T46 - In Vitro Human Neuronal Differentiation Validated As Genomic Model System To Study Major Psychiatric Illnesses

Abstract

Large-scale genetic studies of neuropsychiatric disorders have identified hundreds of susceptibility alleles. An important next challenge is to understand how identified genetic risk loci impact biological pathways and disease. For this, model systems are needed that recapitulate biological pathways in cell types that align with etiological mechanisms underlying the disease. We therefore aim to investigate an in vitro model of neuronal differentiation using human neural stem cells (hNSC) and evaluate its potential for studying molecular pathways important for major psychiatric illnesses based on genome-wide disease risk identified through GWAS. We differentiated WA09, a widely used hNSC line with standardized lab protocols, to a neuronal lineage across 30 days and assayed genome-wide gene expression profiles at seven time points in at least triplicates. We first used transition mapping (TMAP) together with publicly available brain transcriptome datasets to investigate overlap in transcriptional profiles between in vitro neuronal differentiation and in vivo human brain development. We furthermore setup a statistical pipeline tailored to time-series expression data that identifies genes with non-constant expression over time and subsequently performs soft clustering according to similarity in expression patterns using an empirical Bayes approach and fuzzy c-means clustering, respectively. We next applied stratified LD score regression, a statistical method that partitions heritability from GWAS summary statistics, to estimate how each unique identified cluster contributes to heritability of major psychiatric illnesses using results statistics of large GWAS (>5,000). We first show that in vitro transcriptional profiles significantly match in vivo human neurodevelopmental stages and cellular laminae of human cortical regions, which emphasizes the relevance of this model for studying brain function. We identified >30% of genes to be differentially expressed with high probability. These genes are important for in vitro neuronal differentiation and group to 10 clusters with distinct expression patterns. Biological annotation of these clusters highlights pathways and mechanisms important for neuronal differentiation, such as transcription factor activity, RNA processing, and synaptic transmission. Finally we show that differentially expressed genes are significantly enriched for heritability of schizophrenia, bipolar disorder, and major depressive disorder. There is no enrichment for height and Alzheimer’s disease, which is a late onset neurodegenerative disease. The enriched heritability partitions into specific clusters that can be distinct or shared between the three disorders with the strongest signal for schizophrenia in a cluster that contains genes related to synaptic function and transmission. In summary, we present an in vitro model of neuronal differentiation that at baseline activity shows overlap in transcriptional profiles with human cortical development. These profiles are enriched for genome-wide disease risk of major psychiatric diseases that partitions to specific identified gene clusters. This functional model is robust and simple and allows for genomic manipulations across an isogenic background in a controlled environment. This study validates WA09 neuronal differentiation as an in vitro genomic tool to study major psychiatric illnesses and provides directions for GWAS functional follow-up studies.