Centrosomes are cellular organelles present in most animal cells, and are formed of two main components: the centrioles and the pericentriolar material (PCM). Centrosomes perform a variety of functions: they are the main microtubule organising centre in the cell, and are important localisation hubs for kinases involved in regulating the cell cycle. Hundreds of proteins are thought to localise to centrosomes, but work in the last decade has narrowed down this list to a handful of proteins that are thought to be essential for centrosome structure and function in Drosophila. Asl, Ana2, DSas-4, DSas-6 and Sak have been identified as essential components for centriole duplication, while Cnn and DSpd-2 are thought to be important in establishing the PCM. However, the relative position of these 7 components in the pathway of centrosome assembly in Drosophila embryos remains elusive, as a genetics analysis of this process is hampered by the absence of centrioles in most mutant embryos for these proteins. In this thesis I elucidate the pathway of centrosome assembly in Drosophila by using SAPs (DSas-6/Ana2 particles that form in Drosophila unfertilised eggs upon moderate expression of DSas-6 and Ana2) as proxy models of centrosomes. I show SAPs are very similar to centrosomes in composition and dynamics but unlike centrosomes are able to form even in the absence of some of the essential centriolar components. SAP analysis in the absence of each of the main centrosome components reveals that: Sak is not required for the recruitment of downstream components; DSas-4 is necessary for Ana2 and DSas-6 to interact; Asl is the most upstream component of the PCM recruitment pathway, followed by DSpd-2; it is likely that there is an additional PCM recruitment pathway. I then take advantage of some of these results to examine how centrosome formation is potentiated after egg activation. My work allows me to propose an improved description of the pathway of centrosome formation in Drosophila.

The purpose of this thesis was to study mitosis in Drosophila, from a network biology perspective. The primary aim was to develop and test a network-based prediction model that could integrate available data in public databases (like Flybase) and, based on that, predict potential mitotic proteins. The approach taken to design the protein interaction network included the use of a priori knowledge about the microtubule composition of the mitotic spindle and the higher likelihood of microtubule-associated proteins (MAPs) to have a putative mitotic function. The design also included the integration of different complementary datasets, from gene expression and functional RNAi screens to cross species conservation of MAPs for fitting a network-based model for predicting mitotic proteins. I begin with the creation of the MAP interactome based on a MAP dataset in Drosophila. This initial network was extended by transferring homologs and interologues of MAP datasets from four other species, i.e. human, mouse, rat and Arabidopsis. These proteins were then used as seed proteins to conduct a virtual pull-down experiment, by adding indirect interactors into the network, i.e. proteins that directly bind to two or more MAPs within the network, which completed the MAP interactome. Data from genome-wide studies in Drosophila were gathered for each node in the MAP interactome. These ‘layers’ of data were then used as features to fit a prediction model that could score each node in the network, based on the likelihood of its role in mitosis. The final model performed with 96% accuracy after 10-fold cross validation and was used to rank all the proteins in the MAP interactome. By analysing the top 100 high scoring predicted mitotic proteins, a highly connected cluster of 33 proteins was identified that was subject to experimental validation in the lab. The first approach was to conduct an in vitro analysis using an RNAi screen to test for any spindle, chromosome or centrosome phenotypes upon gene knockdown. After two independent RNAi screens, around 80% of the proteins produced mutant mitotic phenotypes strongly supporting the results of the MAP prediction model. The second approach was to conduct an in vivo analysis by expressing GFP- fusion constructs of selected genes from the subcluster. These were expressed in Drosophila early embryos to study their subcellular localization during interphase and mitosis. A variety of localizations were observed ranging from chromatin and microtubules to more generic cytoplasmic localizations. These results suggested not all predicted proteins were co-localizing with microtubules, and therefore might not necessarily be microtubule associated proteins but can possibly be functioning as microtubule associated regulator proteins. Proteomics analysis of a subset of these genes showed a large proportion of false positive interactions but also picked new interactions between member proteins that highlighted a module within the subcluster. The RNAi hits from the in vitro analysis and the members of the module within subcluster-16 from the in vivo analysis provide interesting subjects for further characterization.

Carbonic anhydrase IX (CA9) is strongly induced by hypoxia and its overexpression is associated with poor therapeutic outcome in cancer. The function of CAIX is to catalyze the reversible hydration of CO2 to bicarbonate and a proton. This helps hypoxic tumours to maintain a more neutral intracellular pH (pH_i) promoting survival, but produces a more acidic extracellular (pH_e) which promotes invasion and metastasis. Recent evidence has expanded on the role of hypoxia and CAIX by relating them to stem cell niches. In this study, taking advantage of the transmembrane location of CAIX, we show for the first time, a direct marked heterogeneity in response to hypoxia within each tumour cell population studied, associated with major biological differences. Based on CAIX expression pattern under hypoxic conditions, we identify, isolate and characterize two distinct populations of tumour cells, one that express CAIX and the other that does not. Interestingly, we discover that the CAIX positive population is enriched with cells expressing cancer stem cell markers. These include ALDHA1, IGF1, LIN28 and genes involved in epithelial-mesenchymal transition (EMT) and multi-drug resistance (i.e. WNT2, TWIST1, and ABCC2). Accordingly, CAIX+ve cells show higher self-renewal capacity and form tumours significantly faster compared to the CAIX-ve population. Importantly, functional suppression of CAIX in vitro and in vivo, in two breast cancer cell lines resulted in the downregulation of breast cancer stem cell signatures, suggesting that CAIX is not just a marker of stemness but also a regulator of stemness. The molecular mechanism underlying the differential expression of CAIX in the two populations is not HIF-1α-dependent, but instead driven by hypoxia-induced reorganization of chromatin structure. In line with this, we provide experimental evidence showing that the genomic locus encoding CA9 has a more “open” and transcriptionally active chromatin structure in CAIX+ve cells, and a condense and transcriptionally silent chromatin structure in the CAIX-ve cells. Given that HIF induces the transcription of CA9 by binding to hypoxia response elements (HREs) in its promoter we show a significant reduction in binding of HIF to the CA9 promoter of the negative population. We suggest that the reduce HIF binding is a result of the compact chromatin structure of CA9 promoter of the negative cells. Analysis of the transcriptome of the positive and negative populations suggests a symbiotic relationship between these two subpopulations and their environment, likely required to promote tumour growth. This is based on the following observations: Firstly, we identified that CAIX-ve cells express high levels of cytokines and based on this, we suggest that the cytokines secreted by CAIX-ve cells may transmit paracrine signals that regulate the CAIX+ve cells, thus providing a wider hypoxia tolerant microenvironment to protect the stem cell population. Secondly, we identified a metabolic heterogeneity between the CAIX+ve and CAIX-ve cells. The CAIX+ve cells show an upregulation of genes implicated in oxidative phosphorylation, TCA cycle and fatty acid synthesis. Whereas in CAIX-ve cells there is an upregulation of genes implicated in autophagy and mitophagy. Given the above together with the upregulation of oxidative phosphorylation and TCA cycle in the CAIX+ve cells, we proposed the existence of a metabolic symbiosis between the CAIX+ve and CAIX-ve cells. We postulate that the catabolic process such as autophagy and mitophagy in the CAIX-ve cells may results in the overproduction of high-energy metabolites such as lactate, glutamine and ketone bodies which in turns they are been utilized by CAIX+ve cells to fuel mitochondria respiration. Finally, we also demonstrated that in the CAIX+ve cells mTORC1 signaling is upregulated, and contributes to the regulation of CAIX expression. Given the role of mTORC1 in stem cell maintenance and EMT as well as the interdependence of mTORC1 and CAIX expression in the CAIX+ve cells we suggest that mTORC1 signaling may be the critical factor by which CAIX regulates stemness. Interestingly, the subpopulations show a differential sensitivity to HDAC inhibitors, NaBu and SAHA as treatment of MCF7 breast cancer cell line and HCT116 colon cancer cell line leads to elimination of the CAIX+ve population. This is not driven by the downregulation of HIF-1α, the major transcriptional regulator of CAIX. In contrast, we demonstrate that SAHA causes downregulation mTORC1. This suggests that SAHA-induced downregulation of CAIX expression could be due to its effect on mTORC1 pathway. Of wider significance, our findings show that tumours are not homogenous in their response to hypoxia, and distinct signal transduction networks regulate different populations of cells within the tumour. This highlights the need for the utilization of biomarkers, which reveal distinct functional hypoxia profiles of human cancers, and permit the stratification of tumours. Furthermore, the identification of epigenetic regulation of the histones in response to hypoxia for highly selective gene regulation, provides a connection between the epigenetic mechanisms under environmental stress and cancer progression, and is model for development of novel epigenetic cancer therapeutic drugs.

Eukaryotic cilia are complex, highly conserved microtubule-based organelles with a broad phylogenetic distribution. Cilia were present in the last eukaryotic common ancestor and many proteins involved in cilia function have been conserved through eukaryotic diversification. The evolution of these ciliary functions may be inferred from the distribution of the molecular components from which these organelles are composed. By linking protein distribution in 45 diverse eukaryotes with organismal biology, I define an ancestral ciliary inventory. Analysis of these core proteins allows the inference that the cenancestor of the eukaryotes possessed a cilium for motility and sensory function. I show that the centriolar basal body function is ancestral, whereas the centrosome is specific to the Holozoa, and I use this information to predict a number of roles for proteins based on their phylogenetic profile. I also show that while remarkably conserved, significant divergence in ciliary protein composition has occurred in many lineages, such as the unusual centriole of Caenorhabditis elegans and the transitional changes throughout the land plants. I exemplify this divergence through ultrastructural studies of the fern Ceratopteris richardii and the liverwort Marchantia polymorpha both of which have cilia that exhibit a number of distinctive morphological features, the most conspicuous of which is a general breakdown of canonical microtubule arrangements. Cilia have also been lost multiple times in different lineages: at least twice within the land plants. During these evolutionary transitions proteins with ancestral ciliary functions may be lost or co-opted into different functions. I have interrogated genomic data to identify proteins that I predict had an ancestral ciliary role, but which have been maintained in non-ciliated land plants. I demonstrate that several of these proteins have a flagellar localisation in protozoan trypanosomes and I use expression data correlation to predict potential non-ciliary plant roles.

The case-control design is one of the most commonly used designs in genome- wide asociation studies. When we increase the sample size of either the controls or, more importantly, the cases, the power of whatever test we use will certainly increase. However increasing the sample size, means that addi- tional individuals need to be genotyped and this implies extra financial costs. However, nowadays with the emergence of genetic studies, a large number of genetic data are available at low or no extra cost. Even though those data may not be completely relevant to the current study, they can still be used to increase the probability to identify true associations. Furthermore, additional information, non-necessarily genetic, can also be used to improve the power of a method. In this thesis we extend the case-control design in order to take ad- vantage of such types of additional data and/or information. We discuss three designs; the case-cohort-control, the kin-cohort and the super-case– case–control–super-control designs. For each of these, we present methods that are adjusted or modified versions of standard case-control methods but we also propose novel ones developed with those extended designs in mind. Ultimately, we describe how those methods can be used in order to increase the power of association tests, especially compared to similar methods of the case-control design.

This thesis is a study of the detection of separation among the sample points in binary logistic regression models. We propose a new algorithm for detecting separation and demonstrate empirically that it can be computed fast enough to be used routinely as part of the fitting process for logistic regression models. The parameter estimates of a binary logistic regression model fit using the method of maximum likelihood sometimes do not converge to finite values. This phenomenon (also known as monotone likelihood or infinite parameters) occurs because of a condition among the sample points known as separation. There are two classes of separation. When complete separation is present among the sample points, iterative procedures for maximizing the likelihood tend to break down, when it would be clear that there is a problem with the model. However, when quasicomplete separation is present among the sample points, the iterative procedures for maximizing the likelihood tend to satisfy their convergence criterion before revealing any indication of separation. The new algorithm is based on a linear program with a nonnegative objective function that has a positive optimal value when separation is present among the sample points. We compare several approaches for solving this linear program and find that a method based on determining the feasibility of the dual to this linear program provides a numerically reliable test for separation among the sample points. A simulation study shows that this test can be computed in a similar amount of time as fitting the binary logistic regression model using the method of iteratively reweighted least squares: hence the test is fast enough to be used routinely as part of the fitting procedure. An implementation of our algorithm (as well as the other methods described in this thesis) is available in the R package safeBinaryRegression.

The mammalian E2F family of transcription factors plays a crucial role in the regulation of cellular proliferation, apoptosis and differentiation. E2F7 and E2F8 are the most recently identified family members and have unusual features that distinguish them from other members in the E2F family, including two distinct DNA-binding domains that bind to DNA in a DP-independent manner. E2F7 and E2F8 have been shown to be transcriptional repressors. However, the mechanism by which E2F7 represses E2F responsive gene expression remains to be elucidated. The results presented here provide the first insight into the E2F7-mediated transcriptional mechanism. E2F7 was shown to contain a CtBP binding motif and form a complex with CtBP in both HeLa and MCF7 cells. An E2F7 deletion mutant lacking the CtBP binding motif was unable to form a complex with CtBP and repress the transcription of E2F target genes in luciferase assays, suggesting that this motif is essential for E2F7-dependent repression. Furthermore, the E2F7-CtBP complex was shown to be stable under different types of damage, such as following etoposide and UV treatment, and under different cell redox states. Interestingly, however, E2F7 was unable to repress the transcriptional activity of E2F target genes following treatment with the CtBP substrate MTOB. Moreover, E2F7 endogenous immunoprecipitations showed that E2F7 forms a complex with the chromatin-modifying enzymes HDAC1, HDAC2 and LSD1 and the co-repressor CoREST. Interestingly, via chromatin immunoprecipitations, E2F7 was shown to recruit these co-repressors to a subset of E2F-responsive promoters where they affect the activity of these promoters in a target gene-specific manner. Furthermore, results presented here suggest that CtBP could play a dual role in E2F7 function, not only being involved in E2F7-mediated repression but also in the repression of E2F7 itself as siRNA mediated CtBP depletion was shown to cause an upregulation of E2F7 protein levels. These results implicate a repertoire of co-repressors in a target gene-specific E2F7 repression mechanism, and as such define a new level of complexity in cell cycle control.

Haematopoietic stem cells (HSC) are strictly regulated by intrinsic regulators and extrinsic signals from the microenvironment. Nov (CCN3), a matricellular protein of the CCN family, has been reported as a suppressor gene in solid tumours and chronic myeloid leukaemia (CML). Recent study identified Nov as a positive regulator in human cord blood CD34+ stem cells. However, the functions of Nov in haematopoiesis and adult HSC remain largely unknown.

Two-pore channels (TPCs, gene name Tpcn) have recently been identified as endolysosomal cation channels modulated by the potent calcium (Ca2+) releasing messenger nicotinic acid adenine dinucleotide phosphate (NAADP). Gene knockout (KO) and RNA knockdown studies have implicated TPCs in fundamental cellular processes, including secretion, of insulin in pancreatic islets, and differentiation, of skeletal myoblasts and osteoclasts. Investigations of Tpcn1 and Tpcn2 mRNA expression have indicated widespread tissue distribution, but a lack of suitable antibodies has impeded study of the endogenous proteins. In this study, an anti-TPC1 antibody was purified from immune sera and used in immunoblotting investigations to demonstrate TPC1 protein expression in a wide range of mouse tissues, with highest expression levels observed in kidney, liver and adipose tissue. Endogenous mouse TPC1 was demonstrated to be glycosylated, with apparent differences in the extent of glycosylation in different tissues based on the indicated molecular weight before and after treatment with a deglycosylating enzyme, which may have implications for the functional regulation of channel activity. Given the increasing prevalence of type 2 diabetes and obesity, an understanding of the molecular basis of glucose homeostasis and adipose tissue formation and function is an important scientific goal. Tpcn KO mice have been developed; in both Tpcn1 KOs and Tpcn2 KOs, impaired pancreatic β-cell Ca2+ signalling and reduced insulin secretion from the whole pancreas were demonstrated. However, the whole-animal phenotype has not been extensively researched. In this study, intraperitoneal glucose tolerance tests were conducted in Tpcn KO mice. These indicated that glucose homeostasis was not significantly affected in Tpcn2 KOs or Tpcn1/2 double KOs (DKOs), and only mildly impaired in Tpcn1 KOs, despite the defects previously observed at the cellular and tissue level. In addition, body composition was investigated in Tpcn1 KO, Tpcn2 KO and Tpcn1/2 DKO animals using magnetic resonance spectroscopy and time domain-nuclear magnetic resonance. Single Tpcn KOs were found to have lower adipose tissue levels as a percentage of body composition, while Tpcn1/2 DKOs were shown to have increased bodyweight but normal body composition. To investigate potential roles for TPCs in adipose tissue formation, Tpcn expression during adipogenesis was investigated using an in vitro multipotent mesenchymal stem cell line model of adipogenic differentiation. Tpcn2 mRNA levels were demonstrated by quantitative PCR to be transiently increased during the early stages of adipogenic differentiation, and cyclic AMP (cAMP) was identified as the factor that induced this upregulation. Lentiviruses were developed to express fluorescently-tagged TPCs, and overexpression of TPC2 was demonstrated to partially overcome the requirement for the cAMP-inducing agent in the medium used for the induction of adipogenesis. Collectively, these data suggest that TPCs may play a role in the formation and/or function of adipose tissue.

This thesis focuses on multilocus methods designed to detect single nucleotide polymorphisms (SNPs) that are associated with disease using case-control data. I study multilocus methods that allow for interaction in the regression model because epistasis is thought to be pervasive in the etiology of common human diseases. In contrast, the single-SNP models widely used in genome wide association studies (GWAS) are thought to oversimplify the underlying biology. I consider both pairwise interactions between individual SNPs and modular interactions between sets of biologically similar SNPs. Modular epistasis may be more representative of disease processes and its incorporation into regression analyses yields more parsimonious models. My methodological work focuses on strategies to increase power to detect susceptibility SNPs in the presence of genetic interaction. I emphasize the effect of gene-gene independence constraints and explore methods to relax them. I review several existing methods for interaction analyses and present their first empirical evaluation in a GWAS setting. I introduce the innovative retrospective Tukey score test (RTS) that investigates modular epistasis. Simulation studies suggest it offers a more powerful alternative to existing methods. I present diverse applications of these methods, using data from a multi-stage GWAS on prostate cancer (PRCA). My applied work is designed to generate hypotheses about the functionality of established susceptibility regions for PRCA by identifying SNPs that affect disease risk through interactions with them. Comparison of results across methods illustrates the impact of incorporating different forms of epistasis on inference about disease association. The top findings from these analyses are well supported by molecular studies. The results unite several susceptibility regions through overlapping biological pathways known to be disrupted in PRCA, motivating replication study.