Mapping and Resolving Genome-wide Gene Regulatory Networks in Human Hematopoietic Stem and Progenitor Cells

Regulation of gene expression is crucial in establishing cell identity and function. Chromatin accessibility and binding of transcription factors to target sites in DNA and the assembly of protein complexes that regulate gene transcription is one of many levels of control. The protein products of such transcription may be components of a regulatory network that in turn influences the transcriptional output of itself and other genes. These gene regulatory networks (GRNs) establish and maintain cell-type-specific gene expression while their dynamic remodelling contributes to cell trajectories that direct differentiation of stem cells to more mature cell states. Human blood stem cells residing in the bone marrow continuously repopulate billions of mature circulating blood cells including red blood cells, white blood cells and platelets, each day throughout postnatal life. These haematopoietic stem cells (HSC) undergo a series of differentiating events to give rise to more mature and less stem-cell-like “progenitor” populations enroute to the production of mature circulating cells. Corruption of GRNs may lead to aberrant proliferation and differentiation of these progenitors resulting in bone marrow failure and/or leukaemia. As such, studying these networks in stem and progenitor populations in the bone marrow could provide vital information regarding normal blood cell development in humans. My study explores the interplay of a heptad of transcription factors- FLI1, ERG, GATA2, RUNX1, TAL1, LYL1 and LMO2, that play important roles during blood development. To map the binding patterns of these factors and construct gene regulatory networks in various primary stem and progenitor populations, I fractionated cells including HSCs, common myeloid progenitors (CMP), granulocyte monocyte progenitors (GMP) and megakaryocyte erythroid progenitors (MEP) using fluorescence-activated cell sorting and extracted chromatin from these populations for downstream assays. These included a) chromatin immunoprecipitation (ChIP) and ChIPmentation for genome-wide identification of active (H3K27ac, H3K4me3) and inactive (H3K27me3) histone marks, and transcription factor and co-factor (including CTCF and PU.1) binding. b) HiC to broadly classify the higher order 3D genome structures including compartments and topologically associated domains and c) H3K27Ac HiChIP to identify active looping structures between regulatory and promoter regions. I have shown that the heptad transcription factors exhibit shared and distinct patterns of binding across the stem and progenitor populations, with a preference for binding to regulator-like regions. These transcription factors exhibit combinatorial binding, with the binding of all seven factors showing highest significance. Interestingly, across lineage determining genes such as GATA1 and MPO, I noticed a dynamic accumulation of the heptad transcription factors at gene promoters as cells became more differentiated. Higher order genome architectures were conserved across the four cell types, while chromatin loops between gene promoters and distal regulatory regions showed cell type specificity. On resolving the regulatory network of the seven individual transcription factor genes, I found an interconnected network displaying combinatorial binding that was asymmetric across the four stem and progenitor populations. I was able to connect candidate distal gene regulatory regions with specific gene promoters and relate differential transcription factor binding to differential gene expression in relevant cell populations. Furthermore, I noticed patterns of binding that changed along the differentiation arc across the regulatory regions of genes expressed in mature cells. For example, at gene loci expressed in monocytes/granulocytes, there was an increase in GATA2, ERG, LYL1 and LMO2 and a decrease in TAL1 binding in GMP with respect to other populations. In contrast, at regulatory regions of genes expressed in erythroblasts/megakaryocytes, I noticed an increase in TAL1, GATA2, LYL1 and LMO2 and a concomitant decrease in ERG binding in MEP. Finally, by combining datasets generated in this study, I clustered 85,100 accessible regions present in HSPCs based on their regulatory potential and used transcription factor binding in stem and progenitor populations as a scaffold to map usage of candidate gene regulatory regions during hematopoietic stem cell differentiation. Taken together, my study provides a comprehensive characterisation of the genome wide gene regulatory landscape in rare human blood stem and progenitor cells. The results of this study constitute an important framework for accurate analysis of aberrant regulatory networks in leukemic cells and assist in devising better therapeutic strategies.