In this thesis, a novel case-based reasoning (CBR) approach to radiotherapy treatment planning for brain cancer patients is presented. In radiotherapy, tumour cells are destroyed using ionizing radiation. For each patient, a treatment plan is generated that describes how the radiation should be applied in order to deliver a tumouricidal radiation dose while avoiding irradiation of healthy tissue and organs at risk in the vicinity of the tumour. The traditional, manual trial and error approach is a time-consuming process that depends on the experience and intuitive knowledge of medical physicists. CBR is an artificial intelligence methodology, which attempts to solve new problems based on the solutions of previously solved similar problems. In this research work, CBR is used to generate the parameters of a treatment plan by capturing the subjective and intuitive knowledge of expert medical physicists stored intrinsically in the treatment plans of similar patients treated in the past. This work focusses on the retrieval stage of the CBR system, in which given a new patient case, the most similar case in the archived case base is retrieved along with its treatment plan. A number of research issues that arise from using CBR for radiotherapy treatment planning for brain cancer are addressed. Different approaches to similarity calculation between cases are investigated and compared, in particular, the weighted nearest neighbour similarity measure and a novel non-linear, fuzzy similarity measure designed for our CBR system. A local case attribute weighting scheme has been developed that uses rules to assign attribute weights based on the values of the attributes in the new case and is compared to global attribute weighting, where the attribute weights remain constant for all target cases. A multi-phase case retrieval approach is introduced in which each phase considers one part of the solution. In addition, a framework developed for the imputation of missing values in the case base is described. The research was carried out in collaboration with medical physicists at the Nottingham University Hospitals NHS Trust, City Hospital Campus, UK. The performance of the developed methodologies was tested using brain cancer patient cases obtained from the City Hospital. The results obtained show that the success rate of the retrieval mechanism provides a good starting point for adaptation, the next phase in development for the CBR system. The developed automated CBR system will assist medical physicists in quickly generating treatment plans and can also serve as a teaching and training aid for junior, inexperienced medical physicists. In addition, the developed methods are generic in nature and can be adapted to be used in other CBR or intelligent decision support systems for other complex, real world, problem domains that highly depend on subjective and intuitive knowledge.

Automatic detection of protein-protein interactions (PPIs) in biomedical publications is vital for efficient biological research. It also presents a host of new challenges for pattern recognition methodologies, some of which will be addressed by the research in this thesis. Proteins are the principal method of communication within a cell; hence, this area of research is strongly motivated by the needs of biologists investigating sub-cellular functions of organisms, diseases, and treatments. These researchers rely on the collaborative efforts of the entire field and communicate through experimental results published in reviewed biomedical journals. The substantial number of interactions detected by automated large-scale PPI experiments, combined with the ease of access to the digitised publications, has increased the number of results made available each day. The ultimate aim of this research is to provide tools and mechanisms to aid biologists and database curators in locating relevant information. As part of this objective this thesis proposes, studies, and develops new methodologies that go some way to meeting this grand challenge. Pattern recognition methodologies are one approach that can be used to locate PPI sentences; however, most accurate pattern recognition methods require a set of labelled examples to train on. For this particular task, the collection and labelling of training data is highly expensive. On the other hand, the digital publications provide a plentiful source of unlabelled data. The unlabelled data is used, along with word cooccurrence models, to improve classification using Gaussian processes, a probabilistic alternative to the state-of-the-art support vector machines. This thesis presents and systematically assesses the novel methods of using the knowledge implicitly encoded in biomedical texts and shows an improvement on the current approaches to PPI sentence detection.

The discovery of stem cells has led to rapid advances in the field of regenerative medicine. Their unique properties, including the ability to self-renew and differentiate make them ideal for the repair/replacement of tissues that have been damaged as a result of disease or injury. Mesenchymal stem cells in particular represent a highly valuable pool of adult stem cells for such regenerative applications due to their accessibility, and potential as an autologous patient derived autologous nature However current methods for the in vitro expansion of high quality autologous mesenchymal stem cells results in spontaneous differentiation of the stem cell population and a loss of differentiation capacity over time. In vivo, it is the stem cell niche that provides stem cells with the appropriate cues required to maintain stem cell self-renewal. It is proposed that by mimicking these cues using biomaterials, that the self-renewal of mesenchymal stem cells can be controlled in vitro. In this study, a novel nanopit topography was investigated for its effects on the maintenance and growth of mesenchymal stem cells in vitro. To investigate this, three main aspects of mesenchymal stem cell state were examined in response to this novel nanotopography: maintenance of the stem cell phenotype over time including expression of stem cell markers and differentiation potential over time, changes in signalling pathways associated with differentiation and lastly, the metabolic profile of stem cells. As a result of this study we have identified a novel nanopit topography, which in the absence of chemical supplements, provides a substrate that is conducive to the maintenance of mesenchymal stem cells. Small RNAs have also been implicated in the regulation of signalling pathways and the metabolic state of stem cells. Furthermore, the ability to produce nanotopographically-patterned substrates using current standard techniques provides an inexpensive, high throughput method for the production of novel tissue culture plastics suitable for the maintenance of mesenchymal stem cells.

In order for cells to proliferate, a certain size has to be reached, which depends primarily on the rate of translation. RNA polymerase (pol) III plays a key role in protein synthesis by catalysing the production of small, untranslated RNA molecules such as transfer (tRNA) and 5S ribosomal RNA (5S rRNA). Indeed, recent evidence suggests that tRNAiMet production is limiting for translation and proliferation in some cell types. Therefore, the rate of pol III transcription plays a fundamental role in cellular growth and proliferation. Regulation of pol III output is mediated via a number of different mechanisms that can alter the activities of the transcription factors which are responsible for directing pol III transcription. Work presented in this thesis aimed at investigating the mechanisms behind the regulation of pol III transcription by the protein kinase CK2.

Nociception involves detection of tissue damage by specialized receptors; nociceptors. These convey information to the first synaptic relays in the dorsal horn of the spinal cord. Within the dorsal horn itself are the dorsal horn neurons which can be divided into two broad classes, based on their axonal projections: projection neurons and interneurons. The neurokinin 1 receptor (NK1r), the main target for substance P, is expressed by most projection neurons and many interneurons in the dorsal horn. These NK1r-expressing neurons show a bimodal size distribution in lamina I. The 1st part of the project tested the hypothesis that large NK1r-immunoreactive cells in this lamina are projection neurons, while the small cells are interneurons. Rats were anaesthetised and received injections of tracers into two supraspinal areas that are likely to label all contralateral lamina I projection neurons. The rats were re-anaesthetized and perfused 3 days later and 1341 NK1r-positive cells were analysed, of which 441 were retrogradely labelled. Cross-sectional soma areas of projection neurons were larger than those of cells that were not retrogradely labelled. This difference was highly significant. Nearly all (99.4%) of the NK1r-expressing cells that were not retrogradely labelled had soma areas <200 microm2, while only 9.8% of the retrogradely labelled NK1r-expressing cells had somata <200 microm2. These results provide a means of distinguishing lamina I NK1r-expressing projection neurons from interneurons based on their soma sizes without the need of retrograde tracing surgeries. Lamina I contains another population of projection neurons that lack or weakly express the NK1 receptor and consists of very large cells: giant cells, which are coated with the glycine and gamma-aminobutyric acid (GABA) receptor associated protein, gephyrin. There is also a group of large NK1r-expressing projection neurons with cell bodies in laminae III-IV and dendrites that pass dorsally to enter lamina I. Extracellular signal-regulated kinase (ERK) is expressed in dorsal horn neurons and is activated (phosphorylated) by noxious stimuli. In the 2nd part of the project, ERK phosphorylation in NK1r-expressing neurons as well as in lamina I giant cells was investigated following different type of noxious stimuli. Anaesthetised rats received noxious cutaneous, deep or visceral stimuli. They remained anaesthetized for 5 min after the end of the stimulus, and were then fixed by perfusion. Spinal cord sections were immunoreacted to reveal NK1r, gephyrin and phosphorylated ERK (pERK). Among the NK1r-expressing lamina I neurons, pERK was detected in both projection (somata >200 microm2) neurons and interneurons, with a significantly higher proportion in the larger cells, after all types of noxious stimulation. There was no significant difference in the frequency of pERK expression between the three morphological classes (fusiform, pyramidal and multipolar) of lamina I NK1r-expressing projection neurons after these stimuli. Most of the giant cells contained pERK after noxious cutaneous stimuli, but few did so following noxious deep stimulation. Only a few of laminae III-IV NK1r-expressing projection cells contained pERK after noxious deep or visceral stimulation, and the labelling in these was very weak. Results from the present study indicate that different types of neurons have different roles in conveying nociceptive information. The superficial dorsal horn (SDH) is also a vital area for modulating nociception and contains large number of excitatory and inhibitory interneurons. Glutamate, released by primary afferents and local excitatory neurons, acts on G-protein-coupled metabotropic glutamate receptors (mGlus). Group I mGlus (mGlu1 and mGlu5) are strongly expressed in the SDH. It has been reported that intrathecal administration of the mGlu1/5 agonist 3,5-dihydroxyphenylglycine (DHPG) induces spontaneous nociceptive behaviours, which are ERK-dependent. In the 3rd part of the project, ERK phosphorylation in mGlu5-expressing neurons following the administration of DHPG was investigated. Anaesthetized rats underwent a laminectomy procedure. DHPG or saline was applied to their exposed lumbar cord for 8 minutes after which they were perfused. Sections from the lumbar spinal cord were immunoreacted to reveal mGlu5, pERK and one of various markers for excitatory or inhibitory interneurons. Following DHPG (but not saline), numerous pERK-positive cells were seen in the SDH, particularly lamina II, and the great majority of these were mGlu5-positive. ERK phosphorylation was detected in both inhibitory and excitatory mGlu5-expressing cells, suggesting that type I mGlus have a complex role in nociceptive processing.

Modern day medicine is on the brink of a new age of therapy, which aims to harness the natural power of molecular biology for disease treatment. This therapy could include replacement of dysfunctional genes that cause disorders such as cystic fibrosis (Lommatzsch and Aris, 2009), or silencing the overexpression of genes that cause disorders such as cancer (Pelengaris and Khan, 2003). In both examples, the treatment of these genetic diseases lies in the delivery of synthetic nucleic acids into diseased cells, the former being called gene replacement therapy (Dobson, 2006a), and the latter being called RNA interference (RNAi) therapy (Whitehead et al., 2009). While these techniques have long been in use as genetic research tools for gene transfection or silencing in vitro, their translation for use in clinical disease treatment has yet to be achieved. The main problem facing the development of these novel therapies is the specific delivery of nucleic acids into diseased cells within the body. It is hoped that nanoparticles (NPs) can be used to overcome this problem, by acting as vehicles to transport nucleic acids through the body for specific delivery into diseased cells. This feat can be aided by the attachment of additional functional molecules such as cell penetrating peptides (CPPs), targeting peptides, additional drug types and molecules for imaging during treatment. Many different NP design strategies are currently under development. It is essential for new designs to be extensively tested for toxicity and efficiency in human cells before they can be successfully released into the clinic. As part of this effort, this PhD project has investigated two different NP design strategies for drug delivery: 1) the use of a magnetic field (MF) and a CPP to increase the delivery of iron oxide magnetic NPs (mNPs) to cells grown in tissueequivalent 3D collagen gels, and 2) gold NPs (AuNPs) for the delivery of siRNA to silence the c-myc oncogene for cancer treatment. In the first investigation, a MF and the CPP penetratin were found to increase mNP delivery to cells grown in 3D. In the second investigation, AuNPs were assessed in a range of different cell types (grown in 2D) for their performance in 4 main areas; cellular toxicity, cellular uptake, c-myc knockdown and effect on the cell cycle.

cAMP is a secondary messenger that is involved in a variety of signalling pathways through its effectors including EPAC, PKA and ion channels. cAMP signalling regulates processes such as memory, muscle contraction and inflammatory responses. PDE enzymes offer a mechanism to negatively regulate elevated cAMP levels elicited by activators of adenylyl cyclase. Studies have shown that cAMP signalling is compartmentalised through binding of PDEs to A-kinase anchoring proteins (AKAPs) that scaffold PKA regulatory subunits. In this study post-translational-modification of PDE4 isoforms is investigated. SUMOylation is a relatively newly identified post-translational modification that is known to regulate the structure and function of its substrates. PDE4 isoforms of the PDE4A and 4D subfamilies are SUMOylated by an E3 ligase, PIASy. SUMOylation alters the rolipram sensitivity and potentiates the PKA mediated activation of the isoforms whilst it confers protection from ERK-mediated inhibition of PDE4 activity. SUMOylation alters the association of PDE4 isoforms with binding partners like β-Arrestin, AKAP18 δ and UBC9. Rolipram is an archetypal PDE4 specific inhibitor. In this study it is shown that in cells expressing a GFP tagged form of PDE4A4 undergoes redistribution into accretion foci upon chronic treatment with rolipram. Data suggests that foci formation requires protein turnover and is regulated by signalling pathways such as PI3 kinase pathway, p38 MAP kinase pathway and PKC pathways. Further, the Immunomodulatory drug Thalidomide® also inhibits foci formation. PDE4 isoforms have isoforms specific N-terminal regions, which play a crucial role in sub-cellular localisation and protein-protein interactions. It is shown here that PDE4D5 interacts with a novel RhoGAP called ARHGAP21 which has been previously reported to bind β-arrestins. This interaction is independent of GAP activity of ARHGAP as well as PDE4 activity. Previous reports have indicated a role of β-Arrestin, PDE4 and ARHGAP21 in regulation of actin cytoskeleton dynamics. Hence complex β-Arrestin-PDE4-ARHGAP21 may play a crucial role in regulating actin dynamics.

Background: Rheumatoid arthritis (RA) and psoriatic arthritis (PsA) are two distinct forms of chronic auto-immunity; understanding the transcriptomic profiles of key leukocyte subsets implicated in these arthritides could improve the diagnosis and treatment of patients. Current microarray analyses of samples derived from RA and PsA patients have examined the genetic profiles of whole blood or diseased tissue which, although informative, can mask the genetic contributions of individual cell types. Monocytes and macrophages are a cellular subset known to play a major role in PsA and RA through the production of pro-inflammatory chemokines, cytokines and destructive proteinases. Aim: To define the transcriptome in CD14+ cells separated from the blood and synovial fluid of PsA and RA patients, and to then compare and contrast that signature in health and disease. Thereafter to define the relevant activities of selected novel moieties described in the foregoing analysis. Methods & Results: The transcriptomic profiles of healthy, RA and PsA CD14+ blood cells were remarkably similar - few genes could distinguish diseased from healthy CD14+ cells. Comparison of the genetic signature of the RA and PsA synovial fluid CD14+ cells revealed that just over 50% of the differentially expressed genes were shared between the two disease groups. Furthermore, analysing the canonical pathways in the synovial fluid cells compared to the matched peripheral blood of both patient groups surprisingly revealed Liver X receptor (LXR) activation pathway as the most significantly upregulated pathway: this pathway has been previously shown by our group to play a pro-inflammatory role in arthritis. Examination of specific upregulated mRNAs in the synovial fluid CD14+ cells from both disease types revealed two novel genes that had not previously been associated with arthritis, the lysosomal enzyme legumain and the cell surface molecule plexin A1. Legumain was demonstrated to be present in RA and PsA CD14+ cells by RNA and protein analysis and was physiologically active. Incubation of CD14+ cells with patient synovial fluid under hypoxic conditions also potentiated legumain expression. Plexin A1 was confirmed to be expressed at the mRNA level within RA synovium. siRNA knockdown of plexin A1 suggested that it may play a pro-inflammatory role within macrophages since subsequent treatment of these macrophages with LPS resulted in decreased TNFα production. However, investigations into the identity of the specific ligands for plexin A1 in arthritis, known as semaphorins, were inconclusive. I finally generated microarray data to evaluate the transcriptome of macrophages activated via cel contact with activated T cells. Such cells shared only a small percentage of genes with those dysregulated in the RA and PsA synovial fluid derived CD14+ cells suggesting that this model at the time points chosen may not be an appropriate in vitro representation of articular macrophages. An imaging system of this in vitro model was also established to visualise the dynamic nature of the T cell – macrophage interactions and demonstrated that variables such as duration or method of T cell activation could alter the number and duration of interactions between the two cell types. Conclusions: These studies demonstrate that the CD14+ cells isolated from the blood are similar transcriptomically between healthy controls and RA and PsA patients. The synovial fluid CD14+ cells from RA and PsA patients exhibit substantial overlap in terms of their genetic profile. Two novel molecules expressed by diseased patients namely plexin A1 and legumain have been identified and their preliminary characteristics in the context of synovitis have been defined.

Kinesins are molecular motors that use energy from ATP hydrolysis to transport cargoes along microtubule tracks. There are at least 14 families of kinesins with different structural organisations but all kinesins have a motor domain that is the catalytic core for ATP hydrolysis and the binding site for microtubules. Most kinesins have a stalk domain, which facilitates oligomerisation, and a tail domain that is implicated in cargo binding and regulation. Depending on their structural organisation, each kinesin is suited for different functions. Some are involved in transporting vesicles and organelles in cells, while others are essential for axonal transport in neurons. Still others are involved in intraflagellar transport in cilia. Lastly, a group of kinesins participate in different steps of mitosis. One such kinesin is the human mitotic kinesin Eg5. It is a homotetrameric kinesin that is made up of a dimer of anti-parallel dimers. By cross-linking anti-parallel microtubules and moving towards their plus ends, Eg5 slides them apart and establishes the bipolar spindle. When Eg5 is inhibited by antibodies or siRNA, cells arrest in mitosis with non-separated centrosomes and monoastral spindles. Prolonged mitotic arrest eventually leads to apoptotic cell death. For that reason, Eg5 is a potential target for drug development in cancer chemotherapy with seven inhibitors in Phase I and II clinical trials. The first inhibitor of Eg5 was discovered in a phenotype-based screen and is called monastrol. Since then, several classes of inhibitors, such as ispinesib (a clinical trial candidate) and Strityl- L-cysteine (STLC), have been discovered. To develop more potent inhibitors, we employed a structure-based drug design approach. By determining crystal structures of the Eg5 motor domain in complex with various inhibitors, we can understand the interactions between the inhibitor and Eg5; thus, analysis of the structure-activity relationship (SAR) can help us to improve their potency. Consequently, these inhibitors could complement or act as alternatives to taxanes and vinca alkaloids, which are successful cancer chemotherapeutics currently used in the clinic, but have the tendency to cause neurotoxicities and develop resistance in patients. Here, I report the crystal structures of Eg5 in complex with three monastrol analogues, STLC, and four STLC analogues separately. Based on the crystal structures with monastrol analogues, I identified the preferential binding mode of each inhibitor and the main reasons for increased potency: namely the better fit of the ligand and the addition of two fluorine atoms. Next, the crystal structure of Eg5-STLC indicates that the three phenyl rings in STLC are buried in a mainly hydrophobic region, while the cysteine moiety of STLC is solventexposed. In addition, structures of Eg5 in complex with STLC analogues, which have meta- or para-substituents on one or more of the phenyl rings, reveal the positions of the substituents and provide valuable information for the SAR study. In short, these structures reveal important interactions in the inhibitor-binding pocket that will aid development of more potent inhibitors. To understand the molecular mechanism of inhibition, I examined the structure of the Eg5-STLC complex, which revealed an unprecedented intermediate state, whereby local changes at the inhibitor-binding pocket have not propagated to structural changes at the switch II cluster and neck linker. This provides structural evidence for the sequence of drug-induced conformational changes. In addition, I performed isothermal titration calorimetry to determine the thermodynamic parameters of the interaction between Eg5 and its inhibitors. The structural information and the thermodynamic parameters obtained help us to gain a better understanding of the molecular mechanism of inhibition by an Eg5 inhibitor. While there is a large amount of information about the motor domain of Eg5, less is known about the stalk domain, which facilitates oligomerisation. A prediction program showed that the first ~100 residues of the stalk domain have a high probability of forming a coiled-coil structure, while the middle ~150 residues have a low probability. Using analytical ultracentrifugation, I showed that the Eg5 stalk364-520 domain exists predominantly as a dimer with a sedimentation coefficient of 1.76 S. The purported coiled-coil quaternary structure is backed-up by circular dichroism data, which showed that Eg5 stalk364-520 domain contains about 52 % helical content. Finally, the low resolution solution structure of Eg5 stalk364-520 domain was determined by small angle X-ray scattering, which revealed an elongated structure that is ~165 Å in length. Together, these data give us a glimpse into the structural characteristics of the Eg5 stalk364-520 domain. Besides gaining a better understanding of Eg5, I decided to investigate the molecular mechanism of autoinhibition in conventional kinesin (later known as kinesin-1). As the founding member of kinesins, it was first discovered to be involved in axonal transport. When not transporting cargo, kinesin-1 is autoinhibited to prevent squandering of ATP. Although it is widely accepted that the tail binds to the motor domain to keep it in a folded autoinhibited state, the molecular mechanism remains unclear and several mechanisms have been proposed. Here, I report the crystal structures of the Drosophila melanogaster kinesin-1 motor domain dimer and the dimer-tail complex. The dimer, which exhibits ~180° rotational symmetry between the monomers, provides valuable structural information for modeling the motility of kinesins on microtubules. By comparing the free dimer with the dimer-tail complex, we observe that the motor domains have considerable freedom of movement in the absence of tail binding. However, in the dimer-tail complex, a ‘double lockdown’ at both the neck coil and the tail interface freezes out major movements. This could prevent conformational changes, such as neck linker undocking. Data from our collaborator (David Hackney) showed that a covalent cross-link, which mimics double lockdown of the dimer, prevents ADP release. Together, we propose a ‘double lockdown’ mechanism, in which cross-linking at both the coiled-coil and tail interface prevents the movement of the motor domains that is needed to undock the neck linker and release ADP. In short, the structures shed light on the autoinhibition mechanism, reveal important residues at the dimer-tail interface, invalidate other proposed mechanisms, and open up the possibility that other kinesins may be regulated by the same mechanism.

This is a general investigation of the SPECT imaging process. The primary aim is to determine the manner in which the SPECT studies should be performed in order to maximise the relevant clinical information given the characteristics and limitations of the particular gamma camera imaging system used. Thus the first part of this thesis is concerned with an assessment of the performance characteristics of the SPECT system itself. This involves the measurement of the fundamental planar imaging properties of the camera, their stability with rotation, the ability of the camera to rotate in a perfect circle and the accuracy of the transfer of the information from the camera to the computing system. Following this the performance of the SPECT system as a whole is optimised. This is achieved by examining the fundamental aspects of the SPECT imaging process and by optimising the selection of the parameters chosen for the acquisition and reconstruction of the data. As an aid to this a novel mathematical construct is introduced. By taking the logarithm of the power spectrum of the normalised projection profile data the relationship between the signal power and the noise power in the detected data can be visualised. From a theoretical consideration of the available options the Butterworth filter is chosen for use because it provides the best combination of spatial frequency transfer characteristics and flexibility. The flexibility of the Butterworth filter is an important feature because it means that the form of the actual function used in the reconstruction of a transaxial section can be chosen with regard to the relationship between the signal and the noise in the data. A novel method is developed to match the filter to the projection data. This consists of the construction of a mean angular power spectrum from the set of projection profiles required for the reconstruction of the particular transaxial section in question. From this the spatial frequency at which the the signal becomes dominated by the noise is identified. The value which the Butterworth filter should take at this point can then be determined with regard to the requirements of the particular clinical investigation to be performed. The filter matching procedure can be extended to two dimensions in a practical manner by operating on the projection data after it has been filtered in the y direction. The efficacy of several methods to correct for the effects of scatter, attenuation and camera non-uniformity are also investigated. Having developed the optimised methodology for the acquisition and reconstruction of the SPECT data the results which are obtained are applied in the investigation of some specific clinical problems. The assessment of intractable epilepsy using 99mTc-HMPAO is performed followed by the investigation of ischaemic heart disease using 99mTc-MIBI and finally, the diagnosis of avascular necrosis of the femoral head using 99mTc-MDP is studied. The SPECT studies described in this thesis make a significant contribution to patient management.