The mitotic spindle is required for the accurate segregation of an equal number of chromosomes, into two separate daughter cells. The movement of chromosomes is carried out by kinetochore fibres (K-fibres) of the spindle apparatus, these extend from the cell pole and terminate at attachment on the kinetochore. K-fibres are composed of bundles of individual kinetochore microtubules (kMTs). Microtubule (MT) bundling is coordinated by inter-MT bridges, which are thought to promote MT stability and organisation. The inter-MT bridges in K-fibres have never been fully characterised. Clathrin is a triskelion-structured protein with a well-characterised role in membrane trafficking, where it forms the ‘coat’ of clathrin-coated vesicles. More recently clathrin has been shown to have an unrelated function during mitosis: a period of the cell cycle when membrane trafficking is dormant. The depletion of clathrin by RNAi results in a number of mitotic defects, all of which are a consequence of destabilised K-fibres. It has been proposed that a TACC3/ch-TOG/clathrin complex physically cross-braces kMTs to stabilise K-fibres. This project set-up correlative light electron microscopy (CLEM) methods allowing mitotic, clathrin-depleted cells to be analysed from the level of light microscopy to electron microcopy. A dual-axis ultrastructural methodology was also developed allowing K-fibre bundles and inter-MT bridges to be measured. This project used CLEM to reveal that a clathrin complex functions as a population of short inter-MT bridges in K-fibres. The depletion of clathrin or TACC3 resulted in: 1) a loss of kMTs, 2) a decrease in kMT organisation and 3) a loss of a population of short inter-kMT bridges. Immunogold labelling confirmed that clathrin was a component of the inter-kMT bridge. The literature typically refer to inter-MT bridges as electron dense ‘struts’ that cross-link MTs, however, a few classic EM studies have described bridges as ‘filamentous’ with ‘multiple projections’. A highly resolved view of bridge structures was sought. Using high-pressure freezing with freeze-substitution, tomography and 3D rendering, this project revealed that a novel structure was present that cross-linked kMTs, we have termed this structure ‘inter-kMT mesh’. The mesh was found exclusively between kMTs and appeared to promote kMT organisation through multiple contacts with MTs. The mesh responded to manipulated levels of TACC3 expression, resulting in disorganised K-fibre bundles, misshapen kMTs and an enhanced electron-dense inter-kMT mesh. Collectively, these new data contribute to the existing knowledge of inter-MT cross-linkers in K-fibres by providing novel insights into their structure and how they confer kMT stability and organisation.

There are currently 30 extracellular proteins known to form pathogenic amyloid within the human body. These proteins result in a wide range of diseases described as amyloidoses. These amyloidoses can be either systemic, affecting multiple organs, or localised to a single tissue. Almost all systemic amyloidoses can have cardiac manifestations and cardiac involvement correlates with poor prognosis. Cardiac infiltration is most commonly associated with AL amyloidosis, with 50 % of patients presenting with heart failure at the time of diagnosis. In addition to the myocardium, amyloid deposits are frequently reported in the surrounding vasculature. The most common form of localised amyloid – aortic medial amyloid (AMA) - is primarily localised to the internal elastic laminae of the ascending aorta. It is estimated to be present in 97 % of the Caucasian population of 50 years of age but despite its prevalence there remains very little information available about the structural, biophysical and aggregation properties of the main protein component, a 5.5 kDa peptide termed medin. Clinically, it has been suggested that AMA may have a role in aortic aneurysm and dissection through binding to the elastic structures of the aorta and reducing elasticity. This work aimed to further our molecular-level understanding of amyloid aggregates formed by proteins, known to affect the cardiovascular system; medin, and a model peptide for AL amyloidosis, SMA. Initially, this work investigated the structural and biophysical properties of three medin derived peptides with a view to identifying key amyloidogenic regions that could be targeted therapeutically. Furthermore, this work established that small ii medin peptides were not suitable structural models for full length medin. It was therefore necessary to develop a reliable method for the production of full-length medin. This work describes for the first time a procedure for the expression and purification of soluble medin in both unlabelled and 13C/15N labelled forms in E.coli for future biophysical and structural investigations. The recombinant medin was then characterised using a variety of techniques commonly used for the characterisation of amyloid proteins. These results indicated that medin aggregation proceeds via a nucleation dependent mechanism with a lag time of 30 hours and forms amyloid-like fibrils visible by transmission electron microscopy. Structural and biophysical studies were coupled with computational techniques to generate a structural model of medin fibrils, stabilised by a salt bridge between residues D25 and K30. Salt bridge mutations gave rise to aggregates with altered aggregation mechanisms and morphology. It is not known what causes some immunoglobulin light chain fragments involved in AL amyloidosis to become pathogenic. It is possible that structure underlies the differences. This work demonstrates a protocol for producing isotopically labelled immunoglobulin light chains, in particular a model protein SMA, and subsequent NMR analysis to detect inter-nuclear contacts within aggregates. Collectively this work has identified key regions and features of medin that could be targeted in future for therapeutic or diagnostic purposes. Furthermore, it paves the way for further structural studies of aggregates formed by immunoglobulin light chains.

Protein ubiquitylation represents a versatile mechanism to regulate multiple cellular processes, including protein turnover, transcription, cell signaling, membrane trafficking and DNA damage repair. The attachment of ubiquitin moieties to a target substrate is sequentially catalysed by three enzymes, namely E1, E2 and E3, and this process can be reversed by the deubiquitylases (DUBs). This research project seeks to decipher the involvement of DUBs in regulation of the components of the adherens junction (AJ), E-cadherin and β-catenin, in particular, which are the core components of the AJ. It is found in MCF7 cells, a breast cancer cell line, that E-cadherin undergoes constitutive turnover via the lysosomal pathway, as indicated by the accumulation of an 80kDa fragment of E-cadherin degradative fragment following treatment of MCF7 cells with Folimycin, a lysosomal v-ATPase inhibitor. Using the ratio of the full length and 80kDa degradative fragment as a biochemical readout for the trafficking status of E-cadherin, a siRNA human DUB library screen was performed and USP38 was identified as a leading candidate. siRNA depletion of USP38 led to increase in the full length to 80kDa E-cadherin ratio. Notably, the loss of USP38 led to significant loss of total E-cadherin level (sum of full length E-cadherin and 80kDa E-cadherin fragment). The loss of E-cadherin at the protein level following USP38 knockdown was not accompanied by a decrease in its mRNA level or increase in degradation rate, suggesting that translational effects could operate. Preliminary characterisation of this enzyme showed that it has both nuclear and cytosolic localisation. Mapping of nuclear localisation determinants revealed a role for a 46 amino acid insertion within the core catalytic domain, and the last 100 amino acid on C-terminus of USP38. In a separate siRNA DUB library screen performed on A549 cells, the established tumour suppressor BAP1 was identified as a DUB regulating β-catenin. siRNA depletion of BAP1 resulted in a loss of β-catenin and loss of cell-cell contacts, whereas its overexpression led to an increase in β-catenin level, indicating a positive regulatory role of BAP1 on β-catenin level. To facilitate further screening efforts and to provide an extra layer of validation to my own screens, a human DUB endonuclease-prepared small interfering RNA (esiRNA) library was developed. EsiRNA is a complex mixture of siRNAs generated by random cleavage of long double-stranded RNA and is argued to be able to circumvent the problem of siRNA-induced off-target effects. EsiRNAs were made for 91 out of the currently predicted 93 human DUBs. Quality control experiments were carried out upon a sample of selected DUBs and optimisation of experimental conditions for efficient knockdown in various cell lines was explored.

Dissection of the cellular dynamics and molecular pathways that drive collective cell migration is necessary to better understand cellular rearrangements that underpin normal development, as well as disease states such as cancer metastasis. Border cell migration in the Drosophila ovary has proven to be a good model of invasive cell migration, because of its genetic tractability, and also because recent advances in culturing egg chambers ex vivo have facilitated live cell imaging in this system. The aim of this thesis was to further develop and implement live cell imaging approaches, and to apply these to characterise the role of Pico, the Drosophila Mig10/RIAM/Lpd (MRL) protein in border cell migration. MRL proteins are known to regulate actin dynamics, but their role in epithelial cell migration had not been established. Through careful optimisation, suitable approaches were developed for: medium preparation; dissection and mounting of egg chambers; acquisition of images by confocal microscopy. A fluorescently-labelled reporter strain with improved optical properties was generated to monitor actin dynamics, and a number of other reporters were characterised, either alone or in combination, to determine their behaviour and effect on migration. After trialling several analytical tools and quantitative methods, a streamlined approach to analysing the image data was developed allowing: tracking of border cell migration in four dimensions (XYZ and time) to obtain information about behaviour of the migratory cells; measurement of cellular protrusion dynamics to obtain mechanistic insight into why cellular dynamics might change in different genetic backgrounds. Finally, these approaches were applied to the characterisation of Pico and its interacting partner SCAR, demonstrating that pico affects border cell migration through the modulation of actin protrusion dynamics in a SCAR-dependent manner.

Palmitoylation is a reversible post-translational modification of proteins which involves the addition of the C16 saturated fatty acid palmitic acid to sulfhydryl groups on cysteine residues, forming a thioester linkage. The addition of palmitate allows proteins to associate with different cellular membranes and membrane subdomains. Palmitoylation is catalysed by the DHHC family of palmitoyl-acyl transferases (PATs), named for their characteristic DHHC motif in a cysteine-rich domain (CRD). Reversibility is conferred by palmitoyl-protein thioesterases (PPTs), which cleave the thioester linkage. The study of palmitoylation has recently gathered pace with the development of methods which allow proteome-scale identification of candidate palmitoyl-proteins. Despite the importance of model organisms in several key studies in the field, palmitoylation has barely been studied in the simple eukaryote Caenorhabditis elegans, the nematode worm. This study commenced with the use of the C. elegans genome data to identify its PATs and PPTs, using the DHHC-CRD and homology respectively. The 15 DHHC PATs were officially named using a dhhc-x system and the two previously known PPTs were confirmed as the only ones with homology to other known PPTs. The current knowledge on palmitoylation enzymes and substrates was collated and analysed to predict possible phenotypes resulting from mutation or knockdown of the enzymes and potential substrates. All available C. elegans strains containing a mutation in an individual PAT or PPT were obtained, covering about half of the PATs and both PPTs, and assayed for various gross phenotypes. In addition a complete library of bacteria able to express double-stranded RNA against PAT or PPT genes was sourced and used to perform similar assays using feeding RNA interference (RNAi). A number of small but significant differences were seen both with mutant and RNAi-treated strains, especially in lifespan assays. To test for possible redundancy and compensation amongst the enzymes, double RNAi was performed against selected closely related PATs and both PPTs. This resulted in the largest phenotype seen: a reduction in lifespan after simultaneous knockdown of both ppt-1 and ath-1. As there are no known palmitoyl-proteins in C. elegans, the proteomic approaches acyl-biotin exchange (ABE) and acyl-resin-assisted capture (acyl-RAC) were employed to provide a list of candidates. These were first optimised using rat brain material and the results compared with previous proteomic studies to find that two-thirds of the hits had been previously found. With this validation, both methods were applied to wild-type C. elegans lysates to give 91 hits as putative palmitoyl-proteins. Mutants for the PPT ath-1 were also profiled by ABE, showing 33 hits which were not present in the wild-type profile. These are potential ATH-1 substrates whose lack of depalmitoylation in the mutant leads to their enrichment relative to wild-type. However, further repeats of these analyses are required for rigorous statistical testing. Taken together, this study shows the first characterisation of palmitoylation in C. elegans, encompassing all of the DHHC PAT and PPT enzymes, putative palmitoyl-proteins and potential substrates of the PPT enzyme ATH-1.

The Mig10/RIAM/Lpd (MRL) family of adapter proteins have been recognised in transducing signals derived from growth factor receptors to alterations in cell motility and adhesion via changes in actin dynamics. Reduction in the levels of MRL proteins results in diminished cell division rates, growth retardation, an increase in monomeric (G): filamentous (F) actin ratios, loss of cell migration, and lethality. Conversely, overexpression of MRL proteins reduces the ratio of G:F actin, thereby promoting Serum Response Factor (SRF) signalling, lamellipodia formation, cellular invasion and coordinated cell growth and proliferation. Members of the MRL family all share common structural characteristics, most notably the presence of highly conserved central Ras-association (RA) and Pleckstrin homology (PH) domains in addition to an N-terminal Talin binding site and multiple C-terminal SH3 and proline rich motifs capable of interacting with Ena/VASP and Profilin. This report extends previous work carried out on Pico, the Drosophila MRL homologue, by demonstrating physical interactions with Chickadee (Profilin), Rhea (Talin), Ras, and Protein Phosphatase 1 (PP1). Many of these binding partners were found to co-localise with Pico within highly dynamic membrane ruffles during Drosophila cell spreading, while only Enabled (Ena/VASP) co-localised at the periphery of cells once they had reached a maximal size and spreading had ceased, pointing to the existence of distinct Pico-associated complexes. This work also details the presence of a highly conserved MAPK-binding site adjacent to the RA domain which had not been previously recorded in the literature. Site-directed mutagenesis revealed this MAPK-binding motif to be required for Pico's interactions with the Erk1/2 homologue Rolled, while conservation of MAPK binding ability was demonstrated in both human Lpd and RIAM orthologues. Further analysis showed that Rolled might phosphorylate Pico at a serine residue (Ser 819) previously identified by high-throughput phosphoproteomics, which may in turn promote Pico's interactions with PP1. Wing growth assays performed using site-directed Pico mutants indicated that PP1 plays a role in negative regulation of Pico-mediated growth, although the relevant targets of the phosphatase remain to be identified. Interestingly, reporter gene experiments confirmed that Pico induced SRF-dependent gene expression in ii Drosophila cells while ectopic SRF signalling has been found to increase expression of rolled, suggesting the potential existence of a Rolled/PP1 mediated negative feedback loop regulating Pico functionality.

The Nrf2 system plays an important role in the regulation of the redox state of the cell. Protection from oxidative and genotoxic damage is achieved by Nrf2-mediated expression of antioxidant and phase-2 detoxification enzymes. The Nrf2 network topology possesses several feedback mechanisms that could play a role in the dynamics of Nrf2 protein. To address this issue, a range of molecular reagents were generated, including plasmid and Bacterial Artificial Chromosome (BAC) expression systems. These allowed the localization and dynamics of Nrf2 to be analysed at near endogenous expression levels. Specifically, we generated and characterised an Nrf2-venus BAC stable SK-N-AS cell line using state-of-the-art molecular biology and microscopy techniques including Fluorescence In Situ Hybridization (FISH) and Fluorescence Correlation Spectroscopy (FCS). Single cell imaging showed oscillations in Nrf2 levels that were heterogeneous amongst the cell population. These novel tools were applied to investigate the reciprocal effects of crosstalk between the Nrf2 and NF-κB signalling networks. These interactions have direct implications for a range of physiological and pathological processes including cancer, ageing and inflammation. While systems biology approaches have made important contributions to our understanding of the NF-κB dynamics, we still have a relatively poor understanding of how this system integrates with other regulatory networks in different cell types. Single live-cell imaging of the dynamics of these pathways suggested reciprocal dynamics, where stimulation of Nrf2 delays rather than inhibits NF-κB dynamics and Nrf2 oscillates out-of-phase to robust p65 oscillations. The tools developed during this study will be an important resource for further investigations of the Nrf2 system and its crosstalk with other signalling pathways. The BAC could also be used to generate novel reporter transgenic mice. Such data could be used to inform predictive mathematical models, which have direct implications for rational drug evaluation in either normal or diseased cells and tissues.

The giant muscle protein titin is a central player in cardiovascular health and disease. Titin molecules spanning half of the sarcomere form a filament system in striated muscles. The titin filament is not only an essential structural component of sarcomere, but also plays a central role in myofibril signaling through its kinase domain (TK) and numerous protein ligands. Thus, not surprisingly, mutations in this molecule might have detrimental effects. In this work a structure-driven approach was taken to re-evaluate titin kinase catalytic activity and the pathogenicity of two cardiomyopathy-associated titin mutations. Comparison of recombinant preparations from E.coli and insect cells revealed intrinsic inactivity of TK. It was demonstrated that previously reported phosphotransfer activity towards Tcap (a small Z-disc protein) is due to contaminant kinase activity from insect cells but not TK itself. Next, it was established that the eukaryotic host produces structurally indistinguishable TK from that produced in insect cells, albeit inactive towards Tcap, classical kinase substrates or extracts from mature or differentiating muscles. Structural analysis identified evolutionary conserved residue substitutions converting vertebrate TKs to pseudokinases. The structural context of dilated cardiomyopathy associated mutation was revealed in the crystal structure of TK enclosing neighboring domains A170 and M1. The mutation site resides in a conserved helix located in the linker region between TK and the binding site of ubiquitin E3 ligase MuRF1. Aspartate to valine substitution causes disruption of a conserved hydrogen bond and detachment of the affected helix from TK. In the context of the titin filament, this causes dissociation of the binding site from TK and increases interdomain flexibility. Structural alterations translate into increased MuRF1-mediated degradation of mutant titin fragments through the ubiquitin-proteasome pathway. Speculatively, haploinsufficiency of mutant titin could be a possible pathomechanism leading to dilated cardiomyopathy associated with the mutation under study. Comprehensive analysis of arrhythmogenic left ventricular cardiomyopathy associated titin mutation generated a novel model of pathogenesis. In contrast to previous reports, we demonstrate that mutation does not cause affected domain I10 unfolding, and is structurally compatible. Observed destabilization of the domain was attributed to a disrupted hydrogen bond, causing increased flexibility. A crystal structure of the affected domain flanked by adjacent domains I9-I11 demonstrated that threonine to isoleucine substitution might have detrimental effects on interdomain arrangement, resulting in exposure of a hydrophobic patch. Functionally, differential localization of mutant protein was observed in transgenic muscles. Speculatively, mutation could result in impaired folding of mutant protein and lead to accumulation of degradation-resistant aggregates or cause an increased stickiness to thin filaments as a novel pathomechanism. Results presented in this work demonstrate that TK is a catalytically inactive pseudokinase acting as a molecular scaffold. It was demonstrated that TK and MuRF1 signaling modules are structurally interconnected and genetic perturbation of this link might lead to dilated cardiomyopathy. In similar fashion, genetic alteration of interaction between immunoglobulin domains might cause arrhythmogenic left ventricular cardiomyopathy.

Human embryonic stem cell (hESC)-based therapies will only become viable once we eliminate the use of animal-derived material during ESC scale-up. Some groups have demonstrated the expansion of hESCs in xeno-free systems but the effect on downstream self-renewal and differentiation is poorly understood. Heparan sulfate (HS) is a master regulator of cellular behavior but the role of HS during ESC expansion is unclear, as is the exogenous source of HS in cultures. It has been shown that mESCs synthesise low levels of low-sulfated HS, but it is unclear if culture condition has any impact. In the studies here, three discrete culture conditions were employed for E14 mESC expansion along with immunostaining and RT-qPCR to study marker expression for differentiation to the three lineages and corresponding BM synthesis. SAX-HPLC was used to characterise soluble HS from cells/medium/serum. A varierty of polymers were tested as synthetic alternatives for ESC expansion. It was found that HS-deficient embryoid bodies (EBs) (derived from EXT1-/- mESCs in normal culture conditions) remained in a pluripotent state and lacked a typical differentiation pattern. Furthermore, HS-deficient mESCs could not be maintained in the absence of serum, highlighting a link between serum and HS. EBs derived from E14 mESCs cultured in the absence of serum displayed unusual differentiation patterns, which were rescued by exogenous porcine mucosal heparin (PMH). Feeder cells displayed cell-surface HS but feeder-cell conditioned medium (CM) was predominantly an unsulfated structure. An array of low and highly sulfated HS structures were identified in serum-alone. 10-fold more HS was purified from serum-free feeder-free (-F –FBS) CM compared to the other mESC CM (with/without feeders but in the presence of serum; +/-F +FBS). Furthermore, unlike +/-F +FBS conditions, highly sulfated HS disaccharide UA2S–GlcNS6S was the major constituent in –F-FBS and Sulf2 levels were significantly reduced. Poly-Ɛ-lysine macroporous substrates supported mESC and kidney-derived stem cells (KSCs-GFP) adherence and proliferation, further enhanced by adsorbing RGD or per-sulfated HS structures to the surface of the poly-Ɛ-lysine. The key conclusions from these studies were that serum is a source of HS, without which, mESCs behave uncharacteristically; that synthetic HS-mimetic structures could represent an alternative to serum; and poly-Ɛ-lysine shows great promise to replace current animal-derived coating materials for ESC expansion.

SUMO is a small protein that is ligated to other proteins to regulate their function. Ligation occurs at lysine residues within a SUMO site motif. A wide range of proteins are targets of SUMOylation and in plants SUMO plays a diverse role in many important processes. Processes including development, stress tolerance, hormone regulation, DNA repair and chromatin remodelling are regulated by SUMOylation. SUMO affects protein function primarily by establishing interactions through SUMO interacting motifs (SIMs) in interacting protein partners. SUMO can also alter protein function by blocking access to protein domains and by causing conformational changes to the target. The ability to predict SIMs in plant proteins would be useful for research into the poorly understood mechanisms behind SUMO regulation. Large arrays of synthetic peptides were screened with SUMO to identify SIM peptides. These data were used to characterise the sequence composition of plant SIMs. The plant SIMs were compared and contrasted with human SIMs to highlight the functional differences between these two evolutionary distinct species. The data were used to build a predictor for SIMs using random forest models. A new SUMO site predictor was built using random forest models as well. The SIM predictor was used to identify putative SIM containing proteins in the Arabidopsis thaliana genome and the functional enrichment of these genes was analysed. The role of SUMO in the plant gibberellin (GA) pathway was also investigated. The DELLA protein RGA is a negative regulator of GA signalling and this protein was shown to be SUMOylated. RGA stability is regulated by the GA receptor GID1 and it was demonstrated that GID1a contains a SIM. It was proposed that SUMOylated RGA interacted with GID1a through its SIM which inhibited its function. The model was tested by investigating the binding of SUMO to GID1a and by generating mutants of GID1a that had reduced SUMO a affinity. The results demonstrate that GA signalling can be enhanced by introducing a mutation into the GID1a SIM.