Uncovering the dynamics and regulatory logic of cell fate specification in cerebral organoid data

Yale University, New Haven CT

Investigators

Abstract

A deep understanding of the gene regulatory logic that controls human brain development is essential for un- covering the mechanisms of neurodevelopmental disorders and designing treatment protocols. Here, we com- bine cortical organoids, single cell CRISPR screens, spatial transcriptomics and metabolomics with a suit of state-of-the-art computational approaches developed during the previous funding cycle to decipher the gene regulatory networks (GRNs) that control the establishment, expansion and maintenance of radial glia (RG), the stem cells of the human brain. Human cortex develops from initially uniform neuroepithelium through sequential steps of differentiation and maturation known as neurogenesis. Neurogenesis is fueled by the RG cells which self-renew to maintain their pool size and differentiate to form intermediate progenitors and mature neurons. Prior studies provided a description of human RG cells. However, the mechanisms regulating the emergence, maintenance and differentiation of RG cells remain poorly defined. To study the behavior and regulation of hu- man RG we have generated human embryonic stem cell-derived cortical organoids. We performed scRNA-seq profiling of organoids at three timepoints that represent the establishment, expansion and the maintenance phases of RG development. We identified distinct RG populations as well as major differentiation branches in- cluding intermediate progenitor cells, excitatory and inhibitory neurons, and glial cells. We hypothesize that dy- namic cell type-specific GRNs control the development and function of RG cells. We will decipher these regula- tory networks as follows. In aim 1, we will infer transcriptional networks of RG that control its establishment, expansion and maintenance in organoids. We will construct TF-GRNs for each organoid stage from scRNA-seq data using our newly developed RITINI graph ODE network and in silico perturbations. To validate these predic- tions, we will perform perturb-seq analyses of differentiating organoids with a pool of gRNAs targeting key nodes of TF-GRNs. The key TFs will be individually validated. In aim 2, we will identify external inputs such as signals from ECM or neighboring cells, and mechanical inputs (herein the RG niche) that control RG fate. We will carry out spatial transcriptomics profiling of the organoid cultures. We will enhance our dynamics learning networks MIOflow and RITINI to incorporate spatial context, cell shape and intercellular signaling to account for their ef- fects on RG development. The key findings will be validated with functional assays. In aim 3, we will utilize our newly developed GEFMAP neural network to predict metabolic states of RG and its progeny from single cell omics data. The key findings will be validated by metabolome profiling and functional analyses. Uncovering the molecular mechanisms driving the emergence, maintenance and differentiation of human RG is expected to have profound implications for understanding and treatment of neurodevelopmental diseases that result from defects in the RG cells. The suit of innovative computational approaches developed through this proposal will be applicable to a broad range of biological systems and questions.