Tau Transfer via Extracellular Vesicles Disturbs the Astrocytic Mitochondrial System

Abstract

Tauopathies are neurodegenerative disorders involving the accumulation of tau isoforms in cell subpopulations such as astrocytes. The origins of the 3R and 4R isoforms of tau that accumulate in astrocytes remain unclear. Extracellular vesicles (EVs) were isolated from primary neurons overexpressing 1N3R or 1N4R tau or from human brain extracts (progressive supranuclear palsy or Pick disease patients or controls) and characterized (electron microscopy, nanoparticle tracking analysis (NTA), proteomics). After the isolated EVs were added to primary astrocytes or human iPSC-derived astrocytes, tau transfer and mitochondrial system function were evaluated (ELISA, immunofluorescence, MitoTracker staining). We demonstrated that neurons in which 3R or 4R tau accumulated had the capacity to transfer tau to astrocytes and that EVs were essential for the propagation of both isoforms of tau. Treatment with tau-containing EVs disrupted the astrocytic mitochondrial system, altering mitochondrial morphology, dynamics, and redox state. Although similar levels of 3R and 4R tau were transferred, 3R tau-containing EVs were significantly more damaging to astrocytes than 4R tau-containing EVs. Moreover, EVs isolated from the brain fluid of patients with different tauopathies affected mitochondrial function in astrocytes derived from human iPSCs. Our data indicate that tau pathology spreads to surrounding astrocytes via EVs-mediated transfer and modifies their function.

Keywords: astrocytes; extracellular vesicles; mitochondria; tau spreading; tauopathies.

Figure 1

Tau is shuttled from neurons to astrocytes. (A) Schematic of the microfluidic system used to investigate tau transfer from hippocampal neurons to astrocytes. (B) Confocal micrographs showing 1N3R or 1N4R tau (V5+) and astrocytes (GFAP+) in the somatodendritic and axonal compartments and microgrooves. (C) Histogram showing the Tau-V5 optical density in the somatodendritic compartment. (D) 3D reconstruction of confocal micrographs showing tau (V5+) in an astrocyte (GFAP+) in the axonal compartment. (E) Histogram showing the number of Tau-V5 inclusions detected per astrocyte. The scale bars are 20 µm (B) and 5 µm (D). SD = somatodendritic compartment, AX = axonal compartment, LV = lentiviral vector, NI = noninfected. For (C,E), N  =  cultures/microfluidic chambers/cells: NI: 1/3/32, 1N3R: 1/3/25, 1N4R: 4/3/39. Ordinary one-way ANOVA with Sidak’s multiple comparison test. (^ns p >0.05, **** p < 0.0001).

Figure 2

Neuronal tau is mainly secreted in a free form. (A) Confocal micrographs showing tau (V5+) in rat hippocampal neurons infected by LVs overexpressing 1N3R or 1N4R tau; NI = noninfected. (B) Histogram showing Tau-V5 accumulation (optical density normalized to that in the NI condition) in rat hippocampal neurons overexpressing 1N3R or 1N4R tau. (C) Electron microscopy images of ND-SEVs and ND-LEVs isolated from cultured neurons. The scale bar is 250 nm. (D) Histogram showing the percentage of ND-SEVs with a size between 100–150 nm and greater than 150 nm. (E) Histogram showing the percentage of ND-LEVs with a size between 100–150 nm and greater than 150 nm. (F) Electron microscopy and immunogold labeling (Cter-tau) of ND-SEVs and ND-LEVs isolated from the supernatant of neurons overexpressing 1N4R tau. The scale bar is 100 nm for ND-SEVs and 250 nm for ND-LEVs. (G) Histogram showing Tau-V5 concentration in the ND-FFP, ND-SEV, and ND-LEV fractions from control rat hippocampal neurons (NI) and those overexpressing 1N3R or 1N4R tau. For (B,G), n  =  4 cultures per condition; ordinary one-way ANOVA with Sidak’s multiple comparison test and the nonparametric Kruskal–Wallis test, in (B) and (G), respectively. (^ns p > 0.05, * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001).

Figure 3

Tau is shuttled from neurons to astrocytes via LEVs. (A) Schematic representation of the protocol employed to isolate ND-EVs from neurons infected by LVs overexpressing 1N3R or 1N4R tau. ND-EVs from NI cultures were used as controls. (B) Example of confocal images showing the transfer of ND-FFP, or tau in ND-SEVs and ND-LEVs from neurons overexpressing 1N3R or 1N4R tau. The scale bar is 25 µm. (C) Histogram showing the Tau-V5 concentration in astrocytes 24 h after incubation with the ND-FFP, ND-SEV, and ND-LEV fractions from control rat hippocampal neurons (NI) and those overexpressing 1N3R or 1N4R tau. (D) Histogram showing the tau uptake efficiency 24 h after incubation with the ND-FFP and ND-LEV fractions from control rat hippocampal neurons (NI) and those overexpressing 1N3R or 1N4R tau. For (C) and (D), n = 4 cultures per condition; ordinary one-way ANOVA with Sidak’s multiple comparison test. (^ns p > 0.05, *** p < 0.001, **** p < 0.0001).

Figure 4

Tau isoform-containing LEVs induce mitochondrial dysfunction in primary rat astrocytes. (A) Micrographs of the mitochondria labeled with the biosensor MitoTimer before (BL) and after (6 h and 24 h) treatment with ND-LEVs^CFP, ND-LEVs^1N3R, and ND-LEVs^1N4R. (B,C) Radar charts showing MitoTimer ratio 555/488nm, morphology (surface area, length, number of branches, factor of elongation,), mobility (displacement and speed), and number of event changes (fusion/fission), normalized to BL values and the value of the ND-LEV^CFP group, in astrocytes treated with ND-LEVs^1N3R and ND-LEVs^1N4R. Two-way matched ANOVA followed by post hoc analysis (Tukey’s test) was applied to compare the effect of each treatment with that of ND-LEVs^CFP (* p < 0.05, ** p < 0.01 and *** p < 0.001). A multiple t test was performed on each criterion between ND-LEVs^1N3R and ND-LEVs^1N4R to evaluate the difference in effect between the 3R and 4R isoforms ### p < 0.001).

Figure 5

LEVs from 3R and 4R tauopathies induce differential mitochondrial dysfunction in iPSC-derived astrocytes. (A) Schematic representation of the protocols used to isolate BDF-EVs from patients diagnosed with PiD or PSP and non-demented control subjects (C) and to treat iPSC-derived astrocytes. (B) Micrographs of the mitochondrial system in human iPSC-derived astrocytes expressing the biosensor MitoTimer before (BL) and after (6 h and 24 h) treatment with BDF-LEVs^PiD and BDF-LEVs^PSP. (C) Histogram showing mitochondrial redox state changes in iPSC-derived astrocytes 24 h after BDF-LEVs treatment. (D) Histogram showing mitochondrial elongation in iPSC-derived astrocytes 24 h after BDF-LEVs treatment. N  =  cultures/cells/mitochondria; ND-LEVs^CFP: 3/21/801, ND-LEVs^1N3R: 3/15/958, ND-LEVs^1N4R: 4/14/789, BFD-LEVs^C: 3/19/3883, BFD-LEVs^PiD: 3/21/749, BFD-LEVs^PSP: 4/17/2555. Two-way matched ANOVA followed by post hoc analyses (Tukey’s test) were used to compare the effects of treatments with those of the other treatments (### p < 0.001) and with the nondemented control group (*** p < 0.001). Scale bars are 30 µm.