Adenosine-to-inosine editing of endogenous Z-form RNA by the deaminase ADAR1 prevents spontaneous MAVS-dependent type I interferon responses

Abstract

Nucleic acids are powerful triggers of innate immunity and can adopt the Z-conformation, an unusual left-handed double helix. Here, we studied the biological function(s) of Z-RNA recognition by the adenosine deaminase ADAR1, mutations in which cause Aicardi-Goutières syndrome. Adar1^mZα/mZα mice, bearing two point mutations in the Z-nucleic acid binding (Zα) domain that abolish Z-RNA binding, displayed spontaneous induction of type I interferons (IFNs) in multiple organs, including in the lung, where both stromal and hematopoietic cells showed IFN-stimulated gene (ISG) induction. Lung neutrophils expressed ISGs induced by the transcription factor IRF3, indicating an initiating role for neutrophils in this IFN response. The IFN response in Adar1^mZα/mZα mice required the adaptor MAVS, implicating cytosolic RNA sensing. Adenosine-to-inosine changes were enriched in transposable elements and revealed a specific requirement of ADAR1's Zα domain in editing of a subset of RNAs. Thus, endogenous RNAs in Z-conformation have immunostimulatory potential curtailed by ADAR1, with relevance to autoinflammatory disease in humans.

Keywords: ADAR1; Aicardi–Goutières syndrome; MAVS; MDA5; RNA editing; Z-RNA; Zα domain; influenza A virus; interferon; neutrophil.

Graphical abstract

Figure 1

Mutation of ADAR1-p150’s Zα domain triggers spontaneous type I IFN responses in multiple organs (A) Levels of the indicated mRNAs were analyzed using qRT-PCR in RNA samples extracted from tissues of WT and Adar1^mZα/mZα animals and are shown relative to Gapdh. Each dot represents an individual mouse. N.D., not detectable. (B) Protein extracts from whole lungs from animals of the indicated genotypes were used for western blot with an α-ISG15 antibody. β-Actin served as a loading control. Each lane represents a sample from an individual mouse. (C–E) mRNA levels of the indicated ISGs were analyzed using qRT-PCR from cultured lung fibroblasts (C), BMMCs (D), and MEFs (E) of the indicated genotypes and are shown relative to Gapdh. Each dot represents cells derived from an individual mouse. Pooled data from biological replicates are shown with mean (A, D, and E) or mean ± SD (C) and were analyzed using unpaired t test (^∗p < 0.05, ^∗∗p < 0.01, and ^∗∗∗∗p < 0.0001). See also Figure S1.

Figure 2

Adar1^mZα/mZα lungs display a type I IFN gene signature Total RNA was extracted from lungs of three WT and three Adar1^mZα/mZα mice. Ribosomal RNAs were depleted before random-primed library preparation and RNA sequencing. About 100 million reads were obtained per sample. (Α) Differentially expressed genes were defined as displaying a fold change of ≥2 with an adjusted p value < 0.01. The 89 induced and 10 repressed genes were ordered by decreasing fold change and the data were clustered by sample. ISGs are indicated in bold. (B) GO analysis of induced genes. The top 20 GO terms (biological processes), ranked and ordered by p value, are shown. Diameters indicate the number of induced genes assigned to the GO term and colors show the p value. (C) Detected and differentially expressed REs were assigned to the indicated classes and are shown as pie charts. Differentially expressed REs were identified as having a minimum fold change of 2 and an adjusted p value of less than 0.01. See also Figure S1 and Table S1.

Figure 3

Stromal and hematopoietic cells induce ISGs in Adar1^mZα/mZα lungs (A) The proportion of hematopoietic (CD45+) and stromal (CD45−) cells in WT and Adar1^mZα/mZα lungs is shown. (B) mRNA levels of the indicated ISGs were analyzed by qRT-PCR using RNA extracted from whole lung, or from CD45+ or CD45− cells, and are shown relative to Actb. Fold increases relative to WT samples were calculated. Data points represent individual animals. In (A), data from a representative experiment are shown with mean ± SD. In (B), pooled data from two independent experiments including a total of six animals per genotype are shown with mean and were analyzed using unpaired t test (^∗∗p < 0.01, ^∗∗∗p < 0.001, and ^∗∗∗∗p < 0.0001). See also Figure S2.

Figure 4

Multiple hematopoietic and non-hematopoietic cell types display ISGs upregulation in Adar1^mZα/mZα lungs (A–C) mRNA levels of the indicated ISGs were analyzed by qRT-PCR using RNA extracted from cell populations sorted from lungs of WT and Adar1^mZα/mZα mice and are shown relative to Actb. B, B cells; T, T cells; DC, dendritic cells; Mono, monocytes; Mph, macrophages; NK, natural killer cells; Neut, neutrophils; Eos, eosinophils; Total, whole lung. (D) ISG mRNA levels were analyzed as in (A)–(C) in cell populations sorted from lungs of BM chimeric mice and are shown relative to Gapdh. Each data point represents an individual mouse. Because of the small number of epithelial cells recovered, samples were pooled from multiple mice before RNA extraction (C). Pooled data from two (A and B) or three (D) independent experiments are shown with mean (^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, and ^∗∗∗∗p < 0.0001, unpaired t test). See also Figures S3–S5.

Figure 5

Adar1^mZα/mZα mice are protected from early IAV infection (A) WT or Adar1^mZα/mZα mice were infected intranasally with 0.04 HAU of IAV strain A/X-31. Body weight was monitored daily and is shown as a percentage of starting body weight. (B–F) WT or Adar1^mZα/mZα mice were infected as in (A) or mock-infected using viral growth medium. On day 3 post-infection, lungs and sera were collected. (B) A “lung index” was calculated (lung weight/body weight × 100). (C) Levels of the viral NP and M transcripts were analyzed using qRT-PCR in RNA samples extracted from total lung. Data are shown relative to Actb (NP) or Gapdh (M). (D) Lung viral titers were determined in samples from infected animals by TCID₅₀ analysis and were normalized to lung weight. (E) Levels of the indicated mRNAs were determined as in (C). (F) Serum IL-6 concentrations were analyzed using ELISA. In (A), data from three independent experiments including a total of 15 mice per genotype were pooled (mean + SD; ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, and ^∗∗∗∗p < 0.0001, mixed-effects analysis). In (B)–(F), pooled data from two independent experiments (mock infected, n = 4 mice per genotype; A/X-31-infected, n = 10–16 WT and n = 9–13 Adar1^mZα/mZα mice) are shown. Each dot represents an individual mouse and the mean is indicated (^∗∗p < 0.01, ^∗∗∗p < 0.001, and ^∗∗∗∗p < 0.0001, unpaired t test). See also Figure S6.

Figure 6

ISG induction in Adar1^mZα/mZα mice is MAVS dependent (A and B) Levels of the indicated mRNAs were analyzed using qRT-PCR in RNA samples extracted from tissues of WT and Adar1^mZα/mZα animals that were either MAVS sufficient or deficient. Data are shown relative to Gapdh. (C) Mice of the indicated genotypes were infected intranasally with 0.04 HAU of IAV strain A/X-31. Body weight was monitored daily and is shown as a percentage of starting body weight. In (A) and (B), each dot represents an individual mouse, and pooled data from three independent experiments are shown (^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, and ^∗∗∗∗p < 0.0001, unpaired t test). In (C), data from two independent experiments including a total of 10–12 mice per genotype were pooled (mean with SD).

Figure 7

ADAR1-p150’s Zα domain is required for editing of a subset of RNAs (A) Editing sites were mapped in RNA sequencing reads from three WT and three Adar1^mZα/mZα lung samples (Z > 2.58). The numbers of edited sites were scaled to the total number of reads per sample. Each data point corresponds to an animal, and the mean is shown (ns, not significant; unpaired t test). (B) Editing frequencies for sites detectable in all three WT or Adar1^mZα/mZα samples (left; Z > 2.58) and for differentially edited sites (right; Z > 2.58, >2-fold) are shown as violin plots. Solid horizontal lines show the median and dotted lines indicate quartiles. (C) Editing sites detected in WT or Adar1^mZα/mZα samples and differentially edited sites were matched to annotated genomic features. The percentage of sites is shown for each category. (D) The number of expected and observed editing sites in WT samples are shown for families of REs for which either value exceeded 500. See text for details. (E) The distances of SINEs to their nearest inverted super-family member were determined for all SINEs and for SINEs harboring an editing site detected in WT or Adar1^mZα/mZα samples or containing a differentially edited site. Results are shown as proportions of SINEs with the maximum count set to 100. See also Figure S7.