Alzheimer's disease (AD) is the most common form of dementia, now the leading cause of death in the UK, which affects more than 35 million people worldwide. Genome-wide association studies identified 20 genetic loci associated with disease susceptibility, however these only exerted small effects on risk. Next-generation sequencing is now being employed to identify more of the missing heritability. This project utilised whole-exome sequencing to explore genetic variation using the Brains for Dementia Research (BDR) resource, a well-characterised cohort of neuropathologically confirmed samples. Exome-wide and candidate gene approaches were employed to assess coding variants for association with AD, using single-variant and burden tests. Coding variants in other neurodegenerative disease genes were also analysed as potential susceptibility factors for AD. Furthermore, polygenic risk scores (PRS) were generated to explore the ability to classify case and control individuals based on their genetic profiles. A synonymous variant in PILRA (rs2405442) was nominally associated with 3-fold increased risk of AD, also contributing strongly to PILRA burden. It was previously linked to AD through risk gene ZCWPW1; however, it has not been directly associated until now. Additional variants in GWAS gene ABCA7 (rs3764645, rs3752234, rs3752237, rs4147915) and rare variants in CLU were also implicated, further supporting their roles in AD susceptibility. A variant in PD gene LRRK2 (rs35303786) inferred protection against AD, implicating potential pleiotropy across the two diseases. PRS could distinguish AD cases from controls with 85.3% accuracy and also identified controls with high PRS but no cognitive impairments. This could be useful for identifying individuals at risk of developing AD in the future. We have uncovered tentative associations both in established and newly identified loci; highlighting several interesting candidates for further investigation. Although there remains a large amount of missing heritability, we hope that as the BDR resource grows, we will achieve increased power to detect significant associations with AD.

Myotonic dystrophy type 1 (DM1) is a dominant human neuromuscular disorder caused by a CTG repeat expansion at the 3' end of the DMPK gene. Pathogenesis of DM1 is linked to a toxic gain of function due to mutant RNA and is manifested by nuclear retention of expanded CUG repeats and aberrant splicing. The development of an inducible model for DM1 with uninterrupted CTG repeats could help us to better understand early pathogenic changes in DM1 due to CUG repeat expansion. In the first part of this thesis, I report an inducible C2C12 mouse myoblast cell line in which the pTetOne inducible system was used to express 174 CTG repeats. This resulted in the production of RNA foci in 26% of cells. Efforts to make larger un-interrupted CTG repeats were unsuccessful due to their instability in the E.coli plasmid. In the second part of the thesis, I used the CRISPR/Cas9-induced genome-editing technique to knock-in an inducible promoter into the endogenous DMPK gene in a DM1 fibroblast cell line. For this, I employed two different CRISPR/Cas9 based strategies which exploit homology-directed repair (HDR) and non-homologous end joining (NHEJ). Our results suggest that CRISPR/Cas9 induced knock-in is enabled by non-homologous end joining more efficiently as compared to homology-directed repair. In the last part of the thesis, TruSeq RNA Sequencing was used to quantify the number of mutant DMPK transcripts and other molecular markers of DM1 pathogenesis that could be a valuable tool for the evaluation of the efficacy of therapeutic compounds. The sequencing results reveal the significant low abundance of mutant DMPK transcripts in the cytoplasmic fraction of DM1 lines and confirm the previously reported nuclear retention of mutant DMPK transcripts. We identify six potential genes which are dysregulated in DM1 fibroblasts and the absolute quantification of mutant DMPK transcripts along with these six reported dysregulated genes can be suitable biomarkers for disease severity and therapeutic response in DM1 fibroblasts. Additionally, these assays could be further refined to provide valuable tools to assess therapeutic compounds.

Replication of genetic information is a vital process across all domains of life. Bacillus subtilis is considered the gram-positive model bacterium for studying DNA replication (Escherichia coli has been studied extensively as the gram-negative model) and is most representative of the ancestral phylum of prokaryotes. DNA replication has three distinct stages; initiation, elongation and termination. Replication initiation is the focus of this research and this process occurs at a single origin conserved throughout bacteria, termed oriC. B. subtilis primosomal machinery is formed of replication initiator proteins DnaA, DnaD and DnaB, the helicase loader DnaI, replicative helicase DnaC and primase DnaG. The role of the initiator proteins is to melt the DNA double helix and enable loading of the hexameric ring helicase onto each strand of DNA for bidirectional replication. Initiation is the first stage in DNA replication and despite its importance the molecular mechanisms of replication initiation remain largely unclear. The work presented in this thesis has focussed on the essential interaction between replication initiator proteins DnaA and DnaD, with an aim to characterise their binding interface and reveal molecular details of their mechanisms of interaction during DNA replication initiation. The direct interaction between isolated DnaA domain I and DnaD DDBH2 domain was detected by NMR spectroscopy which was subsequently used to identify the specific residues involved and characterise the nature of the binding interface. The kinetics of the interaction were investigated by SPR and computational techniques were used to model the DnaA-DnaD complex. This structural characterisation of the DnaA-DnaD interaction provides greater understanding of the molecular mechanisms of DnaA and DnaD during DNA replication initiation.

The chloroplast protein CP12 is present in almost all photosynthetic organisms. This protein has been shown to regulate the activity of two enzymes of the Calvin-Benson cycle, namely glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and phosphoribulokinase (PRK). The regulation of these enzymes is achieved by the reversible formation of this multiprotein complex in response to a change in light intensity. In Arabidopsis, there are three CP12 genes, CP12-1, CP12-2 and CP12-3. Expression analysis of these genes suggested that they may have a wider role in non-photosynthetic plastids through the plants’ life cycle and that their function may not be restricted to the Calvin-Benson cycle. The main aim of this study was to determine the functional significance of having three CP12 isoforms and to explore the importance of each individual isoform in vivo. This was done by using Arabidopsis thaliana T-DNA mutant and RNAi transgenic lines with a reduced level of CP12. Our results revealed that single mutant lines did not develop a severe growth phenotype. However, a reduction in the transcript of more than one CP12 gene, in a number of multiple lines, led to a significant reduction in photosynthetic capacity at early stages of development and a severe growth phenotype, including reduced fresh and dry weight, number of leaves and seed yield, as well as affected lateral roots formation. Complementation analysis of CP12-1 in the triple mutant revealed that two out of the three lines rescued the phenotype by showing normal growth and development, confirming the importance of CP12. Our results suggest that the CP12 protein family is essential for normal growth and development and that these proteins are likely to have additional functions apart from the regulation of Calvin-Benson cycle enzymes.

During DNA replication, forks often stall upon encountering obstacles blocking their progression. Cells will act to speedily remove or overcome such barriers, thus allowing complete synthesis of chromosomes. This is the case for R-loops, DNA/RNA hybrids that arise during transcription. One mechanism to remove such R-loops involve DNA/RNA helicases. Here, I have shown that one such helicase, Sen1, associates with replisome components during S phase in the model organism S. cerevisiae. I demonstrate that the N-terminal domain of Sen1 is both sufficient and necessary for the interaction of the protein with the replisome. I also identified Ctf4 as one of at least two replisome interactors of Sen1. By mutational analysis, a mutant of Sen1 (Sen1-3) that cannot interact with the replisome was created. This mutant is healthy on its own but is lethal in the absence of both RNase H1 and H2. Overexpression of the sen1-3 allele from the constitutive ACT1 promoter is able to suppress this synthetic lethality, suggesting that Sen1 travels with replisomes in order to be quickly recruited at sites of R-loops that impair fork progression so as to remove those R-loops. In some cases, cells exploit fork stalling for biologically important processes. This is the case in Sz. pombe, where an imprint prevents complete DNA replication, triggering cell-type switching. This imprint is dependent on Pol1, a component of the replisome. Importantly, a single imprinting-defective allele of pol1 has been identified to date. Using in vitro assays, I have shown that this Pol1 mutant has reduced affinity for its substrates and is a correspondingly poor polymerase. By generating novel alleles of pol1, I have also demonstrated that switching-deficiency correlates with the affinity of Pol1 for its substrates in vivo. Finally, two interactors of Pol1 (Mcl1Ctf4 and Spp1Pri1 ) have been shown to have switching defects. S. cerevisiae and Sz. pombe have similar yet distinct genetic nomenclature conventions. Given that both model organisms were used in this study, it is important to highlight the conventions for both organisms to prevent confusion. In S. cerevisiae, wildtype gene names are expressed as a three letter, uppercase and italic name followed by a number (e.g. SEN1). The three letter name often corresponds to the screen through which the gene in question was originally identified. Mutants are generally designated with the same three letter but in lower case (unless the mutant is dominant) and with an allele designation (e.g. sen1∆, sen1-1 and sen1-2). Because of historical context, the allele designations vary in format (e.g. leu2-3,112 is a mutant of LEU2). Protein names are given as a three letter name with the first letter in uppercase (e.g. Sen1). This is also true for mutant proteins, with the added allele designation (e.g Sen1-1 and Sen1-2). In this study, I have generated constructs of the SEN1 gene and these constructs are referred to as SEN1 (X-Y), where X and Y refer to the first and last residues being encoded for. The corresponding proteins are referred to as Sen1 (X-Y). Different promoters have been used and, where appropriate, the promoters are expressed similarly to their wildtype gene names (e.g. GAL1, SEN1 and ACT1). In Sz. pombe, wildtype gene names are expressed as a three letter, lowercase and italic name followed by a number (e.g. pol1). Mutants are generally designated in the same format but with an allele designation. Like in S. cerevisiae, the allele designation varies widely (e.g. pol1-1, pol1-H4 and pol1-ts13). Additionally, because of the historical context, some (but not all) alleles of pol1 are referred to as swi7 to reflect the fact that they are defective for cell-type switching. Similar to the situation in S. cerevisiae, proteins names are given as a three letter name with the first letter in uppercase for both wildtype and mutants (e.g. Pol1 and Swi7-1). Sometimes, for the sake of comparison, genes or proteins are referred to their S. cerevisiae orthologues (e.g. swi1TOF1 and Swi1Tof1 , respectively). Several protein tags have been used in this study. When written in gene form, they were written in capital letters and italicized, irrespective of the host (e.g. 5FLAG) and when in protein form, they were written in capital, irrespective of the host (e.g. 5FLAG).

Although marine microorganisms drive the major biogeochemical cycles in marine ecosystems, there is a dearth of information on interactions between phototrophic and heterotrophic organisms co-occurring in oceanic waters. The aim of this project was to study these interactions using Synechococcus sp. as the model phototroph – a cosmopolitan and highly abundant member of the picophytoplankton. Heterotrophic bacteria most-frequently present in non-axenic Synechococcus sp. cultures, were identified by PCR screening using primers targeting the 16S rRNA gene. Members of the Nitratireductor, Rhodobacteraceae, Muricauda and Phyllobacteriacae genera were present in more than half of all the cultures tested (Chapter 3). Using a member of the Rhodobacteraceae as the model heterotroph, specific metabolites present in axenic cultures and co-cultures were analysed (Chapter 4). Much lower concentrations of these specific metabolites were present in the milieu of Synechococcus – Roseobacter co-cultures compared to axenic Synechococcus cultures as discovered by LC-MS. Natural product database searches suggest that these may be a group of novel compounds. A Synechococcus sp. WH7803 null mutant in the gene encoding a type III polyketide synthase was constructed (Chapter 5). A targeted exometabolomic analysis showed a decreased production of the metabolites identified above in the mutant strain compared to the wild type. Growth was considerably affected in the T3 PKS mutant and T3 PKS mutant culture supernatants had a stronger negative growth effect on a range of picocyanobacteria and green algal species than the wild type extract. Further research is required to establish the precise biological function of the observed molecules, their biosynthetic pathway and their function in the natural environment. Improving our understanding of interactions between environmentally important microorganisms not only helps us to learn more about how biogeochemical cycles in the ocean function, but can also provide new natural products for use in the pharmaceutical industry.

The question of how the fertilized egg develops into an adult organism is one of the most fundamental ones in Biology. A very important stage in the development of the embryo is cell differentiation, in which unspecialised cells, called stem cells, become specialised ones, such as skin or nerve cells depending on the signals that they receive. This is controlled by a very large network of genes that interact with each other, the state of which defines the characteristics of the cell. With the recent development of experimental techniques that allow us to obtain very detailed information about the changes in cells, new data analysis methods and mathematical models are required for the understanding of stem cell differentiation. A common approach to the mathematical modelling of stem cell differentiation is by means of gene regulatory network (GRN) models describing the gene regulation behind the process. However, the number of variables and parameters in these models rapidly scales up as one tries to study more genes in the network, difficulting its analysis. This thesis aims to assess these problems and it is structured into two main parts. In the first one, which comprises Chapters 3 and 4, we will develop a phenotypic quasi-potential landscape model for vulval development in C. elegans to illustrate how catastrophe theory can be a powerful tool to construct and understand these recently emerging types of models. Moreover, will use advanced statistical techniques to fit the built model to the experimental data. The second part, in Chapter 5, will be devoted to developing a methodology to understand protein expression data in order to reverse engineer the gene regulatory network from it and create a mathematical model that explains such experimental data.

In the first part of the thesis we focus on a hierarchical analysis on multiple, single cell, Nrf2 reporter levels in nucleus and cytoplasm, observed in human endothelial HMEC-1 in vitro cells (Xue et al., 2015a). Nrf2 is a transcription factor that regulates the expression of several defensive genes protecting against various cellular stresses and forms of oxidation. This analysis aims to gain an insight into this essential cellular protective mechanism. We propose a reaction network based on five reactions, including a distributed delay and a non-linear term, for longitudinal measurements of the amount of Nrf2 in nucleus and cytoplasm. The diffusion approximation (DA) is used to approximate this Markov jump process with a stochastic delay differential equation (SDDE). Since this continuous process is only observed at discrete time points, a second approximation, the Euler-Maruyama approximation (EMA) of the DA, is needed to obtain an approximate likelihood for this bivariate process. Furthermore, to make use of multiple single cell data, we embed the model in a Bayesian hierarchical framework. Moreover, a measurement equation, which involves a proportionality constant and a bivariate normal error, for the nuclear and cytoplasmic measurements, is necessary to relate the original unobserved population levels, X, to the observations, Y. This introduces a hidden Markov process for X and a Bayesian analysis is performed, via a data augmentation procedure, to explore the high dimensional posterior space which includes a bivariate latent process X for every cell. We show results obtained on simulation studies, proving the validity of the methodology, and on a real data application, composed of 35 single cell fluorescent xvi levels under the basal condition, and of 36 under the induction by a stimulant, both observed every two minutes for 1.5-7 hours. In the second part of the thesis we describe the analysis of a switch gene model for mRNA populations. We consider a gene that switches, with exponential waiting times, between a more active ON state and a less active OFF state, where the gene transcribes mRNA at a higher and a lower rate, respectively. We observe, via a measurement equation, the mRNA level in each cell, which is assumed to have reached a steady state. We analytically derive the stationary distribution of such a model and infer its parameters from experimental data, again via hierarchical Bayesian inference. The mRNA populations are only observed up to a proportionality constant and with a second source of white noise attributed to the measurement process. As in the previous case, we use a data augmentation procedure to explore the posterior space of the latent data. The analysis is repeated for different levels of induction by tetracycline, a stimulant, which results in increased gene expression. We particularly focus on studying how the stimulation affects the system.

Targeted degradation of cytoplasmic mRNA is of fundamental importance for regulated gene expression in eukaryotic cells. The shortening and removal of the poly(A) tail (deadenylation) is the initial and often rate-limiting step in this process. Two key components in deadenylation are the PAN2-PAN3 and the Ccr4-Not complexes. PAN2-PAN3 is a trimer composed by two regulatory subunits (PAN3) and a catalytic enzyme (PAN2). The Ccr4-not complex is composed of at least eight subunits, constituting four modules, that bind the scaffold protein CNOT1: the N-terminal module (CNOT10, CNOT11); the catalytic nuclease module (Caf1, Ccr4); the CNOT9 module (CNOT9); and the NOT-module (CNOT2, CNOT3). A cellular and biochemical characterisation of the PAN2-PAN3 complex was planned, to further understand its role in eukaryotic cells and determine the deadenylation rates. Similarly, we have started to reconstitute the human Ccr4-Not to understand the contributions of the individual modules to the activity of the complex. Initially, we purified a recombinant nuclease module composed of Caf1, Ccr4, and the anti-proliferative protein BTG2. Next, we reconstituted a core complex including the nuclease module, CNOT9 and a segment of CNOT1 encompassing the MIF4G and DUF3819 domains (CNOT1M). Then, we compared the activity of the core complex and the nuclease module using substrates containing short (A9) and longer (A20 or A50) poly(A) tails, as well as poly(A) tails coated with the poly(A)-binding protein PABPC1. We observed that the CNOT9 module contributes to enhanced deadenylation and increased selectivity towards terminal adenosine residues. Finally, the influence of the melanoma-associated P131L amino acid substitution of CNOT9 was investigated with structural and biochemical approaches.

VEZF1 is a ubiquitously expressed transcription factor with epigenetic roles at gene regulatory elements, including protection from DNA methylation. Genetic ablation of Vezf1 in murine embryos is lethal due to haemorrhages localised in head and neck mainly. Interestingly, Vezf1 seems to be important for the generation of erythroid cells. The role of human VEZF1 in erythroid gene regulation was the primary focus of this study. Previous work found that VEZF1 unexpectedly interacts with the majority of erythroid-specific enhancer elements, including those at the α- and β-globin gene clusters, in the K562 cell line, a long- standing model for erythroid progenitor cells. The co-occupancy of VEZF1 with well characterised erythroid transcription factors and histone marks associated with active enhancers indicated that VEZF1 may support the regulation of the erythroid gene regulatory programme. To understand the role of VEZF1 in erythroid gene regulation and differentiation, we used CRISPR technology to knock out VEZF1 in K562 and hESCs. RNA-seq analysis revealed that VEZF1 does have regulatory roles at several erythroid genes, but it is not essential for the erythroid gene regulatory programme. VEZF1 null K562 cells display increased levels of haemoglobin biosynthesis. Intriguingly, we find that VEZF1 protein levels are progressively downregulated during the in vitro erythroid differentiation of human ESCs. VEZF1 null ESC cultures commit more readily to the erythroid lineage at the onset of erythropoiesis, which then proceeds as normal. VEZF1 is therefore not required for human erythropoiesis and appears to be an early regulatory block to erythropoiesis proceeding effectively. Careful inspection of the cultures at a stage prior to the onset of erythroid differentiation revealed that a sub-population of endothelial precursors was absent in VEZF1 null ESC cultures. We demonstrate that the absence of VEZF1 prevents the efficient differentiation of endothelial cells. The role of VEZF1 in vascular endothelial development is therefore conserved in mammals, although the impact on human endothelial differentiation appears to be more profound. VEZF1 is anticipated to be a key regulator of the vascular endothelial gene regulatory programme.