Defining the gene regulatory roles of non-coding variants in the pathogenesis of autism

ABSTRACT In this Predoctoral Fellowship proposal, I will be trained for a future as a physician-scientist with my own independent research program at the interface of genomics, computational biology, and neuroscience. During my MD/PhD training, I will be co-mentored by two physician-scientists, Drs. John Greally (Medical Genomics) and Pablo Castillo (Neurology), addressing a question that is timely with the imminent widespread use of whole genome sequencing (WGS) in clinical diagnostics – how do we interpret variants in the non- coding majority of the human genome when a patient presents with a medical problem? I will focus on autism, as a condition that represents a substantial proportion of patients seen by medical genetics services, for which there is extensive WGS information from thousands of families. Despite this wealth of research sequencing, only a small minority of individuals with autism receive a positive outcome of diagnostic exome or WGS. I propose that the currently limited diagnostic success rates are mostly due to our inability to interpret pathogenic variants in the non-coding majority of the genome of these patients. By improving our insights into non-coding variants, we will be able to offer diagnostic information to many more families seeking answers than currently. My strategy is to focus on de novo variants (DNVs) in offspring with autism born to unaffected parents. My hypothesis is that DNVs can be pathogenic when they occur in the cis-regulatory regions of cell types mediating autism. The project is therefore based on a computational genomics foundation, using WGS and DNV calls from large datasets from thousands of families who have a member with autism. In my preliminary data, I show that DNVs in individuals with autism are enriched at cis-regulatory loci in glial and neuronal cells in particular, and at genes known to be causative for autism. In my project, I will test these associations more rigorously, and will define a high-confidence set of DNVs for functional testing. Two types of functional testing will be performed. One will test whether the DNVs alter molecular genomic properties, including chromatin accessibility and gene expression. The second will test the physiological properties of the cells. To generate the appropriate cells for testing, I plan to use induced pluripotent stem cells (iPSCs) that are in vitro differentiated to GABAergic neurons and astrocytes. By performing CRISPR-mediated genomic editing in the iPSCs, I can generate cells with the candidate pathogenic DNVs, and test whether they have effects on cellular properties like dendritogenesis, synaptogenesis, and electrophysiology, increasing the confidence that these DNVs have pathogenic effects. I will have the privilege of getting training in sophisticated computational, stem cell and neuroscience techniques, under the guidance of two leaders in their fields, as part of a comprehensive training plan that will equip me to become the independent physician-scientist I aspire to be in my career.