Search

Single cell protein analysis has the potential to comprehensively profile cellular heterogeneity advancing the understanding of cell function, disease progression and drug development. Although it is a possible contributor to the discovery of novel therapy, the protein mapping on the individual cell basis is complex and poses challenges. For example, cancer is a heterogeneous disease which analysis using conventional bulk methods can potentially conceal a dangerous population like Circulating Tumour Cells (CTCs). Analysis of CTCs can only be reliably attained using single cell technologies. The Microfluidic Affinity Capture (MAC) chip is a tool that was developed in our group to study cellular heterogeneity by measuring protein abundance in single cells. Another problem in analysing cancer cells is the difficulty in obtaining samples in a non-invasive manner. This thesis reports on attempts to obtain, concentrate and analyse CTCs from blood using a composite MAC-based device. The ability to analyse single cells has advantages beyond the potential for analysing heterogeneity. Chronic Obstructive Pulmonary Disease (COPD) is a heterogeneous illness that is characterised by a chronic inflammation. Although the immune response in COPD requires investigation, it is complicated by a lack of biomarkers and methods to non-invasively collect samples. The ability to obtain and effectively analyse precious and scarce biomaterial from lungs, is greatly advantageous for both diagnosis and tracking of COPD. Here, we report on a workflow to achieve this using sputum from negative control volunteers and COPD+ patients. Also, we describe the establishment of a novel MAC chip assay targeting FOXO3 protein which is involved in regulation of processes like tumour suppression, inflammation and senescence. Therefore, it has a biomarker potential to monitor and study diseases like cancer and COPD. We developed the protocol to analyse Forkhead box 3 (FOXO3) protein expression in nasal cells from healthy donor samples that were retrieved with a non-invasive tool (NASAM, nasal synthetic absorptive matrix). Nasal cavity is a front line of exposure to inhaled pernicious substances and, it is speculated to reflect anomalous bioprocesses in lungs preceding respiratory disease development and progression. This study provides the first-time quantification of FOXO3 protein in single cells from nasal samples. This work shows the analytical capacity of the MAC chip to study cellular heterogeneity and quantify important biomarkers in single cells of clinically relevant material.

A wide range of cellular processes are inherently stochastic. While stochasticity of gene transcription and translation or cellular growth profiles is well-understood, little is known about the stochastic properties of metabolism. Recent experimental findings strongly suggest that metabolism may indeed be subject to stochastic phenomena, which has questioned the traditional deterministic view of metabolism and casts crucial doubt on the general validity of this modelling paradigm. In this thesis, we examine stochastic aspects of metabolic reactions in detail. We focus on stochastic versions of classic deterministic models for metabolic reactions coupled with well-established stochastic models for gene expression. We incorporate experimental measurements of kinetic parameters in the study, which results in a specific multiscale structure of the presented class of models. In the course of this thesis, we present numerous models with increasing complexity, focussing on three key contributions. Firstly, we present the derivation of an analytical tool to approximate stationary metabolite distributions in closed-form by exploiting the multiple scales. As a result, we propose a strikingly-accurate analytical tool for exploring the parameter space. Secondly, we reveal which parameters have strong impacts on the stationary metabolite distributions and identify conditions for increased coefficients of variation and highly-complex bimodal and multimodal patterns. Finally, we propose a general strategy to obtain closed-form approximations in more complex models, such as multi-step pathways and regulatory processes commonly found in metabolism, such as allostery or end-product inhibition. The results in this thesis lay the groundwork for future studies of metabolic heterogeneity and offer numerous biological hypotheses that could soon be tested in light of recent progress in single-cell measurements of cellular metabolites.

The packaging of DNA in chromatin is an important determinant of transcription, replication and recombination. Prokaryotic DNA is thought to be permissive and accessible to the transacriptional machinery. In contrast, eukaryotic chromatin is thought to be repressive for transcription (Struhl, 1999). To ask whether histones are sufficient for the transition from a permissive (open) to a repressive (closed) transcriptional state, we introduced histones HMfA and HMfB from the methanogenic archaeon Methanothermus fervidus into a bacterial host, Escherichia coli. We found that archaeal histones form dynamic nucleosomal arrays composed of oligomerising histone dimers in the E. coli genome, just as they do in Archaea. The positioning of archeal histones is determined by the underlying E. coli DNA sequence, as well as bacterial transcription and replication during active growth. Archaeal histones are surprisingly well tolerated in rich medium and under a variety of stresses, with the exception of treatments that directly affect cellular DNA topology or cause DNA damage. Even though E. coli lacks chromatin remodeling enzymes to actively reshape the chromatin landscape, archaeal histones have at best weak repressive local effects on transcription. These findings suggest that archaeal histones, similar to bacterial chromatin components, are not as restrictive as their eukaryotic counterparts, and may define a “proto-restrictive” ground state.

The Type III secretion system is a widespread bacterial secreting machine that drives assembly of important motility and virulence structures, with a large human impact on agricultural and human diseases. A conserved core module common to flagella and injectisomes, the type III secretion system uses both a proton motive force (PMF) and energy of ATP hydrolysis to unfold and secrete protein substrates that refold in self-assembling structures like a molecular 3D printer. The mechanism of secretion, and the co-operating roles of the PMF and an associated ATPase are not well understood. This thesis presents three complementary structural biology approaches to advance our understanding of how T3SSs operate. In Chapter 2, the first cryo- EM structure of the major component protein of the type III secretion system is resolved to 3.5Å. The structure of an SsaVc nonameric ring resembles that of the homologous MxiAc nonamer, and a comparison of homologue structures shows a highly conserved D3 domain which forms the subunit interfaces, and variable opening of the large cleft between D2 and D4. The unique acidic C-terminus of SsaV is disordered in our map, and the linker and transmembrane regions that are involved in energising secretion are not resolved. In Chapter 3, the assembly and structure of the ATPase complex is investigated by comparing mutant complexes in situ with tomography. We find that FliH functions as a structural scaffold for the ATPase and can assemble independently to FliY on the C-ring. We find evidence for overlapping functional roles of the ATPase and the substrate chaperones, as well as heavy binding of the C-ring in the absence of FliH; the top candidate identity for the C-ring binding factor is the initial isoprenoid pathway enzyme DXP. In Chapter 4, computational predictions are made for the protein folds of the inner membrane export apparatus proteins. A recent breakthrough in the field allows assessment of some of the predicted folds against experimental structures, confirming some of our predictions of the pore structure. The picture of type III secretion that emerges is one where the ATPase complex is stably-assembled and is dispensable for secretion, as are chaperones, but together contribute assistance to an essential PMF-driven step of substrate processing that occurs at the SctV nonameric cytoplasmic ring.

Interactions with membrane lipids have long been known to affect a wide range of membrane protein properties including folding, stability, and function. Our understanding of the precise molecular basis of these interactions is currently limited. Here, we combine mass spectrometry (MS)-based techniques with molecular dynamics (MD) simulations, in vivo biochemical assays, and other biophysical techniques to characterise how protein-lipid interactions affect the oligomerisation, conformation, and function of the structurally related xanthine/uric acid transporter UapA from Aspergillus nidulans and the boron transporter BOR1 from Saccharomyces cerevisiae (ScBOR1p). Using native MS and lipidomics, we found that UapA requires phosphatidylethanolamine or phosphatidylinositol binding to form the physiological dimer. A putative lipid binding site at the dimer interface was identified using MD simulations. Lipid binding at this site is essential for formation of functional UapA dimers. Similar analysis of ScBOR1p revealed that this protein is primarily monomeric in detergent-based solution and that phosphatidylethanolamine or phosphatidylserine binding is also essential for its dimerization. Mutation of a putative lipid binding site in ScBOR1p results in loss of lipid-induced dimer formation but has no effect on transport function. Thus, the lipids play slightly different roles in UapA and ScBOR1p. Additional research probed the conformational dynamics of UapA in different lipid compositions using hydrogen-deuterium exchange MS (HDX-MS). The results revealed that protein-lipid interactions stabilise the mobile domain of UapA responsible for substrate transport. HDX-MS also revealed how substrate binding stabilises the inward facing conformation of UapA and that specific mutations stabilise the outward facing conformation of UapA. Together, this work represents the first detailed analyses of the interactions between eukaryotic membrane transporters and their associated lipids.

It is widely understood that dysregulation of epigenetic events such as post-translational modifications (PTMs) of histone proteins, are associated with a plethora of diseases, such as cancer. A prevalent and regularly studied PTM is the methylation of lysine residues on histones, mediated by the family of enzymes - histone lysine methyltransferases (HKMTs). However, to date there are considerable challenges in defining unambiguously the methylation targets of a single HKMT. The major reasons for this are: 1) there is considerable redundancy in histone methylation and multiple HKMTs can 'write' the same target methylation modification, 2) HKMTs also often have many non-histone substrates. The majority of HKMTs contain the SET (Su(var)3-9, Enhancer-of-zeste and Trithorax) binding domain which binds the co-factor S-adenosyl methionine (SAM) enabling the transfer of the methyl donor group to the intended substrate. Through computational studies, the group has located an endogenous 'hole' in the SET binding domain of SETD7 making it structurally different to other HKMTs (e.g. G9a). The work presented here shows the synthesis of a small library of SAM analogues to be utilised by SETD7. The analogues have various groups appended at the C-2 position of the adenine ring to compliment the endogenous 'hole' in SETD7. Additionally, these cofactors include an isotopically labelled CD3 reporter group so newly deposited 'heavy' methyl marks can be readily distinguished from endogenous methylation using LC-MS/MS analysis. Initial in vitro and in vivo studies have been performed to demonstrate the validity of this approach for histone targets with the intention of extending to a library of non-histone targets.

Lipid droplets (LDs) are intracellular organelles, which have long been ignored as passive aggregates of lipids. However, recent research firmly established LDs as dynamic structures, actively regulating energy storage and generating signalling molecules. Concomitant with unfolding the regulatory mechanisms of LDs, including the identification of LD-associated proteins, their functional importance is increasingly being recognised in various pathologies such as metabolic diseases and cancer. Early embryogenesis is a tightly orchestrated process, often investigated through in vitro models, such as pluripotent stem cells. Interestingly, dynamic LD enlargement and lipid storage was found to operate in in vivo and blastocyst-derived embryonic stem cells (ESCs). The overall objective of this PhD research project is to specifically investigate the biology and roles of LDs in regulating pluripotent stem cell biology and metabolism. Here, we demonstrate an essential role for LD-associated protein Cidea in promoting LD enlargement in ESCs using gain- and loss-of-function approaches. In vivo, we observe that Cidea expression peaks at the morula-blastocyst transition consistent with the formation of enlarged LDs throughout the embryo. Yet these are mobilised in the pluripotent (epiblast) compartment at the onset of implantation, pointing to enhanced, tissue-specific lipid usage during this key developmental step. Using ESC models, we recognize a possible role for Cidea during the morphogenesis events accompanying epiblast maturation post-implantation. From a metabolic perspective, we find that enforced Cidea expression and LD enlargement correlate with enhanced oxygen consumption and TCA cycle activity with unchanged anaerobic glycolytic capacity. The functional significance of these alterations is further investigated using self-renewing, ESC clonal populations that harbour different LD morphologies, gene expression patterns and metabolic capacities. Integrating RNA-sequencing data, GC-MS and Seahorse-based experiments, we propose that ESCs might transiently enhance their oxidative status to sustain the energy-demanding process of lipid synthesis taking place at the exit of ESC pluripotency.

X-ray crystallography is the most widely used technique for determining 3-D protein structures. Information gained from crystallography is vital to the success of rational drug design and other biotechnological applications. Crystallography relies on the production of high quality protein crystals but obtaining such crystals is a major obstacle to progress. This project tackles this problem by (i) designing, developing and validating new and improved crystallisation methodologies for obtaining high quality crystals; and (ii) applying the new techniques in addition to existing methods, to crystallise medically important proteins in order to facilitate their 3-Dimensional structure determination. Presented here is the design and validation of an improved method to slow down protein crystallisation in order to enable the growth of fewer, larger single crystals of higher quality. The method called the oil-on-drop method is a variation of the Chayen method which places an oil barrier over precipitant reservoirs in hanging drop trials. The variation consists of dispensing the oil directly onto the protein drops. This method was successfully tested and validated using five proteins, three of which are of medical relevance including a methyltransferase and an antibody-peptide complex. A comparison of the oil-on-drop method with existing standard and non-standard techniques for optimising crystal quality such as the application of nucleants and the Chayen method showed the oil-on-drop technique to be superior in many ways. Most importantly, it reproducibly yielded crystals diffracting to higher resolution of three of the proteins tested namely the methyltransferase, the complex and trypsin. In the case of the Roab13 antibody-peptide complex the crystals obtained were of the highest resolution than ever attained previously. In addition, the oil-on-drop method overcomes problems encountered by other methods. It can be used with a wider variety of precipitating agents and can be performed using robots in a high throughput mode.

Bisindole natural products encompass the well-studied indolocarbazoles, such as rebeccamycin and staurosporine, among other minor classes including indolotryptolines and bisindolylpyrroles. Bisindole biosynthesis starts from L-tryptophan and forms highly aromatic bisindole scaffolds in a series of biosynthetic steps, which are catalysed by various core scaffold formation and tailoring enzymes that work in tandem to shuttle highly labile bisindole intermediates to the cognate biosynthetic routes. Due to the demonstrated potent biological activities of bisindoles against disease-relevant clinical targets, new bisindoles and their synthetic analogues could yield promising drug candidates with improved activities. This thesis sets out to implement synthetic biology approaches to discover and diversify bisindoles via a two-pronged strategy – (1) by combinatorial assembly of bisindole biosynthesis pathway variants, or in combination with synthetic chemistry to further diversify the chemical scaffolds, and (2) by genome mining of new bisindole biosynthetic gene clusters (BGCs) using existing bioinformatic tools and experimental validation of the BGCs via heterologous expression. The first aim was achieved by derivatising the antimicrobial bisindole violacein scaffold by precursor-directed biosynthesis, in vivo halogenation and semi-synthesis of cross-coupling violacein derivatives based on a brominated scaffold. The violacein biosynthesis pathway was also refactored and heterologously expressed in Escherichia coli as a proof-of-concept demonstration of rapid pathway assembly using the EcoFlex Golden Gate toolkit with user-customisable genetic parts. The second aim was achieved by genome mining for the uniquely conserved chromopyrrolic acid synthase (CPAS) enzyme homologues, which led to the identification of two new bisindole BGCs (acf and nos). Heterologous expression of these BGCs in Streptomyces coelicolor resulted in the discovery of new bisindoles including arcyriaflavin F. Further investigations into new bisindole BGCs with unusual enzymes, in combination with semi-synthetic and synthetic biology approaches, will likely lead to the discovery of novel bisindoles with unprecedented scaffolds and biological activities.

The physical barrier posed by the plasma membrane greatly restricts the potential of intracellular delivery of macromolecules. Currently available delivery methods suffer from various limitations, including low delivery efficiency or high cytotoxicity. To overcome these issues, stimuli-responsive polymers such as the bio-inspired, pH-responsive PP50 polymer, comprising a poly-L-lysine isophthalamide backbone with hydrophobic L-phenylalanine grafts can be used. In mildly acidic environments, PP50 can permeabilise the cell membrane overcoming the problem of payload entrapment in the endosomes and allowing for efficient delivery of molecules into the cell interior. The work presented herein demonstrates that PP50 is capable of delivering various macromolecular payloads in vitro, such as different-sized dextrans, green fluorescent protein and the apoptotic peptide Bim by simple co-incubation with the desired cargo at pH 6.5. The delivery process was fast, non-toxic and compatible with multiple cell lines tested, including adherent and suspension cell lines, primary human mesenchymal stem cells as well as cells grown as spheroids. In addition, peptide delivery by co-incubation with PP50 was at least 3 times more effective than delivery using other common delivery methods, including poly(ethyleneimine), cell penetrating peptides and electroporation. In addition, novel conjugates of PP50 and different model and functional payloads were developed using a cleavable crosslinker to enable in vivo delivery and release. Model fluorescent payloads such as a peptide-sized PEG and green fluorescent protein were delivered to HeLa cells following conjugation with PP50. Finally, Bim conjugated to PP50 was shown to retain its apoptotic effect in vitro and was demonstrated to be non-immunogenic and well tolerated in a mouse model and exhibited preferential tumour accumulation. Our findings suggest that PP50-mediated payload delivery is a versatile method allowing delivery of various payloads to many different cell lines and can find many potential uses both in vitro and in vivo.

Engineering well-functioning metabolic pathways remains a challenge, largely because designing a metabolic pathway-encoding construct to express proteins at intended levels is limited and often leads to an imbalance of the pathway’s protein concentration profile and therefore a poor performance. Whilst combinatorial and iterative approaches have proven effective for optimising the pathway balance in model organisms, implementing these approaches has typically proven arduous and time-consuming for many other organisms with phenotypes of industrial relevance. Such organisms are often considered as non-model because they are difficult to engineer, often lacking the tractability, fast-growth rates and high transformation efficiencies of model organisms. This project aims to implement rapid prototyping of metabolic pathway-encoding constructs in the tractable organism, Escherichia coli, before transferring the construct for expression in the non-model organism, the cyanobacterium Synechocystis sp. PCC 6803. Synechocystis is a photoautotroph that offers the potential to sustainably produce products from carbon dioxide. However, its slow growth rate, genetic instability and lack of genetic tools make it difficult to engineer. To test the rapid prototyping of metabolic pathway-encoding constructs, genetic parts and a design framework were required. For this, libraries of synthetic promoters and synthetic RBSs were developed, from which portable promoters and RBSs were identified that offered predictive protein expression between the two organisms. A novel multi-part, modular DNA assembly method was developed, termed Start-Stop Assembly, to serve as a framework. Using these tools, variations of the lycopene pathway were compared using current state-of-the-art and novel rapid prototyping designs. In this instance, rapid prototyping did not allow the performance of a pathway in E. coli to be used as a prediction of its performance in Synechocystis, and therefore it offered no benefit to the pathways design. Implementing a combinatorial approach did however improve the proportion of effective and stable pathways in Synechocystis, which led to lycopene being directly produced from carbon dioxide.

This thesis describes my work to develop and apply an automated fluorescence lifetime imaging (FLIM) microscopy-based method for high-content analysis (HCA) of protein- protein interactions (PPIs) at the budding yeast kinetochore in live cells via Förster resonance energy transfer (FRET). FLIM provides a robust readout of FRET, since the fluorescence lifetime is independent of the fluorophore concentration, excitation power and the spectral properties of the sample and instrument, but it requires relatively high signal to noise. Here I demonstrated, for the first time, that FLIM HCA could be applied in live cells to study endogenous proteins labelled with fluorescent proteins and developed new protocols and data acquisition and analysis tools to address the challenges associated with the low copy number of kinetochore proteins (KPs) and high (autofluorescence) background. The openFLIM-HCA platform developed at Imperial College London implements automated multiwell plate FLIM using wide-field, optically-sectioned, time-gated imaging with open source software to analyse FLIM data. I adapted the instrumentation and software tools to map (sub-10 nm) co-localisation of KPs using FRET in live yeast cells and validated this methodology by showing that it can detect known KP interactions. I then explored the potential of this FLIM HCA methodology to query the kinetochore localisation of Rdh54p as a means to investigate how this DNA repair protein promotes the adaptation to the DNA damage-induced cell cycle arrest. I also forced the associations of Rdh54p (and mutants) with most of the proteins in the proteome using the high- throughput synthetic physical interaction (SPI) method to identify associations that affect cell growth. Both the FLIM and SPI data converged on the CTF19/COMA complex at the kinetochore, suggesting that Rdh54p constitutively localises adjacent to this complex, which may have functional significance. The SPI screens also identified hits related to the DNA damage checkpoint that are interesting for future investigations.

Bacterial transporters play fundamental roles in nutrient acquisition and drug efflux. Determining the structures of these transport proteins is essential for understanding the mechanisms bacteria use to ensure cell survival. In E. coli, TauABCD is a taurine-sulfur assimilation system upregulated during conditions of sulfur starvation. The substrate-binding protein TauA is required for the initial step of taurine recognition followed by its delivery to the ABC transporter TauBC. In this work, structures of TauA in complex with taurine and three analogues have been determined. Structural comparison shows that the size of the binding site dictates substrate selectivity of TauA. Desolvation of the TauA binding site is likely to be a key determinant in the binding of taurine and its analogues. Under conditions of nutrient starvation, bacteria also produce and release antibacterial peptides like MccC7/C51. MccC7/C51 hijacks the ABC importer YejABEF to enter sensitive bacteria. However, the mechanism for MccC7/C51 internalisation and the physiological function of YejABEF have yet to be defined. In this work, the structure of YejA has been determined in a peptide-bound state, suggesting YejA is likely to be an oligopeptide-binding protein. Furthermore, interrogation of the bound peptide sequence provides a rationale for prodrug design of novel antibiotics. Finally, the AcrAB-TolC efflux system is a major reason for multidrug resistance in E. coli. AcrB is a major component of the system, responsible for energy transduction and substrate recognition. However, to date, no inhibitor against bacterial efflux pumps has been clinically licensed as a drug. Here, using a combination of transport and MIC assays, one compound has been highlighted as a potential AcrAB-TolC inhibitor. In addition, well diffracting crystals have been produced for AcrB in the conformation favouring ligand binding. These crystals were applied to the XChem fragment screening platform for ligand soaking. AcrB represents the first example of a membrane protein tested using this platform.

This thesis explores the crystallisation of 8 proteins; lysozyme, superoxide dismutase, monoclonal antibody anti-CD20, native state human serum albumin and 4 engineered human serum albumins HSA-0098, HSA- 0354, HSA-0575 and HSA-0311. The application of mesoporous silica nanotemplates for protein crystallisation for both downstream therapeutic protein manufacture and for protein structure determination perspectives are studied in this thesis. Mesoporous silicas were synthesised with controlled pore diameters and narrow pore size distributions to validate the relationship between protein hydrodynamic diameter and nucleant surface porosity previously proposed by Shah et. al. Results confirm that such a correlation influences the propensity of crystallisation demonstrated by all proteins studied here. The thesis reports on the crystallisation of the enzyme superoxide dismutase in concentrations x10 lower than that previously reported in literature, for the enzymes family, and at a x68 lower concentration for the specific enzyme complex CuZnSOD for concentrations as low as 0.2mg/ml. Such low concentrations for protein crystallisation are rarely reported, and indeed are occurring at the lowest levels of supersaturation concentrations. The success of crystallisation was due to the use of mesoporous silica nanotemplates with specific pore size distributions which correlated with CuZnSOD’s hydrodynamic diameter in solution. High quality diffraction crystals with resolution of 2.2Å were produced in solution conditions that failed to grow crystals without silica nanotemplates. A native state polymorph of therapeutic human serum albumins was successfully crystallised to produce diffraction quality crystals with a resolution of 2.8Å. Data was collected for structure determination and successfully solved to reveal a novel polymorphic structure which has never been reported before. The protein revealed a P 21 21 21 space-group unique to human serum albumins. The structure shows a high degree of symmetry with superimposable dimers which sets it apart from the other known structural forms. In addition to crystallisation for structure determination, the crystallisation conditions proved to be very robust and highly reproducible which allowed the construction of a phase diagram to map out the limits of crystallisation within the bounds of the reagent conditions. Further experiments advantageous for process development such as batch to batch reproducibility, product crystal yield were also reported. Attempts to crystallise engineered forms of Human Serum Albumins (HSA) showed varying levels of success. Out of four different engineered samples, one was demonstrated to be crystallisable to diffraction quality of 2.8Å with its resolved structure revealing two octanoate ligands which bind in domain II and IIIA. Other HSA forms showed crystallisation of low crystal quality in the form of spherulites. Nevertheless, these are the first crystals of this kind reported. To understand the interaction occurring between mesoporous nanotemplates and proteins, novel Inverse Liquid Chromatography (ILC) experiments resulted in confirmation that crystallisation does indeed occur within the porous structure as opposed to its surface which has been a common question with this type of porous material mediated crystallisation. The ILC results are two-fold, firstly it demonstrates a novel way to observe actual porosity in a porous material in solution conditions using colloidal gold nanoparticles which are similar to conditions to the crystallisation experiments. Secondly, the resulting pore size distribution can be mapped out before and after protein adsorption which demonstrates unequivocally the protein uptake into the pores. Thus, verifying a key step in the protein crystallisation mechanism of these mesoporous nanotemplates.

Do genes enriched in genetic variations that have been linked to human disease have different biological properties than genes enriched in variants that have never been related to disease? What are the characteristics of genes enriched in rare variants, and can they represent the missing link in disease aetiology? The answers to these questions could shed light on the architecture of pathogenesis process and may improve the future studies of genetic disorders. To that end, we analysed 481,277 rare variants (minor allele frequency < 0.01), 81,822 common variants (minor allele frequency ≥0.01), and 26,884 disease-causing variants occurring in the coding region of 17,975 human protein-coding genes. Three novel sets of genes were identified: genes enriched in rare variants (32 genes), genes enriched in common variants (282 genes), and genes enriched in disease-causing variants (800 genes). Our analysis presented consistent results obtained from several well-established tools which showed that genes enriched in rare variants have far greater similarities in the biological and network properties to genes enriched in disease-causing variants, than to genes enriched in common variants. Thus, genetic variants in these genes are strong candidates for disease, and their identification should prompt further in vitro analyses as they may represent the missing link in disease heritability. All the data used in the analysis in addition to other biological data will be publicly available through the development of a dedicated database and website. We expect the website to become the foundation for understanding the molecular details of the different types of genetic variants, which in turn will be of a great benefit to the medical community.

In plants, the H3K27me3 modification, established by the subunit of the Polycomb Repressive Complex 2 showing a methyltransferase activity, is present at more than 4400 genes, which are mostly transcription factors involved in major plants developmental processes. The first aim of this thesis is to test whether nascent RNA could be present at PRC2 targeted genes and whether these nascents RNA could act as anchor point for cis-elements to recruit PRC2 to a specific locus. The presence of nascent RNA at Polycomb repressed genes was demonstrated in Arabidopsis before adapting a new innovative chromatin tool, called Chromatin Isolation by RNA Precipitation (ChIRP), to isolate and identify proteins interacting with nascent RNA. The fully optimized method revealed nineteen candidate proteins to interact specifically with the nascent RNA at the FLC locus (Chapter 3) but the link to PRC2 recruitment is yet to be shown and should be investigated in the future. The second aim of this thesis is to determine the effects of the H3K27me3 erasers on plant development. The flowering time, the fertility and the embryo to seedling transition were studied in different Arabidopsis plants lacking the different demethylases, REF6, ELF6 and JMJ13. It showed antagonist phenotype suggesting very distinct roles. Flowering time studies identified four potential targets of JMJ13, FLC, SNZ, AGL24 and SOC1 that require further validation (Chapter 4) while the effect of ELF6 and REF6 on the flowering time have been described several years ago. Fertility studies showed JMJ13 promotes stamen elongation while ELF6 suppresses stamen and promotes pistil elongations. Seed analysis revealed that the seed size and weight are negatively affected by the demethylases and that REF6 and ELF6 share redundancy (Chapter 5). Germination rate analysis showed JMJ13 suppresses germination while ELF6 and REF6 promote it and the embryo-to-seedling transition study revealed ELF6 is partially required to generate true leaves (Chapter 6). The characterisation of the effects of the demethylases on Arabidopsis development has started recently and should be investigated to better understand the regulation of plant life cycle by the histone demethylases.

As the field of Synthetic Biology is expanding, there is an increasing need for methods to develop new modes of synthetic gene regulation. A demand for measurement methodologies which provide results that can be interpreted outside a specific context and used to design larger composite devices in diverse cell types is especially urgent. The motivation of my PhD research was to establish new prototyping tools to measure gene regulation, specifically for mammalian applications. Two different experimental approaches were explored. At first, a tissue culture based method was adapted to measure gene expression regulation. However, it was found that this method would not scale to testing many variants, and alternatives needed to be sought. The results from characterising an RNA binding protein which can repress translation (L7Ae) in cells were used as a reference point to compare later obtained results using the second cell-free measurement approach. The hypothesis of this thesis comprises of establishing a new cell-free protein synthesis (CFPS) method as a prototyping tool for mammalian gene expression. The sections involve an initial proof-of-concept phase, the development of a semi-automated workflow, discovery of novel findings using the method, and preliminary work with the aim of establishing a mathematical modelling framework. The hypothesis that CFPS is an appropriate model for gene expression in mammalian cells is proven across the biochemical analysis of transcription and translation steps. The potential of CFPS was demonstrated through prototyping diverse modes of gene regulation, including transcription and translation regulation. Specifically, T7 constitutive promoter variants, IRES constitutive translation-initiation sequence variants, CRISPR/dCas9-mediated transcription repression, and L7Ae-mediated translation repression were characterised. Liquid handling technology was used to automate reaction assembly and software tools were developed to aid the experimental process throughout design, implementation and data analysis. A graphical user interface was designed to enable wide use of the method. The robustness of the method was established through comparison of repeated experiments. In the last section, the feasibility to measure RNA production in real time in CFPS was established. A mathematical model of bacterial CFPS was adapted to the mammalian context, and preliminary parameter inference was carried out, based on experimental data. Overall, the results contribute to the understanding of the biochemistry of L7Ae, CRISPR/Cas9 and transcription in mammalian systems. The established automated methods will impact mammalian synthetic biology by enabling further work at a faster scale than would be possible using traditional experimental approaches.

The rapid growth in the number of human genome sequences is leading to the discovery of numerous single amino acid variants (SAVs), some of which are associated with genetic disease. Understanding these variants at both the sequence level and the structural level is of great benefits particularly for clinical researchers. The first part of this thesis focuses on characterising pleiotropic proteins (proteins associated with multiple disease phenotypes). This study shows that pleiotropic proteins have unique properties: 1) they are enriched in disease-associated SAVs and rare neutral SAVs, 2) they tend to contain disordered regions, 3) they tend to have a higher number of protein-protein interactions, and 4), they are more specific to diseases of certain classes such as neoplasm, nervous system, and congenital malformation. This new understanding of pleiotropic proteins could have important implications for evolutionary studies, genetic studies, and any other related field. At the structural level, the effects of SAVs on protein structures were analysed by remodelling high-quality PDB coordinates of proteins to obtain their corresponding mutant structures. By considering 16 structural features affecting protein stability, 40% of disease-associated SAVs and 11.5% neutral SAVs were shown to cause at least one of the 16 structural effects. The limited number of available experimental PDB structures poses a major challenge in structural analysis. However, incorporating homology models allows more SAVs to be analysed as the residue coverage increases from 17% to approximately 50%. The same analysis was repeated on homology models with sequence identity ranging from 30% to 95%. Similar percentages of disease-associated and neutral SAVs causing structural effects were observed even on models with low sequence identity (<40%). Subsequently, a web server Missense3D was developed to help analyse the structural effect of SAVs on protein structures. A better understanding of genetic variants could help researchers unveil the molecular mechanisms underlying human disease, which could support further research such as in drug design and personalised medicine.

Cell differentiation is fundamental to multi-cellular life because a single unspecialized zygote must give rise to all specialized cell types in the body. Cell fate specification requires the coordinated regulation of different genes across developmental time. How genes are regulated during cell differentiation is a longstanding question in biology. Gametogenesis in budding yeast is a cell differentiation program where a diploid cell gives rise to four haploid gametes and is a tractable model to understand gene regulation during development. Multiple studies in various vertebrate systems suggest that a significant fraction of genes are expressed as different transcript isoforms arising from alternative transcription start sites, or overlap with non-coding RNAs (ncRNAs) of unknown function. Here, I investigated how transcript isoforms or overlapping ncRNAs regulate gene expression during budding yeast gametogenesis. First, I showed that the kinetochore gene NDC80 is repressed by transcription of a 5’ extended mRNA isoform from an alternative promoter during early gametogenesis. This repressive mechanism required the co-transcriptional deposition of histone modifications and the stabilization of nucleosomes at the canonical NDC80 promoter. NDC80 repression could also be rapidly reversed to allow cells to re-enter mitosis if they were no longer starved. Next, I showed how a genome-wide approach called transcript end sequencing (TE-seq) could distinguish the ends of overlapping transcripts. TE-seq analysis showed over a thousand meiotic upstream transcripts or isoforms upregulated at specific stages of gametogenesis. Upstream overlapping transcription was associated with different regulatory outcomes on coding genes, and did not always repress them. TE-seq also identified examples of gene regulation which depended on different chromatin remodellers. Finally, most meiotic upstream alternative transcripts were rapidly repressed upon return to growth, similar to the NDC80 example. These results provide a basis for understanding how mRNA isoforms and ncRNAs contribute to local gene regulation during development in eukaryotes.

Metabolomics is a well-established approach for investigation of the metabolic state of an organism usually conducted via high-throughput methods and focusing on quantification and identification of small molecules. A popular analytical technique used in metabolomics is 1H NMR spectroscopy. The data obtained in NMR experiments contains a wealth of information on metabolites in a sample and their chemical structure. To help uncover this information and find patterns in the data, statistical and machine learning methods must be applied. The work presented in this thesis demonstrates applications of probabilistic generative modelling, with particular focus in Latent Dirichlet Allocation (LDA), as a tool for information recovery in 1H NMR data sets obtained in metabolomics research. LDA is an example of a topic model. The model is based on a generative process which can be thought of as a source of the data. Topics are latent variables which select co-occurring metabolites in a sample. In turn, NMR spectra can be represented in the latent variable space. We present applications of LDA in three scenarios. (1) How LDA can be used to simulate NMR spectra; such spectra demonstrate that LDA is a valid model for NMR data and also provide synthetic data for evaluation of statistical models. (2) Unsupervised learning with LDA to uncover patterns in the NMR data; we use synthetics and real NMR data with knowledge of key biomarkers from a prior study and conclude that LDA was successful in the recovery of useful topics. (3) Supervised learning with SLDA and combined latent variable models with ElasticNet regression where we investigate NMR data from The Multi-Ethnic Study of Atherosclerosis (MESA) study which is paired with clinical variables such as BMI. The goal was to examine if topics can be informative about clinical outcomes.

G protein-coupled receptors (GPCRs) present themselves as an attractive class of membrane protein for use within eukaryotic whole-cell biosensors due to their responsiveness to an extensive and diverse range of ligands. Despite being a comparatively simple organism, Saccharomyces cerevisiae has the complex machinery necessary for coupling heterologous GPCRs to a cellular output, thus enabling this highly amenable chassis with the sensing abilities of higher eukaryotes. Although examples are beginning to emerge within the field of synthetic biology, GPCR signalling in yeast remains an underutilised foundation for the creation of new biosensors. Current designs are often limited due to a mismatch between the input concentrations to which these biosensors respond and the application requirements. Here we present a new platform for rationally tuning GPCR-based biosensors in Saccharomyces cerevisiae to address these limitations. While previous efforts to manipulate GPCR signalling in yeast have involved a top-down approach or overlaying complexity, we sought to identify the minimal requirements to achieve fully tuneable behaviour from the bottom-up. Using genome engineering, we constructed an insulated, modular GPCR signal transduction system to study how the response to stimuli can be predictably tuned using synthetic tools. By systematically refactoring the system, we delineated the contributions of a minimal set of components, identifying robust and straightforward design rules for tuning the sensitivity, leakiness, and signal output. Using these principles, we then established novel community-based approaches for tuning the final dose-response property – the Hill slope. This work enables the development of diverse yeast biosensors that are well-suited to their applications demands, while also providing a framework to guide the reprogramming of GPCR-based signalling in more complex systems.

Cell membranes are a complex mixture of lipids and proteins, all of which vary in size, shape and composition. The ability for a cell to control its membrane, and regulate its contents, is in large part due to the regulation of local composition, curvature and tension. These changes to membrane properties can in turn alter the behaviour of membrane bound proteins. Understanding how these membranes, and their fundamental properties, can influence proteins in vitro is essential for building true synthetic cells, and building artificial cellular networks. In this thesis, work is presented which investigates the effect of membrane composition on the activity of the mechanosensitive channel of large conductance (MscL) from E.Coli. Using different model membrane systems, and many activation mechanisms of the protein, a greater understanding of MscL function has been gained. The first reconstitution, and activity measurement, of MscL into droplet interface bilayers (DIBs) is presented. Being able to measure MscL activity in the DIB model membrane has then lead to whole new avenues of research. The DIB system allowed for electrophysiological recording of single MscL channels, exploring the ability to see in-depth changes to MscL activity as DIB composition changes. Then, by linking multiple DIBs together, a network of bilayers connected by MscL channels has been presented. The first time a synthetic bilayer network linked by a gated ion channel has been demonstrated. Finally, a new device has been built allowing the generation of a measurable amount of shear force inside a DIB, and MscL activity was measured. It is hoped that the results of this thesis, will lead to further studies of more complex biological systems, deeper mechanistic studies of membrane proteins, and further synthetic networks of membrane protein interactions.

This thesis describes the development of a new type of two-dimensional mass spectrometry: partial covariance-based two dimensional mass spectrometry, or pC-2DMS. Using a commercially available linear ion trap mass spectrometer and the statistical technique of partial covariance mapping, the ability to identify fragment ions originating from the same or consecutive decomposition pathways of the same molecule is demonstrated. As will be shown, this information is beneficial to analytical and mechanistic studies of trapped biopolymers. After initial development on peptide molecules, the generality of the pC-2DMS technique has been confirmed by extension to whole protein molecules as well as DNA and RNA oligonucleotide sequences. This work represents the first time partial covariance mapping has been used for structural analysis, the first application of a covariance mapping technique to trapped ions, the first time a covariance mapping technique has been applied to molecular species as large as molecules on the order of kilodaltons and the first demonstration of the correlation of secondary consecutive fragmentation products by a covariance mapping technique. Within pC-2DMS, the total ion count is used as a single fluctuating parameter in the partial covariance formula to suppress spurious correlations due to scan-to-scan fluctuations in a number of different experimental parameters, which cause quasi-uniform fluctuations across all of the tandem mass spectrum. The applications demonstrated in this thesis include the resolution of mixtures of different combinatorially modified structural isomers (shown here for structural isomers of histone H4 fragment 4-17), an orders-of-magnitude reduction in the false positive rate for matching fragment ion signals in database search algorithms, resolution of charge state for highly charged fragment ions without the need for high mass accuracy resolution of the isotopic envelope and the in silico separation of strongly overlapping fragment ion spectra from different co-isolated and co-fragmented molecular species.

Sebaceous and skin lipids from the outermost layer of the epidermis, the stratum corneum (SC), play a vital role in regulating the hydration level of our skin. Skin lipids constitute the first layer of defence against toxic particles entering the body, and the first barrier to overcome when cosmetic and therapeutic creams are applied to the skin. This thesis gives insights into the properties of the skin surface at the microscopic level by introducing a model of the sebum, the oily layer covering the SC. Molecular dynamics simulations are used to investigate the structural and thermodynamic properties of the sebum in bulk and at air and water interfaces, and to study its interactions with water and SC lipids. The sebum model accurately reproduces experimental results, such as density, surface tensions and surface pressure-area isotherms. Calculating the free energy barrier to water penetration indicates that the sebum is permeable to water diffusion via a network of clustered triglyceride glycerol groups, suggesting that the sebum may play a significant role in the water regulation function of the skin. The model is also used to obtain a molecular understanding of SC lipid-sebum interactions. It is shown that sebum triglycerides interact strongly with epidermic lipids and extract them from the bilayer. Using microsecond simulations, the mechanisms associated with the extraction processes are identified and the corresponding free energies are quantified. Further, it is shown that the skin surface has a low affinity for water (contact angle: approximately 100 degree ), highlighting a striking difference between SC lipids and phospholipids. Heat transfer from a nanoparticle and the conformations of an antimicrobial peptide are also studied, with the eventual, future goal of combining these with the sebum-SC model. Finally, an outreach activity to improve the general public’s understanding of lipid molecules and their amphiphilic structure or “split-personality”, is described.

Guanine-rich sequences are primarily found in the promoter region of several proto-oncogenes and in the telomeric region; they have been shown to readily fold into quadruple-stranded structures, termed G-quadruplexes, in vitro. Mounting experimental evidence suggests that these structures are involved in key biological processes like replication, transcription and telomere maintenance. A wide range of high-affinity binders has been presented in the literature, but the development of small molecule binders has been mostly limited to conventional batch synthesis. In addition, screening efforts for new potent binders are mostly restricted to 96-well plate-based techniques with reduced throughput. The research presented in this thesis aimed to overcome both shortcomings by providing new technology and new synthetic approaches, the key findings are the following: 1) A “one-pot” as well as an “in flow” synthesis approach was developed for metal salphen based G4 binders. 1H-NMR spectroscopy proved full conversion of a range of Ni(II) salphen complexes. The FRET melting data of the “one pot” prepared compounds (which do not require any purification before testing) were in good accordance with the conventionally synthesised compounds. 2) A microfluidics-based FRET melting platform that enables the real-time and on-line assessment of G-quadruplex binders in continuous flow was developed. The performance of the platform was tested with four commercially available G4 binders BRACO19, TMPyP4, PDS and PhenDC3, the non-binder TMPyP2 as negative control and three new Ni(II) salphen complexes. 3) Segmented flow microfluidics was exploited to develop a pressure-driven approach for generating monodisperse water-in-oil droplets which relied on integrating a quartz pipette (pore size = 3 – 10 μm) into a microfluidic device and pushing the dispersed phase at a constant flow rate through the pipette while varying the carrier phase flow rate. Droplet volumes ranged from 231 fL to 261 pL for a single device when varying the carrier phase flow rate from 1 to 500 μl/min. Using this droplet generation method high-throughput fluorescence-based assays in droplets, namely the FRET melting assay and the Job’s method were carried out, to shine light on the stabilising effect of several G4 binders (PDS, BRACO19, TMPyP4) and on the binding stoichiometry of a novel Pt(II) salphen-based “switch-on” ligand. 4) Furthermore, the integration of the biological target (dimeric quadruplex DNA) into the in situ synthesis was explored in a target-guided synthesis approach. Four different alkyne modified Ni(II) salphen complexes and three different diazide PEG linkers served as building blocks for the in situ “click” reaction. Successful protocols to analyse the mixtures by LC-MS were developed. However, integration of the target inhibited the reaction rendering the “target-guided synthesis” approach unsuccessful.

The cellular membrane is a complex biophysical environment that occupies a central position in many fundamental biological processes. In association with the membrane there are a wide variety of proteins, macromolecular actors with crucial roles in driving those processes. Being a primary point of contact between the inside and outside of cells, membrane proteins are targeted by the majority of approved therapeutic drugs. New methods that can unveil the mechanisms by which these molecules function are therefore crucial in deepening our biomolecular understanding and can provide rational approaches for the design of novel therapeutic approaches. Conformational dynamics are an important factor in driving the behaviour of membrane proteins. Receptors, channels and other membrane-embedded proteins function by populating distinct structural states and interconverting between them. However, the molecular detail of these heterogeneous processes is extremely challenging to elucidate with the current analytical tools, generally tailored for homogeneous protein states. This thesis focuses on the methodological development of approaches to accurately characterise the dynamical behaviour of membrane proteins. The methods outlined combine the use of experiments and theoretical sampling to generate ensemble descriptions of membrane proteins. Specifically, the experimentally-restrained molecular dynamics approach is further developed for the incorporation of solid-state NMR observables. The accuracy of this method is validated in the context of membrane proteins and subsequently applied to the study of three different biological systems: the cardiac regulatory protein phospholamban, the influenza proton channel M2 and the G-protein coupled receptor CXCR1. The generated ensembles reveal some important conformational transitions in these membrane proteins, and highlight the importance of combining theory and experiments in order to achieve an accurate, detailed description of biomolecular dynamics in lipid membranes.

Work outlined in this report describes the development a biological device which can precisely control the type, speed, and spread of mutations directed towards a target gene of interest. The intention was to continuously search through a vast DNA sequence landscape by constructing a self- controlling system in vivo, which can direct the evolution of the target protein using a non-rational approach, in an effort to identify new or improved functionality, whilst circumnavigating the need for complex and unreliable rational protein design. I describe the creation of a novel fusion protein, consisting of the mutagenic protagonist, Activation Induced Deaminase (AID), tethered to a highly targetable transcription factor (T7 RNA Polymerase). By specifically mutating only the intended target gene this system is able to increase the number of interesting mutations i.e. those which have a direct consequence on protein function. Furthermore, results show that reducing background mutations helps maintain a viable host cell. Initial results suggest that when this mutator protein is expressed in an E.coli strain with a limited base excision repair (BER) pathway, there is a ~1500 fold increase in the frequency of mutation events. Although early iterations of the system primarily showed transition mutations (C-T 75.5%, G-A 24.5%) were by far the most common, later improvements suggest a broader spectrum of mutations can be achieved by the adaptation of error-prone repair mechanisms. I also present a novel DNA assembly method named BASIC, which was designed with automation in mind, in order to streamline construction and reiteration of expression plasmids, reporter devices and the numerous fusion proteins required throughout the project. BASIC has shown exceptional performance, achieving ~90% accuracy for assemblies containing 7 DNA fragments of varying lengths. With principle mutator and construction methods in place, further work highlighted here describes the creation and characterisation of non-natural biological parts in the form of hybrid promoters/operators responsive to the inducer Tetracycline. I provide a set of variants which achieve a range of expression strengths with tight transition between uninduced (off) and induced (on) states. Estimations of relative promoter strength and critical inducer concentrations were incorporated into a deterministic model of the overall mutator circuit using a number of ordinary differential equations, including both my own parameterised kinetic functions and those taken from the literature. Although simplified, such computational models quickly revealed limitations of early circuit designs. This included the overly engineered solution to premature stoppage of the mutator device in the form of an AND gate. In its current form, the mutator system will be used to select for novel forms of the LasR molecule which can bind or associate with a number of small AHL molecules required for quorum sensing activity and activation of the downstream transcriptional activation. More specifically this could be used to 3 screen for species specific variants, and future inclusion in biosensor circuits. From a wider perspective, the mutator system could be introduced alongside alternative display screening methods depending on the target protein and its functional activity to accelerate evolution through multiple rounds of mutation and selection. Furthermore, BASIC assembly has already been widely cited by the synthetic biology community and has been integrated into the work flow of industry collaborators. It is expected that the move towards complete automation of molecular biology lab work, and the increasing integration of liquid handling platforms across biotech, the approach will become increasingly popular.

Successful fold recognition can be readily achieved by current homology detection methods if there exists sufficient sequence similarity between the protein of interest and known structures. However, when such sequence signal does not exist or is too remote to be detected, the task of fold recognition and tertiary structure prediction becomes immensely difficult due to the enormous conformational space to explore. In this thesis, a novel contact threading approach, PhyrePower, which avoids the sequence constraint by identifying structural similarity through protein contact map matching is presented. By combining eigendecomposition and classic dynamic programming algorithms, PhyrePower can effectively utilize the information in predicted residue-residue contacts even in the presence of considerable noise. The potential of this approach is exemplified by its ability to identify a similar structure for 36% of the proteins in a validation set where structural relatives were completely undetectable by the state-of-the-art profile HMM-HMM method. With a simple alignment refinement, a model with a TM-score above 0.5 against the native structure was constructed for 18% of the proteins by PhyrePower. Traditionally, predicted protein contact maps were used as distance constraints in template-free modelling. A widely used molecular mechanics package, TINKER was thoroughly optimised in this work to represent the upper limit on the performance of such contact-assisted methods and a model with a TM-score above 0.5 against the native structure was produced for 11% of the proteins in the validation set. A comparative evaluation between the two methods demonstrated PhyrePower's ability to succeed on the challenging modelling task where an advanced sequence-based method and a direct modelling approach both failed and thus, highlighted PhyrePower's added value in protein folding.

Cysteine-reactive small molecules are used as chemical probes of biological systems and as medicines. By forming covalent bonds with target proteins they achieve distinct potency and selectivity profiles that are redefining the scope of potentially druggable targets. Identifying high-quality covalent ligands requires comprehensive kinetic analysis to distinguish selective binders from pan-reactive compounds. This thesis describes quantitative irreversible tethering, a general method for screening cysteine-reactive small molecules based upon the maximisation of kinetic selectivity. A high-throughput assay is described that facilitates robust kinetic analysis of electrophile-thiol conjugation, effective for both small molecule- and protein-derived thiols in combination with any irreversibly thiol-reactive ligand. Crucial to its success is its ability to account for variations in intrinsic reactivity observed within electrophilic fragment libraries by comparing reactivity between target- and control-thiols. The method is applied prospectively to discover covalent fragments that target two distinct regions of the clinically important cell cycle regulator cyclin-dependent kinase 2. Crystal structures of the inhibitor complexes validate the approach and guide further optimisation. The power of this technique is highlighted by the identification of cyclin-dependent kinase 2-selective probes whose novel mode-of-action may offer future therapeutic potential.

In this thesis, the structures of proteins involved in carbon fixation by enzymes of the 3 hydroxypropionate (3 HP) cycle of Chloroflexus aurantiacus, and the acetone carboxylase from Xanthobacter autrophicus were investigated. The crystal structures of C-terminal and N-terminal domains of malonyl CoA reductase, and mesaconyl C1-CoA hydratase were solved. Preliminary structural information was obtained for mesaconyl-CoA transferase and mesaconyl-C4-CoA hydratase, the right-hand side 3 HP cycle enzymes. The mechanism of 3 HP formation from malonyl CoA, catalysed by malonyl CoA reductase was studied using ligand bound structures, site-directed mutagenesis, and kinetics. The ATP dependent acetone carboxylation mechanism was examined structurally. Acetone carboxylase was natively purified from Xanthobacter autotrophicus and the AMP and acetate bound structure was determined at 1.9 Å. This thesis has applications in the biotechnology of carbon fixation.

This work presents a methodology for simulating ionic and electronic grain boundary transport through thin films, scales or membranes with columnar grain structure. In the model developed the grain structure is idealized as a lattice of identical hexagonal cells - a honeycomb pattern. Reactions with the environment constitute the boundary conditions and drive the transport between the surfaces. Time-dependent simulations solving the Poisson equation self-consistently with the Nernst-Planck flux equations for the mobile species are performed. In the resulting Poisson-Nernst-Planck system of equations, the electrostatic potential is obtained from the Poisson equation in its integral form by summation. An object oriented C++ code has been implemented to solve the system of equations numerically. First, the model is used to interpret alumina membrane oxygen permeation experiments, in which different oxygen gas pressures are applied at opposite membrane surfaces and the resulting flux of oxygen molecules through the membrane is measured. Simulation results involving four mobile species, charged aluminum and oxygen vacancies, electrons, and holes, provide a complete description of the measurements. Second, the model is extended to simulate internal oxidation and stress generation within a thin film of alumina at conditions of high-temperature metal oxidation. The steady-state stresses predicted are compatible with experimental measurements of lateral growth stresses in alumina scale growth experiments. The hypothesized p-n ionic transition within the alumina grain boundaries is observed. In limiting cases the more general simulations are closely related to the Wagner theory of metal oxidation. The Wagner theory assumes local ionic equilibrium while the simulation results demonstrate the possibility of a significant deviation from local Schottky equilibrium within an oxide scale at conditions of steady-state growth. The deviation is related to stress generation, and limits the applicability of the Wagner theory.

The activity of lysine methyltransferases (KMTs) is crucial for the regulation of chromatin-related processes, such as gene expression. Since all KMTs catalyse the transfer of a methyl group from the same SAM cofactor, no current technology is able to directly assess which KMT has performed each observed methylation in a cellular context. This information would be particularly informative in the development of more specific treatments for epigenetically-regulated diseases, such as cancer. To address this technology gap, this work aims at developing a "methylome" profiling assay based on engineering synthetic SAM cofactors specific towards a given KMT. Our proposed computational design strategy takes aligned crystal structures of KMTs as input to identify unique cavities in wild type KMTs that could accommodate complementary synthetic "bumps" on SAM. SAM analogues with sterically matching "bumps" are then designed, and molecular dynamics simulations are utilised to predict their binding mode and affinity with the target KMT to rank the designs. The highest-ranked SAM analogues are synthesised and include an isotopically-heavy methyl donor group (CD3), which retains the biological-consistency of cellular methylation and renders new methylation events detectable via MS. The ability of the KMT to utilise the synthesised SAM analogues as a cofactor for methylation is finally assessed using in vitro assays and quantified via LC-MS/MS proteomics. The results of this work identified bumped SAM analogues that were selective to certain KMTs lacking a post-SET domain. Specifically, a natural pocket was identified in the SET domain of SETD7 that does not exist in post-SET containing KMTs, such as G9a. Complementary bumped SAM analogues were designed and synthesised for experimental validation. A computational screening approach using docking, molecular dynamics and binding affinity estimation revealed that the binding mode and affinity for some of these bumped SAM analogues were comparable to the one of native SAM for SETD7, but not for G9a. Subsequent in vitro experiments using recombinant G9a and SETD7 with our engineered SAM analogues and nucleosome substrates revealed cofactor selectivity towards SETD7. These results were consistent with the predictions of the computational screening approach, which supports the feasibility of our proposed design methodology.

Phosphatidylinositol phosphates (PIPs) are membrane phospholipids and have been studied owing to their crucial role in a wide range of cellular functions. Amongst the seven naturally occurring PIPs, PI(4,5)P2 has been primarily investigated and linked to cellular events related to cell growth, proliferation and apoptosis. Changes of cellular PIP levels and a disruption of the equilibrium between the different PIPs has been linked to the development of a host of diseases, such as Alzheimer's, bipolar disorder or cancer. The development of a chemical tool which is capable to control these PIP levels can thus have the potential for diagnostic and therapeutic application. Previously, it was shown by our group that small molecules which bind to PIPs can be employed to disrupt protein-PIP interactions. By binding to the PIP headgroups, the small molecule prevents the protein from interacting with PIPs, resulting in an apparent reduction of target PIP available and as a consequence, proteins are not recruited to the membrane and downstream signalling pathways can be disrupted. Following from these results, the aim of this project was to investigate the potential of these small molecules to be developed into optical probes and also further expand the library of small molecules for PIP targets. A series of small molecule-receptors were synthesised based on previous leads, incorporating the same recognition motifs for PIPs. Additionally, fluorophores were attached to the scaffold with the aim of studying the host-guest interaction as well as to monitor the cellular uptake of the new receptors. Additionally, we investigated the effect of introducing solubilising substituents on the receptor to facilitate both synthesis and application of these receptors. Moreover, a receptor based on a tripodal scaffold was synthesised, which differed from previously investigated receptor arrangements. In vitro investigations showed that the inclusion of an optical unit changes both binding affinity and PIP-binding selectivity compared to previous small molecule-receptors. Additionally, it was shown that the tripodal scaffold, containing an extra target recognition motif, is capable of binding PIP with higher affinity compared to receptors based on one set of recognition motifs and additionally, PIP-binding selectivity is shifted compared to previously investigated lead compounds. Preliminary results suggest this tripodal receptor is able to enter cells and bind its preferred PIP target in vivo. Combined with its in vitro binding affinities, this suggests that the tripodal arrangement has the potential to generate potent PIP binders, which could be further investigated both for diagnostic or therapeutic application.

Microfluidic Antibody Capture (MAC) chips are small devices capable of quantifying biomarkers in single cells. These devices offer an all-optical approach for cell manipulation, lysis and single molecule quantification of a specific protein. This thesis details various developments to this device, both in terms of throughput and improvements to the single molecule counting process. The tumour suppressor protein p53 is a central hub for cellular stresses such as DNA damage, overproliferation and ribosomal biogenesis stress. Under stressed conditions p53 brings about the expression of a host of downstream effectors ultimately leading to DNA repair, temporary cell cycle arrest, senescence or apoptosis. The specifics of how p53 can lead to a number of different cell fate decisions are still unknown and require the development of quantitative biochemical techniques. In this thesis MAC chips are used to quantify p53 in single cells under a number of conditions. The chip data is used to create a quantitative model of p53 expression. This involved the use of stochastic simulation techniques such as the Gillespie algorithm and Approximate Bayesian Computation (ABC). These simulations determined that differences in p53 expression are best described as changes in the p53 degradation rate. This agrees with previous reports describing the p53-MDM2 relationship and its associated negative feedback loop. Lastly, attempts were made to obtain absolutely quantifiable data from the MAC chip platform. This involved calibrating the platform with known amounts of recombinant p53. By providing absolutely quantifiable data to the model of p53 expression the simulations could potentially provide real, biologically relevant parameters.

The most successful methods for predicting the structure of a protein from its sequence rely on identifying homologous sequences with a known structure and building a model from these structures. A key component of these homology modelling pipelines is a model combination method, responsible for combining homologous structures into a coherent whole. Presented in this thesis is poing2, a model combination method using physics-, knowledge- and template-based constraints to assemble proteins using information from known structures. By combining intrinsic bond length, angle and torsional constraints with long- and short-range information extracted from template structures, poing2 assembles simplified protein models using molecular dynamics algorithms. Compared to the widely-used model combination tool MODELLER, poing2 is able to assemble models of approximately equal quality. When supplied only with poor quality templates or templates that do not cover the majority of the query sequence, poing2 significantly outperforms MODELLER. Additionally presented in this work is PhyreStorm, a tool for quickly and accurately aligning the three-dimensional structure of a query protein with the Protein Data Bank (PDB). The PhyreStorm web server provides comprehensive, current and rapid structural comparisons to the protein data bank, providing researchers with another tool from which a range of biological insights may be drawn. By partitioning the PDB into clusters of similar structures and performing an initial alignment to the representatives of each cluster, PhyreStorm is able to quickly determine which structures should be excluded from the alignment. For a benchmarking set of 100 proteins of diverse structure, PhyreStorm is capable of finding over 90% of all high-scoring structures in the PDB, and over 80% of all structures of moderate alignment score.

Photosynthesis is the process by which autotrophic photosynthetic organisms utilise light energy to assimilate CO2 into biomass, releasing O2 into the atmosphere as a by-product. Even though photosynthetic reactions were crucial in the “Great Oxygenation Event” of our atmosphere 2.4 billion years ago, these greatly limit crop yields. Hence, increasing the photosynthetic efficiency of light conversion into biomass has become a crucial practice to feed the increasing global population. Rubisco is a fundamental enzyme in the carbon reactions of photosynthesis, which fixates atmospheric carbon dioxide into biomass. However, due to its slow turnover (3 molecules of CO2 fixed per second) and inhibition of carboxylation reactions by oxygenation, Rubisco is a major bottleneck of carbon fixation in photosynthesis. Rubisco form I from higher plants is the most abundant form of Rubisco on earth. It is a complex enzyme consisting of 8 large and 8 small subunits, which forms a hexadecameric structure with a mass of 550 kDa in higher plants. Due to Rubisco’s multimeric nature, targeted mutagenesis experiments to investigate more efficient catalysts in higher plants is extremely challenging. Furthermore, its eight active sites are located in the interface of the large subunits, a feature which further complicates the understanding of the events occurring in the active sites responsible for Rubisco’s catalytic inefficiencies. While molecular dynamics (MD) simulations can be used to investigate these inefficiencies, previous studies are limited by a 50 ns time frame, thereby lacking the ability to adequately capture the underlying structural dynamics. For the first time, this thesis presents 17 long MD simulations, ranging from 500-1500 ns, using 13 different structures of Rubisco form I from three distinct organisms (Synechoccocus, spinach and Chlamydomonas). It provides evidence of the suitability of this technique in inspecting the impact of different mutants of Rubisco on the RuBP substrate’s binding affinity. For this purpose, results were compared to existing experimental data of mutant forms of Rubisco. The essays hereby reported demonstrate that, after long MD simulations of Rubisco, the resulting binding affinity ranking of the substrate to different mutants is consistent with previous experimental work. Moreover, the simulations reveal an allosteric behaviour of the substrate binding between the eight active sites of Rubisco, and verify the influence of Rubisco’s structural elements on the binding affinity of its substrate.

The core goal of synthetic biology as a discipline is to design, develop and characterize biological parts in order to precisely control cellular behaviour. Much of the research in this field has been focused on the development of gene regulatory networks, namely switches and oscillators. The study of synthetic gene oscillators has attracted significant attention in the past decade due to their intriguing dynamics and relevance in controlling inflammatory, metabolic and circadian signalling pathways. Additionally, the precise expression dynamics and molecular mechanisms that underlie the mammalian circadian clock structure are not fully understood. The work presented herein regards the design and implementation of a tuneable mammalian synthetic gene oscillator with a novel biological structure. To this end, an approach based on a combination of in silico design and in vivo part validation, in conjunction with a comparative analysis of previously implemented synthetic gene oscillators, was taken when assembling the proposed system. The topology of the system relies on a delayed negative feedback loop, consisting of the coupled regulatory activities of the transcription regulators LacI, tTA, and Gal4. The numerical solution and stability analysis of an ODE-based model describing the dynamics of the system are indicative that the proposed system is capable of generating sustained oscillations across a wide range of parameter values. The biological parts that comprise the system have been monitored and validated in HEK293T cells through time-lapse fluorescence microscopy and image analysis. The in vivo performance of the proposed mammalian synthetic gene oscillator was also assessed in the HEK293T cell line, and monitored using time-lapse fluorescence microscopy. Damped fluorescence oscillations were observed: these could be tuned by a differential IPTG concentration gradient and abolished by doxycycline. The proposed mammalian synthetic gene oscillator provides valuable insight into the gene expression regulatory processes leading to oscillatory behaviour, and has the potential to foster progress in future synthetic biology-based therapies.

In the mouse and human embryo the onset of differentiation is accompanied by a dramatic series of changes at the level of gene expression, epigenetics, signalling, and metabolism. During this period the epiblast becomes hypersensitive to death stimuli and defective cells are eliminated. Cell competition has been proposed as a quality control mechanism that compares cell fitness levels within a cell population and eliminates those cells that are less fit than their neighbours, whilst promoting the proliferation of the fitter cells. In Drosophila, the transcription factor dMyc plays a key role in cell competition. This thesis examines the role of c-Myc, the mouse dMyc homologue, in mammalian cell competition, and demonstrates that c-Myc is key determinant of cell fitness in a number of developmental contexts. We find that in mouse embryonic stem cells (ESCs), and early post-implantation embryos, defective cells decrease c-MYC expression and are eliminated by apoptosis specifically when surrounded by high-MYC expressing neighbours. We also find that defective cardiomyocytes have low c-MYC expression during cardiac regeneration. Furthermore, we show that over-expression of c-Myc in ESCs is sufficient to turn cells into super-competitors, therefore capable of eliminating their wild-type neighbours, and that loss of c-Myc during early post-implantation development leads to reduced embryo size. Together these results indicate that relative c-MYC levels determine the outcome of cell competition during mammalian embryogenesis. We have also analysed what triggers cell competition during early development. Metabolomic analysis of wild-type and defective ESCs revealed that upon differentiation (that triggers cell competition), defective ESCs have significantly lower levels of glycolysis, TCA cycle activity and amino-acid synthesis, indicating a lower metabolic activity. These differences in cell metabolism suggest that during early development cell competition is acting as a metabolic selector mechanism, and eliminating those cells that are less metabolically active.

The work presented in this thesis is aimed to develop and evaluate methodologies for noninvasive measurements of metabolism using fluorescence microscopy. The use of 3D cell cultures in biomedical research is increasing and these require appropriate tools and techniques to provide quantitative readouts for image-based studies. Fluorescence lifetime imaging microscopy (FLIM) can provide robust readouts in complex optical samples and here I have investigated its application to map changes in the response of genetically expressed biosensors utilising Förster resonance energy transfer (FRET) in spheroids. In particular, I adapted a FRET biosensor for the activity of a key metabolic enzyme, AMP activated protein kinase (AMPK), by substituting the donor fluorescent protein ECFP for mTurquoise2, in order to improve its performance in FLIM-based assays. I developed spheroid cultures expressing FRET biosensors and studied these using quantitative FRET readouts. To take account of possible influences of the microenvironment of 3D culture on the fluorescence lifetime measurements, I generated spheroids expressing simple fluorescent proteins and expressing an inactive mutation of the FRET biosensor. I evaluated the new AMPK FRET biosensor, demonstrating improved performance for fluorescence lifetime readouts, and compared dose responses for a direct activator of AMPK with the biosensor expressed in “2D” monolayer cultures and in spheroids, consistently observing a uniform response. In contrast, the dose response of an indirect activator of AMPK in spheroids presented a spatially varying AMPK activation. I further explored the application of FLIM to map the readout of a genetically expressed FRET biosensor for glucose and again observed a spatially varying response in spheroids. I then explored cell specific AMPK activities using FRET biosensors in prostate cancer cells and bone marrow stromal cells with a spheroid system of tumour stromal interactions. I also used biosensors for ATP and glucose concentration in a similar manner and undertook measurements of oxygen consumption rates using a metabolic flux analyser. I observed changes in metabolism that indicate the prostate cancer cells were metabolically benefitting from the interaction with bone marrow cells.

Clostridium difficile is a Gram-positive spore-forming, anaerobic bacterium that is the leading cause of gastrointestinal infection acquired after antibiotic therapy. The symptoms of Clostridium difficile infection (CDI) are primarily attributed to the action of two toxins Toxin A (TcdA) and Toxin B (TcdB), which have been extensively studied. However, how C. difficile successfully colonizes the gut remains poorly defined. It is likely that the cell wall of C. difficile plays a key role in the interaction and colonization of host gastrointestinal tissues. The outermost layer of the cell wall, the Surface layer (S-layer) is a paracrystalline array composed of a complex of two S-layer proteins (SLPs); the high molecular weight (HMW) SLP and low molecular weight (LMW) SLP, derived by post-translational cleavage of the precursor protein SlpA. Previous studies concluded that SlpA in a variety of strains is not glycosylated. SlpA shows sequence variation between strains and 12 distinct slpA cassettes have been recently described. Type 11 cassette strains possess an unusual genotype; adjacent to slpA, they possess a cluster of genes similar to those observed in other species. The protein products of these genes collectively and sequentially glycosylate surface proteins.   In this study, SlpA of type 11 cassette strain Ox247 was demonstrated to be glycosylated using a combination of nuclear magnetic resonance, mass spectrometry and genetic analysis. The putative structure of the Ox247 glycan was determined collaboratively using mass spectrometry and nuclear magnetic resonance. The SlpA of Ox247 is glycosylated only on the LMW-SLP via a threonine residue, of which this O-linked glycosylation parallels other bacterial S-layer glycosylation examples including Geobacillus stearothermophilus and Paenibacillus alvei. The cluster of 19 genes adjacent to slpA was genetically characterised as an operon. Insertional inactivation of genes predicted to initiate glycosylation (orf2) and to ligate the glycan to the protein (orf19) extracellularly resulted in a S-layer deficient in glycosylation and migration of the LMW-SLP on SDS-PAGE at its predicted MW of 20 kDa. However, upon glycosylation, the LMW SLP shows an aberrant migration to 50 kDa, representing the addition of sugar mass to the LMW-SLP. Insertional activation of genes predicted to be glycosyltransferases produced several protein species bearing varying length glycans as confirmed by mass spectrometry, and differential migration on SDS PAGE at MWs between 25-40 kDa. The data obtained is consistent with the S-layer glycosylation model proposed in Geobacillus stearothermophilus, where the glycan is synthesised inside the cell, translocated via an ABC transporter through the membrane to the targeted protein and ligated via an O-linkage. Creation of partially glycosylated mutants suggests that the specificity of this ABC transporter is not restricted to the full-length glycan. This study provides the first functional characterisation of S-layer glycosylation in a Gram-positive pathogen. A broad range of in vitro experiments including antibiotic resistance, motility, aggregation, sporulation, biofilm formation, toxin production and adhesion to mammalian cells, coupled with in vivo analysis using a mouse model were performed to analyse the lifestyle and pathogenesis of C. difficile cells with this post-translational modification. Our data indicates that SlpA glycosylation affects cell division, adhesion and biofilm production.

The survival and proper functioning of an organism depends on the faithful replication of its genetic material. The yeast checkpoint kinase Mec1 and its integral partner Ddc2 (human ATR and ATRIP respectively) play a pivotal role in initiating checkpoint signalling as a response to DNA damage. Mec1 is activated by cell cycle speci c activators that act through conserved aromatic residues located in their unstructured C-terminal regions. Since the molecular details of the activation mechanism of Mec1 are not fully understood yet, we studied the structure of the Mec1-Ddc2 complex and its interactions with the activator Dpb11 protein. Our ndings provide the rst insights into the overall architecture of Mec1-Ddc2 and serve as a basis for a proposed framework for the activation of Mec1. Our single-particle cryo-electron microscopy reconstruction shows Mec1- Ddc2 exists as a dimer of heterodimers and associates through intertwining arm regions consisting of the N-terminal Mec1 HEAT repeats of the adjacent monomers. The Ddc2 subunit, which extends from these repeats, further stabilises the oligomerization interface. The head of the structure accommodates the kinase domain located at the conserved C-terminus of Mec1. Due to a head-to-head dimer conformation, the kinase domains face each other although they are fully separated indicating they do not cause structural impediments that would block substrate access. The interactions between the N-terminus of Ddc2 and the C-terminal tail of Dpb11 mediate the recruitment of the activator to the checkpoint complex. The Ddc2 subunit, which neighbours the kinase domain, not only provides structural support but also facilitates Mec1 activation by bringing Dpb11 into close proximity to the active site. We propose a model where the activator stimulates the kinase activity via multiple interactions between the Mec1 and Ddc2 subunits of the complex that trigger small allosteric changes within the kinase domains. We hope such insights will pave the way to a full mechanistic understanding of this important signalling pathway.

Lateral inhomogeneity in biological membranes has been linked with many cellular functionalities including protein sorting and signal transduction. Fluid phase coexistence has been extensively studied by modelling membranes as bulk mesophases and as giant unilamellar vesicles (GUVs). However, the basis for microdomain formation in cells remains uncertain, and this is thought to be due to the small domain size and the highly dynamic nature of the cell membrane. The application of high pressure technology offers an ideal biophysical tool for the study of phase behaviour in model membranes both in and out of equilibrium. By coupling high-pressure technology with fluorescence microscopy we have been able to simultaneously induce and visualize phase separation in GUVs. This allows the structural dynamics (including domain size and morphology in individual vesicles) to be studied, which ideally compliments small angle x-ray scattering (SAXS) measurements of bulk mesophase properties. We employ high pressure technology to induce thickness mismatch and therefore alter the line tension between coexisting liquid domains, and to study the pressure effects on the lateral structuring of membranes containing general anaesthetics. The ability to trigger rapid phase separation using pressure-jumps across the phase boundary has been used to study the dynamic evolution of structural changes, with time-resolved microscopy and SAXS giving an insight into transition kinetics, energetics and mechanisms.

Systems biology takes a mechanistic, relational approach to the study of biological processes, commonly finding expression in mathematical models. Hypotheses about systems can be tested when formulated as models, and promising avenues for further study identified. A model sufficiently faithful to the system under study can be used to guide experiments, to probe the system in silico, and to learn about emergent features not evident from the static picture of the system. In this work, three contributions to the modelling community are proffered. First, a computational package is presented that implements an algorithm for the validation and parametrisation of a model. In validation, we are asking how likely we were to make some observation, given the model, or, equivalently, how able the model is to explain the data. The subsequent two contributions concern noise in biological systems. Biological systems display inherent variability, or noise, due to the stochastic mechanisms through which biochemical processes occur. This variability can be critical to the behaviour of a system and to the fates of individual cells. With this in mind, the second contribution is the development of a methodology to model protein-dependent population dynamics. The idea is to model cell population dynamics that result of noisy intracellular protein dynamics. The method's application is demonstrated in population-level models of a protein-dependent cell cycle and yeast antibiotic resistance. Given an appreciation of the pivotal effects of noise, the third and final contribution is a study of the mechanism of noise propagation. I present an analysis of the contributions of biochemical reaction motifs to the creation and transmission of noise that ultimately manifest in observations of biological systems. This study points to specific processes that enhance or attenuate noise, with the aim of beginning to unravel the flow of noise through a system.

The 3D-packaging of β-cells into spheroid micro-organs termed islets of Langerhans leads to a gain-of- function due to heightened intercellular communications. Dissociated β-cells fail to respond coordinately to input, secrete less insulin relative to cell mass and display elevated basal hormone levels suggesting connectivity is imperative for insulin secretion. β-cells are a heterogeneous population in regards to both metabolism and insulin secretion. Recent studies performed by this group and others, have revealed functional differences between hundreds of individual β-cells monitored in situ in the intact islet. A role for specialized β-cells in orchestrating these dynamics has long been postulated, including the presence of putative 'pacemakers'. However, whether particular subsets of functionally different cells drive the behavior of others has so far been difficult to prove empirically. Over the past decade, optogenetics has allowed reversible control of neuronal activity with light. In parallel, photopharmacology has enabled the development of synthetic photoswitches including JB253, a photoswitchable sulfonylurea which has been shown to optically control β-cell activity in vitro. Using this all-optical interrogation of individual β-cells in situ, alongside functional connectivity analysis this study set out to probe the topology that regulates population glucose responsiveness and understand islet heterogeneity in the context of insulin release. Mapping of islet functional architecture revealed the presence of highly-connected hub cells with pacemaker properties. Silencing of hubs via optically-controlled hyperpolarisation, abolished coordinated islet responses to glucose, whereas specific stimulation of hubs using JB253 restored communication patterns. Hubs were metabolically-adapted and their population size was reduced by both pro-inflammatory and glucolipotoxic insults to induce widespread β-cell dysfunction. Therefore, it appears the β-cell population within the pancreatic islet is wired by hubs, whose failure may contribute to T2D. Application of the photoswitchable insulin secretagogue, JB253, in vivo suggests that glycaemia can be optically controlled in vivo.

The 5’-phosphatases are a family of metalloenzymes involved in the regulation of phosphoinositide signalling in cells and also implicated in diseases such as cancer, diabetes and neurodegenerative disorders. Even so,only a handful of modulators have been reported to target these enzymes. In this thesis,efforts towards the design of zinc(II)complexes for 5’-phosphatase inhibition are described with further analysis of the interaction between these enzymes and zinc(II). The validation of previous reports that zinc(II) can inhibit the 5’-phosphatases is described and, alongside published evidence that calcium(II) displaced the magnesium(II) co-factor from the 5’-phosphatase active site, was the basis for design and synthesis of zinc(II) complexes. Synaptojanin 1 was chosen as the representative of the 5’-phosphatases for this study due in part to its importance in neurodegenerative disease and the paucity of reported inhibitors for this enzyme. Iterations of design (with the aid of computational docking), synthesis and testing in vitro with a series based on a zinc(II) cyclen motif yielded poorer than anticipated inhibition of synaptojanin 1. Investigation into the mechanism of zinc(II) mediated inhibition of the 5’-phosphatases using a combination of biochemical, biophysical and computational techniques revealed that, rather than displacing magnesium(II), zinc(II) more likely bound a second metal binding region within the 5’-phosphatase active site. This proposed zinc(II) binding site incorporated residues recently postulated to be important for a 5’-phosphatase mechanism of action involving a “moving” magnesium(II) co-factor. As zinc(II) was inhibitory, the results described in this thesis support this mechanism. Furthermore, the weak inhibitory activity of zinc(II) cyclen towards synaptojanin 1 was found to be via binding to the phosphoinositide substrate I(1,4,5)P3. Considering the tractable synthesis and good aqueous solubility of substituted zinc(II) cyclens reported herein, these motifs might hold promise towards modulation of phosphoinositide signalling by direct binding to phosphoinositides.

Chemotaxis allows flagellated bacteria to navigate in complex chemical environments, following nutrients and escaping toxins. The sensory system made up of chemoreceptors is constantly monitoring the extracellular concentrations of nutrients and toxins, while the signalling pathway processes and transmits the external information to the flagellated motors for movement. In the case of Escherichia coli, the chemotaxis pathway has been extensively characterised experimentally using genetics, biochemistry, and a wide range of imaging tools. This makes E. coli an ideal model organism for quantitative analysis and modelling. Several remarkable properties of the E. coli chemotaxis pathway have been summarised in terms of design principles. However, the swimming behaviour remains poorly understood, even for genetically identical cells in the artificial conditions normally used in a laboratory. Here, I propose an interdisciplinary approach, which combines theory, computational simulations, and experimental data from my collaborators, to study E. coli chemotaxis from an information-theoretic point of view. I demonstrate that the E. coli chemotaxis pathway is designed to optimally transmit environmental information over a certain range of concentrations and gradients. To do so, I develop a theory that identifies both the responses and the environmental conditions that transmit maximal environmental in- formation. Interestingly, when maximal information is transmitted, the behaviour characterised in terms of the drift velocity towards the nutrient is also maximised. A new design principle is proposed: maximal information transmission leads to maximal drift. Furthermore, the energetic cost of chemotaxis is much lower than the energy consumed to maintain the biological signalling pathway. Hence, thermodynamics does not seem to set constraints on information transmission and drift. However, to fully capitalise on my results, a closer connection with single-cell experiments is suggested.

von Willebrand factor (VWF) is a multidomain glycoprotein that has well-established functions in haemostasis and thrombosis. The lack of a constraining disulphide bond results in unfolding of its VWF A2 domain under rheological forces in the vasculature. This unfolding of the VWF A2 domain is an important aspect of VWF function, as it facilitates the proteolysis and regulation by the plasma metalloprotease ADAMTS13. In this thesis, I investigate the structural elements of the VWF A2 domain that control its transition from folded to unfolded state. Dye binding measurements made during the temperature-induced unfolding of a recombinant VWF A2 domain fragment revealed that its calcium binding site (CBS) and vicinal disulphide bond (VicCC) both stabilise the VWF A2 domain and do so in a cooperative manner. Disruption of the CBS and VicCC in full-length recombinant VWF (FLVWF) results in its spontaneous cleavage by ADAMTS13. Temperature-induced unfolding of the VWF A2 domain was also used to examine the role of N-glycosylation on VWF A2 domain stability. Production and analysis of truncated glycan variants of the VWF A2 domain indicated that the N1574 glycan most likely stabilises the VWF A2 domain through an interaction of its first sugar moiety (GlcNAc) with VWF A2 residue Y1544. Stabilisation provided by N1574-GlcNAc impedes interaction with ADAMTS13. The triad of structural features of the VWF A2 domain, the CBS, VicCC and N1574-glycan, therefore collectively provides folding stability. I have also begun an investigation into the consequence of genetic mutation of residues within the VWF A2 domain. I introduced 34 mutations (that naturally occur and are described within the von Willebrand Disease, VWD, mutation database) into FLVWF, the isolated VWF A2 domain, an A1A2A3 domain fragment and a newly developed VWF A2 FRET construct. Preliminary results suggest that the latter construct may be informative in understanding the molecular basis of a subtype of VWD. Collectively, my findings help to clarify how the functions of VWF are controlled by conformational changes in the VWF A2 domain.

As the physical boundary separating the cell interior from the outside world, the plasma membrane is arguably one of the most important cellular organelles. Along with the proteins embedded within it, the plasma membrane plays a part in a number of vital roles including intercellular communication, cell motility and homeostasis. Despite this, comparatively little is known about the exact structure of the plasma membrane. The plasma membrane is a lipid bilayer that is widely believed to undergo transient phase separation into domains of different viscosities, which may be responsible for certain membrane properties. By using molecular rotors (organic fluorophores whose fluorescence properties depend on the viscosity of their surrounding environment) in conjunction with fluorescence lifetime imaging microscopy (FLIM), it is possible to generate a viscosity map of a heterogeneous biological system. In this thesis, we use molecular rotors to investigate the viscoelastic properties and phase behaviour of a range of lipid bilayer systems. We begin by using molecular rotors based on a boron-dipyrrin (BODIPY) core to investigate membrane viscosity within model systems, finding that different lipid phases display different viscosities, and visualising lipid phase separation within giant vesicles. We go on to use BODIPY rotors to image viscosity within the plasma membranes of live Escherichia coli bacteria at different temperatures, finding that they display a much higher degree of membrane ordering than previously observed in eukaryotic cells. Next we explore synthetic alternatives to the conventionally used BODIPY rotors in order to address their shortcomings, before using molecular rotors based on a thiophene moiety to determine the viscosity of the plasma membranes of live human cell lines. Our results from these studies suggest that phase-separation may indeed occur within the plasma membrane.

Bacterial glycoconjugates are a group of highly complex macromolecules that are deeply implicated in various biological processes. ey can be crudely divided into several subgroups mainly including bacterial glycolipids and glycoproteins. eir structural complexity once hindered the elucidation of the biological roles they are playing. Based on the development of modern chromatography, mass spectrometry and nuclear magnetic resonance, research described in this thesis analysed the structures of multiple glycoconjugates such as glycolipids and glycoproteins. Lipopolysaccharides are characteristic bacterial glycolipids. e Pseudomonas aeruginosa lipid A and Helicobacter pylori lipopolysaccharide was investigated by a combination of chemical, MS and/or NMR strategies. For the lipid A, it was proved to be modi ed by 4-amino- arabinose, which is associated with bacterial antibiotic resistance. For the lipopolysaccharide, the speci cities of mutiple putative glycosyltransferases, including HP1284, HP1283, HP1578, HP0102 and WaaL ligase, involved in its biosynthesis were characterised. e structural data suggest Helicobacter pylori WaaL ligase has a relaxed speci city in vivo that transfers an un- expected long O-antigen to a short core-oligosaccharide. In addition, our data shed light on the regulation of bacterial O-antigen biosynthesis. e heavily glycosylated mbrial-associated protein 1 of Streptococcus parasanguinis was subjected to in-depth glycomics and glycoproteomics, which helps determine the speci cities of multiple putative glycosyltransferases such as GALT1/2 and GLY. More speci cally, GALT1 was proved to be a bi-functional glycosyltransferase that transfers a glucose and an N-acetyl- glucosamine. GALT2 and GLY was deteremined to transfer a glucose and a rhamnose, respectively. Based on these conclusions, our study helps elucidate the glycosylation pathway of mbrial-associated protein 1. e recent characterisation of bacterial oligosaccharyltransferases has the potential to revolutionise the production of glycoconjugate vaccines. e putative O-linked oligosaccharyltransferases system in Burkholderia thailandensis was investigated based on a zwitterionic hydrophilic interaction liquid chromatography mass spectrometry method, which helps expand the toolbox of glyco-engineering. By using a simpilied liquid chromatography- mass spectrometry method, multiple glycoproteins in Coxiella burnetii were semi-identi ed. During these studies, many classical chemistries were resurrected to inspire novel chemical strategies and facilitate the instrumental analysis. We hope the methods included in this thesis can contribute to further research where comprehensive structural characterisations of complex bacterial glycoconjugates are required.

Imaging of protein tyrosine kinase activity requires negligibly intrusive, molecularly precise optical probes to provide spatiotemporal insights on conformational dynamics appertained to receptor dimerization and activation, which govern the engagement of downstream signalling networks in cancer and other pathologies. We present two sets of constructs with genetically encoded, site-specifically incorporated, bioorthogonal reporters that can be selectively labelled with exogenous fluorogenic probes to monitor the structure and function of an oncogenic tyrosine kinase, fibroblast growth factor receptor 1 (FGFR1). In this study, the conformational changes in the kinase domain of FGFR1 (FGFR1K) were monitored by imaging technologies that afford analysis at great spatial and temporal resolution. First, a novel set of recombinant constructs was engineered allowing the incorporation of BCNK, a bicyclononyne-containing, lysine-like unnatural amino acid (UAA) into the activation loop or the kinase insert region of FGFR1K. BCNK incorporation was accomplished by genetic code expansion, in which an evolved tRNA/synthetase pair from Methanosarcina species directed the site-specific installation of the UAA in response to an amber codon by orthogonal translation. We report here for the first time the efficient labelling of FGFR1K via fluorogenic bioorthogonal Diels-Alder cycloadditions using a tetramethylrhodamine dienophile, Tet1-TAMRA-X. We developed a single molecule Förster resonance energy transfer (smFRET) assay that uses an alternating laser excitation (ALEX) scheme in association with a surface immobilization strategy to examine changes in the conformational states of FGFR1K. FRET efficiency distributions and interdye distances provide evidence of large-amplitude domain motions in the tyrosine kinase domain and can be correlated to autophosphorylation-dependent changes in conformational states of FGFR1. The second construct, GyrB.FGFR1K3Y.TC, comprises a coumermycin-induced artificial dimerizer (GyrB), FGFR1K exhibiting wild-type like intrinsic kinase activity and a TC motif that enables fluorescent labelling with the biarsenicals FlAsH-EDT2 and REAsH-EDT2. We conceived bimolecular systems for in vitro studies by implementing time-resolved fluorescence lifetime imaging technologies. Our approach allowed examination of the interaction between GyrB.FGFR1K3Y.TC and N-terminal Src homology 2 (nSH2) domain of phospholipase C-γ (PLCγ), a downstream effector of FGFR1, fused to mTurquoise Fluorescent Protein (mTFP). Furthermore, we were able to probe FGFR1 dimerization by artificially inducing dimerization of GyrB using a chemical agent. We report here phosphorylation-dependent FRET readout of complex formation between mTFP.nSH2 and GyrB.FGFR1K3Y.TC as well as artificial dimerization of GyrB.FGFR1K3Y.TC.