Pathogenic tau-induced transposable element-derived dsRNA drives neuroinflammation

Abstract

Deposition of tau protein aggregates in the brain of affected individuals is a defining feature of "tauopathies," including Alzheimer's disease. Studies of human brain tissue and various model systems of tauopathy report that toxic forms of tau negatively affect nuclear and genomic architecture, identifying pathogenic tau-induced heterochromatin decondensation and consequent retrotransposon activation as a causal mediator of neurodegeneration. On the basis of their similarity to retroviruses, retrotransposons drive neuroinflammation via toxic intermediates, including double-stranded RNA (dsRNA). We find that dsRNA and dsRNA sensing machinery are elevated in astrocytes of postmortem brain tissue from patients with Alzheimer's disease and progressive supranuclear palsy and in brains of tau transgenic mice. Using a Drosophila model of tauopathy, we identify specific tau-induced retrotransposons that form dsRNA and find that pathogenic tau and heterochromatin decondensation causally drive dsRNA-mediated neurodegeneration and neuroinflammation. Our study suggests that pathogenic tau-induced heterochromatin decondensation and retrotransposon activation cause elevation of inflammatory, transposable element-derived dsRNA in the adult brain.

Fig. 1.. Retrotransposon lifecycle and toxicity.

(A) Retrotransposons mobilize via a copy-and-paste mechanism, where activation of retrotransposons within genomic DNA involves transcription of retrotransposon RNA, reverse transcription into cDNA, and insertion into a new location within the genome. Depending on the subclass of retrotransposon, the element may encode for proteins needed to facilitate the copy-and-paste mechanism, such as capsid, protease, reverse transcriptase, integrase, and envelope proteins (2). (B) While retrotransposition can create novel insertions, toxicity can also result from (i) retrotransposon RNA, (ii) retrotransposon proteins, (iii) dsRNA generated from nascent retrotransposon transcripts, (iv) episomal retrotransposon DNA, and (v) DNA double-strand breaks from failed retrotransposon insertions.

Fig. 2.. dsRNA and dsRNA-sensing machinery are elevated in human tauopathy and localize to astrocytes.

(A) Visualization of dsRNA in postmortem human frontal cortex via J2 immunostaining. DAPI, 4′,6-diamidino-2-phenylindole. (B) Quantification of (A). A.U., arbitrary units. (C) Visualization of dsRNA and astrocytes in postmortem human frontal cortex via J2 and GFAP coimmunostaining. (D) Quantification of (C); J2 fluorescence within GFAP-positive cells. (E) Visualization of MDA5 and astrocytes in postmortem human frontal cortex via MDA5 and GFAP coimmunostaining. (F) Quantification of (E); MDA5 fluorescence within GFAP-positive cells. Scale bars, 50 μm. n = 8 biological replicates. **P < 0.01; ***P < 0.001, one-way analysis of variance (ANOVA). Braak 0, little to no detectable pathological forms of tau in the entorhinal region; Braak II/III, pathological forms of tau detected in the entorhinal region as far as the occipitotemporal gyrus; Braak V/VI, pathological forms of tau detected in the entorhinal region extending into the occipital lobe and neocortex. Human cases are described in table S1.

Fig. 3.. dsRNA and dsRNA sensing machinery are elevated in astrocytes of the rTg4510 mouse model of tauopathy at 6 months.

(A) Immunofluorescence-based detection of dsRNA in cortex of control and rTg4510 mice using the J2 antibody. (B) Quantification of (A). (C) Immunofluorescence-based detection of dsRNA and astrocytes (GFAP) in control and rTg4510 mouse cortex. (D) Quantification of GFAP-positive astrocytes in (C). (E) Quantification of the co-occurrence of dsRNA and GFAP in GFAP-positive astrocytes in control and rTg4510 mouse cortex. (F) Immunofluorescence-based detection of MDA5 and astrocytes in control and rTg4510 mouse cortex. (G) Quantification of MDA5 in (F). (H) Quantification of the co-occurrence of MDA5 and GFAP in GFAP-positive astrocytes in control and rTg4510 mouse cortex. All mice were aged 6 months. Scale bars, 50 μm. n = 6 biological replicates. *P < 0.05; **P < 0.01; ****P < 0.0001, unpaired t test.

Fig. 4.. dsRNA is elevated in neurons and astrocyte-like glia of tau transgenic Drosophila and causally mediates neurotoxicity.

(A) J2/K2 ELISA-based quantification of dsRNA levels in heads of control and tau transgenic Drosophila. (B) Immunofluorescence-based detection of dsRNA in control and tau transgenic Drosophila using the J2 antibody. (C) Quantification of (B). (D) Localization of dsRNA in elav-positive neurons in control and tau transgenic Drosophila. (E) Localization of dsRNA in repo-positive astrocyte-like glia in control and tau transgenic Drosophila. (F) Quantification of (E). (G) ELISA-based quantification of dsRNA in heads of tau transgenic Drosophila and tau transgenic Drosophila with genetic pan-neuronal overexpression of Dicer-2 (Tau+Dcr-2^OE). (H) TUNEL-based quantification of neurodegeneration in brains of tau transgenic Drosophila and tau transgenic Drosophila with genetic pan-neuronal overexpression of Dicer-2. All experiments were performed at 10 days of adulthood; n = 20 biological replicates for (A) and (G), where each replicate consists of total RNA lysates from six pooled heads analyzed in triplicate. n = 10 biological replicates in (B) to (F); n = 6 biological replicates for (H). Scale bars, 10 μm. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001, unpaired t test.

Fig. 5.. Identification of retrotransposons that form dsRNA in brains of tau transgenic Drosophila.

(A) NanoString-based quantification of retrotransposon transcripts in J2-immunoprecipitated RNA relative to input. Each dot is a retrotransposon. A positive log₂ fold change (x axis) indicates increased dsRNA formation of that retrotransposon in tau transgenic Drosophila. For retrotransposons that are significantly enriched in the dsRNA fraction (−log₁₀ of P < 0.05, labeled in green), individual NanoString nCounter counts per sample are provided in (B). n = 7 biological samples; each sample is composed of five female and five male heads at day 10 of adulthood. Code set probe sequences are included in table S2. *P < 0.05, multiple unpaired two-sample t test with Welch correction.

Fig. 6.. Pan-neuronal expression of heterochromatin decondensation is sufficient to elevate dsRNA levels in the Drosophila brain.

J2/K2 ELISA-based quantification of dsRNA levels in control and Drosophila with pan-neuronal RNAi-mediated reduction or loss of function of (A) Su(var)205, (B) Su(var)3-9, (C) rhino, (D) piwi, (E) aubergine, and (F) Argonaute 3. All experiments were performed in heads of 10-day-old adult flies; n = 20 biological replicates, each biological replicate is a pool of three female and three male heads. n.s., not significant; *P < 0.05; **P < 0.01, unpaired t test.

Fig. 7.. Tau- and heterochromatin decondensation–induced elevation of dsRNA are causally associated with innate immune activation.

(A) Schematic of the four immune pathways assayed via NanoString for gene expression, illustration created using BioRender.com. A full list of Drosophila genes and their human homologs can be found in table S3. PGRP-SA, Peptidoglycan recognition protein SA; Socs36E, Suppressor of cytokine signaling at 36E; Tab2, TAK1-associated binding protein 2; Tak1, TGF-β activated kinase 1. (B to F) Normalized gene expression counts (nCounts) of innate immune gene transcripts quantified by NanoString gene expression assay using a custom code set. Gene expression is reported for (B) Tau transgenic, (C) Tau + Dicer-2 overexpression (tau + Dcr-2^OE), (D) Su(var)205^RNAi, (E) Su(var)3-9^RNAi, and (F) rhi^RNAi Drosophila as compared to control, which was set to one, with the exception of (C), in which transcript levels in tau transgenic Drosophila were set to one. Genes are listed by pathway in order from upstream ligand to downstream antimicrobial peptide and color-coded by immune pathway: Toll in blue, IMD in green, Jak/STAT in purple, and RNAi in yellow. n = 6 biological replicates; each replicate consists of a pool of three female and three male heads. Code set probe sequences are listed in table S4. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001, multiple unpaired two-sample t test with Welch correction. AGO2, Argonaute 2.