We report a novel AGS mouse model in which a single nucleotide replacement encoding an Adar D1113H mutation in the catalytic domain of ADAR1 resulted in early-onset IFN-signaling pathway activation in the brain, leading to mild microgliosis and astrocytosis at 8 weeks of age. In 8-month-old mutant mice, calcification in the DGM was observed in addition to microgliosis and astrocytosis. We demonstrated that blocking the RNA sensing pathway by deleting the cytosolic RNA receptor MDA-5 not only normalized ISG expression, but also prevented brain pathology in ADAR D1113H mice, including astrocytosis and microgliosis. Thus, this study demonstrated that an AGS-associated genetic mutation is a causative factor of the early-onset neuropathogenesis of AGS. The study illuminated the molecular consequence of this mutation in RNA transcript editing, as well as the signaling pathway underlying the disease development.
Mutations in seven protein-coding genes have been found to be associated with AGS. However, mouse models bearing these mutations did not sufficiently capture the clinical and neurological characteristics of AGS . Mouse models carrying ADAR1 mutations have been reported recently [2, 20,21,22,23,24]. The pathology changes observed in these models were more striking in the peripheral tissues than in the brain [20,21,22, 24]. ISGs were expressed in the mutant brains in some of these models [2, 24, 29], but they did not cause early-onset encephalopathy. Various ADAR1 knockout [17, 32,33,34,35,36] and mutant mouse models, including E861A [18, 28], W197A  and Y177A [20, 22], were used for ADAR1 mechanistic studies, whereas these mutations have not yet been identified in AGS patients. AGS-associated ADAR1 mutations such as Adar P195A (equivalent to Adar P193A in humans)  and K948N (equivalent to Adar K999N) [2, 24] were recently reported in mouse models. In AGS, the Adar P193A mutation is present as a compound heterozygous mutation. Mice bearing analogous heterozygous Adar P195A and a P150 null allele exhibited ISG upregulation in the brain; however, brain pathology was not reported, whereas homozygous Adar P195A did not differ from WT in its phenotype . Adar K999N mutation caused tissue injury in young mice in the liver, lung, kidney, and spleen, but not in the brain; mild neurologic changes and microscopic calcification were observed only in the brains of aged mice . Adar D1113H is a mutation found in AGS patients. In our mouse model, this mutation caused robust ISG expression preferentially in the brain and caused early-onset neuroinflammatory responses with astrocytosis and microgliosis.
Our RNA ISH studies on the mutant mouse brain demonstrate widespread multifocal expression of ISGs that does not obey neuroanatomical or neurophysiological boundaries. In addition, different cell types (e.g., neurons and microglia) in the same region show selective expression of distinct ISG. While ISG-15 was expressed in widespread neurons and ependymal cells, CXCL-10 expression was preferential to microglia. This specific ISG expression pattern suggests a microenvironmental stimulus stochastically initiating the ISG expression, since the same gene mutation is in all the cells. However, the regulatory mechanism is currently unknown.
Despite markedly elevated mRNA levels of ISG expression showing a clear “interferon signature” within the nervous system and that elevated protein levels are usually associated with increased coding mRNA levels as shown in Adar K999N mice , the gross brain morphology was impacted and there was no evidence of significant inflammatory cell infiltration, neuron loss, or other remarkable morphology changes. This was the case despite markedly elevated CXCL10, which has been associated with leukocyte infiltration in several CNS disorders , including when artificially expressed in astrocytes . Rather than inflammatory cell infiltration, we observed a mild neuroglia response, including both astrocytosis and microgliosis in 8-week-old mutant mice. These pathologic changes occurred earlier than those reported at one year of age in the Adar K948N mice . We also observed DGM calcification in the Adar D1113H mutant mice at 8 months of age.
However, Adar D1113H mice did not capitulate all the clinical and pathological features of AGS. Although early-onset and brain-predominant inflammation had developed in the mutant mice, severe brain injury manifested by progressive loss of motor ability, spasticity, dystonia, and obvious basal ganglia calcification was not observed. The reason for the difference is not currently known. Since the genetic features of Adar D1113H mutation in this mouse model is the same as that in AGS patients, the phenotypic difference might be due to immune response and environmental differences between mice and humans. The mice used in this study were kept in specific pathogen-free (SPF) conditions, which might have prevented environmental factors from exuberating the phenotype.
In AGS patients, most of the mutations in the ADAR1 gene fall in the catalytic domain, and all the homozygous Adar mutations, e.g., the three mutations coding Adar K999N, W1112F, and Adar D1113H, are only found in the catalytic domain [1, 10]. Although Adar D1113H is predicted to be a critical amino acid in the catalytic domain , decreased editing levels were only observed in mir 381 and BLCAP coding RNAs among the tested sites of known ADAR1 edited RNA substrates in Adar D1113H mutant mice, whereas the editing levels of intron + 60 sites on the receptor GIRAs, A and B sites on 5HTR2c and the editing site in Ube2o mRNAs remained the same as those of the control mice. In contrast, Adar K999N mutation showed a significant editing level decrease in more tested editing sites, while ISG expression in Adar K999N mutant mice was not as high as in Adar D1113H mutant mice. Considering that a large number of noncoding sequences are targets of RNA editing, we cannot exclude that there are some untested editing sites that were significantly affected by Adar D1113H mutation. A genome-wide measurement of the editing needs to be carried out to determine whether a specific group of RNAs that lost or gained editing with this mutation resulted in ISG expression. It is possible that editing in some specific RNA transcripts inhibits the RNA sensing signaling pathway. Recently, a study suggested that deficient editing of a small group of RNAs is sufficient to activate the IFN pathway . This Adar D1113H mutant mouse model will be a useful tool for identification of the specific RNA molecules involved in AGS development.
Previous studies showed that deletion of MDA-5 completely rescued the lethality of the catalytic null Adar1E861A/E861A mutant mice [18, 28] and prevented ISG expression in some Adar mutant mice [20,21,22,23,24, 29]. However, blocking RNA sensing by deletion of MDA-5 or MAVS was not sufficient to prevent the mortality of Adar knockout mice, including Adar1∆7–9 , Adar1∆2–13 knockout  or P150−/− mice in which MAVS deletion delayed perinatal death by a few weeks [19, 30]. The Adar D1113H mutant mouse is a novel model with profound ISG expression in the brain with early-onset neuropathologic changes. We demonstrated that ISG expression in Adar D1113H and Adar K999N mutant mouse models depended on MDA-5. Deletion of the MDA-5-encoding Ifih1 gene normalized ISGs in Adar D1113H mice and confirmed the restoration of ISGs in Adar K999N mice as previously reported in the Adar K948N (equivalent to Adar K999N) mouse model .
Our study provides evidence that MDA-5 deletion not only normalized ISG expression but also prevented brain pathology in Adar D1113H mice, including astrocytosis and microgliosis. The mechanisms linking these pathologic changes to MDA-5-mediated ISG expression are important to elucidate because they could provide a roadmap for treatment in AGS.