Search

Most adaptive immune lymphocytes are in the lymphoid tissues. However, studies of human lymphocytes to date have primarily investigated cells from the blood, and lymphoid tissues remain poorly studied. This thesis explores the hypothesis that there is local variation of B cell subsets between the tissues that are micro-anatomically different and receive antigenic stimulation via different routes. Matched samples of appendix, mesenteric lymph nodes (mLNs), and spleen from human cadaver were analysed by mass cytometry (cytometry by time of flight [CyTOF]) for deep phenotypic profiling and tissue-specific subset variation. Populations of B and T cells that had significant abundance differences between the three tissues were identified in an unsupervised manner. The subsets that varied significantly in abundance between tissues included: germinal centre (GC) B cells, marginal zone (MZ) B cells, IgM-only B cells, IgA+ B cells, follicular helper T cells and regulatory T cells, as well as CD4+ and CD8+ effector memory cells. These differentially abundant subsets were further characterised by their different marker expression profiles between tissues, and results were confirmed using a second supervised analysis method. Most notably from the unsupervised analysis and subsequently the supervised analysis, MZ B cells were further divided into two populations based on their tissue-specific abundance and distinct phenotype including differences in CCR7 and BAFF-R expression. IgM-only B cells also included two populations with different phenotype and tissue-specific abundances. Single-cell RNA sequencing was used to further investigate the differentiation, functional, and dissemination differences of these subsets between lymphoid tissues. The two MZ B cell populations were transcriptomically distinct and had significant tissue-specific differential abundances. Within each tissue, B cell clones from the CCR7+ MZ B cell subset were significantly associated with IgM-only subset whereas the other CCR7- MZ B cell subset had a greater association with activated naive and double negative B cells. Between the paired tissues, members of the same B cell clone were observed in multiple tissues. Clones of CCR7+ MZ B cells and IgM-only B cells were more disseminated between tissues than other subsets including the CCR7- MZ B cells. On the other hand, transitional and naive B cells only had significant correlation within the same tissue and not between tissues implying local maturation in tissues. Overall work in this thesis identifies regional variation in B cell subset abundance, tissue based local B cell differentiation and also clonal links between B cell subsets in distant human tissues sites.

Solid organ transplantation is the standard treatment for end-stage organ disease. However, its impact is short-lived due to allograft recognition. Adoptive transfer of regulatory T-cells (Tregs) protects from allograft rejection, and the safety has been demonstrated in clinical trials. Although antigen-specific Tregs are superior to polyclonal Tregs, they are still MHC-restricted. As well as antigen specificity, genetic engineering offers a solution to the unmet need of improving Treg function for clinical therapy as CARs are customisable building blocks. Tregs expressing chimeric antigen receptors (CAR) confer antigen-specificity in a non-MHC restricted manner. Thus, two major approaches have been explored in this thesis. The first approach was applying fourth-generation CAR technology, i.e. those that release an additional cytokine into the respective microenvironment. Therefore, this thesis focuses on design, construction, and investigation of an IL-10-releasing HLA-A2- targeting CAR Treg in protecting an allograft. Additionally, the ability to track Tregs in vivo such as inclusion of the sodium-iodide symporter (NIS) within the constructs, would inform us of Treg biodistribution and longevity, aiding therapy development and personalised monitoring. Herein, the IL-10 HLA-A2 NIS TagRFP CAR was designed and constructed. Tregs were transduced and their potency was compared to control Tregs in vitro. The IL-10 A2 CAR Treg was shown to be superior to the HLA-A2 CAR Treg. A skin transplant model using skin from an HLA-A2 expressing transgenic mouse was performed. The IL-10 CAR Treg functions potently in vitro, however the skin transplantation and imaging model need further optimisation to reliably assess CAR Treg efficacy and detection. The second approach was to comprehensively analyse a panel of different costimulatory molecules in second-generation CARs to identify one that may be optimal for Treg function. Herein, I performed a pilot study comparing five endodomains to the CD3ζ-CD28 molecule - CD27, OX40, 4-1BB, ICOS, DR3. There are no significant differences in function or phenotype, although incorporating a DR3 endodomain preliminarily seems to suggest a weaker suppressive capacity. To conclude, this is the first attempt in creating a fourth-generation CAR Treg, and an independent validation of CAR Tregs with different signalling domains. This study introduces a broad range of possibilities in manipulating Tregs, with clinical implications for transplantation and autoimmunity.

Low density granulocytes (LDGs) are a poorly understood class of immune cells found in patients with chronic inflammatory diseases including psoriasis and systemic lupus erythematosus (SLE). Research completed at the National Institutes of Health (NIH) revealed that in the context of SLE, LDGs release higher levels of type 1 interferons, undergo increased NETosis, and accordingly drive inflammation. Meanwhile, advances in mechanical phenotyping at the University of Cambridge have driven hypotheses of neutrophil trafficking and immune function being intimately linked to cellular biomechanical properties (e.g. density, stiffness, morphology). This thesis analyses the intersection of immune cellular biomechanical phenotypes and their function. Specifically, it focuses on the role of neutrophils and LDGs in inflammatory diseases. In this thesis, real-time deformability cytometry (RT-DC) was optimised as a high-throughput mechanical phenotyping technique for the analysis of neutrophils. This enabled development of a protocol to recover purified neutrophils to their whole blood mechanical phenotype. Neutrophil biomechanical properties were analysed by RT-DC, lattice light-sheet microscopy, confocal microscopy and scanning electron microscopy. Neutrophil immunologic functions (e.g. NETosis, macropinocytosis) were imaged using florescence microscopy. To analyse the contribution of biomechanical properties to neutrophil trafficking, a novel microfluidic microvasculature mimetic was developed. An endothelial flow assay was used to image neutrophils interacting with endothelial cells. Finally, the complete proteomes and phosphoproteomes of LDGs and normal dense neutrophils (NDNs) were obtained from five healthy donors and five SLE patients. Several key insights were gained. Firstly, hypotonic lysis and magnetic column-based isolation techniques are damaging to neutrophil biomechanical properties, but purification of neutrophils retaining their biomechanical properties can be achieved by using gradients and column-free magnetic systems followed by recovery at 37 degrees Celsius. Secondly, the biphasic biomechanical kinetics of neutrophil priming were described; cells contract briefly before immediately expanding. The expansion phase was determined to be macropinocytosis dependent. Thirdly, SLE LDGs are phenotypically rougher than autologous SLE NDNs or healthy LDGs. This appears to impact their microvasculature trafficking abilities, as SLE LDGs were increasingly trapped in the narrow channels of a three- dimensional microvasculature mimetic. These results suggest a role for biomechanical properties in modulation of neutrophil trafficking, indicating that SLE LDGs may be increasingly retained in microvasculature networks, similar to what has been described for primed neutrophils. Finally, unbiased proteomics quantified 4109 proteins and 875 phosphoproteins in four neutrophil subsets (healthy unstimulated NDNs, healthy primed NDNs, SLE NDNs, and SLE LDG). This shed new light into neutrophil heterogeneity at the protein level and to my knowledge, is the first proteomic profile of the SLE LDG. In addition to findings pertaining to SLE LDG biology and function, differential phosphorylation of proteins associated with cytoskeletal organisation were identified in SLE LDGs relative to SLE NDNs, suggesting a phosphoproteomic explanation for the SLE LDGs' distinct biomechanical phenotype. When taken together, this work could have important pathogenic implications in the context of SLE manifestations in various organs and the development of small vessel vasculopathy.

Introduction. Macrophages play a key role in atherosclerotic plaque instability, they engender plaque rupture, leading to thrombosis and myocardial infarction. Their detection in coronary artery plaques would enable stratification of patients for better treatment of "vulnerable" phenotypes. Intra-coronary Optical Coherence Tomography (OCT) enables visualisation of vulnerable plaque morphology and has the potential to detect the signal variance created by increased numbers of macrophages. Aims. Our aim was to enhance the detection of macrophages using OCT in combination with intracellular contrast agents; and to demonstrate it in-vitro, ex-vivo and in-vivo. This was first endeavoured using clinically approved ultra-small superparamagnetic particles of iron oxide (USPIOs): an MRI contrast agent which is readily uptaken by macrophages in coronary plaques. The second approach was to design a novel OCT contrast agent based on different lanthanide ions with either a carbonate or fluoride structure. Methods. A "phantom" artery model was designed and 3D printed for standardised in-vitro imaging with intra-coronary OCT. Particle-laden macrophages were tested in this system to determine OCT contrast enhancement. Cell viability assays were performed to determine toxicity of novel lanthanide particles. Aortic plaques of Apo-lipoprotein E knockout mice, treated with USPIO or saline were OCT imaged and sectioned for histological analysis. Patients presenting with acute coronary artery disease were randomised to receive USPIO or saline treatment, followed by OCT imaging of the culprit plaque and non-culprit control. The local pixel Standard Deviation (SD) of OCT images were processed in Matlab as a metric of contrast enhancement. A local SD hotspot visualisation tool was developed in Matlab for use with the phantom, murine and human OCT images. Results. In all 7 phantoms, USPIO-laden macrophages significantly increased the mean SD compared to untreated macrophages (p<0.001). USPIO treated mouse plaques (n=3) showed significantly higher mean SD than control plaques (n=3) (p<0.05). USPIO treated patients (n=8) demonstrated a significantly increased mean SD compared to control patients (n=10) in culprit and non-culprit plaques (p<0.05). Within patients the mean SD was significantly higher for culprit compared to non-culprit plaques in both USPIO and control groups. The initial OCT screening of 10 novel lanthanide particles revealed the carbonate structures to be most efficacious for contrast enhancement compared to the fluoride ones. Phantoms made with particle-laden macrophages revealed the Gadolinium carbonate particle with Polyacrylic acid coating to have the highest mean SD increase compared to untreated macrophages. This particle also showed low levels of cell toxicity. Conclusions. USPIO administration was associated with a significant increase in OCT contrast in a novel in-vitro phantom model, an ex-vivo atherosclerotic mouse study and in a human clinical trial. USPIO enhances the detection of macrophages and may facilitate patient stratification. Additionally, novel OCT contrast agents have been made which show great potential for further enhancing macrophage detection in this same context.

It is well established that the brain pulsates with each cardiac cycle, yet relatively little is known about how these brain tissue pulsations (BTPs) are influenced by different physiological factors. Previous research suggests that measurement of BTPs could be used as a potential marker for brain pathology and impaired cerebral haemodynamics, however, existing methods used for BTP measurement are often costly, require an expert user, and are not suitable for continuous neuromonitoring. The aim of this thesis is to use transcranial tissue Doppler ultrasound to investigate factors that affect healthy brain tissue pulsations, and to aid the development of a prototype device developed at University of Leicester, in collaboration with Nihon Kohden (Japan), for clinical measurement of BTPs. An ultrasound phantom is used to investigate the impact of blood pressure on BTPs, along with a systemic state-space model of intracranial pressure dynamics. Numerous healthy volunteer studies are carried out to characterise healthy BTPs, and determine which factors influence BTP amplitude. The system used was capable of obtaining quality data from numerous positions on the head, including the forehead. BTP signals were successfully detected in all volunteers studied, and the system was well tolerated. BTP amplitude was found to vary greatly between individuals, and with probe position, and increases with depth into the brain. A strong association between BTP amplitude and pulse pressure was observed in all studies. This thesis introduces a new technique for measuring BTPs which has the potential to be used clinically to give real-time information on brain motion at the patient's bedside. BTP amplitude is strongly influenced by pulse pressure, which is a variable that can be greatly affected in critically unwell patients. Reference data from healthy subjects has also been generated, which can be compared to pathological cases in future work.

Introduction and aims: The most commonly used approaches for the analysis of time-to-event (TTE) outcomes impose an assumption of proportional hazards (PH), such that the hazard ratio (HR) is assumed to be constant over time. Meta-analysis of TTE data is most commonly based on extracting or estimating the HR from individual trials, and so again assumes PH. Methods are available for assessing the validity of the PH assumption, however, the assumption is not always checked or reported for validity. This is a problem for meta-analysis, where different assumptions may have been made in the analysis of each included study. The aim of this thesis is to investigate how often the PH assumption is assessed within Randomised Controlled Trials (RCTs) and meta-analysis, including understanding the impact of non-PH on meta-analyses. This is of particular importance as current research has focused on alternative methodology, without knowing what impact non-PH may have on results. Methods: The thesis summarises the results from a novel systematic review (SR) of the reporting of meta-analysis of TTE outcomes that have assumed PH, and how often the results of the PH assumption were reported. Two further SRs of PH assumption reporting within RCTs and Single Technology Assessments were also performed. A survey was also conducted targeted at the UKCRC network of registered clinical trials units, to understand what is done routinely rather than what is reported within RCTs. A simulation study was undertaken to assess the suitability of different modelling approaches for meta-analysis of TTE data in situations where PH is valid and invalid. Results: All of the reviews on reporting of the PH assumption within SRs and RCTs, highlighted the poor reporting of the PH assumption. Only 33 out of 123 (27%) SRs and only 12 out of 106 (11%) RCTs reported the PH assumption. For the simulation study, meta-analytic datasets were simulated for twelve scenarios. Across scenarios, parameters controlling the Weibull distribution, the censoring level (25% and 75%), the time-dependent log HR (None, 0.1 and 0.5) and whether the treatment effect across studies is homogeneous or heterogenous were varied. The simulated datasets were analysed using Cox, Weibull, Accelerated Failure Time, and Flexible Parametric models. In situations where PH is valid, all models performed well as expected. However, as soon as the time-dependent log HR of 0.1 was introduced, the Cox and Weibull model could not cope. The best performing model in all cases was the Flexible Parametric Model. Conclusions: The work of this thesis has provided a detailed insight into the poor reporting of the methods used to assess the PH assumption as well as the lack of reporting of the results of assumption checking. The work of this thesis highlighted the lack of reporting guidelines as there is no mention of the PH assumption in the CONSORT or PRISMA guidelines, Cochrane handbook or the ICH E9 guidelines. Recommendations that can be used by trialists, reviewers and 'consumers' of reviews on how to approach the PH assumption have been provided in the thesis.

The computational approach to image acquisition and analysis plays an important role in medical imaging and optical coherence tomography (OCT). This thesis is dedicated to the development and evaluation of algorithmic solutions for better image acquisition and analysis with a focus on OCT retinal imaging. For image acquisition, we first developed, implemented, and systematically evaluated a compressive sensing approach for image/signal acquisition for single-pixel camera architectures and an OCT system. Our evaluation outcome provides a detailed insight into implementing compressive data acquisition of those imaging systems. We further proposed a convolutional neural network model, LSHR-Net, as the first deep-learning imaging solution for the single-pixel camera. This method can achieve better accuracy, hardware-efficient image acquisition and reconstruction than the conventional compressive sensing algorithm. Three image analysis methods were proposed to achieve retinal OCT image analysis with high accuracy and robustness. We first proposed a framework for healthy retinal layer segmentation. Our framework consists of several image processing algorithms specifically aimed at segmenting a total of 12 thin retinal cell layers, outperforming other segmentation methods. Furthermore, we proposed two deep-learning-based models to segment retinal oedema lesions in OCT images, with particular attention on processing small-scale datasets. The first model leverages transfer learning to implement oedema segmentation and achieves better accuracy than comparable methods. Based on the meta-learning concept, a second model was designed to be a solution for general medical image segmentation. The results of this work indicate that our model can be applied to retinal OCT images and other small-scale medical image data, such as skin cancer, demonstrated in this thesis.

Self-monitoring of blood glucose plays a crucial role in diabetes care. Current methods apply invasive means, which causes pain and discomfort to the people with diabetes. Therefore, non-invasive methods are in high demand. Reverse Iontophoresis (RI) is a technique for non-invasive transdermal glucose monitoring. Thereby, electric current is applied across the skin to extract glucose from the interstitial fluid into a gelreservoir, where it is measured with an in situ glucose sensor. One of the key problems in RI based wearable sensors, is the glucose build-up in the gel-reservoir, which results in poor correlation between the extracted glucose and blood glucose level. Activated Carbon (AC) has excellent adsorption capabilities to organic molecules and could therefore act as a glucose binding agent in the RI devices leading to more accurate glucose readings and ultimately better correlation between the extracted and blood glucose levels. The overall aim was to study and compare glucose fluxes in in vitro RI experiments using the developed AC-integrated electrodes to the standard electrodes without AC. The extracted glucose was quantified using glucose assay kits. Furthermore, this study involved embedding a biosensor in the electrode system to allow direct glucose measurements. In vitro RI experiments using the newly developed AC-integrated electrodes showed increased glucose fluxes of 61% in the high glucose concentration level in the diffusion cell. Besides, the extracted glucose correlated well with increasing glucose concentrations in the diffusion cell (R2 =0.90). Thus, confirming AC's suitability as a glucose binding agent in the RI environment. In vitro RI experiments with an integrated mediated enzymatic glucose sensor in the electrode system showed limited success for electrochemical glucose quantification of the extracted glucose. However, a thorough electrochemical characterisation of the developed glucose sensor using cyclic voltammetry was performed, which may provide good basis to make improvements for future glucose sensors.

In a world of emerging technologies and medications, sensing and detection is of paramount importance for the accurate diagnosis of illness and disease. This project explores the development of a point-of-care biosensor for the detection of both cortisol and human growth hormone, basing their capture and detection on two main antibodies. There is currently no single sensor that could simultaneously detect both cortisol and growth hormone, two essential hormones that are produced by the hypothalamic-pituitary-adrenal axis. However, there is a market for such a device, which would prove useful in, not only the non-invasive diagnosis of diseases such as Addison's disease, Cushing's syndrome and dwarfism, but also in identifying tumours in the hypothalamus, pituitary and adrenal glands. The hypothalamus-pituitary-adrenal axis is complex and dependent on many factors however, generally, increased cortisol production is the hypothalamus-pituitary-adrenal axis's physiological response to an external stressor. Growth hormone is also regulated through the hypothalamus-pituitary-adrenal axis and secreted from the anterior pituitary gland. Hence a multiplex point-of-care sensing device for the quantification of both cortisol growth hormone would be multifaceted for the unintrusive diagnosis of hypothalamus-pituitary-adrenal axis activity and the practical uses of stress indicator would be diverse and include the medical and pharmaceutical sector as well as sport and performance. A range of sensing is explored including enzyme-linked immunosorbent assay, cyclic voltammetry and impedance methods. It is intended that the final biosensor is based on electrochemical impedance spectroscopy due to the high sensitivity, accuracy and versatility associated with impedance. The final ensemble was based on inexpensive screen-printed electrodes and was able to detect cortisol between 30-300 ng/mL and recombinant growth hormone for the range of 5-30 ng/mL.

Chimeric Antigen Receptor (CAR) T cell therapy has emerged as a treatment for haematological malignancies with currently two products approved by the FDA in the USA and EMA in Europe (Kymriah™ and Yescarta™). Recent clinical studies demonstrated that the presence of less differentiated and long-lasting subsets such, as T Central Memory (TCM) cells in the CAR-T products is required to improve their clinical efficacy. Most manufacturing protocols use media containing high glucose concentration when expanding CAR-T cells ex vivo. Effector subsets primarily utilise glucose and glycolysis as a primary metabolic pathway, therefore, the hypothesis was that culturing T cells in a glucose deprived environment would limit T cell differentiation and effector subsets growth, producing TCM enriched products. It was demonstrated that a 2-stage feeding strategy, where CAR-T cells are activated and transduced in the presence of glucose and fed with a glucose-free medium during the expansion phase, would consistently produce a TCM enriched CAR-T product. Improved in vitro functionality and proliferation capability was demonstrated using TCM enriched CAR-T therapy compared to the CD19-specific CAR-T cells generated using a standard protocol where a high glucose medium was used throughout the ex vivo expansion. The impact of different bioreactor operating conditions on T cells growth and phenotypic composition was further investigated. Two different levels of Dissolved Oxygen (dO2; 25% and 90%), pH (6.9 and 7.4), and shaking speeds (100 rpm and 200 rpm), as well as the interaction between them were assessed. The results demonstrated that the optimal culture condition for generating a high number of CD8+ TCM cells is a combination of 200 rpm, 25% dO2, and pH of 7.4. This thesis highlights the importance of investigating the impact of metabolites in the medium and bioreactor parameters during the T cell therapies manufacturing in order to improve the quality and quantity of the final product.

CAR-T immunotherapies present a novel therapeutic modality for the treatment of various blood tumours. However, the development of such immunotherapies requires the manufacture of large numbers of CAR-T cells (2-6 x 108 total viable CAR-T cells per dose), which remains a major translational and commercial bottleneck due to the manual, small-scale, and often static culture systems used for their production. Such systems are easy to use in a pre-clinical research settings, but are not efficient when a higher number of doses need to be produced. Furthermore, there is a general concern that primary T-cells are shear sensitive and do not grow in agitated systems, such as stirred-tank bioreactors. This doctoral thesis aims to demonstrate that primary human T-cells and CAR-T cells can be cultivated in stirred-tank bioreactors at different scales (15 ml, 250 ml, and 1 L), which can be used for both autologous and allogeneic products. Furthermore, data in this thesis shows that the growth of T-cells and transduced CAR-T cells was significantly better in stirred-tank bioreactors than in T-flask static culture. At agitation speeds of 200 rpm and greater (up to 500 rpm) in the ambr 250 stirred-tank bioreactor, the CAR-T cells were able to proliferate effectively, reaching viable cell densities of > 5 x106 cells ml-1 over 7 days. This is comparable with current expansion systems and significantly better than static expansion platforms (T-flasks and gas-permeable culture bags). Importantly, the cell quality and potency was assessed at the end of the expansion and was equivalent to the one presented by the cells grown under static conditions as control. It was demonstrated that higher agitation rates, corresponding to higher power inputs lead to a better proliferation. This improvement is likely due to the inability at the lower agitation rates to effectively suspend the Dynabeads used to activate T-cells. Importantly, from the data obtained, there is no indication that T-cells prefer being grown under static conditions or are sensitive to fluid dynamic stresses within a stirred-tank bioreactor system at the agitation speeds investigated.

The importance of mechanical forces within the immune system has long been recognised through their roles in cell motility, cell adhesion and in the engulfment of pathogens. Additionally, mechanical forces have been found to regulate lymphocyte activation. In the case of B lymphocytes, mechanical forces promote extraction of cell-presented antigens in cell-cell contacts called immune synapses. It is hypothesised that the mechanical forces during antigen extraction promote affinity discrimination. This is required for B lymphocytes to produce high affinity antibodies and thus efficiently eradicate pathogens. This process is particularly important during presentation of antigens bound in immune complexes, which is mediated by complement receptor 2 (CR2) on follicular dendritic cells (FDCs). However, the mechanics of immune complex presentation and extraction remain poorly characterised. This thesis focuses on three main goals. Firstly, this work aims to develop an approach for measuring receptor ligand interactions under applied forces using a novel adaptation of the single molecule technique of magnetic tweezers (MT). Secondly, this thesis aims to apply the newly developed approach to study the mechanics of the interactions of CR2 with its ligand, C3dg, and with an anti-CR2 monoclonal antibody. These interactions will act as models of B cell pulling of immune complexes from FDCs. Finally, this in vitro approach is complemented by studying cellular expression of CR2 and measuring its interactions with C3dg at the surface of live cells. Through this work, I have developed a comprehensive, MT-based workflow for measuring lifetimes of single molecule interactions under externally applied forces. Although this method proved inefficient to study the fast force-mediated dissociation of single C3dg molecules from CR2, through studies of CR2 interactions with a monoclonal antibody, I have demonstrated the general applicability of this method for protein-protein interactions. I have also extended this method to use with live cells, where I found an enhanced binding of C3dg by a longer variant of CR2, CR2L, expressed on FDCs. This is the first functional difference of CR2L discovered to date and suggests that improved binding of immune complexes by FDCs via expression of CR2L may enhance B cell and antibody responses.

N-methoxybenzylphenethylamines (NBOMe) were first synthesised and published in Ralf Heim's thesis in 2003,1 detailing the selectivity of these compounds towards serotonin receptors (5-HT2A). This thesis details the full synthesis of seven phenethylamines (10 - 93 % yield as their hydrochlorides) and their transformation into the corresponding NBOMe's (27 - 74 % yield as their hydrochlorides) with full characterisation to compare against literature. The synthesis of 21 novel derivatives (containing either a 3-pyridylmethyl-, 5-fluoro-3-pyridylmethyl- or 4-methoxy-3- pyridylmethyl- moiety) was achieved via reductive amination (3 - 18 % yield as their hydrochlorides), which were evaluated in Signal Amplification By Reversible Exchange (SABRE) Hyperpolarisation experiments in their free-base forms. A GC-MS method has be developed and achieves separation of eight NBOMe's in 11 minutes with Limits of Detection (LoD) and Quantification (LoQ) of 6.2 - 10.8 µg mL-1 and 18.8 - 32.6 µg mL-1 respectively. The developed method has been validated against simulated blotters impregnated with the NBOMe's and directly compared against ATR FTIR and presumptive (Marquis and Lieberman) colour tests. SABRE Hyperpolarisation was explored using the 3-pyridylmethyl-, 5-fluoro-3- pyridylmethyl- or 4-methoxy-3-pyridylmethyl-containing derivatives, and achieved 1H and enhancements, via a low field (60 MHz) NMR, of 2x - 192x - where the best enhancement was observed in the 25B-NBOMe derivative where the 2- methoxybenzyl- group was replaced with a 4-methoxy-3-pyridylmethyl- moiety. These preliminary results indicate the potential for further development of these compounds as possible agents for medical imaging.

Image segmentation has a critical role in medical diagnosis systems as it is mostly the initial stage, and any error would be propagated in the subsequent analysis. Certain challenges, including Irregular border, low quality of images, small Region of Interest (RoI) and complex structures such as overlapping cells in images impede the improvement of medical image analysis. Deep learning-based algorithms have recently brought superior achievements in computer vision. However, there are limitations to their application in the medical domain including data scarcity, and lack of pretrained models on medical data. This research addresses the issues that hinder the progress of deep learning methods on medical data. Firstly, the effectiveness of transfer learning from a pretrained model with dissimilar data is investigated. The model is improved by integrating feature maps from the frequency domain into the spatial feature maps of Convolutional Neural Network (CNN). Training from scratch and the challenges ahead were explored as well. The proposed model shows higher performance compared to state-of-the-art methods by %2:2 and %17 in Jaccard index for tasks of lesion segmentation and dermoscopic feature segmentation respectively. Furthermore, the proposed model benefits from significant improvement for noisy images without preprocessing stage. Early stopping and drop out layers were considered to tackle the overfitting and network hyper-parameters such as different learning rate, weight initialization, kernel size, stride and normalization techniques were investigated to enhance learning performance. In order to expand the research into video segmentation, specifically left ventricular segmentation, U-net deep architecture was modified. The small RoI and confusion between overlapped organs are big challenges in MRI segmentation. The consistent motion of LV and the continuity of neighbor frames are important features that were used in the proposed architecture. High level features including optical flow and contourlet were used to add temporal information and the RoI module to the Unet model. The proposed model surpassed the results of original Unet model for LV segmentation by a %7 increment in Jaccard index.

Understanding what is happening in endoscopic scenes while it is happening is a key problem in Computer-Assisted Interventions (CAI). It is through semantic information extraction that we will be able to characterise and eventually improve on what is the current clinical practice. In this thesis, we focus on three main topics, real-time segmentation of surgical tools, automatic generation of synthetic instrument segmentation labels, and tissue classification. In most endoscopic scenes, surgical instruments play a crucial role. Instrument segmentation has a wealth of potential applications. It is already, or in some cases is bound to become, an essential building block of many computer-assisted clinical systems. These applications require real-time segmentation. Two approaches are introduced to achieve it, one multiplexing deep learning with optical flow and another employing a lightweight model and deep supervision. In keyhole surgery, surgeons cannot manipulate the endoscope themselves. Their hands are occupied with other instruments. Additional human camera operators pose a number of disadvantages. As an exemplar application of tool segmentation, a method that employs it as a building block for autonomous robotic endoscopy is introduced. Throughout our experimentation, we notice that even though we are able to achieve real-time segmentations, their quality is limited by the scarcity of labelled data, which leads to a poor generalisation to different endoscopic domains. We propose a self-supervised mechanism to automatise the generation of synthetic instrument segmentation labels. Equally important to segmenting the surgical instruments is to understand the rest of the surgical scene. As an exemplar addressing such need, we propose the first open dataset for endoscopic detection of early squamous cell neoplasia and provide a baseline method to distinguish between normal and abnormal video frames. However, in a clinical context, we are also required to explain the behaviour of our network. We propose a method to display network activation that allows us to study whether the input features looked at by the network align with the endoscopic markers employed in clinical practice.

One of the rate-limiting steps for the cytosolic delivery of endocytosed therapeutics is endosomal escape. An increase in endosomal release would increase the therapeutic output of current therapeutics, as well as allowing the use of novel therapeutics that currently cannot be successfully delivered to the correct cellular location. The cellular mechanisms involved in maintaining endosomal integrity and endosomal release are not fully understood and our work here aims to increase our understanding of endosomal release.

Immunofluorescent staining continues to be a crucial technique for determining the cellular localisation of proteins of interest. However, the need to fix the cells prevents its use in time-resolved approaches and can create imaging artefacts. In this project I was developing a new method for immunofluorescence in living cells. The B-subunit of cholera toxin (CTB), which has the ability to non-toxically induce endocytosis upon binding to the cell surface, was used to deliver antibody mimetics (called Affimers) into the endomembrane system of human cells.

The adaptive immune system provides antigen-specific immune responses through B and T lymphocytes (B and T cells) with specialised cell surface receptors. Antigens are developed de novo for specific antigen-mediated immune responses using recombination from vast repertoires and affinity maturation for diversification. Somatic hypermutation (SHM) is a B cell specific affinity maturation process whereby point mutations are intentionally inserted into the genomic sequence to marginally alter the confirmation of immunoglobulin (Ig) receptors. SHM is part of normal B cell development and is primarily restricted to germinal centres (GC). SHM has therefore been reported in lymphoproliferative disorders (LPDs) from GC origin. Strong prognostic links and clinical indications based on SHM status have been described in chronic lymphocytic leukaemia (CLL) were patients with mutated Ig present improved prognosis and longer overall survival. Currently, standardised SHM testing is performed by PCR amplification of clonal Ig sequences prior to Sanger sequencing (SSeq). However, next generation sequencing (NGS) presents a promising alternative as it also allows high sample throughput and investigation of structural variation (SV) and mutation analysis. Several NGS applications have been proposed for clonality and SHM status reporting but these still rely on initial PCR amplification of clonal sequences, limiting analysis of other molecular risk factors. Targeted NGS-capture represents a viable approach to assess IG/TR rearrangements, other SV, single nucleotide variants (SNV) and Indels. However, SHM status has not yet been reported using targeted NGS-capture platforms. The purpose of this thesis is to investigate the use of NGS-capture techniques and novel bioinformatics analysis to accurately report SHM status in accordance with standardised SSeq applications. To achieve this a novel analytical programme was developed: VCF-SoMAtic, which reports the frequency of somatic mutations in rearranged IGH genes while accounting for polymorphic variants and sequencing artefacts. Identification of clonally rearranged genes was required for this analysis so the effect of wet lab strategies, including probe targeting and NGS read length, were investigated. NGS-capture was shown to accurately detect clonal VDJ recombination by targeting one side of the rearrangement and using lower read lengths, down to 75 bp. NGS SHM analysis using the 1.5 kb genomic sequence from the clonally rearranged joining gene towards the enhancer region (IGHJ-E) was found to significantly correlate with SSeq IGHV%. Novel bioinformatic analysis of IGHJ-E and introduction of a stringent 99.8% mutational threshold in tested LPD cohorts found between 88.16-97.44% concordance (90.3% in all 175 samples) with SSeq SHM status. Poor SHM stratification of mutated CLL was observed when IGHV genes were analysed using NGS, which was due to poor alignment of captured reads. In summary, we have created and applied a novel bioinformatic application with targeted NGS to successfully assess SHM status in a range of LPD subtypes using novel analysis of IGHJ-E.

The study of solvated proteins by terahertz (THz) frequency electromagnetic radiation, characterized by submillimeter wavelength signal beams, suffers from strong absorption by water. Yet, water is vital to normal protein function; hydration plays a formative role in a protein acquiring its operational molecular structure. To decouple the traditional conflation of water spectra with that contributed by the protein in conventional transmission or absorption spectroscopy, this thesis innovates the experimental methodology of circular dichroism (CD) spectroscopy; water being achiral. Dealing with the heavy signal loss from absorption by water is countered by the gain afforded by using a state-of-the-art vector network analyser (VNA), that is interfaced with a quasi-optical (QO) circuit, containing the sample of dissolved protein. This apparatus has been employed to co-study CD protein conformational changes conventionally acquired in the far-UV at the discrete frequency of 222 nm. This thesis explores the first acquisition of a continuous THz CD spectra of a protein as it responds to variation in the temperature, salinity and pH of its host aqueous medium, in order to lay foundations for interpreting associated spectra in relation to the given conformational state the protein adopts in response. Experimental work in this thesis was performed over the mm and sub-mm waveguide bands: WR7 (110 - 170 GHz) and WR3 (220 - 325 GHz), respectively; and, THz Time-Domain Spectroscopy (TDS). (200 - 1400 GHz). A QO-based, sub-THz CD spectrometer is demonstrated on the sample protein, myoglobin, over the waveguide bands of WR7 and WR3. In so doing, an equivalent methodology is propounded for acquiring cross-polar spectra that eliminates the need for having to physically rotate the receive-horn through 90˚. A significant source of systematic-error is thereby eliminated (and, no less, random-errors, that may occur when cabling is moved). The 'zero-crossing' points of CD spectra are characterized as being correlated with traditional far-UV CD spectra in the given environmental parameter settings of, for example, pH and/or sample concentration level. For the extending application, the sub-THz CD QO system studies three solvated proteins namely, Myoglobin, Bovine Serum Albumins, and Papain at WR3. Water is used as a buffer, since it contributes no CD signal. Secondary structures of the proteins consist of different amounts of α-helices and β-sheet forms. Different optical path-lengths and concentration levels are considered in order to compare their efficiency in providing sub-THz CD signatures by different structures of proteins. The application of THz TDS is to study the dielectric properties of the three proteins. Dielectric properties of solvated proteins are strongly influenced by the flexibility of protein secondary structures, hydration bonds, and water-protein interaction. By analyzing the protein's dielectric behaviour, explanation of protein CD behaviour will be sought for the sub-THz frequency domain.

Increasing the static magnetic field (B0) strength of magnetic resonance imaging (MRI) scanners improves the signal-to-noise ratio (SNR). Higher SNR may be used for more sensitive, faster, and higher resolution imaging. Despite its potential, the biggest challenges for ultra-high field MRI is inhomogeneity in the radio-frequency field (B1+), which causes spatially varying flip angles when using fixed-frequency pulses, leading to undesirable perturbations in signal and contrast. The most flexible solution to B1+ inhomogeneity at UHF is parallel transmission (pTx). A pTx enabled system consisting of more than one RF transmit channels which may be independently driven. This thesis seeks to extend the methodology and applications of pTx magnetic resonance imaging (MRI) in the brain at 7 tesla. Specifically, this thesis focuses on two spin labelling techniques: arterial spin labelling (ASL) that measures cerebral perfusion and T2-Relaxation-Under-Spin-Tagging (TRUST) that measures venous oxygenation. The ASL study demonstrates that labelling efficiency can be improved by the incorporation of pTx static shimming in the labelling pulses, but indicates that the temporal SNR does not yet exceed optimised 3 tesla protocols in the absence of further developments. The TRUST study investigates the use of spatially-selective pTx pulses in a modified TRUST sequence to measure regional oxygen extraction fraction (OEF) in the brain. Results show that localised saturation was achieved by repeatedly exciting (using pTx pulses) and saturating spins outside the region-of-interest.

Mice in which the autophagy protein ATG16L1 was selectively ablated in T cells develop spontaneous intestinal inflammation, characterized by loss of Foxp3+ Treg cells and a selective expansion of Gata3+ Th2 cells. This phenotype was most pronounced in the colonic lamina propria. Thus, a lack of autophagy simultaneously reduces anti-inflammatory Treg cells and increases pro-inflammatory Th2 effector cells in the intestine. This discrepancy could be partly attributed to an altered metabolic programme employed by Atg16l1-deficient Treg cells, with increased glycolysis and impaired fatty acid metabolism. In contrast, the metabolic profile of in vitro polarized Th2 cells seemed to be independent of autophagy. The aims of this thesis are to investigate the mechanistic consequences of altered autophagy in different T cell subsets and also to explore the link between the dominant type 2 immune responses and the observed intestinal pathology.

Histopathology is the study and diagnosis of disease via tissue microscopy and it is currently the 'gold-standard' in formally diagnosing many types of disease including cancers. Due to increasing workloads on pathologists, there is a growing need for automated image analysis pipelines that are able to filter out obviously benign samples. However, these algorithms are sensitive to factors of variation such as the staining and scanning conditions with which a tissue specimen is processed that differ significantly across institutions. In this thesis, novel machine learning algorithms are developed using state-of-the-art deep learning techniques that enable the creation of automated histopathology image analysis pipelines that are more robust to frequently occurring artifacts and variation in appearance. A novel neural network architecture called the Dual-Channel Auto-Encoder (DCAE) is devised that learns representations of histopathology images that are invariant to the effects of differing digital slide scanners whilst holding enough discriminative power to delineate various anatomical structures within tissue. This method achieves a 50% improvement in SSIM score on tissue masks derived from the DCAE features compared to related methods. Next, a novel unsupervised approach to detect and remove artifacts using techniques from generative adversarial networks is presented. This approach allows artifacts to be removed prior to the application of stain separation which is demonstrated experimentally to improve the stain colour estimation performance. Finally, a more sophisticated model of stain separation called Tissue-Dependent Stain Separation (TDSS) is developed which incorporates contextual information such as texture and colour to make more informed estimates of stain colour conditioned on tissue sub-types and thus producing better separations of stains compared to state-of-the-art methods. The TDSS model achieves an 8.81% reduction in mean-squared error in a stain separation task compared to existing state-of-the-art-methods. These contributions represent a significant step forward towards robust, fully automated histopathology image analysis pipelines.

The germinal centre (GC) response is an essential component of adaptive immunity, crucial for the generation of high-affinity antibodies, long-lived memory and plasma cells and is the basis for most current vaccines. While the dynamics of cognate B and T lymphocytes during the GC response are extensively studied, the roles of resident tissue macrophages in this process are less well understood, particularly in the spleen. A unique compartment to the spleen, called the marginal (MZ), is known to be important for adaptive immunity and harbours at least two distinct MZ-resident macrophage subsets. Both MZ macrophage populations are essential for capturing and preventing systemic dissemination of pathogens. However, their individual functions and potential role in adaptive immunity are not established. Here we used pharmacologic and genetic approaches to deplete SIGN-R1+ MZ macrophages and showed that these cells were specifically required for the optimal development of germinal centre responses in the spleen. SIGN-R1+ macrophages were dispensable for the initial GC establishment but were required for optimal maintenance of the response. The GC defect could be partially corrected by boosting T follicular helper (Tfh) cell responses or by inducing Tfh help prior to macrophage ablation. Furthermore, in the absence of SIGN-R1+ macrophages, 33D1+ DCs, a key population involved in Tfh priming, was displaced from the interfollicular regions to the MZ. Reconstituting the SIGN-R1+ subset restored DC positioning and rescued the GC phenotype. This work advances our understanding of how the splenic marginal zone regulates adaptive immunity, highlights the functional diversification of MZ macrophages, and provides the first genetic mouse model allowing for the specific depletion and modification of SIGN-R1+ cells.

Immunoglobulin E (IgE), pivotal for the allergic response, is an important and proven drug target for allergic diseases, including asthma. Recent insights into the molecular properties of IgE, such as the allosteric regulation of binding to the FceRI and CD23 receptors and the calcium dependence of the interaction with CD23, together with recent innovations in biological therapeutics that exploit the natural pathway for IgG and albumin recycling through the neonatal Fc receptor (FcRn), suggest an alternative mechanism for eliminating IgE. A panel of novel anti-IgE fusion proteins was developed that combines the ability of IgE to bind CD23 with the FcRn binding properties of IgG. The fusion proteins thus have the potential to remove IgE and be recycled to target more IgE antibodies. This thesis describes biophysical techniques and cellular-based assays that investigate the properties and potential mechanisms of action of these novel anti-IgE fusion proteins. Surface plasmon resonance (SPR) was used to assess the stoichiometries and kinetics for the interactions between the novel anti-IgE molecules and IgE. Basophil degranulation assays suggested that these molecules do not induce cellular degranulation, nor do they induce degranulation of cells in which IgE is pre-bound to FceRI, therefore these anti-IgE molecules are unlikely to pose a safety risk. The novel molecules do, however, prevent IgE binding to the cell. Furthermore, flow cytometry revealed that CD63 and CD203c, biomarkers of inflammation, were not upregulated by IgE when in complex with the anti-IgE molecules, and the interaction between IgE and FceRI was inhibited in a concentration-dependent manner. A novel cell-based recycling assay was developed to evaluate the ability of the anti-IgE molecules to be recycled to the extracellular supernatant, after delivering IgE to a cellular degradation pathway. This assay confirmed recycling of the anti-IgE molecules into the extracellular supernatant, with intracellular retention of IgE. Transcytosis assays also revealed that IgE could not transcytose a cell monolayer when in complex with the anti-IgE molecules, and was instead retained intracellularly. Finally, the intracellular fate of IgE was tracked using confocal microscopy, which revealed that IgE was delivered to the lysosomal degradation pathway. It is anticipated that such reagents could be a starting point to inform the design of therapeutics able to catalyse the degradation of IgE without themselves being consumed.

The inflammatory response is driven by a highly accurate and kinetically complex transcriptional program that is controlled by the stimulus-regulated usage of thousands of cis-regulatory elements (i.e. enhancers and promoters). Stimulus-activated transcription factors (TFs) involved in this response, as well as their interplay with lineage-determining TFs, have been extensively characterised. However, the role of different families of coregulators that are recruited by those TFs in most cases remains to be elucidated. In light of this, I set out to dissect the role of a specific family of co-regulators, namely the ISWI family of ATP-dependent chromatin remodellers, in modulating the transcriptional response to inflammatory stimuli such as lipopolysaccharide (LPS) in bone marrow-derived mouse macrophages. By combining transcriptomic and other genomic experiments and analyses, I have found that ISWI complexes present in innate immune cells are involved in the regulation of different components of the transcriptional program of murine macrophages. On one hand, I found that ISWI complexes are strongly associated with the regulation of the interferon response. By ChIP-seq analysis, I show that distinct ISWI subunits are recruited to regulatory regions bound by Interferon Regulatory Factors (IRFs), where they control the epigenomic landscape and thus regulate the interferon gene expression program. On the other hand, by ATAC-seq analysis I also identified a large set of genomic regions which are maintained in a repressed state in the presence of the ISWI catalytic subunit SMARCA5. However, these regions are not occupied by SMARCA5, indicating that in this case it is not the direct action of the ATPase which controls the chromatin state.

Pre-formed antibodies to non-self Human Leucocyte Antigens (HLA) represent a major barrier to kidney transplantation. HLA antibodies capable of causing a positive flow cytometric crossmatch (FCXM) against donor lymphocytes are associated with an increased risk of antibody-mediated rejection (AMR). The registration of incompatible (i.e. antibody positive) HLA with the UK kidney allocation system reduces the risk of a positive FCXM and early AMR. HLA antibodies in patient sera are routinely defined using sensitive high-throughput Luminex® single antigen bead (SAB) assays, but these antibodies are not always capable of causing a positive FCXM. As such, the 'clinical relevance' of HLA antibodies identified solely by SAB assays is debated. UK guidelines state that a negative pre-transplant FCXM equates to standard immunological risk to transplant. Therefore, the use of SAB assays to define incompatible HLA potentially restricts opportunities for patients to receive a kidney transplant offer, particularly for those who have HLA antibodies directed against ≥85% of HLA in the donor population. The aim of this project was to establish whether innovative application of alternative commercial methods for HLA antibody characterisation could improve prediction of clinical relevance for individual antibodies. Over a four-year period, 8% (58/730) of crossmatches were performed against patient sera containing donor-specific HLA antibodies (DSA) identified using the One Lambda LABScreen™ Single Antigen Bead (LABScreen) assay. However, an unexpected negative FCXM was seen for 66% (38/58) of sera containing LABScreen DSA. Retrospective analysis demonstrated that FCXM outcome could not be reliably predicted by LABScreen based on the number of DSA, DSA specificity, or assay output value. Additional HLA antibody testing was performed using Immucor LIFECODES® LSA (LIFECODES), Immucor LIFECODES® C3d Detection (LIFECODES-C3d) and BAG Healthcare HISTO SPOT® HLA AB Screen/ID (HISTO SPOT) assays. These three assays improved accurate prediction of a negative FCXM from 0% to 89%, although positive FCXM prediction decreased from 100% to 68%. Consideration of HISTO SPOT and LABScreen results alone improved prediction of a negative FCXM from 0% to 68% and positive FCXM prediction to 88%. LIFECODES-C3d positivity demonstrated an association with a positive FCXM independent of LABScreen output value, contrary to previous reports. Antibody-epitope analysis using HLAMatchmaker did not offer further FCXM predictive value but enhanced understanding of the validity of putative antibody targets identified by each assay, particularly for sera with complex antibody reaction patterns. The HISTO SPOT assay demonstrated high concordance with FCXM outcome for this cohort. This represents the first report supporting application of the HISTO SPOT assay in assessment of clinical relevance for pre-transplant HLA antibodies. A clinical case study for a patient with high panel HLA antibody reactivity illustrated that a small decrease in the number of registered incompatible HLA increased the proportion of compatible UK donors. The use of HISTO SPOT, LIFECODES-C3d and HLAMatchmaker analysis, alongside the routine LABScreen assay, offers a novel strategy for improved clinical risk stratification of pre-transplant HLA antibody. Implementation of this testing strategy could improve access to HLA compatible transplantation, by reducing the number of incompatible HLA registered with the UK allocation system, and thereby decreasing waiting time, particularly for patients who are currently considered HLA incompatible with more than 85% of the UK donor population.

The development of highly accurate, quantitative automatic medical image segmentation techniques, in comparison to manual techniques, remains a constant challenge for medical image analysis. In particular, segmenting the pancreas from an abdominal scan presents additional difficulties: this particular organ has very high anatomical variability, and a full inspection is problematic due to the location of the pancreas behind the stomach. Therefore, accurate, automatic pancreas segmentation can consequently yield quantitative morphological measures such as volume and curvature, supporting biomedical research to establish the severity and progression of a condition, such as type 2 diabetes mellitus. Furthermore, it can also guide subject stratification after diagnosis or before clinical trials, and help shed additional light on detecting early signs of pancreatic cancer. This PhD thesis delivers a novel approach for automatic, accurate quantitative pancreas segmentation in mostly but not exclusively Magnetic Resonance Imaging (MRI), by harnessing the advantages of machine learning and classical image processing in computer vision. The proposed approach is evaluated on two MRI datasets containing 216 and 132 image volumes, achieving a mean Dice similarity coefficient (DSC) of 84:1 4:6% and 85:7 2:3% respectively. In order to demonstrate the universality of the approach, a dataset containing 82 Computer Tomography (CT) image volumes is also evaluated and achieves mean DSC of 83:1 5:3%. The proposed approach delivers a contribution to computer science (computer vision) in medical image analysis, reporting better quantitative pancreas segmentation results in comparison to other state-of-the-art techniques, and also captures detailed pancreas boundaries as verified by two independent experts in radiology and radiography. The contributions' impact can support the usage of computational methods in biomedical research with a clinical translation; for example, the pancreas volume provides a prognostic biomarker about the severity of type 2 diabetes mellitus. Furthermore, a generalisation of the proposed segmentation approach successfully extends to other anatomical structures, including the kidneys, liver and iliopsoas muscles using different MRI sequences. Thus, the proposed approach can incorporate into the development of a computational tool to support radiological interpretations of MRI scans obtained using different sequences by providing a "second opinion", help reduce possible misdiagnosis, and consequently, provide enhanced guidance towards targeted treatment planning.

Interleukin-10 (IL-10) is a key immuno-regulatory cytokine with potent anti-inflammatory effects. IL-10 contributes to immuno-suppression by inhibiting the presentation of antigen and decreasing the production of pro-inflammatory cytokines from innate cells such as dendritic cells and macrophages which in turn can suppress the activity of adaptive T cells. However, much is still unknown about the precise mechanisms through which this cytokine exerts its effects. This thesis used protein engineering techniques to explore the molecular mechanisms underpinning IL-10's activities. IL-10 occurs naturally as a dimeric molecule which binds two receptor subunits: IL-10Rα and IL-10Rß. It has very poor affinity for the IL-10Rß receptor subunit and is unable to bind IL-10Rß in the absence of IL-10Rα (Walter, 2014). Using error-prone yeast display we were able to generate a variant of IL-10 which could directly bind to the IL-10Rß subunit, with affinities in the low micromolar range. This variant was recombinantly expressed in both the monomeric and dimeric conformations. Using these engineered variants in combination with the wild type monomeric and dimeric IL-10, allowed us to examine the contribution of both receptor affinity and stoichiometry to IL-10's activities. Single particle imaging of the receptor subunits on the surface of live cells showed that increased affinity for IL-10Rß enhanced heterodimerisation of IL-10Rα and IL-10Rß for both the monomeric and dimeric variants. This increased receptor complex assembly translated into increased STAT1 and STAT3 activation by the high affinity monomer compared to the wild type monomer. The high affinity dimer showed a similar maximal STAT activation compared to the wild type dimer however, it also showed favourable activity at lower concentrations. This resulted in enhanced biological activities by the high affinity dimer compared to the wild type at sub-saturating concentrations e.g., enhanced suppression of LPS-induced IL-6 secretion. In order to view differences between the variants on a more global level, RNA sequencing was performed on human primary monocytes and CD8 T cells, treated with the monomeric and dimeric wild type and high affinity variants. The high affinity monomer had enhanced regulation of almost all transcripts compared to the wild type monomer however, it failed to reach dimeric expression levels. The wild type dimer showed poorer gene regulation when used at low doses, particularly when looking at suppression of pro-inflammatory cytokines and chemokines in monocytes. The high affinity dimer showed enhanced regulation of subsets of genes at low doses compared to the wild type, suggesting that improved receptor affinity could be used to overcome dose-dependent limitations. These data demonstrate that both stoichiometry and affinity play a large role in IL-10's activities. A structure of the complete IL-10/IL-10Rα/IL-10Rß complex is lacking, hindering our understanding of the molecular interactions in this cytokine-receptor complex. We used our high affinity variants to attempt crystallisation of the receptor complex. While a stable complex was successfully purified by gel filtration, no crystals were generated. However, work done here lays the foundation for future attempts using alternate technologies such as cryogenic electron microscopy. IL-10 has also been shown, under certain conditions, to enhance the tumour targeting ability of CD8 T cells (Naing et al., 2018) however much is still unknown about this process. We used proteomics as well as RNA sequencing to examine the action of IL-10 in human CD8 T cells. This study showed that IL-10 can downregulate exhaustion markers in these cells. In addition, it highlighted downregulation of several enzymes in glycolysis as well as an upregulation of several mitochondrial components, indicating that IL-10 may be regulating metabolic processes in CD8 T cells. While these results are only considered preliminary, they highlight key areas for future work.

While efficient germinal centre (GC) formation is the cornerstone of long-lasting humoral immunity against pathogens, dysregulated GC responses can lead to pathology. GCs are highly dynamic and complex microenvironments, and despite significant progress made during the past few decades, many questions about the key molecules involved in GC regulation remain unanswered. The work presented in this thesis investigated the role of the cytokine interleukin 21 (IL-21) in the regulation of GC responses. IL-21 has been shown to modulate GC responses in wildtype mice, and elevated IL-21 levels have been implicated in several autoimmune diseases, including type 1 diabetes, rheumatoid arthritis and systemic lupus erythematosus. We have previously demonstrated that CTLA-4-/- mice develop spontaneous exacerbated GC responses as well as produce markedly elevated levels of IL-21. In this thesis, we used IL-21R-/-CTLA-4-/- animals to investigate the effects of IL-21 signalling in these chronic GCs and showed that IL-21 considerably increased the scale of CTLA-4-/- GC responses. More importantly, we also demonstrated that, in CTLA-4-/- mice, IL-21 significantly regulated GC B cell polarisation, and IL-21R deficiency led to a profound reduction in the GC dark zone compartment. Further analyses suggested that IL-21 supported the GC dark zone by promoting GC B cell selection and proliferation. Additionally, we demonstrated that our key findings in the CTLA-4-deficient model also applied to immunisation-induced transient GCs in wildtype animals. Furthermore, using wildtype and IL-21R-/- mixed bone marrow chimeric mice, we showed that dark zone GC B cell homeostasis was largely dependent on B cell-intrinsic IL-21 signalling. The findings presented in this thesis provide further insights into IL-21-dependent GC regulation. These will help advance our understanding of the generation of humoral immunity and may aid the development of novel and more targeted therapeutic strategies for patients suffering from a broad range of autoimmune conditions.

Electrical Impedance Tomography (EIT) allows reconstructing conductivity changes into images. EIT detects fast impedance changes occurring over milliseconds, due to ion channel opening, and slow impedance changes, appearing in seconds, due to cell swelling/increased blood flow. The purpose of this work was to examine the feasibility of using EIT for imaging a gyrencephalic brain with implanted depth electrodes during seizures. Chapter 1 summarises the principles of EIT. In Chapter 2, it is investigated whether recent technical improvements could enable EIT to image slow impedance changes upon visual stimulation non-invasively. This was unsuccessful so the remaining studies were undertaken on intracranial recordings. Chapter 3 presents a computer modelling study using data from patients, for whom the detection of simulated seizure-onset perturbations for both, fast and slow impedance changes, were improved with EIT compared to stereotactic electroencephalography (SEEG) detection or EEG inverse-source modelling. Chapter 4 describes the development of a portable EIT system that could be used on patients. The system does not require averaging and post-hoc signal processing to remove switching artefacts, which was the case previously. Chapter 5 describes the use of the optimised method in chemically-induced focal epilepsy in anaesthetised pigs implanted with depth electrodes. This shows for the first time EIT was capable of producing reproducible images of the onset and spread of seizure-related slow impedance changes in real-time. Chapter 6 presents a study on imaging ictal/interictal-related fast impedance changes. The feasibility of reconstructing ictal-related impedance changes is demonstrated for one pig and interictal-related impedance changes were recorded for the first time in humans. Chapter 7 summarises all work and future directions. Overall, this work suggests EIT in combination with SEEG has a potential to improve the diagnostic yield in epilepsy and demonstrates EIT can be performed safely and ethically creating a foundation for further clinical trials.

The draining lymph node (dLN) is the anatomical site in which adaptive immune responses are initiated following vaccination. It is increasingly recognised that the dLN also serves an important innate barrier function and that inflammatory stimuli (including vaccine adjuvants) drive cardinal aspects of the innate immune response within the dLN. The characterisation of these intranodal innate immune processes and their impact upon the concurrently developing adaptive immune response is therefore central to the design of novel vaccines and adjuvants. Neutrophil and monocyte infiltration is a cardinal feature of the innate immune response. This phenomenon is studied within the dLN in the context of two key innate immune pathways; cyclooxygenase-dependent prostanoids and type I interferons. These processes were studied using a murine skin immunisation model following challenge with killed E. coli (KEC), which induced the rapid and sequential infiltration of neutrophils and monocytes into the dLN. These infiltrating myeloid cells were major expressers of cardinal prostanoid synthases (cyclooxygenase-2, microsomal PGE synthase-1 and thromboxane synthase), as well as important interferon-stimulated genes such as CXCL9 and CXCL10. Notably, cyclooxygenase inhibition during their infiltration did not modulate the developing humoral immune response. In contrast, type 1 interferons drove the differential upregulation of CD69 by different lymphocyte subsets and the acute production of interferon-γ by dLN NK cells; processes that play important roles in the retention and activation of T cells. In vitro evidence suggests that these processes are driven by interferon-stimulated monocytes, a hypothesis supported by the markedly increased expression of type I interferon stimulated genes by dLN monocytes in vivo. In conclusion, this thesis highlights the role of infiltrating myeloid cells as unappreciated orchestrators of type I interferon-driven innate immune pathways in the dLN. This finding informs hypotheses that assert that inflammatory monocytes drive Th1 T cell responses in the dLN.

Among its pleiotropic effects, vitamin D has immunoregulatory properties that help to maintain immune homeostasis. Multiple factors make haematopoietic stem cell transplant (HSCT) recipients at higher risk of vitamin D deficiency, and this (even prior to the stem cell infusion) can impact adversely over the course of HSCT. Owing to the lack of consensus, a cut-off to define vitamin D deficiency has not been established yet and clinical practice may vary across different HSCT units. To address this, one of the chapters of this thesis has examined the current management of vitamin D deficiency in the allogeneic HSCT setting, confirming the highly heterogeneous practice across the EBMT affiliate centres, including those from diverse geographical locations and dedicated to patients from different age. Graft-versus-host disease (GvHD) is a major cause of morbidity and mortality after allogeneic HSCT. To confirm or rule out this disease, a biopsy result from the affected tissue may be delayed for several days or even weeks. GvHD biomarkers are promising diagnostic tools that can speed up this process, predict outcomes in the early post-HSCT phase and monitor response to immunosuppression, minimising the detrimental effect of this therapy on HSCT recipients. Due to this, an observational study will explore the role of three of these biomarkers (elafin, REG3α and ST2), alongside vitamin D, in the context of patients with acute and chronic GvHD on immunosuppressive therapy. The final study moves away from vitamin D and biomarkers although it is still linked to the graft-versus-host reaction. Infusion of donor lymphocytes (DLI) is an effective adoptive immunotherapy that approximately 25% of post-HSCT patients have received in the UK1. DLI enhances graft-versus-leukaemia effect but its main side effect is GvHD. This chapter describes a single-centre experience treating 100 patients with DLI after reduced intensity conditioning (RIC) HSCT for mixed chimerism (MC) or relapse of the primary disease. It aims to determine factors implicated in achieving full donor chimerism (FDC) or disease remission, as well as their impact on other outcomes post-DLI, such as survival, relapse post-DLI or GvHD, in order to improve survival and quality of life in these patients. We found that patients with younger donors were less likely to develop acute GvHD and subsequently it contributed to a greater survival, which has been previously reported in HSCT recipients but never in those receiving subsequent DLI.

CD4+ T cell interactions with dendritic cells (DC) are pivotal in adaptive immune responses and play an important role in both protective immunity (e.g. infection) and autoimmune diseases (e.g. rheumatoid arthritis). As such, there is increasing interest in discovering molecules that would promote/disrupt this interaction. Taking advantage of transcriptomic data generated in our lab using a murine model of inflammatory arthritis, we aimed to identify molecules that can control these interactions. We hypothesised that blockade of some of these molecules will disrupt these interactions, affecting T cell activation and/or migration of immune cells that promote pathology. This will not only further our basic understanding of disease processes but may also highlight potential therapeutic targets for human disease. The F11 receptor (F11R) gene, which encodes the junctional adhesion molecule-A (JAM-A), was identified in a gene list previously generated in the laboratory. F11R was upregulated in non-migratory leukocytes from inflamed joints in comparison to migratory immune cells. To investigate the role of JAM-A on CD4+ T cell activation, we employed in vitro and in vivo assays with cells from transgenic T cell receptor (TCR) mice (OTII). Treatment with an antagonistic anti-JAM-A monoclonal antibody (mAb) in vitro attenuated CD4+ T cell activation and proliferation and decreased T cell differentiation towards the Th1 subset in comparison with cells treated with its isotype control. In vivo, anti-JAM-A mAb treatment impaired CD4+ T cell proliferation in comparison with cells from mice treated with its isotype control. However, prophylactic treatment with anti-JAM- A mAb did not impact clinical disease in an acute, breach of self-tolerance model of arthritis. These findings are the first to describe a role for JAM-A during CD4+ T cell-DC interactions, but do not support JAM-A as a therapeutic target for rheumatoid arthritis (RA). Future work to evaluate the effects of treatment with JAM-A antagonists in late models of RA, in which autoimmune components may play a bigger role in clinical arthritis, or in models of other autoimmune diseases will further our understanding on JAM-A contribution to human disease.

Antibody plays an important role in the protection against many different infections. Here the role of antibody in the protection against P aeruginosa infection was further investigated. Patients suffering from chronic lung infection with P. aeruginosa can produce antibodies that protect the bacterium from the bactericidal activity of serum. The effect of the removal of these antibodies on the P. aeruginosa population was investigated. One clonal lineage appeared to be eradicated from a patient with a multi-lineage infection. However, the treatment did not promote strain replacement in either patient suggesting that a single strain was able to persist within a specific niche in the lung. The role of antibody in the protection against P. aeruginosa infection by sera from healthy individuals was further investigated. Killing of P. aeruginosa was complement-dependent and in some instances, complement alone was sufficient to elicit killing. P. aeruginosa-specific antibodies enhanced the killing of some strains and blocked the killing of others. Inhibitory antibody was identified in healthy serum, suggesting that healthy individuals can produce inhibitory antibodies without an active infection. The mechanism of inhibitory antibodies is not fully understood. A transposon library was constructed in a strain of P. aeruginosa isolated from a non-cystic fibrosis bronchiectasis patient and used to define the essential genome. This was the first library to be constructed in a multi-drug resistant respiratory isolate and consisted of 577,494 unique mutants. The library can be used as a tool to provide further insight into the mechanism of inhibitory antibodies and has the potential to lead to the development of new treatments and diagnostics for patients with inhibitory antibody.

Purpose: Dynamic susceptibility-contrast magnetic resonance imaging (DSC-MRI) is widely used for cerebral perfusion measurement, but dependence on operator input leads to a time-consuming, subjective, and poorly-reproducible analysis. Although automation can overcome these limitations, investigations are required to further simplify and accelerate the analysis. This research focuses on automating arterial voxel (AV) and brain tissue segmentation, and model-dependent deconvolution steps of DSC-MRI analysis. Methods: Several features were extracted from DSC-MRI data; their AV- and tissue voxel- discriminatory powers were evaluated by the area-under-the-receiver-operating-characteristic-curve (AUCROC). Thresholds for discarding non-arterial voxels were identified using ROC cut-offs. The applicability of DSC-MRI time-series data for brain segmentation was explored. Two segmentation approaches that clustered the dimensionality-reduced raw data were compared with two raw−data-based approaches, and an approach using principal component analysis (PCA) for dimension-reduction. Computation time and Dice coefficients (DCs) were compared. For model-dependent deconvolution, four parametric transit time distribution (TTD) models were compared in terms of goodness- and stability-of-fit, consistency of perfusion estimates, and computation time. Results: Four criteria were effective in distinguishing AVs, forming the basis of a framework that can determine optimal thresholds for effective criteria to discard tissue voxels with high sensitivity and specificity. Compared to raw−data-based approaches, one of the proposed segmentation approaches identified GM with higher (>0.7, p < 0.005), and WM with similar DC. The approach outperformed the PCA-based approach for all tissue regions (p < 0.005), and clustered similar regions faster than other approaches (p < 0.005). For model-dependent deconvolution, all TTD models gave similar perfusion estimates and goodness-of-fit. The gamma distribution was most suitable for perfusion analysis, showing significantly higher fit stability and lower computation time. Conclusion: The proposed methods were able to simplify and accelerate automatic DSC-MRI analysis while maintaining performance. They will particularly help clinicians in rapid diagnosis and characterisation of tumour or stroke lesions, and subsequent treatment planning and monitoring.