Intraneuronal tau aggregation induces the integrated stress response in astrocytes

ABSTRACT Progressive aggregation of tau protein in neurons is associated with neurodegeneration in tauopathies. Cell non-autonomous disease mechanisms in astrocytes may be important drivers of the disease process but remain largely elusive. Here, we studied cell type-specific responses to intraneuronal tau aggregation prior to neurodegeneration. To this end, we developed a fully human co-culture model of seed-independent intraneuronal tau pathology, which shows no neuron and synapse loss. Using high-content microscopy, we show that intraneuronal tau aggregation induces oxidative stress accompanied by activation of the integrated stress response specifically in astrocytes. This requires the direct co-culture with neurons and is not related to neurodegeneration or extracellular tau levels. Tau-directed antisense therapy reduced intraneuronal tau levels and aggregation and prevented the cell non-autonomous responses in astrocytes. These data identify the astrocytic integrated stress response as a novel disease mechanism activated by intraneuronal tau aggregation. In addition, our data provide the first evidence for the efficacy of tau-directed antisense therapy to target cell autonomous and cell non-autonomous disease pathways in a fully human model of tau pathology.

High-content microscopy analysis workflow for nuclear, morphological and presynaptic measurements.
(A) Schematic representation of the high-content screening and analysis workflow to distinguish neurons and astrocytes in human co-culture and to quantify parameters of neuronal morphology and presynapses, based on DAPI, MAP2 and SYP1 (immuno)staining. Neuronal nuclei are distinguished from astrocytic nuclei by their relatively smaller size, higher DAPI intensity and larger MAP2 overlap in the soma, as is also shown in Figure   1C. Astrocytic nuclei are obtained by subtracting the number of neuronal nuclei from the total number of nuclei. By normalization against neuronal nuclei, morphology measures per neuron are obtained, and the number of presynaptic puncta are normalized for dendrite length to calculate presynapse density. See MAP2-negative (astrocytic) nuclei in red) (i) and dendrites are detected (traces in red) (ii). Presynapses that fall within a dendrite region (red outline) (iii) are detected (selected puncta in green) (iv). (green) or with transduction of EGFP-tagged FTDtau 1 (grey) as determined by density (H) and fluorescence intensity (I). Bar graphs show the mean ± SEM, data points represent mean/well. N = 4 independent experiments, n = 14 wells were analysed for B-I. Statistical significance was assessed by Nested T test. ns not significant. For each dataset, Supplementary Table S1 lists the exact p values and total number of cells analysed.

◄ Timing and specificity of FTDtau 1+2 aggregation
Levels of total and aggregated tau were quantified in untransduced and EGFP-tagged FTDtau 1 -or FTDtau 1+2transduced co-cultures at week 1, 2.5 and 4. (A) Schematic representation of the protocol to measure intraneuronal tau levels and aggregation in a human neuron-astrocyte co-culture over time. Representative widefield images captured by automated microscopy at week 1, 2.5 and 4 of PFA-fixed (B) and MeOH-fixed (D) co-cultures without or with transduction for EGFP-tagged FTDtau 1 -or FTDtau 1+2 . Each panel consists of 4 widefield images, showing EGFP direct fluorescence (green), immunostainings for MAP2 (magenta) and cell nuclei visualized using DAPI (blue). (C and E) Quantification of EGFP mean fluorescence intensity from B showing total tau levels and EGFP mean fluorescence standard deviation (SD) from D showing aggregated tau levels, respectively, in untransduced (green), FTDtau 1 -(grey) or FTDtau 1+2 -(red) transduced co-cultures at week 1, 2.5 and 4. Bar graphs show the mean ± SEM, data points represent mean/well. N = 3 independent experiments, n = 12 wells and N ≥ 2 independent experiments, n = 8 wells were analysed for C and E, respectively. Shapiro-Wilk normality testing followed by two-way ANOVA with Tukey's post-hoc analysis. *p<0.05, ***p<0.001, ****p<0.0001, ns not significant. Supplementary Table S1 lists the exact p values and total number of cells analysed.

Tau aggregation does not affect the morphological development of neurons.
Co-cultures were transduced for EGFP-tagged FTDtau 1 or FTDtau 1+2 and immunostained for MAP2 to assess neuron morphology at week 1, 2, 3 and 4 by high-content microscopy. Co-cultures were transduced with mCherry-(A and B) or EGFP-tagged (C and D) FTDtau 1 or FTDtau 1+2 and oxidative stress and ISR activation were measured in neurons and astrocytes at 8 weeks. Shown are representative zoomed widefield images captured by automated microscopy of co-cultures showing MAP2 immunostaining (grey) and nuclei visualized by DAPI (blue) that were detected by automated microscopy (red outline). The ROS probe CellROX green (A and B) and ISR target ATF4 (C and D) were detected by direct-and immunofluorescence, respectively, in the nuclei of neurons (A and C) and astrocytes (B and D), in the absence (Ctrl) or presence of menadione (Men) or tunicamycin (TM).

Tau overexpression does not induce oxidative stress and ISR activation in neurons and astrocytes
Oxidative stress (A and B) and ISR activation (C and D) were quantified in neurons and astrocytes by automated microscopy in FTDtau 1 -transduced and untransduced co-cultures at 8 weeks. (A and B) Oxidative stress in co-cultures without (green) or with transduction with mCherry-tagged FTDtau 1 (grey), treated with 100 µM menadione (Men) or vehicle control (Ctrl) 3.5 hours before loading with CellROX green.
Immunostaining was performed for MAP2, cell nuclei were visualized using DAPI. Oxidative stress was determined by quantifying CellROX green intensity in neuronal (A) and astrocytic (B) nuclei. (C and D) ISR activation in co-cultures without (green) or with transduction with EGFP-tagged FTDtau 1 (grey), treated 48 hours with 10 µg/mL tunicamycin (TM) or vehicle control (Ctrl). Immunostaining was performed for MAP2 and ATF4, cell nuclei were visualized using DAPI. ISR activation was determined by quantifying ATF4 intensity in neuronal (C) and astrocytic (D) nuclei. Bar graphs show the mean ± SEM, data points represent mean/well. The lowest and highest value per experiment were set to 0 and 1, respectively. N = 3 and N = 1 independent experiments, n = 12 and n = 4 wells were analysed for A, B and C, D, respectively. After Shapiro-Wilk normality testing, statistical analysis was by two-way ANOVA and Tukey's post-hoc analysis. ****p<0.0001, ns not significant. Supplementary Table S1 lists the exact p values and total number of cells analysed.

Supplementary Figure S9
Oxidative stress and ISR activation in astrocytes by intraneuronal tau aggregation is not transferred via conditioned co-culture media.
Monocultured astrocytes were exposed to conditioned media of 8 week-old untransduced co-cultures, or cocultures expressing EGFP-tagged FTDtau 1 or FTDtau 1+2 . After one week, oxidative stress and ISR activation were measured by automated microscopy using CellROX green fluorescence and ATF4 nuclear fluorescence in B and D is shown in (C) and (E), respectively. (F-I) Monocultured astrocytes exposed to conditioned media of untransduced (green) co-cultures or co-cultures expressing EGFP-tagged FTDtau 1 (grey) or FTDtau 1+2 (red). Representative zoomed widefield images captured by automated microscopy showing actin (phalloidin; grey) and nuclei visualized by DAPI (blue) that were detected by the analysis (red outline) as a merged image, and CellROX green (F) and ISR target ATF4 (H) are shown as separate greyscale images. Automated quantification of CellROX green and ATF4 nuclear fluorescence in F and H is shown in (G) and (I), respectively. For C, E, G and I mean values of vehicle and untransduced controls were set to 1, respectively. N = 1, n = 4 and N = 3, n = 6 independent experiments and wells were analysed for C, E and G, I, respectively. Bar graphs show the mean ± SEM, data points represent mean/well. After Shapiro-Wilk normality testing, significance was assessed using Student's T-test (C and E) and Nested One-way ANOVA with Tukey's post-hoc analysis (G and I). ***p=0.001, ****p<0.0001, ns not significant. Supplementary Table S1 lists the exact p values and total number of cells analysed.

Supplementary Figure S10
◄ MAPT antisense therapy prevents tau aggregation in human co-culture. Total tau levels in conditioned media of untransduced co-cultures and co-cultures expressing EGFP-tagged FTDtau 1 or FTDtau 1+2 , in the absence (untreated) or presence of scramble or MAPT AON treatment at week 4 and 6. After 8 weeks of co-culture, tau levels were measured using an MSD S-Plex Total Tau assay.
Conditioned media from N = 3 independent experiments was pooled from n ≥ 3 wells per experiment. Bar graphs show the mean value per independent experiment ± SEM. Shapiro-Wilk normality testing followed by two-way ANOVA with Tukey's post-hoc analysis. *p<0.05, ns not significant. Supplementary Table S1 lists the exact p values.