Genome-encoded cytoplasmic double-stranded RNAs, found in C9ORF72 ALS-FTD brain, propagate neuronal loss

Abstract

Triggers of innate immune signaling in the CNS of patients with amyotrophic lateral sclerosis and frontotemporal degeneration (ALS/FTD) remain elusive. We report the presence of cytoplasmic double-stranded RNA (cdsRNA), an established trigger of innate immunity, in ALS-FTD brains carrying C9ORF72 intronic hexanucleotide expansions that included genomically encoded expansions of the G4C2 repeat sequences. The presence of cdsRNA in human brains was coincident with cytoplasmic TAR DNA binding protein 43 (TDP-43) inclusions, a pathologic hallmark of ALS/FTD. Introducing cdsRNA into cultured human neural cells induced type I interferon (IFN-I) signaling and death that was rescued by FDA-approved JAK inhibitors. In mice, genomically encoded dsRNAs expressed exclusively in a neuronal class induced IFN-I and death in connected neurons non-cell-autonomously. Our findings establish that genomically encoded cdsRNAs trigger sterile, viral-mimetic IFN-I induction and propagated death within neural circuits and may drive neuroinflammation and neurodegeneration in patients with ALS/FTD.

Copyright © 2021 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of Science. No claim to original U.S. Government Works.

Figures

Fig. 1.. CdsRNA in ALS/FTD C9ORF72 brains is coincident with TDP-43 cytoplasmic inclusions, is comprised of G4C2 repeats, and increases PKR quantities.

(A) Immunofluorescence staining with anti-dsRNA and anti-p-TDP43S409/S410 antibodies on frontal lobe and motor cortex sections from ALS/FTD C9ORF72 patients (orange arrowheads= dsRNA and p-TDP43S409/S410 positive cells, cyan arrowhead= dsRNA and p-TDP43S409/S410 negative cells). (B) Reversed transcribed-PCR amplification with primers to the region of the C9ORF72 expansion from cdsRNA immunoprecipitation from frontal cortex and cerebellum from ALS/FTD-C9ORF72 patients (n=3 biological replicates). (C) Quantitation of Western blots of PKR from ALS/FTD-C9 (n=8 biological replicates) or control frontal cortices (n=6 biological replicates). (D) Anti-phospho-PKRT451 immunostaining of motor cortex from an ALS/FTD C9ORF72 patient.

Fig. 2.. CdsRNA induces IFN-I signaling and neuronal cell death in differentiated human neural cells in vitro and mouse neurons in vivo.

(A) Western blots showing induction of p-STAT1Y701, MDA5, and PKR in differentiated human neurons 24 hours post-transfection with a dsRNAmi. (B) Quantitation of neuronal viability using the CellTiter-Glo assay 48 hours post transfection with a dsRNAmi at 0.25 (p=0.4807, n=4 biological replicates), 0.5 (p=0.891, n=4 biological replicates), 1 (p=0.0031, n=4 biological replicates), and 2 μg/mL (p=0.0005, n=4 biological replicates). (C) Quantitation of cell survival in WT or MDA5 knockout cell lines treated with 2 μg/mL dsRNAmi. (D) Quantitation of cell survival in differentiated ReNcell VM cells pre-treated with the FDA-approved JAK inhibitors, ruxolitinib, baracitinib, and tofacitinib, then treated with 2 μg/mL dsRNAmi. (E) Schematic showing mice expressing GFP in mature OSNs were instilled with AAV9 viral constructs expressing either sense or antisense GFP. (F) Coronal sections of OE stained by immunofluorescence using an anti-dsRNA in 4 month old mice instilled AAV9 expressing sense or antisense GFP, and coronal sections of mouse OE stained by RNA in situ hybridization with a digoxigenin labeled probe for OMP in control or mice instilled with AAV9 expressing sense or antisense GFP. Scale bar= 50 μm (G) Quantitation of height of OE positive for OMP staining (p<0.0001, n=4 pairs of animals, 4 months old) and counts of OMP-positive cells (n=3, p=0.028) each normalized to controls.

Fig. 3.. Neuronal cdsRNA produced from a structural genomic lesion in the Nd1 mouse line markedly activates IFN-I signaling.

(A) Color coded schematic of the transgenic construct, sequencing reads from RNA-seq after dsRIP mapped below, and a cartoon schematic of the structure of the integrated transgenic sequences and their chromosomal locations (flanking sequences) (to scale). Double slash (//) represents regions of the transgene where Sanger sequence did not annotate. (B) Coronal sections of mouse OE stained by tyramide amplified immunofluorescence with an anti-dsRNA antibody. Scale bar= 25μm. (C) PCR amplification of transgenic hAPP from cDNAs generated from RNA isolated by dsRIP, but not from immunoprecipitation with an isotype (IgG2a) and concentration-matched antibody, or dsRIP immunoprecipitates preincubated with a synthetic dsRNA mimetic. (D) Pathway schematic modified from Ingenuity Pathway Analysis software showing genes belonging to IFN-I pathway that are unregulated (red) or unchanged (gray); fold-change in expression is shown for individual genes. (E) Coronal sections of mouse OE stained by RNA in situ hybridization with probes for Stat1, IFI44, and OASL2 in control or Nd1 transgenic mice. Scale bar = 50μm (F) Western blots showing the induction of p-STAT1Y701, PKR, MDA5, and reduction of MAVS proteins in control or Nd1 transgenic mice.

Fig. 4.. IFN-I signaling propagates across multiple synapses in the mouse brain in response to neuronal restricted, genomically encoded cdsRNA expression.

(A) Schematic showing synaptic connections made by axons of OSNs (green) with dendrites of mitral cells (Mi; orange) in the OB, and granule neuron (Gr; maroon) synaptic connections with mitral cells, but not OSNs. (B) Coronal section of the OB from Nd1 mice stained by immunofluorescence for GFP (transgene), highlighting that axons of OSNs terminate on the surface of the OB (4x). (C) RNA in situ hybridization with a probe to GFP (transgene) on a coronal section containing both OE and OBs. (D) RNA in situ hybridization with probe to OASL2 in coronal sections of OBs from Nd1 transgenic mice.

Fig. 5.. Neuronal cdsRNA evokes non-cell autonomous neuronal death of adjacent and connected neurons.

(A) Quantitation of the number of cleaved-caspase3 positive cells in coronal sections of the OE in Nd1 mice normalized to littermate controls (green bar) or for mice treated with doxycycline for three months (light green bar); Nd1 p=0.011, n=5 biological replicates; Nd1 + dox p=0.25, n=3 biological replicates. (B) Quantitation of OSNs (all mature neurons labeled with GFP) relative to all cells in the OE quantified by FACS analysis, in Nd1 (green bar) and littermate controls (blue bar); p<0.0001, n=4 biological replicates. (C) Coronal sections of mouse OE stained by RNA in situ hybridization with probes for GFP (transgene), OMP (mature neurons) in control or Nd1 transgenic mice. Scale bar = 50μm. (D) OBs were smaller (outlined in red) in Nd1 relative to controls and OBs were significantly smaller by weight (E); p= 0.001, n=4 biological replicates. (F) Coronal sections of mouse OBs stained by immunofluorescence with an antibody for γH2AX in control and Nd1. (G) Quantitation of cells with pan-nuclear γH2AX staining per coronal section of OB normalized to the area of the OB in Nd1 relative to littermate control mice. p < 0.0001, n=3 mice, at lease 6 OB sections per mouse.

Fig. 6.. Propagation of cdsRNA between synaptically connected neurons in vivo.

(A) Whole mount images of mice feed Doxycycline (DOX) for 30 blocks transgene expression (green puncta). (B) PCR of transgenic sequence from immunoprecipitates using an anti-dsRNA antibody in mice fed DOX for 30 days. (C) Immunofluroescence with an anti-MDA5 antibody on sections of olfactory epithelia from Nd1 and littermate control mice, and quantification of integrated intensity, n= 4 biological replicates. (D) PCR of hAPP (transgenic sequence) from DsRIP assay on olfactory bulbs from Nd1 transgenic mice. (E) Immunohistochemistry with an anti-dsRNA antibody in olfactory bulbs from Nd1 transgenic mice. Scale bar = 50μm. (F) Immunofluorescence staining for MAVS in OB of Nd1 and control mice, and quantification of integrated intensity, n= 4 biological replicates.