSC = amino acid motif (valine-glycine-serine-cysteine). (B and C) Representative flow cytometry analysis (B) and the quantification (C) of the percentage of MC1+cells in D21(7+14) 3 μg/ml K18-seeded 4R-P301S neurons infected with lentivirus (GFP+) containing shRNAs targeting UBA5 and UFM1. n=6–9 from two independent experiments. ***p < 0.001, ****p < 0.0001, mixed model analysis, Dunnett’s multiple comparisons test. (D and E) Representative immunofluorescence images (D) and quantification (E) of the percentage of DAPI+/MC1+ cells in D21(7+14) as in B and C. n=10 from two independent experiments. ****p < 0.0001, one-way ANOVA, Dunnett’s multiple comparisons test. Scale bar, 50 μm. (F) Representative immunoblot of relative free UFM1 D21(7+14) 3 μg/ml K18-seeded 4R-P301S-HaloTag detergent-fractionated neuronal samples. Triton soluble (soluble), SDS-soluble (insoluble). (G) Quantification of relative free UFM1 in Triton-soluble and SDS-soluble fractions of control vs. K18-seeded neurons normalized to GAPDH. n=3 biological replicates. **p < 0.01, unpaired t-test of SDS-soluble fraction. (H and I) Representative immunoblot (H) and quantification (I) of lysates from D14 4R-P301S neurons transduced with control, UBA5, or UFM1 shRNA lentivirus stained with antibodies for t-Tau (HT7), UBA5, UFM1, and GAPDH. n=9 from three independent experiments performed in triplicate. ****p < 0.0001, one-way ANOVA, Dunnett’s multiple comparisons test. (J) Quantification of total Tau in lysates from D14 4R-P301S neurons transduced with control, UBA5, or UFM1 shRNA lentivirus. n=9 from three independent experiments performed in triplicate. **p < 0.01, ***p < 0.001, one-way ANOVA, Dunnett’s multiple comparisons test.

Figure 7.. UFMylation is functionally relevant in a Tau spreading mouse model and in tauopathy human brain tissues

(A) Representative immunofluorescence images of dentate gyrus of 3-month-old PS19 mice stereotaxically injected with either control shRNA or Uba5 shRNA lentivirus (GFP+), harvested 4 weeks after injections and stained with GFP and UBA5 antibodies, and DAPI. Scale bar, 100 μm. (B) Quantification of Uba5 immunofluorescence normalized fold-change intensity in control shRNA and Uba5 shRNA treated groups. n= 3 mice per group, triplicate measurements per mouse. ****p<0.0001, mixed model analysis. (C) Diagram of PS19 mouse brain indicating lentivirus (spreading) and K18-Tau fibril (seeding) injection sites. (D) Representative immunofluorescence images of the contralateral hippocampus (CA3) from 2.5–3.5-month-old mice stereotaxically injected with K18 fibrils and either control or Uba5 shRNA lentivirus, harvested 4 or 6 weeks after injections and stained with GFP antibody and DAPI. Scale bar, 200 μm. (E) Representative images of MC1 immunostaining in the contralateral CA3 after control shRNA or Uba5 shRNA lentivirus injections. Scale bar, 200 μm. (F) Quantification of relative MC1-positive signal from the hippocampus of control and Uba5 shRNA treated groups, normalizing the spreading side to the seeding side. n=32 shCtrl and 24 shUba5 replicates, 10 shCtrl and 8 shUba5 mice. *p<0.05, mixed model analysis. (G) Representative immunofluorescence images of human PSP post-mortem tissue stained with MC1 and UFM1 antibodies, and DAPI. Scale bar, 25 μm. (H) Quantification of UFM1 immunofluorescence normalized fold-change intensity in MC1+ and MC1- cells in human PSP post-mortem tissue. n=120 MC1- cells and 100 MC1+ cells, six male and two female cases. ****p < 0.0001, mixed model analysis. (I) Representative immunofluorescence images of human AD post-mortem tissue stained with MC1 and UFM1 antibodies, and DAPI. Scale bar, 25 μm. (J) Quantification of UFM1 immunofluorescence normalized fold-change intensity in MC1+ and MC1- cells in human AD post-mortem tissue. n=106 MC1- cells and 91 MC1+ cells, four male and three female cases. ****p < 0.0001, mixed model analysis.