Lineage, cell state and cell identity in the developing human brain

Abstract

Organisms are complex systems consisting of thousands of distinct types of cells that perform a wide array of functions, yet for the most part share the same DNA. To regulate the expression of genes, the genome contains a vast amount of complex regulatory sequences that bind transcription factors and control the activation of the transcriptional machinery. This gene regulatory process is key in the diversification of cell types and development in general. The brain is a particularly complex tissue consisting of thousands of distinct neurons. In this thesis I present two papers and a preliminary data section that are all aimed at developing and applying high throughput single-cell methods for the description of cellular heterogeneity in the brain.

Paper I describes the development of a novel probe-based spatial transcriptomics method that increases throughput and extends application of smFISH to high autofluorescence tissues. This is achieved through capture of RNAs on a slide through electrophoretic transfer, followed by removal of tissue.

In Paper II we used scATAC-seq and single-cell multiomics to generate an atlas of chromatin accessibility and gene expression in the first trimester developing human brain. Using this atlas, we identified key transcription factors in the development of the nervous system, for instance members of the NFI family that play a role in maturation in separate branches of the tree of neural lineages. We also trained a machine learning model to predict enhancer specificity between different immature neuron types and used this model to explain temporal dynamics in the regulation of Purkinje neuron genes. Finally, we used our dataset to predict cell types that might be involved in psychiatric disorders, finding that midbrain-derived GABAergic neurons might be disproportionately at risk in major depressive disorder.

Finally, in the preliminary data section we introduce a new protocol for combined long term live imaging and in situ sequencing of human tissue samples. We applied this method to the first trimester cortex and to patient-derived glioblastoma samples. In the glioblastoma samples we were able to track nearly 1,000 cells that we could then also annotate based on gene expression. Interestingly we found that many cells were migrating along a blood vessel and that tumor cells that are similar to neural progenitor cells tended to move greater total distances than other tumor cells.