Immune-mediated adverse drug reactions are difficult to predict and can be severe in nature. Recently observed genetic associations highlight the importance of specific human leukocyte antigen alleles in the development of certain reactions. The mechanisms underlying antigen formation and subsequent T-cell activation require further investigation. The drugs abacavir (HLA-B*57:01) and ximelagatran (HLA-DRB*07:01 and HLA-DQA*02:01) represent compounds associated with skin and liver reactions respectively, for which a HLA association has been reported. In order to investigate the mechanism of HLA-restricted T-cell activation a cohort of 400 healthy volunteers was established. Both functional lymphocytes and DNA were isolated and stored. Following sequence-based HLA-typing twenty-six individuals expressing HLA-B*57:01 were identified (1 homozygote, 25 heterozygotes) and 101 individuals (10 homozygotes, 91 heterozygotes) expressing HLA-DRB*07:01 were identified. T-cells from these volunteers were utilised in in vitro assays. The first assays employed had low sensitivity and were unable to detect any drug-specific T-cells either by proliferation or cytokine secretion. Seventy-four CD8+ abacavir-specific T-cell clones however were generated from 3/3 volunteers expressing HLA-B*57:01. These clones secreted an array of cytokines and cytotoxic mediators (IFN-γ, Granzyme B, perforin, Fas ligand) in response to drug incubation. Chemically reactive metabolites are frequently associated with adverse drug reactions. The metabolism of abacavir in both liver and immune cell preparations was therefore assessed. In human liver cytosol, abacavir was metabolised to three isomeric carboxylic acids (48±15% turnover at 20h). This reaction proceeded via a reactive aldehyde metabolite that could be trapped with methoxylamine. Metabolism was blocked by the addition of 4-methylpyrazole, an alcohol dehydrogenase inhibitor (1000µM = 91.7±3.9% inhibition; p<0.05). Low-level carboxylic acid formation could be detected in an S9 fraction (2%) and cytosol (3.6%) generated from antigen-presenting cells suggesting that T-cell clones are exposed to a small amount of the aldehyde metabolite during cellular assays. The parent drug however is the predominant chemical entity present. Cross-reactivity at the MHC-TCR interface was determined with a number of abacavir analogues. The interaction was found to be highly specific with only deuterated abacavir stimulating T-cell clones at a similar level to abacavir. Dihydro abacavir only stimulated clones when used at high concentrations. The alternative enantiomer of abacavir (1S,4R) and carbovir did not stimulate any clones at any concentration investigated. This work further highlights the involvement of T-cells in drug hypersensitivity reactions and the importance of previously described HLA associations. In addition, chemical restriction exists at the HLA-TCR interface and is key to the activation of abacavir-specific T-cell clones. The metabolism of abacavir to a protein-reactive metabolite can occur in antigen-presenting cells therefore the potential of reactive intermediates to activate T-cells in a HLA-restricted manner should be investigated.

Drug-induced liver injury (DILI) is a serious adverse drug reaction (ADR) that is frequently encountered during drug development, representing a major cause of drug attrition. Furthermore, DILI is also a serious concern in the clinic, accounting for approximately half of all acute liver failure cases. Paracetamol overdose (acetaminophen; APAP) accounts for the majority of DILI-associated cases of ALF encountered in patients. The identification and development of novel biomarkers of DILI that are sensitive, specific, and rise early during hepatotoxicity are urgently required in the clinic and in the laboratory. Two liver-specific microRNAs (miRNAs) have recently been described that serve as sensitive and early markers of APAP-induced acute liver injury (APAP-induced ALI) in a mouse model. Together with the superior liver-enrichment of some liver-enriched miRNAs, these potential markers need to be assessed in patients for the clinical promise. Moreover, further work is warranted to test these potential markers in alternative pre-clinical models with other compounds to gain a better understanding regarding sensitivity of release, mechanism of release and circulatory kinetics. Furthermore, in separate work discussed in this thesis, the mammalian biological clock has been found to exert a powerful influence on the physiology of mammalian systems. This regulation hinges on the complex interplay between the clock genes and their products that oscillate over a twenty-four hour period and promote a diurnal variation in numerous output pathways. Emerging evidence suggests that the efficacy and toxicity of many drugs follow a diurnal rhythm and that this may be at least partly attributable to the clock-mediated regulation of drug targets and pathways of drug metabolism. APAP and FS represent two compounds that elicit hepatotoxicity in the mouse through two distinct mechanisms. Both APAP and FS are known to exhibit circadian variation in their toxicology and/or pharmacology. However, little is known about the molecular mechanisms that govern these differences in circadian variation. Two liver-enriched miRNAs (miR-122 and miR-192) were tested alongside serum ALT activity, the gold-standard marker of ALI, for sensitivity and time of release in a mouse model of APAP-induced ALI. At 2 hours after APAP administration, miR-122 (ΔΔCt 75.0, P=0.02) was significantly higher compared to controls (ΔΔCt 4.1) while ALT levels were in the normal range (21 U/L) indicating earlier release of miR-122. In a sensitivity study, miR-122 was not more sensitive than ALT at a 300 mg/kg dose of APAP compared to controls (mean values 300 mg/kg vs 0 mg/kg: ALT = 491 U/L vs 38.1 U/L; miR-122 = ΔΔCt 572.9 vs ΔΔCt 209.4). In patients, serum miR-122 and miR-192 were substantially higher in APAP-induced ALI patients, compared to healthy controls (median ΔΔCt miR-122: 1,265 versus 12.1, P < 0.0001; miR-192: 6.9 versus 0.44 P < 0.0001). A heart-enriched miR-1 showed no difference between APAP-ALI patients and controls, whereas miR-218 (brain-enriched) was slightly higher in the APAP-ALI cohort (ΔΔCt 0.17 versus ΔΔCt 0.07 P = 0.01). In a cohort comprised of patients who presented early (median time of presentation since APAP overdose = 8 hours), miR-122 was significantly raised in patients who develop ALI (> 3 x ULN serum ALT activity) compared to those that did not (median ΔΔCt 3.48 vs ΔΔCt 0.16, P<0.0001). In contrast, presentation ALT levels were not different between patients who developed ALI compared to those that did not (median ALT = 21 U/L vs 19 U/L). Moreover, miR-122 was significantly raised in patients who develop coagulopathy (INR > 1.3) compared to those that did not (ΔΔCt vs 3.48 vs ΔΔCt 0.17, P=0.0004). In contrast, presentation ALT levels were not different between patients who developed ALI iv compared to those that did not (median ALT = 21 U/L vs 19 U/L). In chronotoxicity studies, both APAP exhibited greater toxicity in the evening (mean ALT = 12785, 66% survival) compared to morning (mean ALT = 380, 100% survival) whereas FS showed greater toxicity after morning administration (mean ALT=561, 100% survival) compared to evening administration (mean ALT = 69.2, 100% survival). Circadian variation in APAP-induced ALI was associated with 38% lower (P=0.003) GSH levels and 20% higher (P=0.024) Cyp2e1 levels at 21:00h compared to 09:00. This work confirms that miR-122 is released earlier than ALT in a young mouse model of APAP-induced ALI. Furthermore, it is shown for the first time that circulating liver enriched miRNAs are higher in patients following an APAP overdose. Also, plasma miR-122 is raised at emergency room presentation when serum ALT activity is in the normal range. Further clinical development of blood-based miR-122 is warranted, this work suggests that miR-122 analysis at the point of hospital admission can predict risk of subsequent liver injury in patients. Finally, APAP and FS exhibit circadian variation in their toxicity in a mouse model associated with circadian variation with genes involved in drug metabolism and drug detoxification. Profiling of the hepatic proteome over the circadian phase is now warranted.

Drug hypersensitivity reactions are a significant cause of patient morbidity and mortality. They are difficult to predict in the clinic and during the drug development process. This is because mechanisms have not been fully elucidated. In fact, recent data using T-cells cloned from hypersensitive human patients have questioned the long-standing hapten concept, which states that a drug must bind irreversibly to protein to initiate a T-cell response. The work described herein was performed to investigate the chemical and cellular mechanisms of β-lactam-induced skin and liver injury and in particular to understand the relationship between drug-protein adduct formation and the activation of antigen-specific T-cells. Β-lactam antibiotics, such as piperacillin, provide the cornerstone of treatment and reduce the rate of decline in lung function in patients with cystic fibrosis, but use is limited by a high frequency of hypersensitivity reactions. Using the lymphocyte transformation test, drug-responsive lymphocytes were found in approximately 70% of clinically diagnosed piperacillin hypersensitive patients. By cloning over 400 antigen-specific CD4+, CD8+ and CD4+CD8+ T-cells, the T-cell proliferative response and cytokine secretion to piperacillin was shown to be concentration-dependent and highly drug-specific. Mass spectrometry revealed irreversible binding of piperacillin to selective lysine residues on albumin and a synthetic albumin conjugate stimulated piperacilin-specific clones via a processing-dependent pathway. These results describe the cellular processes that underlie piperacillin hypersensitivity. Approximately 20% of hypersensitive patients with cystic fibrosis develop multiple reactions that restrict therapeutic options. To explore the mechanistic basis of multiple β-lactam hypersensitivity, albumin binding profiles and T-cell responses against three commonly prescribed drugs; piperacillin, meropenem and aztreonam, were studied. PBMC responses were characterized using the lymphocyte transformation test and IFN-γ ELIspot. Clones were generated and found to proliferate and release cytokines following stimulation with all three drugs. However, crossreactivity with the different drugs was not observed. Each compound formed distinct haptens with lysine residues on albumin, which may explain the highly drug-specific T-cell response. These data indicate that multiple β-lactam reactions are instigated through priming naïve T-cells against the different drugs. The role of the adaptive immune system in reactions that target liver has not been defined. For flucloxacillin, a delay in the reaction onset and identification of HLA-B*57:01 as a susceptibility factor are indicative of an immune pathogenesis. Thus, flucloxacillin-responsive CD4+ and CD8+ T-cells were characterized from patients with liver injury. Clones expressed the gut-homing chemokine receptors CCR4 and CCR9 and secreted IFN-γ, Th2 cytokines and cytolytic molecules following drug stimulation. In contrast to the piperacillin clones, flucloxacillin clones were activated with several structurally-related β-lactam antibiotics. Furthermore, naïve CD8+ T-cells from volunteers expressing B*57:01 were activated with flucloxacillin when dendritic cells presented the drug antigen. Activation of CD8+ cells from patients and volunteers was processing-dependent and restricted by HLA-B*57:01, which effectively links the genetic association to the iatrogenic disease. In conclusion, the studies described herein provide novel insight into the way in which β-lactam antibiotics interact with protein and activate T-cells that are thought to be the ultimate mediators of drug hypersensitivity reactions in skin and liver.

Drug Induced Liver Injury (DILI) is associated with high morbidity and mortality rates. It is the leading cause of acute liver failure, accounting for 50% of all cases. Moreover, DILI is the most frequent cause of post marketing drug withdrawal and is often cited as a cause of compound attrition during the drug-development process. The High Mobility Group Box 1 (HMGB1) protein is an important inflammatory mediator which alters the immune system to tissue stress and injury. HMGB1 has been implicated in the pathogenesis of multiple inflammatory diseases including immune-mediated DILI. It has been identified as a potential biomarker of hepatic injury and a target for therapeutic intervention. Research is required to elucidate the pro-inflammatory role of HMGB1. HMGB1 has been reported to interact with a diverse range of endogenous (IL-1, DNA, nucleosomes, CXCL12) and exogenous (LPS) molecules to promote inflammation. However, the mechanisms responsible for these synergistic interactions remain poorly defined and therefore, the overall aim of this work was to characterise the interactions of HMGB1. Specifically, this work has focused on the interaction with IL-1β, which is of particular interest since both molecules often co-exist at the site of inflammation. The interaction between HMGB1 and IL-1β was investigated using combined cellular and Nuclear Magnetic Resonance (NMR) methodologies. LPS-free, isotopically labelled recombinant HMGB1 (full length protein, amino acids 1-215) and IL-1β proteins were expressed and purified from BL21 (DE3) cells. To facilitate these studies, three additional HMGB1 mutants were sub-cloned from the HMGB1 plasmid: Δ30 (1-185), A box (1-85) and B box (89-163). The recombinant proteins were characterised using Mass Spectrometry (MS) and NMR spectroscopy. Synovial fibroblasts were isolated from synovial tissue obtained from rheumatoid arthritis patients undergoing joint replacement surgery. Cells were treated with HMGB1 (full length, Δ30, A box or B box) alone or combination with IL-1β. Cell supernatants were collected after 24hr and IL-6 levels were quantified by ELISA. Untreated fibroblasts or cells treated with any HMGB1 construct, or IL-1β alone had no detectable IL-6 release (<9.375pg/mL). In contrast, full length HMGB1, the Δ30 and the B box domain (but not the A box domain) all acted in synergy with IL-1β to substantially enhance IL-6 production. In one patient, HMGB1, Δ30 and the B box in combination with IL-1β induced IL-6 levels of 28,473 ± 127pg/mL, 18,491 ± 2388pg/mL and 18,710 ± 2792pg/mL, respectively. The synergistic interaction was mediated via the Interleukin-1 receptor (IL-1R) and could be inhibited when the cells were pre-treated with 5µg/mL anakinra, a selective IL-1R antagonist but not detoxified LPS, a TLR4 receptor antagonist. To investigate if there is a direct interaction between HMGB1 and IL-1β, a comprehensive biophysical analysis was performed using NMR methodologies. However, despite performing the experiments in various ways, no evidence of a direct interaction between IL-1β and either full length HMGB1, Δ30 or the B box was detected. This suggests that the synergistic interaction between HMGB1 and IL-1β is mediated via an alternative cellular mechanism in which HMGB1 is required. In conclusion, the work presented in this thesis has identified that the B box domain of HMGB1 is critical for the synergistic effect observed with IL-1β. However, this is not due to the formation of a binary complex between HMGB1 and IL-1β. Instead, it would appear that the synergistic effect is mediated via an alternative cellular mechanism in which HMGB1 is required and additional proteins are involved. Future work could focus on discovering what these other proteins might be. These findings help to elucidate the pro-inflammatory role of HMGB1 and provide a novel an insight into HMGB1 biology.

The non-nucleoside reverse transcriptase inhibitor nevirapine (NVP) is widely used in the treatment of human immunodeficiency virus infection in many sub- Saharan countries. However, NVP-treated individuals have a 5% risk of developing hypersensitivity reactions (HSRs), ranging from maculopapular exanthema to life-threatening severe cutaneous eruptions and hepatotoxicity. A number of clinical (gender and CD4+ T cell count) as well as immunogenetic factors (MHC class I and class II alleles) have been associated with NVP hypersensitivity, but the predictive value is poor. Gene expression analyses provide an unbiased method for determining which immune-related mechanisms are involved in the pathogenesis of NVP hypersensitivity. mRNA microarray analysis demonstrated that expression of CD177, a neutrophil specific antigen, was significantly increased in our cohort of NVP-treated patients from Malawi (p ≤ 0.001), but also after in vitro treatment of NVP-hypersensitive patients from Liverpool (p < 0.05). Interestingly, CD177 protein expression did not increase in patient samples treated with NVP in vitro. A case-control study (n = 288) of NVP-hypersensitive and tolerant patients showed that none of the investigated CD177 polymorphisms were associated with NVP-induced HSRs. Of the clinical factors analysed, only CD4+ T cell count was significantly associated with NVP hypersensitivity (p < 0.001). The polymorphisms rs45441892 and rs10425385 in CD177 have previously been associated with an increased proportion of CD177-positive granulocytes. We confirmed these results in our NVP-naïve, healthy volunteer cohort (n = 35) for rs10425385, but not rs45441892. Previous reports have identified HLA-C*04:01 as a susceptibility marker for NVP-induced HSRs in patients from Malawi. Expression levels of miR-148a, a microRNA known to downregulate HLA-C expression, were analysed in serum samples from NVP-hypersensitive and tolerant individuals (n = 96). However, paradoxically, a statistically significant increase in miR-148a expression was found in NVP-hypersensitive patients at the time of reaction (p = 0.008). Additionally, the serum expression levels of 84 miRNAs were analysed in hypersensitive and tolerant patients treated with NVP (n = 24). Twenty-one miRNAs were differentially expressed in tolerant and hypersensitive samples. Of these miRNAs, miR-205 showed the highest increase in NVP-hypersensitive patients (p < 0.01). Besides HLA-C*04, several other HLA-alleles have been reported as risk factors for NVP hypersensitivity. A summary meta-analysis of published data indicated that four HLA-allelotypes (HLA-B*35, -B*58:01, -C*04, -DRB1*01) might be common risk factors of NVP-induced HSRs for different ethnicities. The results presented in this thesis highlight that various genetic, immunological and clinical factors may contribute to the pathogenesis of NVP hypersensitivity. Further understanding of the complex interactions may ultimately lead to the characterisation of true causative associations facilitating the precise prediction of hypersensitivity reactions to NVP.

Regulator of calcineurin 1 (RCAN1) has been shown to act as a negative regulator of vascular endothelial growth factor (VEGF)-signalling in endothelial cells. Two isoforms are detectable in cells, RCAN1.1 and RCAN1.4, produced by alternative splicing of the RCAN1 mRNA. In this study it was demonstrated that only RCAN1.4 is induced in human dermal microvascular endothelial cells (HDMECs) in response to VEGF. Using siRNA-mediated gene silencing this work shows that RCAN1 depletion leads to a reduction in VEGFR-2 internalisation following VEGF-A stimulation. RCAN1 depletion also leads to a reduction in cell polarity and cytoskeletal reorganisation in response to VEGF. siRNA-mediated silencing of RCAN1 led to an inhibition of VEGF-A mediated migration of HDMECs. These effects of RCAN1 knockdown appear to be specific to VEGFR-2, as no apparent effect on hepatocyte growth factor receptor (HGFR) internalisation and cytoskeletal reorganisation and migration in response to HGF was observed in HDMECs. Over- expression of RCAN1.4 isoform using adenoviral mediated gene delivery resulted in increased migration of HDMECs in the absence of ligand. This effect was insensitive to the calcineurin inhibitor cyclosporine, indicating that RCAN1.4 was not operating through the classical calcineurin/NFAT pathway to regulate cell migration. Instead, the RCAN1.4 effect was sensitive to a small molecule VEGFR-2 kinase inhibitor, which blocked cell migration. This study also examined the role of phospholipase D (PLD) in endothelial cell function. By utilising siRNA for PLD1 and PLD2 it was shown that both PLD1 and PLD2 are required for VEGF-A mediated proliferation, migration and tubular morphogenesis in HDMECs. Overall, the data presented in this thesis defines a novel role for both RCAN1 and PLD in regulating in endothelial cell function. Both proteins could be potential therapeutic targets to regulate vascular function.

Nrf2 is a transcription factor that, under conditions of chemical stress, is able to evade its cytosolic repression and translocate to the nucleus to initiate the transcription of a battery of cytoprotective genes, such as those involved in the detoxication of xenobiotics. Nrf2 has previously been shown to afford protection against chronic and acute renal injury, yet, relatively little is known about the mechanism by which Nrf2 affords this protection, and the extent of its transcriptional roles in the kidney. This thesis seeks to further our understanding of the physiological, pharmacological and toxicological roles of Nrf2 in the kidney. Using an iTRAQ-based proteomic approach to quantify protein expression levels in the kidneys of Nrf2+/+ and Nrf2-/- mice, acutely treated with vehicle or the potent Nrf2 inducer CDDO-Me (3 mg/kg), we demonstrated that 189 proteins were differentially expressed in the Nrf2-/- mouse kidney, compared to Nrf2+/+, and 42 proteins were differentially expressed in the CDDO-Me treated Nrf2+/+ mouse kidney, compared to vehicle. The key finding was that the kidneys of Nrf2-/- mice are deficient in proteins that mediate cellular redox balance, the metabolism of a range of xenobiotics, and the regulation of core metabolic processes, including energy metabolism and the synthesis and recycling of amino acids. Functional demonstration of a reduction in energy metabolism was demonstrated by assessing total NADPH and GSH, of which Nrf2-/- mouse kidneys had 35% and 30% less than their Nrf2+/+ counterparts, respectively. A single acute dose of CDDO-Me failed to augment the expression of proteins, other than Nqo1, that were shown to be regulated by Nrf2 at the basal level in the mouse kidney, however qPCR analysis of these kidneys revealed that CDDO-Me has an effect at the transcriptional level which has not fully translated within the timeframe of this study. In summary, we have provided evidence that Nrf2 regulates the expression of an array of proteins that contribute to cell defence and the maintenance of homeostasis in the kidney, supporting current interest in Nrf2 as a novel therapeutic target in a number of renal diseases. MicroRNAs are a recently discovered RNA-regulatory element that show promise in their use as biomarkers of physiological and pathological events. In order to provide insight into the microRNAs under Nrf2 control in the kidney, we performed an unbiased microRNA array analysis on kidney homogenates from Nrf2+/+ and Nrf2-/- mice, treated with vehicle or CDDO-Me, and then validated several promising microRNA candidates using targeted qPCR analysis. Of particular note are miR-466h-3p, the expression of which was significantly increased in the CDDO-Me treated Nrf2+/+ mouse kidney and decreased in the Nrf2-/- mouse kidney, compared to their respective controls, and miR-28c and 144, which were both significantly decreased in the CDDO-Me treated Nrf2+/+ mouse kidney, and increased in the Nrf2-/- mouse kidney. This novel analysis represents the first step in characterising the renal Nrf2 microRNA-ome, which could reveal novel mechanisms of Nrf2 function and markers of its activity that could translate to the clinic. Recent interest in the use of CDDO-Me as a therapeutic intervention for late-stage chronic kidney disease has culminated in a phase III clinical trial (BEACON), which was subsequently terminated due to unforeseen adverse cardiac events, of which the cause has yet to be identified. In order to determine whether the drive to produce more potent Nrf2 inducers has inadvertently led to the generation of inherently more toxic compounds, the relationship between potency towards Nrf2 and toxicity was evaluated for CDDO-Me and related triterpenoids, and other classes of Nrf2 inducer. Using a rat H4IIE-ARE8L luciferase reporter cell line to determine in vitro therapeutic indices, it was discovered that within the compounds tested an increase in potency toward Nrf2 of four magnitudes results was associated with an increase in toxicity of only two magnitudes, resulting in a relative increase in in vitro safety. This data indicates that it is possible to generate potent Nrf2-inducers that are not inherently toxic, and suggests that therapeutic targeting of Nrf2 continues to hold promise as a novel treatment for a range of diseases. In summary, by using a proteomic approach we have identified an array of renal Nrf2-regulated proteins that contribute to various cytoprotective and metabolic processes in the kidney, supporting current interest in the therapeutic targeting of Nrf2 as treatment for renal disease. Additionally, the microRNAs under Nrf2 regulation in the kidney have also been identified, and represent the first step in fully characterising the Nrf2 microRNA-ome. Finally, it was shown that the drive to produce more potent Nrf2 inducers has not led to the generation of inherently more toxic compounds; indeed an increase in potency is associated with a relative increase in in vitro safety, suggesting that the targeting of Nrf2 is still a promising therapeutic route. Overall, the work presented in this thesis has furthered our understanding of the physiological, pharmacological and toxicological roles of Nrf2 in the kidney.

Drug-induced liver injury (DILI) is a major cause of attrition during drug development, and also a leading cause of hospital admissions. In order to identify novel biomarkers that may improve the diagnosis and treatment of DILI, well characterised hepatotoxins such as acetaminophen (APAP) are used. High mobility group box 1 (HMGB1) protein is a ubiquitously expressed protein involved in the facilitation of gene transcription. However, extracellular HMGB1 acts as a key mediator of inflammation. During APAP-induced liver injury (AILI), HMGB1 is passively released from necrotic hepatocytes in a fully reduced, hypoacetylated isoform, or actively released from immune cells following acetylation of HMGB1 at key lysine residues within nuclear localisation sequences. Extracellular HMGB1 can bind to a range of receptors including TLR2, TLR4, TLR9, CXCR4 and RAGE to mediate cytokine-inducing or chemotactic effects and has been shown to be a more sensitive and predictive serum biomarker of AILI outcome than ALT. Additionally, anti-HMGB1 antibodies have been used to successfully attenuate injury in animal models of inflammatory conditions, including DILI. Mechanistic understanding of HMGB1 regulation and function is essential for this emerging biomarker of AILI. A murine monoclonal anti-HMGB1 antibody (2G7) has previously been used to attenuate AILI. A chimeric anti-HMGB1 monoclonal antibody (h2G7) was investigated, alongside 2G7 and a human isotype control antibody (E2) in an experimental model of AILI. C57BL/6 mice dosed with 530mg/kg APAP by intraperitoneal (i.p.) injection, followed by antibody or PBS treatment i.p. at 2hr post-APAP and euthanisation at 10hr post-APAP showed that 300µg h2G7 treatment afforded protection from AILI by significantly attenuating serum ALT, miR-122 and chemo/cytokine increases seen in the isotype control group. Liver histology and MPO revealed reduced neutrophil infiltration and activity in anti-HMGB1 treated animals. No significant differences were seen between 2G7 and h2G7 animals, and there was no effect on hepatic GSH in antibody-treated animals. Mechanism of action studies using Fc region modified forms of h2G7 ruled out complement-mediated and FcγR-mediated effects of h2G7 in the AILI model, as these did not significantly affect the h2G7 effect on AILI. Redox modulation of HMGB1 function had been proposed as a mechanism of regulation of proinflammatory activity of HMGB1. Investigations into the cytokine-inducing capability of different redox isoforms of HMGB1 in THP-1 and RAW264.7 cells demonstrated that HMGB1 needs to be in a Cys23-Cys45 disulphide conformation with a reduced thiol at Cys106 in order to induce cytokine release via NF-κB activation. During apoptosis, HMGB1 can be oxidised by activated caspases. Oxidation has previously been shown to regulate the structure of proteins and facilitate other post-translational modifications (PTMs). LC-MS/MS characterisation of HMGB1 from APAP overdose patients with high levels of apoptosis revealed ubiquitination at Lys30 (HMGB1-Ub). Ubiquitination was shown in vitro not to direct HMGB1 for proteasomal degradation, but HMGB1-Ub was detectable in apoptotic microparticles (MP), membrane vesicles that are shed from a cell during late apoptosis. In summary, HMGB1 was shown to be critical for the coordination of AILI, as antibody-mediated inhibition lead to a reduction in hepatic injury. Chimeric anti-HMGB1 antibody was effective at attenuating liver injury and advances anti-HMGB1 antibody treatment towards clinical utility. HMGB1 redox regulation of inflammatory signalling and the identification of HMGB1-Ub demonstrate the importance of PTMs in governing HMGB1 function. HMGB1 shows promise as an acute, sensitive and predictive biomarker of AILI in man. Through the use of in vitro and in vivo systems, the work presented in this thesis expands our current understanding of the mechanistic and functional roles of HMGB1 in DILI and has further defined mechanisms that can be used to inform the clinical situation.

Epilepsy is one of the most common chronic neurological disorders, and the symptomatic form is characterised by the occurrence of spontaneous recurrent seizures following a neurological insult. One third of epilepsy patients are resistant to antiepileptic drugs (AEDs), therefore the development of novel treatments is required. Glutamate is implicated in epilepsy however, only a few glutamate receptor antagonists have been successful in epilepsy trials. An indirect means of modifying glutamate-mediated excitation, such as targeting nitric oxide (NO), might be a reasonable alternative approach. Selectively targeting NO signalling pathway by two drug interventions was investigated in this thesis. Post-synaptic density 95 blocking peptide (PSD95BP) is a protein that uncouples GluN2 subunits of the N-methyl-D-aspartate (NMDA) subtype of glutamate receptor from neuronal NO synthase to prevent downstream neurotoxic signalling of NO. In contrast, 1400W is a highly specific inducible NO synthase inhibitor that binds to the guanidine recognition site of the enzyme, competing with L-arginine and preventing neurotoxic production of NO. These compounds are effective in animal models of stroke and were tested here in a C57BL/6J mouse model of epileptogenesis. Kainic acid (KA) is widely used to induce status epilepticus (SE) in animals and the resulting neuropathology mimics that seen in humans with temporal lobe epilepsy. Repeated low dose administration of KA via intraperitoneal injection every 30 minutes consistently induced generalised seizures but was associated with inter-animal variability in KA sensitivity, acute seizure severity and mortality rate. Extra-dural telemetry electrodes were implanted in mice for electroencephalography (EEG) recordings. Two algorithms, measuring spike frequency and EEG coastline respectively, were used to quantify epileptiform activity. Mice that received drug interventions following KA-induced SE had significantly lower mean spike frequency and fewer extended coastline epochs per day than the control group at both 7 and 14 days after the initial insult. Label-free proteomics quantification showed significant changes to the hippocampal protein profile as a result of both PSD95BP and 1400W administration following KA-induced seizures. Drug treatment, singly or in combination, also reversed the effects of KA on the expression of both transforming growth factor β1 and inwardly rectifying potassium channel 4.1 in the hippocampus. Surgical implantation of extra-dural electrodes significantly lowered the seizure threshold to KA and was associated with an increase in brain expression of pro-inflammatory cytokines, suggesting that careful consideration is required in studies involving intracranial surgery to assess epileptogenesis or AED effects. Current research investigating novel therapeutic agents is focusing on non-NMDA glutamate receptors, such as α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid and kainate receptors, as potential AED targets. There are no compounds targeting the NO signalling pathway currently being investigated for the treatment of epilepsy. The data reported in this thesis give cause for optimism but further studies are needed to fully investigate the antiepileptic and possible antiepileptogenic properties of PSD95BP and 1400W and true efficacy will ultimately require clinical evaluation.

The suitability of the electrically stimulated guinea-pig MP-LM preparation, mouse isolated vas deferens and urinary bladder for the study of cannabinoid receptor ligands was investigated. Cannabinoid receptor agonists produced concentration-related inhibition of the contractile responses in all three tissue preparations, demonstrating high potency, chemical- and stereo-selectivity. The rank order of the inhibitory potency of the cannabinoid agonists in all three tissue preparations correlated with their binding affinity for specific cannabinoid CB1 binding sites in rat brain tissue. These results suggested a receptor-mediated mechanism of action for cannabinoid receptor agonists via prejunctional functional cannabinoid CB1 receptors in these three models, in the absence of an antagonist. The endogenous cannabinoid receptor ligand anandamide, also produced concentration-related inhibitory effects in all three tissue preparations. However, anandamide was found to be metabolically less stable in the guinea-pig MP-LM preparation. SR141716A, a potent, CB1 selective cannabinoid receptor antagonist was found to attenuate the inhibitory effects of cannabinoid receptor agonists investigated in all three tissue preparations. This provided further evidence for a receptor-mediated mechanism of action for cannabinoid receptor ligands in these three tissue preparations. However, further studies with SR141716A suggests that, it may be acting as an inverse agonist rather than a pure antagonist in these three preparations. Finally, this study was further extended to characterise some novel cannabinoid receptor ligands in the guinea-pig ML-LM preparation and mouse isolated vas deferens.