Autism spectrum disorders (ASD) encompass neurodevelopmental diseases that share core deficits in verbal and nonverbal language, reciprocal interactions, and stereotyped and repetitive behaviors. Unfortunately, the molecular etiology of ASD remains incompletely understood. Excitingly, the recent advent of next generation RNA sequencing has now enabled whole-genome characterization of RNA regulation, expression, and modification in ASD. One such RNA modification, that is highly prevalent in mammalian synapses yet not studied in ASD, is RNA editing. Thus, in this dissertation, we perform a comprehensive spatiotemporal and first genome-wide study of RNA editing in ASD across multiple implicated brain regions, genetic etiologies, and developmental time points spanning fetal development to adulthood. Collectively, we uncovered general trends and regulatory mechanisms of RNA editing relevant to neuronal tissue.
We first characterized RNA editing in the largest cohort of ASD postmortem brains to date. Strikingly the ASD patients exhibited convergent trends of global downregulated RNA editing (hypoediting) affecting synaptic development and transmission genes. The global hypoediting was observed across multiple brain regions and multiple syndromic forms of ASD including dup15q11.2-13.1 duplication syndrome patients and Fragile X syndrome patients. Network analyses and experimental work demonstrated that Fragile X proteins, FMRP and FXR1P, regulated many of the dysregulated editing sites in human brain.
Since postmortem brains only provide postnatal time windows for studying ASD, we next used organoid models to characterize the landscape of RNA editing over ASD fetal brain neurodevelopment. We generated the first large-scale dataset of hundreds of organoids modelling cerebral cortex development (cortical spheroids) over multiple time periods and encompassing a myriad of penetrant autism-susceptibility mutations. RNA editing gradually increased over cortical spheroid development both in control spheroids and ASD spheroids. However, at all developmental timepoints, the ASD cohort again displayed global trends of hypoediting. Functional enrichment analyses implicated the hypoedited RNA editing in cellular development and proliferation of radial glia, intermediate progenitors, and newborn neurons.
Throughout these ASD studies, we encountered difficulties running common statistical analyses due to distinctive properties of RNA editing data. Thus, we lastly developed a statistical framework to handle RNA editing data based on a beta-binomial distribution. We developed a method, called REDITs (RNA editing tests), to handle important RNA editing analyses including significance of case-control differences and regression associations with covariates. REDITs had demonstrably higher sensitivity and specificity on simulated and real RNA editing datasets when compared to the most alternative methods used in the RNA editing field.