ADAR1 haploinsufficiency and sustained picornaviral RdRp dsRNA synthesis synergize to dysregulate RNA editing and cause multi-system interferonopathy

Abstract

Sensing of viral double-stranded RNA (dsRNA) by MDA5 triggers abundant but transient interferon-stimulated gene (ISGs) expression. If dsRNA synthesis is made persistent by transgenically expressing a picornaviral RNA-dependent RNA polymerase (RdRp) in mice, lifelong MDA5-MAVS pathway activation and marked, global ISG upregulation result. This confers robust protection from viral diseases, but in contrast to numerous other chronic MDA5 hyperactivation states, the mice suffer no autoimmune or other health consequences. Here, we find that they further confound expectations by being resistant to a strong autoimmunity (lupus) provocation. However, knockout of one allele of Adar breaks the autoinflammation-protected state of RdRp^tg mice and results in a severe disease that resembles interferonopathies caused by MDA5 gain-of-function protein mutations. Adar^+/- mice are healthy, but Adar^+/- RdRp^tg mice have shortened lifespan, stunted growth, premature fur graying, poorly developed teeth, skeletal abnormalities, and extreme ISG elevations. A-to-I edits are both abnormally distributed and increased (numbers of genes and sites). These results, with a nucleic acid-triggered and MDA5-wild-type model, illuminate the ADAR1-MDA5 axis in the regulation of innate immunity and establish that viral polymerase-sourced dsRNA can drive autoinflammatory disease pathogenesis.

Importance: RNA virus double-stranded RNAs (dsRNAs) are important pathogen-associated molecular patterns that are sensed by the RIG-I-like receptor MDA5, which triggers an acute innate immune response involving many interferon-stimulated genes (ISGs). One key to a healthy innate immune system is that MDA5 does not sense endogenous dsRNA. This is normally ensured by dsRNA duplex-disrupting ADAR1 editing of host dsRNAs. Picornavirus RdRp^tg mice have an unusual constitutive MDA5 activation state, with very high lifelong MDA5-mediated ISG expression that confers robust protection from diverse lethal viruses. Importantly, and in contrast to numerous other chronic MDA5 hyperactivation states, the mice develop no autoinflammatory consequences. If we delete one ADAR1 allele, however, which by itself is well tolerated, the mice develop a multisystem disease that resembles the human interferonopathy Singleton-Merten syndrome. In contrast to other MDA5/ADAR1 disease models, the MDA5 and ADAR1 proteins are both wild type in this dsRNA-driven model.

Keywords: ADAR1; MDA5; RdRp; autoimmunity; autoinflammation; innate antiviral immunity; interferonopathy; lupus; picornavirus.

Fig 1

RdRp^tg mice resist SLE-like disease induction in the BM12 model of lupus. Ten-week-old mice were challenged with 100 million BM12-derived splenocytes. Mice were harvested 14 days post-challenge for analyses. (A) Splenomegaly. The weights of BM12 and sham-injected mice’s spleens were measured. (B and C) Germinal center B cells, plasma cells, and Tfh cells. RBC-lysed, single-cell suspensions of splenocytes from BM12 or sham-injected animals were generated for flow staining and germinal center B cells and plasma B cells (B), or follicular helper cell populations (C) were measured by flow cytometry according to the gating described in Materials and Methods. (D) Anti-nuclear antibodies. Sera from BM12 splenocyte-injected or sham-injected mice were used to measure anti-double-stranded DNA (dsDNA) antibodies and anti-SmAg antibodies by quantitative ELISA. Data in panels A–D are from n = 7, 6, 8, and 8 mice for WT (PBS), RdRp^tg/– (PBS), WT (BM12), and RdRp^tg/– (BM12) groups, respectively. (E) T_reg cells. RBC-lysed single-cell suspensions were generated from spleens of untreated, 10-week-old WT or RdRp^tg mice (n = 5 for each group), and T_reg subsets were determined by flow cytometry. Data were analyzed using one-way analysis of variance (ANOVA), followed by Tukey tests for (A–D) and an unpaired Student’s T test for (E). *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Data points represent individual animals, and graphs show means with SD.

Fig 2

ADAR p150 and ADAR p110 levels in RdRp^tg mice and effects of ADAR knockdown on RdRp-dependent ISG expression. RNA isolated from brains of 4- to 5-week-old WT or RdRp^tg/– mice (n = 5 for WT, 6 for RdRp^tg/–) was used to measure (A) ADAR p110 and (B) ADAR p150 transcripts by quantitative PCR (qPCR). Data points represent individual animals, and graphs show means and SD. (C) Immunoblotting for ADAR p110 and ADAR p150 in 4- to 5-week-old mouse WT and RdRp^tg/– brain. n = 3 animals per genotype. (D) A549 cells with inducible Theiler’s virus RdRp (Tet-on system). Data are shown from two representative knockdowns. (E) RNAs isolated from the parallel knockdowns done in panel D were used for qPCR analysis to determine relative levels of the ISGs OAS and ISG15 mRNAs. mRNAs were harvested 54 hours after ADAR-targeting small interfering RNA (siRNA) addition and 48 hours after doxycycline (dox) addition. Comparisons were made between dox-treated and dox-untreated cells and expressed as fold mRNA changes induced by dox. Transfection controls (cells receiving transfection reagents but no siRNA) showed similar levels of ISG induction after dox treatment as control siRNA-transfected cells, indicating a lack of contribution of transfected siRNAs to immune activation. Data are means and SD of triplicate biological replicates with three technical replicates each. NS: not significant; *P < 0.05, **P < 0.01, and ****P < 0.0001; unpaired Student’s T test.

Fig 3

Phenotypic and histological differences in RdRp^tg/– Adar^+/– mice. (A) Size and coat color differences in 5-week-old, littermate RdRp^tg/–, and RdRp^tg/– Adar^+/– mice. (B) Weight differences between 4- and 5-week-old littermate RdRp^tg/– and RdRp^tg/– Adar^+/– mice (n = 9 and 5, respectively). (C) Dental developmental differences between RdRp^tg/– and RdRp^tg/– Adar^+/– mice. (D) Survival of animals from each group, followed for 20 weeks, and shown as Kaplan-Meier plot. (E) hematoxylin and eosin (H&E)-stained tissue sections from kidney, brain, lung, liver, heart, and spleens from RdRp^tg/– and RdRp^tg/– Adar^+/– littermate mice. Pathological grading revealed no significant scoring in RdRp^tg/– Adar^+/– animals. (F) Blood urea nitrogen from serum from 4- to 5-week-old animals in WT, Adar^+/–, RdRp^tg/–, and Adar^+/– RdRp^tg/– mice. n = 5, 6, 5, and 8 for WT, Adar^+/–, RdRp^tg/–, and Adar^+/– RdRp^tg/–, respectively. (G) Anti-SmAG antibodies from serum from 4- to 5-week-old animals as measured by quantitative ELISA. Four outlier high values were seen, but two were in the WT group. n = 15, 7, 10, and 9 for WT, Adar^+/–, RdRp^tg/–, and Adar^+/– RdRp^tg/–, respectively. Anti-dsDNA antibody levels from matched serum were below the limit of detection for all animals tested. Where not indicated, all data and tissue sections come from a mix of male and female mice. Data in panel B were analyzed by two-way ANOVA, where ****P < 0.0001. Data in panel D were analyzed using a log-rank test (Mantel-Cox), where ****P < 0.0001. Data in panels F and G were analyzed by one-way ANOVA, where *P < 0.05; data points represent individual animals, and graphs show means and SD.

Fig 4

Skeletal and dental features of WT, RdRp^tg/–, Adar^+/–, and Adar^+/– RdRp^tg/– mice. Femurs from 6-week-old animals were manually de-fleshed and used for all analyses. (A) Femur lengths of WT, Adar^+/–, RdRp^tg/–, and RdRp^tg/– Adar^+/– mice (n = 8, 6, 10, and 17 animals, respectively). (B) Representative µCT images, distal femur sections, and whole bone. (C) Cortical bone volume fraction (BV/TV), n = 8, 7, 10, and 17. (D) Trabecular number, n = 8, 7, 9, and 17. (E) Trabecular separation, n = 8, 7, 9, and 17. (F) Stiffness, n = 8, 7, 10, and 15. (G) Maximum load, n = 8, 7, 10, and 17. (H) Cortical total mineral density, n = 8, 7, 10, and 17. (I) Modulus, n = 8, 7, 10, and 17. (J) Ultimate stress, n = 8, 7, 10, and 17. Mice with gray fur are indicated by light-blue symbols. Data were analyzed first by using a ROUT test to remove outliers (Q = 1%), which resulted in the removal of two mice from one group in one panel (the RdRp^tg/– Adar^+/– group in panel F, stiffness testing; hence, there are 15 mice as opposed to the 17 for this genotype in the other panels). A one-way ANOVA comparing RdRp^tg/– Adar^+/– to each other group was used, followed by a Tukey test where *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Data points represent individual animals, and horizontal bars indicate means.

Fig 5

RNA-seq analysis reveals a clear pattern of ISG upregulation suggestive of an interferonopathy. RNA-seq from neuronal tissue from n = 4 age-matched (5 weeks) mice per group. Where possible, littermate controls were used for analysis. Two male and two female mice were used per group. (A) Multi-dimensional scaling plot of RNA expression profiles in WT-WT, Adar^+/–, RdRp^tg/–, and RdRp^tg/– Adar^+/– mice. (B) Heatmap of differentially expressed ISGs across the four genotypes (Fig. S3 shows an enlarged version with gene names annotated). FPKM values were log transformed with one pseudocount to facilitate visualization. (C) Summary table of differentially expressed genes (DEGs) and proportions of which are known ISGs as determined by the Interferome database. (D) Differentially expressed genes that are shared across all group comparisons. (E) RdRp mRNA abundance in the four genotypes. (F) Canonical molecular pathways from IPA that are significant across group comparisons.

Fig 6

ISG and leukocyte subset differences in RdRp^tg/– Adar^+/– mice. (A) qPCR of three representative ISGs (Ifit1, Isg15, and Oasl2) in brain tissue of 4- to 5-week-old WT, Adar^+/–, RdRp^tg/–, and RdRp^tg/– Adar^+/– mice (n = 5, 5, 6, and 5, respectively). (B–E) Flow cytometry analysis of cellular populations derived from the spleens of 4- to 5-week-old mice. Single-cell, RBC-lysed solutions were prepared for use in analysis. (B) BST-2 expression on the main immune cell subsets in the spleen, including T cells (CD3⁺), B cells (CD19⁺), DCs (CD11c⁺), granulocytes (Ly6G⁺), and NK cells (NKp46⁺). n = 7, 7, 4, and 4 for WT, Adar^+/–, RdRp^tg/–, and RdRp^tg/– Adar^+/–, respectively. (C) Monocyte populations across all four groups. (D) BST-2 expression on monocyte/DC cell subsets, including monocytes, cDC1s, cDC2s, and monocyte-derived DCs (moDCs). (E) Activation (CD80/86 expression) of monocyte/DC subsets, including monocytes, cDC1s, cDC2s, and moDCs. C–E: n = 12, 7, 10, and 8 for WT, Adar^+/–, RdRp^tg/–, and RdRp^tg/– Adar^+/–, respectively. For all graphs shown, data were analyzed using a one-way ANOVA followed by a Tukey test to determine significance, where *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Data points represent individual animals with the mean and SD shown as bars.

Fig 7

Analysis of the A-to-I editing reveals increased editing in RdRp^tg/– Adar^+/– mice. Changes in A-to-I edits were determined via comparison of RNA-seq data with whole-exome DNA sequencing. (A) Pie charts showing the proportion of editing sites within a given RNA element for each mouse genotype. (B) Table summary of the distribution of overall numbers of edited genes and sites for each genotype and the breakdown of locations of the edit sites among different RNA elements. For each of the four genotypes, four animals were sequenced, and to be counted, a gene must have been edited in all four animals sequenced, but editing can occur anywhere in the gene. For sites, identical sites must be edited in all four animals of a genotype. (C) Venn diagrams showing the overlap of edited genes (top) or sites (bottom) among the four genotypes. (D and E) Immunoblot analysis from two animals per group (D) and gene expression fold change (E) of three Adar^+/– RdRp^tg/– uniquely edited genes in the RNA-seq data set (as compared to WT) in 5-week-old neuronal tissue. Rig-I, Isg15, and Zbp1 were evaluated as they are each ISGs that are highly upregulated in mice expressing RdRp and are key regulators of the antiviral response. (F) Expression of Adar isoforms in neuronal tissue of 4- to 5-week-old mice from all four genotypes, as determined by western blot (n = 2 per group).