Gene Expression Dynamics in Single Cells

Abstract

Dynamic regulation of gene expression is central to fate choice and differentiation. Genes drive changes in cell state and also serve as markers of mature cell types and immature intermediates. Advances in single-cell RNA-seq (scSeq) have made it possible measure the expression of thousands of genes in tens of thousands of cells in a single experiment. Single-cell RNA sequencing (scSeq) opens up a new terrain in the study of differentiation and fate choice, but it has an important limitation: existing methods kill cells in the process of measurement, and thus prevent following a cell’s state over time. The inability to track cells prevents a full understanding of the dynamic events that accompany differentiation, or how variation in the state of progenitor cells predisposes their sub- sequent fate choices. These limitations are particularly evident in hematopoiesis – the process of steady-state blood production in bone marrow. scSeq is increasingly used to characterize the heterogeneity of hematopoietic stem and progenitor cells, but the functional consequences of this heterogeneity have been difficult to ascertain. Here, several methods are described for inferring single-cell gene expression dynamics. Chapter 1 presents a visualization method called SPRING that facilitates human inference of dynamics. A more direct approach to infer dynamics from high dimensional data is described in chapter 2 based on the principle of population balance. This analysis also highlights several properties of gene expression dynamics that cannot be inferred from a single-cell snapshot, including: whether unmeasured variables contribute significantly to fate choice; the degree to which cells in the same gene expression state also have the same velocity in their state; and where fate commitment occurs in gene expression space. Chapter 3 describes an experimental approach for tracking cell state over time that involves clonally marking cells, letting them divide, measuring one sister right away, and measuring the other sister later on. Applied to hematopoiesis, this method partially addresses the questions raised in Chapter 2, demonstrating the existence of hidden variables during differentiation and mapping fate probability across the gene expression landscape. It also opens up new directions for perturbational analysis of hematopoietic fate choice, and may pave the way for linking cell state and fate in other systems.