Submitted Abstract
The ultimate goal of tissue engineering and regenerative medicine is to create functional human tissue that can replace or trigger repair mechanisms in tissues or organs that fail due to illness, age, congenital abnormalities or traumatic injury. Although substantial progress has been achieved with novel experimental techniques for cellular conversion, characterizing the cells of interest for cell replacement therapy, and reproducing the same phenotype in vitro still poses problems. In order to guide experiments, computational tools have been developed to identify driver gene combinations that trigger transitions between cellular phenotypes. However, the questions of fidelity (the similarity between the differentiated cells and the desired cellular phenotype) and efficiency (what portion of the original cell population is successfully converted) remain a challenge. These properties of cellular transitions are utmost important for regenerative medicine in order to produce the necessary quality and quantity of cells, yet hard to tackle computationally. This is because current computational methods are predominantly, based on available bulk population studies, where the resolution of the data limits the optimization of fidelity and efficiency of cellular transition. Moreover, if cellular phenotypes are too close to each other, they are possibly indistinguishable at the transcriptomic level, and hence difficult to distiguish them at the gene regulatory level. In this regard, epigenetic information might hold the key to infer Gene Regulatory Networks (GRNs) for these cell types.Recent developments in profiling gene expression at single-cell level provide rich datasets reflecting heterogeneity within a cellular population. The high resolution and statistical descriptive power of this data allows classification of cells into distinct subpopulations and the reconstruction of subpopulation-specific GRNs. Moreover, this type of analysis opens the doors to the characterization of genes that may act synergistically in order to drive cellular transitions between cell subpopulations, which would be impossible in the case of bulk data. Therefore, the goal of this grant is to develop a computational framework that makes use of single-cell transcriptomics and subpopulation-specific epigenetic data in order to design cellular differentiation strategies with high efficiency and fidelity. The proposed computational framework is general and can be applied to any cellular system. Further, we will demonstrate its applicability for inducing dopaminergic neuron development with potential applications to Parkinson’s disease.In particular, we will reconstruct subpopulation-specific gene regulatory networks by integrating the transcriptome from single-cell RNA-sequencing with epigenetic marks in the same samples followed by a topological and information theory-based analysis to (1) identify the steps of human ventral midbrain development; (2) identify transcription factors critical for development, maintenance and subtype specification of human midbrain dopaminergic (mDA) neurons; (3) apply such factors to improve differentiation of embryonic stem cells and long-term neuroepithelial stem cells.