Genomic profiling has revealed that acute myeloid leukaemias (AMLs) tend to have a relatively low mutation rate, but that many different driver genes may contribute to the development of the disease, complicating efforts to identify similarities between cases. Three AMLs (of whom 2, WEHI-AML-1 and WEHI-AML-2, were siblings) underwent detailed genomic analyses using different molecular techniques supplemented by sequencing data from international databases. The results from the siblings identified a novel inherited predisposition to AML involving the base excision repair protein, MBD4. AMLs deficient in MBD4 have a unique mutational signature characterised by elevated CG>TG mutations and all 3 acquired mutations in driver genes in a conserved order. The importance of MBD4 in maintaining genomic integrity was confirmed in other cancers, through the analysis of large public cancer datasets, and a genetically modified mouse model. The third AML, WEHI-AML-3, had DDX3X-MLLT10, a translocation that has only been reported in T-acute lymphoblastic leukaemias. As expected, analyses of the three AMLs at different clinical timepoints, showed the leukaemias underwent clonal evolution, providing further insight into their disease biology. For example, the leukaemia in WEHI-AML-3 acquired new mutations resulting in mechanisms for therapy resistance and subsequent poor clinical outcome. The results from the 3 AMLs have broader implications for other haematological malignancies and cancers.

Tumour Necrosis Factor (TNF) is one of the most potent pro-inflammatory cytokines and it is secreted in response to danger signals, such as those caused by pathogen infection. High levels of TNF have been associated with many chronic and inflammatory diseases, including rheumatoid arthritis (RA), inflammatory bowel disease (IBD) and psoriasis. To prevent high TNF levels and uncontrolled inflammation, Tnf mRNA is degraded when not required, making post-transcriptional regulation a central mechanism to control Tnf expression. Post-transcriptional control operates through cooperation between cis-elements present in the 3’UTR and trans-acting proteins such as RNA-binding proteins. Until recently, knowledge about post-transcriptional regulation of Tnf was limited to the role of the AU-Rich Element (ARE), and to a lesser extent to that of the Constitutive Decay Element (CDE). In 2015, our group identified a New Regulatory Element (NRE), which changed the view on the post-transcriptional regulation of Tnf. Importantly, we have discovered a cooperative mechanism between two or more elements to regulate Tnf mRNA stability in vitro. In this thesis, I have translated our previous in vitro observations, in vivo, using the CRISPR/Cas9 technology to generate mice with deletions of one or two regulatory elements in the Tnf 3’UTR. I showed that the variety of phenotypes of the mice changed greatly in severity, including a concomitant deletion of the ARE and the CDE causing embryonic death. This suggests that not only different cis-elements cooperate to destabilise Tnf mRNA efficiently, but the mechanisms involved also appear to operate in a cell- and tissue-specific manner. Furthermore, I aimed to characterise a new trans-acting RBP called ZC3H12C that was previously identified by our group as involved in the post-transcriptional regulation of Tnf in vitro. To study the physiological role of ZC3H12C and the consequences of its loss, and its expression in vivo, I engineered a mouse deficient in Zc3h12c, in which the green fluorescent protein GFP that can be used as a marker of expression replaces ZC3H12C. Zc3h12c-deficient mice are found as adults at the expected Mendelian frequency and look outwardly normal. In particular, they do not present with any phenotype related to an excess of TNF (like cachexia or arthritis), even at an advanced age. However, loss of Zc3h12c causes aberrations in the structure of secondary lymphoid tissues, and hypertrophic skin-draining lymph nodes with supernumerary B cells and inflammatory dendritic cells in ageing mice. Flow-cytometry analysis of our GFP-reporter mouse showed that dendritic cells (DCs) are the immune cell type expressing ZC3H12C the most. RNA-seq analysis on splenic DCs suggested that loss of Zc3h12c affected the anti-viral immune response. Accordingly, when challenged with chronic LCMV, Zc3h12c-deficient mice presented with an abnormally exaggerated immune response. I characterised the impact of the loss of Zc3h12c in the context of psoriasis to confirm previous Gene Wide Association Studies (GWAS) suggesting that Zc3h12c was one of the risk genes involved in psoriasis incidence in human. I found that loss of Zc3h12c did not impact psoriasis’ development, but this observation could be due to the limits of the psoriasis model used in my study. I further characterised the role of Zc3h12c in skin homeostasis by mimicking the Toxic-Epidermal Necrolysis disease using subcutaneous injection of SMAC-mimetics to induce TNF-dependent cell death in the skin. In this context, loss of Zc3h12c appeared to be beneficial and reduced the lesions and the cell death induced by the SMAC mimetic compound. Finally, to evaluate the potential role of TNF in this phenotype, I generated mice lacking both Tnf and Zc3h12c. While double-deficient (DKO) mice never developed lymphadenopathy, around 30% of the Tnf/Zc3h12c DKO mice developed severe multiorgan inflammation, including pancreatitis, myocarditis, otitis, myositis, pyelonephritis, anaemia, extramedullary haematopoiesis and bone marrow failure. Histopathological analysis suggested that concomitant loss of Tnf and Zc3h12c rendered the mice immunocompromised and potentially sensitive to the opportunistic pathogen Pasteurella pneumotropica, for which they tested positive. To evaluate the role of the TNF-TNFR2 signaling in the phenotype, and given the widely known role of TNFR2 in autoimmunity development, I generated Tnfr2 and Zc3h12c double-deficient mice. While I failed to observe a single Tnfr2/Zc3h12c DKO mouse falling sick, I also observed an absence of disease development in the Tnf/Zc3h12c DKO mice and this coincided with the clearance of P. pneumotropica from the animal facility. These observations raise new questions on the role of Tnf and Zc3h12c in the control of immune responses and inflammation, and further investigation will have to be conducted. Overall, my work suggests that Zc3h12c might be a risk factor in the context of anti-TNF treatment leading to autoimmunity in some patients.

The persistence of a replication-competent HIV reservoir necessitates life-long antiretroviral adherence and precludes the possibility of a HIV cure via conventional therapy alone. Furthermore, recent clinical studies have made it increasingly clear that the predominant strategy for reservoir elimination, enforced transcriptional reactivation, does not diminish the size of the latent reservoir or reduce the time to viral rebound following treatment interruption. A novel approach seeks to purge the HIV reservoir by activating apoptotic pathways in latently infected cells and shifting the balance away from survival and towards cell death. Several lines of evidence implicate Bcl-2 family proteins in the long-term survival of memory CD4+ T cells – the major reservoir for HIV. Bcl-2 antagonism thus represents a viable strategy for sensitizing latent cells to death and delaying viral rebound. The development and clinical progression of BH3-mimetics, which induce apoptosis by binding pro-survival Bcl-2 homologs, has resulted in a well- characterised class of inhibitors with relatively few unknowns regarding toxicity, side effects and dosage. In this thesis, I hypothesise that there are apoptotic blocks in place, specifically a greater dependence on pro-survival Bcl-2 proteins, which prevent a minority of infected CD4+ T cells from dying during active infection. I hypothesise that latently infected cells are distinct from other infected or healthy cells, and that this pro-survival phenotype allows them to persist in such a way that renders them susceptible to pro- apoptotic therapeutics which target the intrinsic pathway, such as BH3-mimetics. In Chapter 3, I infect primary human CD4+ T cells with HIV in vitro to assess the ability of BH3-mimetics to kill actively infected cells. I demonstrate that ABT-737 and Venetoclax, but not the Mcl-1 inhibitor S63845, preferentially kill activated, HIV infected CD4+ T cells in the setting of productive viral replication. These results shed light on the pro-survival role of Bcl-2 proteins during active HIV infection, and inform our progression into a preclinical model of HIV latency. Chapter 4 uses a humanized mouse model of HIV latency to further interrogate the importance of Bcl-2 pro-survival proteins in reservoir survival. I investigate the ability of Venetoclax, a clinically-approved Bcl-2 antagonist, as well as S63845, a preclinical Mcl-1 inhibitor, to delay viral rebound following analytical treatment interruption. This work provides the first compelling evidence that BH3-mimetics, either as monotherapy or in combination, can eliminate latently infected cells in vivo. In Chapter 5 I perform a tat/rev Induced Limiting Dilution Assay (TILDA) on CD4+ T cells from latently infected mice in order to quantify the impact of Venetoclax on the magnitude of the latent HIV reservoir. I confirm the existence of an inducible reservoir in our mouse latency model, although I do not observe a significant effect of Venetoclax treatment as measured by TILDA. I also use single-cell RNA sequencing to characterize peripheral CD4+ T cells from ART-suppressed human donors following Venetoclax treatment ex vivo, arriving at the suggestion that Venetoclax may target CD4+ T cells that are enriched for a gene signature associated with activation and cell metabolism. This work lays the foundation for furthering our understanding of which cells may contribute to HIV persistence and which may be susceptible to death- inducing compounds. Overall, this thesis represents a comprehensive assessment of the ability of BH3-mimetics to kill HIV active and latently infected cells, offering a strong justification for the translation of pro-apoptotic therapeutics such as Venetoclax into a clinical setting where reservoir eradication is the goal.

During an adaptive immune response activated B and T lymphocytes undergo rapid clonal expansion and generate extensive cellular heterogeneity. How lymphocytes guarantee the emergence of functional diversity amongst responding cells is not fully understood. In this thesis, the strategies utilised by the adaptive immune system for the diversification of B and T cells is investigated at the cellular, molecular and clonal levels in a quantitative manner. Activated B cell heterogeneity is predominantly driven by two critical programs. Firstly, the differentiation of antibody-secreting cells (ASCs) and secondly, the diversification of antibody isotype by class switch recombination (CSR). The regulation of these two processes was investigated through combined clonal and molecular analysis using a high-throughput proliferative lineage tracing approach to study ASC differentiation and CSR across thousands of clones. Two distinct fate programs emerged. Firstly, the timing of ASC differentiation within clones was strongly correlated. Diversity in commitment to the ASC lineage is established early and could be traced to the naive founder cell, from where it is transmitted to all progeny during clonal expansion. In striking contrast, isotype switching was highly variable across related cells irrespective of common ancestry, revealing a highly stochastic, cell-autonomous process regulated late within activated single cells. Further analysis demonstrated that single cells faced with a choice of two heavy chain isotypes solve the conflict using stochastic selection that is independent of their clonal lineage. As the principle molecular drivers of CSR are well known, their variation amongst single cell within clonal families was measured. Extensive variation was demonstrated in the expression of both activation-induced cytidine deaminase (AID) and the transcription of the germline noncoding RNAs. Furthermore, there was no correlation between AID expression and germline transcription, nor was the expression of distinct germline transcripts correlated. Thus, the net effect of stochastic influences over these two components can account for the single cell autonomy governing CSR. This stochastic molecular mechanism of CSR was developed into a quantitative model that accurately described and predicted B cell fate decisions across cell division and under varying experimental conditions. Quantitative analysis was applied to multi-parameter data of CD8 T cell heterogeneity, generated in response to diverse external stimulation. Using a combination of novel and established analytical techniques, the influence of time and division progression on T cell diversification, and their control by external signals, was accurately measured. The results of this investigation was subsequently used to construct a kinetic model of time- and division-dependent expression patterning for the molecule CD69 under varying external conditions. This model accurately described the expression dynamics of CD69 over time and division and highlighted the utility of a quantitative modelling approach to understanding CD8 T cell heterogeneity. Collectively, the work presented in this thesis represents a set of quantitative principles that describe lymphocyte fate decisions.

Structural maintenance of chromosomes flexible hinge domain containing 1 (Smchd1) is an epigenetic modifier that plays an important role in X chromosome inactivation, autosomal gene silencing, and is also implicated in several diseases in humans (Blewitt et al., 2008, Mould et al., 2013, Gendrel et al., 2013, Lemmers et al., 2012, Gordon et al., 2017, Shaw et al., 2017, Jansz et al., 2018). Thus far the majority of work on Smchd1 has been carried out on its zygotic form. Therefore, I partnered allele-specific genomics with conditional deletion of Smchd1 in the oocyte to test the role of maternally derived Smchd1 in imprinted gene expression. I found that Smchd1 is a novel maternal effect gene involved in genomic imprinting. When Smchd1 is maternally deleted loss of imprinting is observed at ten imprinted genes, without affecting DNA methylation imprints. Additionally, I also discovered seven imprinted genes where zygotic Smchd1 plays a dose dependent role. Interestingly, almost all the imprinted genes affected by the loss of maternal Smchd1, including Xist possess H3K27me3 imprints. This, together with Smchd1’s known role in long range chromatin interactions and function as insulator protein(Chen et al., 2015, Jansz et al., 2018), lead me to hypothesise that maternal Smchd1 may carry out its function secondary to H3K27me3 imprints and establish a chromatin state required for imprinted expression. To narrow down Smchd1’s behaviour in its native environment and how this behaviour relates to its structure and function, I established a series of robust microscopy experiments. I used fluorescence recovery after photo bleaching (FRAP) and Lattice LightSheet microscopy to demonstrate for the first time, Smchd1 dynamics in the inactive X during interphase and mitosis. Through study of our unique gain of function mutant MommeD43 I found that the act of unbinding and reloading likely has a role in Smchd1’s insulation effects. By using 3D-direct stochastic optical reconstruction microscopy (3D-dSTORM) I discovered that Smchd1 is approximately a 40 nm long protein in the nucleus. These novel techniques open up new avenues to explore higher order structures that Smchd1 may form in the nucleus related to its role in chromatin architecture. Together these data not only revealed a new function for maternal Smchd1 but also established novel methods to unravel Smchd1's molecular mechanism.

The functionally and phenotypically diverse cell populations that make up the immune system arise from the expression of a select part of a genome at a given time. The mechanisms governing such differential transcription are still, surprisingly, not fully understood. Only recently has the three-dimensional organisation of chromatin in the interphase nucleus been acknowledged to play a crucial role in modulating transcription. For instance, distal cis-regulatory elements like enhancers can form long-range chromatin loops with promoters to drive transcription, and these chromatin loops are in turn harboured in topologically associating domains (TADs), shielded from interference by outside elements. These three-dimensional structures can be lineage-specific and their roles during cellular differentiation are beginning to be uncovered. The chromosome conformation, or genome architecture, in B- and T-lymphocytes, captured previously as Hi-C data in the lab, has provided immense information about lineage-specific DNA interactions that might be critical during differentiation. Based on this resource, the work herein aimed to develop an approach to identify, characterise and functionally dissect any novel and critical regulatory elements. Using this strategy, I have identified several putative T- and B-cell specific elements and subsequently adopted the CRISPR/Cas9 platform in generating large deletions as to dissect these elements. The approach has identified and confirmed the enhancers of T cell-specific transcription factors Bcl11b and Gata3. Upon closer inspection an uncharacterised long non-coding RNA (lncRNA) Gm13218 was uncovered to associate with the enhancer of Gata3. Given the recent recognition of lncRNAs as important regulator of the 3D genome, I have retrieved the full-length sequence and characterised its expression pattern. It was found that expression of Gm13218 is highly correlated with that of Gata3 during early T cell development in thymus as well as T helper 2 (TH2) cell differentiation. Knockdown and overexpression of Gm13218 transcripts, CRISPR-mediated silencing, activation, demethylation of the locus as well as interference of transcription elongation suggest that Gm13218 may be involved in the establishment, but not the maintenance of Gata3 expression. By utilising Hi-C, RNA-seq, cell division and cell cycle indicators, the spatiotemporal dynamics of genome architecture during B cell activation and terminal differentiation into antibody-secreting cells was examined. It was revealed that genome organisation exhibit two discrete waves of restructuring – the first occurs just prior to the first cell division, with the resulting genome architecture being inherited through the subsequent rapid clonal expansion for many days until the second wave of restructuring upon differentiation into plasmablast. In addition, the first restructuring event was shown to precede the first DNA replication phase, suggesting that genome reorganisation is independent of, and well partitioned from, DNA synthesis and mitosis. In contrast, transcription underwent very early burst and was altered throughout the entire differentiation process. Further analysis suggests that transcription is intricately intermingled with genome organisation in a reciprocal fashion. Overall, the work in this thesis has revealed a number of important findings regarding how the 3D genome controls the development and function of the immune system.

Histone acetylation affects the way DNA and associated proteins are packaged in the cell nucleus and regulate chromatin organisation and gene expression. Acetylation of core histones has been broadly correlated with initiating and maintaining open chromatin, poised or active gene transcription, DNA damage repair, as well as chromosome decondensation during mitosis and meiosis. The acetylation of lysine residues is catalysed by histone lysine acetyltransferases (KATs) and deacetylases (HDACs), which are tightly regulated. Dysregulation of KATs and aberrant lysine acetylation has been associated with tumorigenesis and negative prognoses in a wide range of cancer, presenting a new area of potential therapeutic targets. Potential acetylation targets of KATs catalysing the acetylation of histones have predominantly been studied in cell-free assays, where the enzymes show little substrate specificity. In contrast, histone acetyltransferases appear to acetylate surprisingly specific residues in whole cells. In this thesis I investigate the effects of acute deletion of the MYST lysine acetyltransferase TIP60 (KAT5) using inducible cre-recombinase and CRISPR/Cas9-mediated deletion in human cells and mouse cells, as well as mouse embryos and its potential role in cancer cells. I found that loss of TIP60 caused complete cell growth arrest in human and mouse cells. In the absence of TIP60 cells displayed cell cycle arrest in G1 and G2/M phase with increased endoreplication, accompanied by chromosomal segregation defects. Remarkably, the proliferation arrest caused by loss of TIP60 also occurred in the absence of the tumour suppressors p53, INK4A and ARF and therefore was independent of these. In contrast, cell survival was not affected. Growth arrest independent of major tumour suppressors flags TIP60 a potential target for novel cancer therapeutics. TIP60 was found to be essential for of H2AZ acetylation, particularly, lysine 7 acetylation. In contrast, global chromatin bound H2AZ levels were not reduced. H2A and H4 acetylation was reduced slightly in TIP60 depleted cells. Identifying H2AZ lysine 7 acetylation as a biomarker for TIP60 activity is a major step in developing TIP60 as a drug target. The mRNA levels of 6236 human and 8238 mouse genes, including many metabolic genes, were dependent on TIP60, supporting a role for TIP60 as a key transcriptional co-activator. Characterization of key mechanisms causing chromosomal aberrations and identifying approaches that can be used to reduce likelihood of genome instability due histone acetylation defects will pave the way for better detection of early changes in cancer and development of therapeutic applications in the future. This work represents important steps towards the development of histone lysine acetyltransferases as drug targets. A comprehensive analysis of histone acetylation activity of TIP60 shines light on the potential of its many proposed roles.

The differentiation of B cells into antibody secreting cells (ASCs) and the production of protective antibodies is a critical part of the adaptive immune response to infection. ASCs are also important for the formation of immunological memory which provides protection against reinfection and the generation of ASCs is the goal for almost all current vaccination strategies. Despite the importance of these cells, we still lack a complete understanding of the factors that control B cell differentiation into ASCs, ASC survival and antibody secretion, all of which must be tightly regulated to ensure an optimal immune response. Here, I have developed a CRISPR/Cas9 mediated arrayed screening approach for the identification of novel positive and negative regulators of primary mouse B cell proliferation, survival, differentiation into ASCs and antibody secretion. By interrogating multiple gene sets I have identified all elements within the ASC gene signature that are essential for the in vitro generation of ASCs. I have also identified several novel negative regulators of the B cell differentiation process (AB124611, Arhgef18, A430078G23Rik, Fam43a, Pold1, Ripk3, Rnf130 and Rps6ka5). This work has also uncovered a novel role for 6 genes, (Cdv3, Hspa5, Sec61a1, Selk, Sumo2, Vcp) in driving the proliferation of B cells. One of these genes, Cdv3, has no previous association with proliferation in any cell type and presents an exciting new candidate for further investigation. I have demonstrated that within the ASC gene signature there are 35 genes which are essential for efficient antibody secretion. Interestingly, many of these genes are components of the ER protein processing pathway, however, not all elements of this pathway appear to be essential for antibody secretion. These results raise the possibility of there being a specific pathway for antibody secretion, or that the genes identified in this thesis may represent weak links in the ER protein processing pathway which could potentially be exploited therapeutically to inhibit antibody secretion in disease settings. Finally, I have used an Irf4 deficient mouse model to uncover a novel role for Irf4 in the development of the peritoneal B-1a population. I have shown that Irf4-/- mice lack peritoneal B-1a cells and by examining multiple stages of B-1a cell development I have demonstrated that in the absence of Irf4, B-1a cell development is blocked at the transitional B-1a stage. By employing RNA sequencing to analyse the transcriptional profiles of the remaining Irf4-/- B-1 cells and analysis of previously published ChIP sequencing data, I have revealed a potential role for Irf4 in directly activating the expression of Bhlhe41, a transcription factor that is required for B-1a cell development and homeostasis. Together, the results from this thesis build upon decades of previous work on the genetic regulation of B cell biology. Integrating the novel regulators of B cell proliferation and differentiation that I have identified in this thesis into the current model of ASC generation will improve our understanding of how the decision between undergoing differentiation or maintaining the B cell fate is made. A detailed understanding of how this fate decision is made has far reaching implications for human health and disease as this information can be used to inform vaccine design, reveal the causes of immunodeficiencies or highlight novel avenues for targeting pathogenic ASCs in autoimmunity and cancer.

Regulatory T (Treg) cells are critical for the maintenance of immune homeostasis and peripheral tolerance. Different subsets of Treg cells have recently been described with many studies showing the importance of context-specific differentiation of Treg cells, in particular within non-lymphoid organs. These non-lymphoid organ Treg cells have a fully suppressive Treg cell phenotype with an effector function and are termed effector (e)Treg cells. However, the ontogeny of eTreg cells have not yet been fully described. Additionally, molecular determinants of the eTreg cell program remain incompletely understood. My thesis examines the transcriptional events that regulate the generation of eTreg cells during their thymic development, their homeostasis and response to infection. Using different gene targeted mouse models at steady state and in viral infection models, I studied the intrinsic molecular mechanisms that contribute to eTreg cell differentiation. In particular, I focused on follicular Treg (TFR) cells, which constitute the eTreg cell subset of the germinal centre. The molecular control of eTreg cell fate and function converges on the transcription factors IRF4 and Blimp-1. IRF4 is induced by antigen receptor signals and cooperates with AP-1 factors, BATF and JUN, to regulate transcriptional networks involved in lymphocyte differentiation, function and metabolism. For example, these factors regulate genes important for antibody class switch recombination in B cells and functional differentiation of distinct CD4 T helper (Th) subsets, including Th2, Th9, Th17 and T follicular helper cells. IRF4 expression in Treg cells is critical for effector differentiation, yet the precise mechanisms of how IRF4 regulates the transcriptional program of TFR cells remains unknown. Using a novel transgenic IRF4 reporter mouse we found that IRF4 is highly expressed in TFR cells. Using IRF4 knockout mouse models, we demonstrate that IRF4 is necessary for TFR cell generation in a Treg cell-intrinsic manner. IRF4 controls important aspects of the transcriptional program that drives TFR cell differentiation, including genes essential for Treg cell migration. Furthermore, I identified the transcription factor c-Maf to be essential for TFR cell generation and demonstrated its central role in maintaining a follicular program in Treg cells. Subsequently, using ribonucleic acid (RNA)-sequencing, we generated a “follicular signature” of gene expression from the combined analysis of TFH and TFR cells. Integrated transcriptional analyses showed that in the absence of either IRF4 or c-Maf, the majority of the follicular signature genes were downregulated, indicating that these two transcriptional regulators, aside from Bcl6 are indispensable for follicular TFR cell development. Finally, analyses of IRF4 and c-Maf DNA binding sites, identified by chromatin induced precipitation (ChIP)-sequencing, in combination with open chromatin regions in follicular T cell specific loci, we showed that the precise orchestration of distinct sets of genes is required to promote conserved aspects of the follicular T cell fate. In conclusion, my thesis describes how a key transcriptional network orchestrates fundamental steps in TFR cell differentiation and function, which contributes to the understanding of eTreg cell biology.

The fast development of single cell RNA sequencing (scRNAseq) presents new challenges in data analysis and opportunities for protocol development. To address challenges in data preprocessing, I developed scPipe, an R/Bioconductor package that integrates barcode demultiplexing, read alignment, UMI-aware gene-level quantification and quality control of raw sequencing data generated by multiple protocols. Results from scPipe can be used as input for downstream analyses and can be easily incorporated into R-based pipelines with other tools. In order to compare different computational methods for scRNAseq data, I generated a realistic benchmark experiment that included single cells and admixtures of cells or RNA to create `pseudo cells' from up to five distinct cancer cell lines. Multiple datasets were generated using both droplet and plate-based scRNAseq protocols and processed by scPipe. We found pipelines suited to different types of data for different tasks. Our data and analysis provide a comprehensive framework for benchmarking most common scRNA-seq analysis steps. Finally, I developed single cell full-length transcript sequencing by sampling (FLT-seq), together with the computational pipeline FLAMES to perform isoform discovery and quantification, splicing analysis and mutation detection in single cells. With FLT-seq and FLAMES, I performed a comprehensive characterization of the full-length isoform landscape in single cells of different types and species and found conserved functional modules that were enriched for alternative transcript usage in different cell populations, including ribosome biogenesis and mRNA splicing. The datasets, protocols and tools that I developed and generated are useful resources for the single cell research community.

Colorectal cancer (CRC) remains one of the leading causes of cancer deaths worldwide. Advances in therapy have resulted in significant gains in survival, particularly in the metastatic setting. While the discovery of biomarkers, such as RAS mutations, have helped refine treatment selection to some degree, more accurate biomarkers are urgently needed. Comparison of existing treatments, as well as the evaluation of the efficacy of new therapies, are informed by randomised controlled trials (RCTs), which form the evidentiary backbone of clinical practice guidelines and represent the gold standard of assessment. Despite their high internal validity, they can lack generalisability due to their highly selective inclusion criteria. Prospective, registry-based, randomised controlled trials (RRCTs) have the potential to bridge the gap between RCTs and real-world clinical practice in oncology. The objective of this thesis is to explore how clinical registries can help to advance biomarker research. This thesis applies real-world data to examine the clinical utility and validity of emerging CRC biomarkers, and explores the feasibility of RRCTs in the oncology setting. In a cohort of 99 metastatic colorectal cancer (mCRC) patients, the role of the epidermal growth factor receptor (EGFR) and its ligands, amphiregulin and epiregulin, as potential prognostic and predictive biomarkers for mCRC patients is explored (Chapter 5). This study examines protein expression by immunohistochemistry and includes patients who were not treated with EGFR inhibitors, representing the largest such cohort reported to date. The real-world validity of biomarker trials is explored in Chapter 6, where the characteristics of patients enrolled in these studies are compared to real-world patients. Using an established multi-centre CRC registry as the reference real-world cohort, clinical data was analysed for participants in three types of biomarker trials (retrospective, prospective observational and prospective interventional). This study provides novel insights into recruitment to, and potential validity of, biomarker trials. Finally, Chapter 7 examines the feasibility of an Australian-first RRCT in oncology. This ongoing study is exploring chemotherapy sequencing in first-line treatment of mCRC and leverages an established multi-centre registry as the data collection platform. This study demonstrates the potential of RRCTs to accelerate progress in optimising patient treatments and outcomes. This thesis demonstrates the power of high-quality clinical registries to facilitate prospective randomised trials, while providing opportunities to investigate and validate biomarkers in real-world settings.

Tumor necrosis factor (TNF) is a master inflammatory cytokine that can, depending on the circumstances, promote survival and proliferation or induce cell death. Anti-TNF drugs have proven strikingly successful in treating inflammatory diseases such as rheumatoid arthritis (RA), psoriasis and inflammatory bowel disease (IBD) but it is still unclear exactly why. For a long time, it was thought that they work solely by preventing TNF induced transcription of other inflammatory cytokines, but more recently it has been proposed that one of their major anti-inflammatory functions is to prevent TNF induced death. Therefore, understanding the mechanism by which TNF induced death is regulated may enable the conceptualization of newer or improved approaches in treating a variety of inflammation-associated pathologies. Binding of TNF to its receptor TNFR1 leads to the formation of two distinct signalling complexes. While most previous studies have focused on the membrane-bound, transcription-activating complex (complex-1), the composition and post-translational modifications of the cytosolic, caspase-8-containing, death-inducing complex (complex-2) remain far less well defined. To analyse TNFR1 complex-2 composition at endogenous levels, we decided to generate FLAG tagged caspase-8 knock-in mouse strains. The reagents for the FLAG tag enable very efficient and specific purification and identification of a FLAG tagged protein and its partners. After some preliminary tests and trials, I decided to use a 3x FLAG tag which has been reported to be 20–200 times more sensitive than other FLAG tags in immunoprecipitation and detection assays. Before generating the mouse strains, in Chapter 3 I performed extensive in vitro comparison of N-terminally or C-terminally 3x FLAG-tagged caspase-8 using a doxycycline (Dox)-inducible stably integrated lentiviral system. The results suggested that when expressed above endogenous levels, the expression and killing activity of caspase-8 was unaffected by a 3x FLAG tag. Interestingly, when expressed at physiological levels, C-terminally 3x FLAG tagged caspase-8 appeared to be equivalent to untagged caspase-8 and marginally more efficient in mediating TNF-induced death and complex-2 formation compared to N-terminally 3x FLAG tagged caspase-8. In addition, I immunoprecipitated TNFR1 complex-2 from cells expressing endogenous levels of 3x FLAG tagged caspase-8 and performed a mass spectrometry (MS) analysis. According to this analysis, Tankyrase-1 (TNKS1/PARP5a/ ARTD5), a member of the poly ADP-ribose polymerase (PARP) superfamily, appears to be a novel interactor of complex-2. Based on our in vitro data, we generated N-terminally or C-terminally 3x FLAG tagged caspase-8 knock-in mice using CRIPSR/Cas9 technology and these mice were characterized in Chapter 4. Homozygous N-terminally or C-terminally 3x FLAG tagged caspase-8 knock-in mice were viable, fertile and developed normally, indicating that N-terminally or C-terminally 3x FLAG tagged caspase-8 were expressed and active in vivo, at least to heterozygous caspase-8 levels. As expected, the expression of N-terminal or C-terminal 3x FLAG tagged caspase-8 was detectable in tissue and cells from knock-in mice by Western blot and immunofluorescence stain using an anti-FLAG M2 antibody. The 3x FLAG tagged caspase-8 displayed similar tissue distribution and comparable expression levels as endogenous caspase-8. The cell death assay suggested that the primary cells and transformed cells from 3x FLAG tagged caspase-8 knock-in mice responded similarly as wild-type cells to apoptotic and necroptotic stimulations. Moreover, by performing anti-FLAG immunoprecipitation, I successfully purified endogenous TNFR1 complex-2 from knock-in mice derived cells. These data indicated that 3x FLAG tagged caspase-8 knock-in mouse strains are useful tools to study caspase-8 and caspase-8-containing protein complexes at physiological levels. In Chapter 5, I characterized tankyrases-mediated poly(ADP-ribosyl)ation (PARsylation) as a novel checkpoint that limits TNF-induced cytotoxicity. Using primary cells from the 3x FLAG tagged caspase-8 knock-in mice described in Chapter 4, I found that the enzyme tankyrase-1 (TNKS1/TNKS/PARP5a/ARTD5), which was identified by mass spectrometry in Chapter 3, is recruited to the endogenous TNFR1 complex-2. Western blot data indicates that tankyrase-2 (TNKS2/PARP5b/ARTD6) may also be recruited. Tankyrases are poly ADP-ribose polymerases and belong to an ancient group of enzymes that post-translationally modify proteins with ADP-ribose. I found that during TNF signalling, complex-2 becomes poly(ADP-ribosyl)ated (PARsylated) in a tankyrases-dependent manner. Furthermore, tankyrases-specific inhibitors sensitized cells to TNF-induced cell death, which correlated with increased levels of complex-2. This suggested that normally tankyrases help limit TNF induced death. Mechanistically, I showed that tankyrases may modulate the stability of complex-2 by recruiting the E3 ubiquitin ligase RNF146, that in turn promotes ubiquitylation and degradation of complex-2. Moreover, inactivation of tankyrases dramatically increased the killing of the clinical Smac-mimetic (SM) birinapant in a primary acute myeloid leukemia (AML) model. Taken together, this thesis describes 3x FLAG tagged caspase-8 knock-in mice as new tools to study caspase-8 and caspase-8-containing protein complexes at physiological levels. Furthermore, this study identifies tankyrases-mediated PARsylation as a novel checkpoint in TNF signalling that expands our understanding of how TNF induced death is regulated and provides a rationale to use tankyrases inhibitors for cancer therapy.