Compromised function of the ESCRT pathway promotes endolysosomal escape of tau seeds and propagation of tau aggregation

Abstract

Intercellular propagation of protein aggregation is emerging as a key mechanism in the progression of several neurodegenerative diseases, including Alzheimer's disease and frontotemporal dementia (FTD). However, we lack a systematic understanding of the cellular pathways controlling prion-like propagation of aggregation. To uncover such pathways, here we performed CRISPR interference (CRISPRi) screens in a human cell-based model of propagation of tau aggregation monitored by FRET. Our screens uncovered that knockdown of several components of the endosomal sorting complexes required for transport (ESCRT) machinery, including charged multivesicular body protein 6 (CHMP6), or CHMP2A in combination with CHMP2B (whose gene is linked to familial FTD), promote propagation of tau aggregation. We found that knocking down the genes encoding these proteins also causes damage to endolysosomal membranes, consistent with a role for the ESCRT pathway in endolysosomal membrane repair. Leakiness of the endolysosomal compartment significantly enhanced prion-like propagation of tau aggregation, likely by making tau seeds more available to pools of cytoplasmic tau. Together, these findings suggest that endolysosomal escape is a critical step in tau propagation in neurodegenerative diseases.

Keywords: CRISPR/Cas; Membrane damage; Tau protein (Tau); endosomal sorting complexes required for transport (ESCRT); endosome; fluorescence resonance energy transfer (FRET); functional genomics; lysosome; protein aggregation.

Figure 1.

Tau seeds induce tau aggregation in a FRET-based reporter cell line. A, overview of cellular processes that may control the prion-like tau propagation and aggregation. Question marks represent unknown cellular mechanisms. B, schematic representation of the FRET-based reporter assay to monitor tau aggregation in HEK293T cells. In the absence of tau seeds, fluorescently labeled tau.K18(LM) is monomeric. Exposure to nonfluorescent tau seeds induces aggregation of the reporter, which can be measured by the formation of tau aggregates by fluorescence microscopy or an increase in FRET intensity by flow cytometry. C, induction of fluorescent tau aggregates in a FRET reporter cell line. Representative images of cells treated with PBS (top row), Alzheimer's patient brain extracts after 5 days (second row), or fibrils of recombinant human 0N4R tau after 2 days (third row). For each tau seed, each condition is complexed with (right column) or without (left column) Lipofectamine. Nuclei were counter-stained with Hoechst 33342. D, comparison of intracellular fluorescent tau aggregates from images in C. Integrated density quantification of fluorescent tau aggregates seeded with various tau seeds complexed with (blue) or without (orange) Lipofectamine were quantified and divided by total nuclei per image. n = 3 technical replicates (with at least 50 nuclei per image), error bars represent mean ± S.D., *, p < 0.05; **, p < 0.01 (two-tailed Student's t test for comparison to PBS (no Lipofectamine) control for each tau seeding condition). E, representative flow cytometry plot of FRET reporter cells after 2 day treatment with PBS (left) or tau fibrils (right). F, incubation of recombinant 0N4R tau with heparin and constant agitation at 37 °C induces fibrillization. Fibrillization is monitored using an increase in thioflavin T fluorescence (excitation 440 nm, emission: 485 nm), which occurs in the presence (blue) of heparin (10 μg/ml), but not in the absence (orange) of heparin. Error bars represent mean ± S.D. from n = 3 technical replicates. G, representative negative stain electron micrograph of tau fibrils. H, lysates from FRET reporter cells treated with PBS or tau fibrils for 2 days were fractionated at 1000 × g into soluble (S) or pellet (P) fractions, and subjected to SDS-PAGE and immunoblotting using antibodies against tau and β-actin. I, quantification of % FRET-positive cells using flow cytometry across concentration ranges of tau fibrils (left) or human Alzheimer's patient brain extracts (right). Error bars represent mean ± S.D. for n = 3 technical replicates.

Figure 2.

CRISPRi screen for cellular factors controlling tau aggregation. A, strategy for pooled FRET-based CRISPRi screen. FRET reporter cells stably expressing the CRISPRi machinery (dCas9-BFP-KRAB) were transduced with pooled lentiviral expression libraries of sgRNAs targeting proteostasis genes. Following transduction and selection, cells were treated with tau fibrils and incubated for 2 days. Cells were detached and sorted into FRET-negative and -positive populations by FACS. sgRNA-encoding cassettes were amplified from genomic DNA of the cell populations and their frequencies were quantified using next generation sequencing to identify genes that control tau aggregation. B, volcano plot summarizing phenotypes and statistical significance (by our MAGeCK-iNC pipeline, see “Experimental procedures”) of the genes targeted by the sgRNA libraries. Nontargeting sgRNAs were randomly grouped into negative control “quasi-genes” (gray dots) to derive an empirical false discovery rate (FDR). Hit genes that passed an FDR < 0.05 threshold are shown in blue (knockdown decreases aggregation) or red (knockdown increases aggregation), other genes are shown in orange. Two hit genes of interest are shown in green and labeled. C–E, validation of hit genes CHMP6 and VPS13D. FRET reporter cells transduced with individual sgRNAs targeting two hit genes or a nontargeting control sgRNA, and 5 days after transduction treated for 2 days with tau fibrils. C, % of FRET-positive cells was quantified by flow cytometry. Error bars represent mean ± S.D. of n = 3 technical replicates. *, p < 0.05; **, p < 0.01 (two-tailed Student's t test for comparison to the nontargeting control sgRNA). D, representative immunoblot for the tau fluorescent protein construct in the soluble and insoluble fractions as in Fig. 1H. E, quantification of insoluble/soluble tau ratios from immunoblots in D. Error bars represent mean ± S.D. for n = 3 biological replicates. *, p < 0.05; **, p < 0.01 (two-tailed Student's t test for comparison to the nontargeting control sgRNA).

Figure 3.

CHMP6 knockdown accelerates the prion-like propagation of tau aggregation. FRET reporter cells or CRISPRi-HEK293T cells were transduced with individual sgRNAs targeting CHMP6 or VPS13D, or a nontargeting control sgRNA, and characterized for different phenotypes 5 days after transduction (A) Knockdown of CHMP6 and VPS13D does not impact steady-state levels of the tau-Clover2 construct in FRET reporter cells, as quantified by flow cytometry. B, individual gene knockdown does not impact uptake of tau fibrils. CRISPRi HEK293T cells were incubated with AF555-labeled tau fibrils for 1 h at 37 °C, stringently washed, and red fluorescence representing internalized tau fibrils was quantified by flow cytometry. The bar graph shows normalized fluorescence intensities and standard deviation of n = 3 technical replicates. C, CHMP6 knockdown accelerates prion-like propagation of tau aggregation. Representative fluorescence micrographs of the Tau.K18(LM)Clover2 reporter in cells 1 and 2 days after fibril addition are shown. Nuclei were counter-stained with Hoechst 33342. D, quantification of C. Tau aggregates were quantified by integrated density across the entire image and divided by total nuclei per image. Error bars represent mean ± S.D., where n = 3 images per condition (with at least 50 nuclei per image). *, p < 0.05; ***, p < 0.001 (two-tailed Student's t test for comparison to the values for nontargeting control sgRNA of the same day).

Figure 4.

CHMP6 knockdown compromises endolysosomal membrane integrity. A, time-lapse microscopy of cell entry of tau-AF555 fibrils and resulting aggregation of the cytosolic tau-Clover2 construct. Representative images of CRISPRi-HEK293T cells expressing tau.K18(LM)-clover2 transduced with either nontargeting control (top) or CHMP6-targeting sgRNA (bottom) are shown. Corresponding movies are provided as Movie S1 (control sgRNA) and Movie S2 (CHMP6 sgRNA). B, representative fluorescence microscopy images of CRISPRi-HEK293T cells LAMP1-mNG₁₁ endogenously labeled with the split-mNeonGreen system. Cells were transduced with nontargeting control (top) or CHMP6 sgRNA (bottom) and treated with AF647-tau fibrils and followed by automated time-lapse microscopy for 48 h. Images for the 12-h time point are shown here. Corresponding movies for 48-h time intervals are provided as Movie S4 (Control sgRNA) and Movies S5 (CHMP6 sg1). C, quantification of Tau fibril colocalization with LAMP1 from representative images and movies shown in B and Movies S4 and S5. Time represents start of treatment with tau fibrils and image acquisition. Error bars represent mean ± S.D. for n = 3 technical replicates (with at least 5 nuclei per image). D, CHMP6 knockdown causes endolysosomal vesicle damage. Representative fluorescence microscopy images of CRISPRi-HEK293T cells expressing an EGFP-Galectin3 (EGFP-Gal3) reporter transduced with either control (left) or CHMP6 (right) sgRNA. Nuclei were counter-stained with Hoechst 33342. E, quantification of EGFP-Gal3 puncta divided by number of nuclei in fluorescence microscopy images shown in D. Error bars represent mean ± S.D. for n = 3 technical replicates (with at least 50 nuclei per image). ***, p < 0.001 (two-tailed Student's t test for comparison to the nontargeting control sgRNA). F, knockdown of various ESCRT components increases tau aggregation. FRET reporter cells were transduced with individual sgRNAs targeting ESCRT components or a nontargeting control sgRNA, and 5 days after transduction treated for 2 days with tau fibrils. Error bars represent mean ± S.D. for n = 3 technical replicates. *, p < 0.05; **, p < 0.01 (two-tailed Student's t test for comparison to the nontargeting control sgRNA). G, simultaneous, but not individual, knockdown of CHMP2A and CHMP2B results in endolysosomal damage, monitored as in C. Nuclei were counter-stained with Hoechst 33342. Scale bar = 50 μm. H, quantification of EGFP-Gal3 puncta divided by nuclei in fluorescence microscopy images shown in G. Error bars represent mean ± S.D. for n = 3 technical replicates (with at least 50 nuclei per image). ***, p < 0.001 (two-tailed Student's t test for comparison to the nontargeting control sgRNA). I, simultaneous, but not individual, knockdown of CHMP2A and CHMP2B increases prion-like tau aggregation. % FRET-positive reporter cells transduced with sgRNAs as indicated 2 days after tau fibril treatment. Error bars represent mean ± S.D. where n = 3 technical replicates. *, p < 0.05 (two-tailed Student's t test for comparison to the nontargeting control sgRNA).

Figure 5.

Small molecules damage endolysosomal compartments and phenocopy the acceleration of the prion-like propagation of tau aggregation following CHMP6 knockdown. A, treatment with the lysosomotropic drug LLOME damages endolysosomal vesicles. Representative fluorescence microscopy images of CRISPRi-HEK293T cells expressing the EGFP-Gal3 reporter treated with DMSO (left) or 500 μm LLOME (right) for 24 h. B, quantification of EGFP-Gal3 puncta divided by number of nuclei in fluorescence microscopy images shown in A. Error bars represent mean ± S.D. for n = 3 technical replicates (with at least 50 nuclei per image). ***, p < 0.001 (two-tailed Student's t test for comparison to DMSO control). C, LLOME treatment accelerates the prion-like propagation of tau aggregation. % FRET-positive cells transduced with control (blue) or CHMP6 sgRNA (orange) were analyzed 24 (dark line) or 48 (light line) h following co-treatment with DMSO or increasing concentrations of LLOME and tau fibrils. Error bars represent mean ± S.D. for n = 3 technical replicates. *, p < 0.05; **, p < 0.01 (two-tailed Student's t test for comparison to the values for nontargeting control sgRNA of the same day). D, Lipofectamine treatment damages endolysosomal vesicles. Representative fluorescence microscopy images of CRISPRi-HEK293T cells expressing the EGFP-Gal3 reporter treated with DMSO (left) or 1.25% (v/v) Lipofectamine 2000 (right) for 6 h. E, quantification of EGFP-Gal3 puncta divided by number of nuclei in fluorescence microscopy images shown in D. Error bars represent mean ± S.D. for n = 3 technical replicates. ***, p < 0.001 (two-tailed Student's t test for comparison to no Lipofectamine treatment). F, Lipofectamine treatment accelerates the prion-like propagation of tau aggregation. % FRET-positive cells transduced with control (blue) or CHMP6 sgRNA (orange) were analyzed 24 (dark line) or 48 (light line) h following co-treatment with PBS or increasing concentrations of Lipofectamine 2000 and tau fibrils. Error bars represent mean ± S.D. where n = 3 technical replicates. *, p < 0.05; **, p < 0.01; ***, p < 0.001 (two-tailed Student's t test for comparison to the values for nontargeting control sgRNA of the same day). G–I, CHMP6 knockdown increases the prion-like propagation tau aggregation when seeding with Alzheimer's patient brain extracts. G, representative images of FRET reporter cells transduced with control (left) or CHMP6 sgRNA (right) and treated with Alzheimer's patient brain extract after 5 days. H, quantification of images in G. Tau aggregates were quantified by integrated density across the entire image and divided by the total nuclei per image. Error bars represent mean ± S.D. for n = 3 images per condition (with at least 50 nuclei per image). **, p < 0.01 (two-tailed Student's t test for comparison to nontargeting control sgRNA) (I) % FRET-positive cells 5 days following treatment with Alzheimer's patient brain extract. Error bars represent mean ± S.D., where n = 3 technical replicates. **, p < 0.01 (two-tailed Student's t test for comparison to nontargeting control sgRNA). J, model for the increased tau aggregation via CHMP6 knockdown and generation of damaged endolysosomes.