Single cell -omic landscapes in normal and perturbed haematopoiesis

Abstract

The use of single cell genomics technologies to generate atlases of healthy tissues has served both as a valuable comparator resource for perturbed datasets, and to permit the discovery of novel biological processes in the differentiation of stem cells into mature tissues. Large scale profiling of human cells now yields tissue scale compendia extending to millions of cells. In the field of haematopoiesis, much of our knowledge is derived from the study of model organisms, of which the most commonly used is the mouse. Unbiased single cell datasets of murine haematopoietic stem and progenitor cells have proved invaluable to drive new biological discovery and serve as comparators for perturbation analyses. However, as yet, no high-resolution tissue-scale atlas of whole bone marrow has been generated. Work in this thesis describes the generation of a new single-cell multiomic atlas of healthy adult murine haematopoiesis, incorporating 200,000 single cell transcriptomes and surface proteome information from 138 antibodies. While consensus is growing over best practices for single cell transcriptomic analysis, analysis of single cell surface proteome and multiomic integration remain fields in development. Chapter 3 discusses the establishment of a multiomic bioinformatic analysis pipeline, including benchmarking analysis of single-cell proteomic data and multiomic integration. Unbiased single cell datasets of haematopoietic stem and progenitor cells have provided insights into stem cell biology and early lineage choice. Flowcytometric sorting and transcriptional or functional profiling of more mature oligopotent or unipotent bone marrow constituents has elucidated the later processes in the maturation of multiple haematopoietic lineages. The primary aim of this work was to generate an unbiased tissue-scale dataset of all haematopoietic bone marrow constituents, enriched for rare early cell states, to uncover the full differentiation process from stem cells to mature cell types ready to leave the bone marrow. Chapter 4 discusses the annotation of the multimodal landscape into cell states, leveraging transcriptional and proteomic modalities, information from published datasets and ‘in silico FACS’ to situate cell states in our dataset within the wider body of cell types already described. Densely sampled single cell landscapes have the powerful capacity to permit reconstruction of differentiation hierarchies through computational alignment along an axis of molecular similarity. These differentiation trajectories can then serve as a substrate for discovery of molecular and gene regulatory processes contributing to lineage choice and differentiation. Chapter 5 describes the assignment of cells into 10 differentiation trajectories from stem cells to mature cell types ready to exit the bone marrow. Transcriptional and gene regulatory processes which occur along the axis of differentiation from stem cell to mature cell type are examined, highlighting shared and lineage specific mechanisms. The question of early lineage bias is also addressed through the application of an algorithm to measure stochastic transcriptional noise in progenitors undergoing rapid changes in cell fate. This analysis points to early lineage bias being accompanied by transcriptional noise in lineage-specific transcripts and transcription factors, which may reflect stochastic transcription at accessible chromatin prior to onset of committed gene expression. The role of epitranscriptomic modifications in stem cell function and differentiation are increasingly being recognised. The RNA methyltransferase METTL3 plays an important role in haematopoiesis and is an essential dependency in Acute Myeloid Leukaemia; small molecule inhibitors are currently in early phase clinical trials in cancer therapy, generating interest in the specific effects catalytic inhibition may have on haematopoiesis. As yet, our knowledge of the role of this enzyme in haematopoiesis come from the pleiotropic effects reported in mouse knock-out models. Chapter 6 describes the results of single cell RNA-sequencing experiments elucidating the effects of catalytic inhibition of METTL3 on healthy haematopoietic stem and progenitor cells in vivo. The effects of inhibition of METTL3 on kinetics of RNA metabolism, including transcription, splicing and degradation rates are also examined using two computational approaches. Overall this thesis has applied single-cell multiomic analysis to haematopoietic differentiation in the healthy and perturbed setting to uncover novel gene regulatory processes which may govern lineage bias and commitment in the blood system.