Investigating high dimensional alternative splicing during haematopoiesis

The haematopoietic system is responsible for the daily turnover of trillions of blood cells in a human adult. It relies on the maintenance of a cellular hierarchy, i.e., a select few haematopoietic stem cells (HSCs) retain self-renewal abilities, whereas amplifying progenitors downstream of these HSCs give rise to mature blood cells of different types. Despite being the first stem cells to be isolated around 50 years ago (and the only ones in routine clinical use today), the molecular mechanisms underlying the elegantly maintained haematopoietic balance remain unclear. Over the years, gene-level analyses have helped in understanding that lineage specific differentiation relies on evolving combinations of gene regulatory factors, including lineage specific master transcription factors. More recently, it is becoming evident that a complex regulome, including transcription factors, RNA binding proteins, and different RNA species controls haematopoietic cell states. Genomic analyses of haematopoietic disorders, such as myelodysplastic syndromes, haematological malignancies, and clonal disorders that may develop into malignancies revealed that a subset of core spliceosomal and splicing factor genes are frequently mutated in these conditions. There is a need to elucidate the alternative splicing landscape during both aberrant and normal haematopoiesis in order to understand how mis-splicing disrupts the cellular hierarchy causing uni-/multi-lineage dysplasia. In contrast to the analysis of differential splicing during terminal lineage specific maturation, relatively fewer studies have focussed on differential splicing during normal early haematopoietic differentiation.Alternative splicing is an economic means for increasing protein diversity by generating a variety of mRNA products from a single genetic locus. Classically, the following types of AS events are most commonly analysed in “splicomics” studies: Cassette exon skipping, alternative 5’ or 3’ ss usage, mutually exclusive inclusion of exons, intron retention, mutually exclusive 5’ untranslated regions (UTRs), and mutually exclusive 3’ UTRs. This binary treatment is fair since most genes express a limited number of transcripts and the splicing changes are describable by these patterns. However, a subset of genes can express multiple, even thousands of unique transcripts. The best example being the Drosophila DSCAM gene, which can express > 30,000 isoforms through alternative inclusion of 95 coding exons. The regulatory mechanisms governing these exons’ inclusion do not conform to the generally ascribed mechanisms/ features associated with the classical binary patterns. In humans, multiple early and recent high throughput sequencing backed studies have revealed that significant isoform diversity indeed exists making it important to survey complex patterns and try to understand the regulatory principles behind their generation.In chapter 2 of this thesis, we present a pipeline, ASTA-P for mining, quantification, and differential splicing testing of arbitrarily complex splicing events. We demonstrate the pipeline results using bulk RNA-seq data from two groups of hiPSC derived cranial neural crest cells – pre-migratory neural plate border cells and migratory cranial neural crest cells, sorted according to their expression levels of the transcription factor, SOX10.In chapter 3, we use simulated RNA-seq data and compare the performance of our pipeline with published, commonly used splicing analysis methods and demonstrate that our pipeline provides a good trade-off between discovery and accuracy.       Chapter 4 explores the role of AS during fate decisions in the early stages of haematopoiesis through contrasting mRNA-seq data from a number of binary decision points, e.g., erythroid and megakaryocytic blast separation.Chapter 5 considers the role of AS during the continuous differentiation from neutrophil progenitor to mature neutrophil. Neutrophil differentiation from haematopoietic progenitor cells was studied ex vivo and samples collected over several time points.The final chapter summarises the overall conclusions drawn from this thesis and makes recommendations of future work.