Despite significant advances in our knowledge of the cellular and molecular elements of transplant immunology the 10 year survival probability for all human corneal grafts is 0.73. In some "high-risk" recipients it is as low as 0.37. To date almost all our knowledge about the cellular events during acute corneal graft rejection comes from animal models. In mice, the presence of pre-existing host corneal vascularisation confers "high-risk" status on a graft and has been shown to accelerate rejection. In the first part of this thesis the effect on survival of grafting to an inflamed conjunctival bed was investigated. Using a mouse model of allergic conjunctivitis significantly reduced survival was seen in graft recipients with perioperative conjunctival inflammation. This appeared to be due to the local effects of conjunctivitis rather than systemic effects of allergy/ atopy. Subsequent experiments investigated the effect of perioperative allergic conjunctivitis on the cellular components of both early (surgical trauma-induced, alloantigen-independent) and late (alloantigen-dependent; rejection) post-keratoplasty anterior segment inflammation and demonstrated significant effects on both. Grafts recipients with allergic conjunctivitis had significantly greater early post-operative corneal inflammation and associated corneal and conjunctival lymphangiogenesis. Analysis of graft infiltrating cells during rejection in mice confirmed that large numbers of CD4+ cells, CD8+ cells and macrophages were recruited. Flow cytometric analysis of human aqueous during acute endothelial rejection demonstrated for the first time the presence of CD4+ cells, CD8+ cells and a surprisingly high proportion of macrophages therein. In mouse recipients with allergic conjunctivitis eosinophils were found in both the graft itself and the ciliary body during rejection although the role of these cells during rejection is uncertain. Chemokine analysis during both murine and human corneal graft rejection demonstrated increased expression of the chemokine IP-10 (CXCL-10) suggesting a potentially important role for this protein in the rejection process.

This thesis describes the genetic analysis conducted to investigate the cause of six autosomal dominant macular dystrophies (North Carolina macular dystrophy, MCDR1; North Carolina-like macular dystrophy, MCDR3; North Carolina-like macular dystrophy with progressive sensorineural hearing loss, MCDR4; progressive bifocal chorioretinal atrophy, PBCRA; bull’s-eye maculopathy, MCDR2 and split-hand/foot malformations with associated North Carolina macular dystrophy, SHFM and NCMD) and one cone dysfunction with associated myopia and dichromacy (Bornholm eye disease, BED). The method of using Affymetrix SNP chips was tested for its usefulness in conducting genetic analysis on the macular disorders. The chips were used to see if disorders previously linked to large genomic regions could have their loci refined to help determine where the genetic error for each disorder lies. The genotyping results reveal the analysis is more informative when including genotyping data of affected offspring and their parents. MCDR1 analysis without such samples did not refine the disease locus. The copy number variation (CNV) analysis conducted was novel for the disorders and highlighted interesting regions of CNV, particularly in the SHFM and NCMD analysis at 5p15.33 for which QPCR confirmed loss of one copy of a novel microRNA. For MCDR2 analysis, new families were identified as carrying the mutation Arg373Cys in exon 10 of the PROM1 gene. Genetic analysis conducted on various families with BED has revealed the genetic cause of the disorder. Genetic changes leading to the amino acid combination of leucine, isoleucine/valine, alanine, valine and alanine at residues 153, 170, 174, 178 and 180, respectively, in the L or M opsin genes were consistently found in BED patients and are believed to result in the disease phenotype. Additionally, a family with a BEDrelated phenotype was found to carry a novel mutation in exon 2 of a hybrid M opsin gene that would lead to a Glu41Lys substitution.

Objectives: To determine the effects of diabetic eye disease (DED) on multiple aspects of visual function and self-reported visual ability. To conduct a randomised controlled trial (RCT) determining the effectiveness of low vision rehabilitation in those with DED. Methods: 100 participants with DED completed a comprehensive visual function assessment and completed the Activity Inventory (AI). The AI, scored using Rasch analysis, provided a measure of visual ability (logits). Univariate and multivariate regression examined the relationship between disease severity, visual function and visual ability. Participants were randomised to immediate (within 2 weeks of enrolment) or delayed (3 months after enrolment) intervention. The intervention was a hospital based low vision clinic appointment. The AI was repeated 3 and 6 months after enrolment. Primary outcome was the difference in visual ability between those receiving immediate intervention and those on a waiting list (no intervention control) 3 months after enrolment. Secondary outcomes were the difference in visual ability between those receiving immediate intervention and those receiving delayed intervention 6 months after enrolment and 3 months after intervention (delayed intervention control). Subgroup analyses examined whether severity, visual acuity or scotoma size influenced intervention outcome. Results: Disease severity and visual function were significantly associated with visual ability. Severity was not independently related to visual ability following adjustment for visual function. Stepwise regression revealed that a single measure of acuity explained 38% of the variance in visual ability (p < 0.001). No significant difference between intervention groups existed 3 or 6 months after enrolment or 3 months after intervention. Those with reduced acuity or central scotoma receiving delayed intervention improved between 0.42 and 0.56 logits more than those receiving immediate intervention 6 months after intervention (p = 0.02). Conclusions: Although explaining less than half the variance, a measure of acuity provided the best prediction of visual ability in DED. No overall effect of low vision rehabilitation was found. Exploratory subgroup analyses revealed visual ability improvements equivalent to between a 3 - 4 line acuity increase in those with reduced acuity or central scotoma receiving delayed intervention, indicating the need for further investigation of the effectiveness of low vision intervention in these patients.

This study involved the conversion of class II MHC negative tumour cells, via transduction with the potent class II MHC transcription factor, Class II Transactivator (CIITA) to enhance in de novo expression of MHC class II molecules. CIITA-transfected tumour cells may act as surrogate antigen presenting cells, processing and presenting endogenous tumour specific antigen in an MHC class II restricted manner to CD4+ T cells. CIITA has the major advantage over other cancer immunotherapy strategies, in that CIITA transfection of tumour cells evokes expression of all three MHC class II isotype; avoiding the need to tissue-type the recipient, a limit of previous MHC class I based strategies. MHC class I restricted strategies have proved largely unsuccessful in the past, due to weak CTL responses which have been unable to control tumour growth in vaccinated patients; thought to be due to the fact that MHC class II restricted CD4+ T cell help is required of optimal induction of both humoral and cellular immune responses. Although, CIITA-transfected tumour cells had been shown to be immunogenic, previous in vivo studies indicate that CIITA gene transfection failed to induce tumour rejection in syngeneic mice, due to up-regulation of invariant chain (Ii). Invariant chain is hypothesised to bind to the antigenic peptide-binding cleft of nascent Class II MHC molecules, thus preventing the binding of endogenous tumour antigens. The realisation of this work involved setting various techniques such as cell culture, RNA and protein extraction, RT-PCR, Western blotting, transfection, RNA interference, flow cytometry, immunofluorescence, immune -histochemistry, isolation of CD4+ and CD8+ T cells from lymph nodes, T cell proliferation, cytotoxic T cell and cytokine assays, as well as in vivo studies. Murine 4T1 mammary carcinoma cell line, a poorly immunogenic but highly tumourigenic tumour cell line, which closely mimics metastatic spread of human stage IV breast cancers in BALB/c mice, was genetically modified by transfection with CIITA. This resulted in up-regulation of MHC class II and co-expression of invariant chain, both at the mRNA and protein level. MHC class I expression was also slightly increased after this transfection. The novel technique of RNA interference was employed in this study to inhibit Ii chain expression, previously thought to render MHC class II based vaccine ineffective. However, CIITA-transfected 4T1 tumour cells were rejected in syngeneic BALB/c mice or induced slower tumour without the need to inhibit invariant chain activity. Tumour growth kinetics was found to be dependent on tumour load. Mice rejecting CIITA-transfected cells were found to be resistant to re-challenge with wild-type 4T1 tumour cells, indicating immunological memory. This rejection was also found to be tumour specific. T and B cells were implicated in tumour rejection in studies with BALB/c mice, and immunohistological studies demonstrated that mainly also confirmed by T cell proliferation and CD4+ and CD8+ T cells, as well as dendritic cells were involved in the anti-tumour immune response. The role of CD4+ and CD8+ T cells was also assessed via in vitro studies and this immune response was found to be mediated by Th1 cytokines. Traditional therapies for metastatic tumours, such as surgery, radiation or chemotherapy are invasive, often have quite drastic side-effects and are frequently not effective. However, these results open the possibility of utilising CIITA as a biological tool, utilising the body's own immune system to eradicate or prevent cancer.

Neurofibrillary tangles (NFT) in Alzheimer's disease (AD) are mainly composed of hyperphosphorylated and aggregated wild-type tau. NFTs are decorated by the ubiquitin-like modifier NEDD8, a protein targeted for proteasomal degradation by the NEDD8 Ultimate Buster 1 (NUB1). NUB1 has been shown to reduce synphilin-1 positive inclusions in a model of Parkinson’s disease. Therefore, this study examined the subcellular localisation of NUB1 as well as the effect of NUB1 on tau phosphorylation and aggregation. Furthermore, the effect of reducing NUB1 expression by RNA interference was investigated. Brain sections from AD patients showed that NUB1 and NEDD8 were expressed in the pyramidal neurons of the hippocampus, where the accumulation of NFTs is most abundant. In rat primary cortical neurons, NUB1 and tau co-localised in neurites and signalling structures such as varicosities, suggesting a functional interaction between them. The upregulation of the tau kinase GSK3β in AD leads to increased tau hyperphosphorylation and accumulation. In SK-N-SH neuroblastoma cells, which lack endogenous tau, ectopic wild-type tau formed inclusions when it was co-expressed with GSK3β, and this was enhanced by proteasome inhibition. NUB1 co-localised with both tau and GSK3β and significantly reduced tau inclusion formation. In neuroblastoma cells, NUB1 could interact with both tau and GSK3β, disrupt their interaction, and decrease the GSK3β-dependent phosphorylation of tau. NUB1 can directly bind synphilin-1 and induce its proteasomal degradation. Therefore, the ability of NUB1 to regulate GSK3β degradation was investigated in neuroblastoma cells. The upregulation of NUB1 accelerated the turnover of GSK3β, and the ubiquitin-associated (UBA) domains of NUB1 were necessary for NUB1 to exert its effect. Conversely, the downregulation of endogenous NUB1 by RNA interference increased the stability of endogenous GSK3β. Thus, NUB1 might have a role in tau inclusion formation by modulating GSK3β levels.

Transplantation of ex-vivo cultured limbal epithelial cells (LEC) is an established treatment for total limbal stem cell deficiency. However, this therapy has preceded the scientific understanding of limbal epithelial stem cells (LESC), the LESC niche and the biological mechanism that underpins it. This body of work tested the hypothesis that the LESC niche has a specialised structure and that an understanding of this may lead to improvements in outcomes and understanding of this therapy. The corneal limbus was investigated using 3D confocal microscopy, scanning electron microscopy, wholemount immunofluorescence and in-vivo confocal microscopy. This produced the most comprehensive study of limbal architecture performed to date. The structure of the LESC niche was identified and the effect of age and disease examined. Next, a clinical study of ex-vivo expansion and transplantation of LEC was performed in patients with LESC deficiency. A defined set of objective outcome measures enabled a baseline standard for outcomes of this therapy to be described. Future improvements to this therapy can be assessed using these techniques. Human amniotic membrane (HAM) is the leading candidate for a surrogate LESC niche. The method of HAM processing was investigated and found to be critical to achieve this. A method of significantly improving the yield of LESC in culture on HAM was identified. It remains to be seen whether this will translate into improved clinical outcomes. Finally, a disconnection exists between published data indicating a good clinical outcome and data demonstrating the survival of transplanted cells. This work assessed and demonstrated the feasibility of using quantum dot nanocrystal labelling of transplanted LESC to track cells post transplantation. It is concluded that translation of these findings to modify and improve current treatment protocols may result in improvements in outcomes for patients with LESC deficiency undergoing this therapy.

Background and Aims: Non-infectious posterior segment uveitis is a potentially blinding disease that usually affects people of working age group. Like other immune mediated diseases, uveitis is a complex polygenic disease. Several cytokines have been identified as important regulators of the immune system during, induction, progression and remission of ocular inflammation in uveitis. The work described in this thesis is based on the hypothesis that polymorphisms in chemokine and cytokine genes can predict clinical outcome in non-infectious uveitis. Methods: Functional polymorphisms in sixteen chemokine and cytokine genes were genotyped and there associations were studied in a cohort of British Caucasians suffering with non-infectious posterior segment uveitis. Results: This study has shown that polymorphisms in IL-18, IL-10 & CCR2 genes can influence the susceptibility to certain phenotypes of non-infectious uveitis. Polymorphisms in many genes particularly, IL-1 , IL-6, CCR5 & IL-18 are found to affect the visual outcome and severity of the disease. Conclusion: The identification of these genetic variants that add susceptibility or resistance to uveitis has provided us further insights into the pathogenesis of uveitis. This work will help us identify patients who are at a greater risk of losing sight with this disease. This, in turn would allow tailored aggressive therapy to be given at presentation when vision is still good with a precise aim to prevent significant amount of blindness.

High resolution and colour vision is derived from cone photoreceptors within the retina of the eye. These highly specialised neuronal cells convert energy from incident light into changes in cellular membrane potential. Action potentials are then relayed and processed within the inner retina and the cerebral cortex. Loss of function in the cone photoreceptors is the direct cause of visual loss for millions o f patients. Yet for many of the most common causes of blindness cone photoreceptor dysfunction occurs late on in the disease and is secondary to inherited and/or environmental influences that primarily affect other cell types. Cone photoreceptors are then secondarily affected and it is once their function is lost that the patient becomes visually disabled. The aim of this thesis was to evaluate therapies targeting the molecular steps of cone photoreceptor death rather than the underlying pathology. If effective at slowing cone photoreceptor degeneration and preserving function, such therapies could be applicable to large numbers of patients with a variety of underlying defects. There will however be continued stimuli for cell death as the primary defect has not been corrected and so it is important to determine the magnitude and duration of any treatment effect. The experiments of this thesis were performed in an animal model of inherited retinal degeneration derived to allow repeated in vivo assessments of cone photoreceptor function and survival. Gene therapy and intra-ocular injections were used to evaluate proteins with contrasting modes of action.

This thesis describes a programme of work designed to investigate the molecular pathways in the retinal response to hypoxia and to develop new therapeutic strategies for ischemic retinal disorders. The first hypothesis investigated was that ischaemia-induced angiogenesis can be controlled by targeting the oxygen-sensing pathway at the level of hypoxia-inducible transcription factor (HIF) translation. This was addressed by RNA interference (RNAi) in vitro and in murine oxygen-induced retinopathy (OIR). The second hypothesis investigated was that murine OIR offers a model of ischemic neurodegeneration in which to investigate endogenous mechanisms of neuroprotection and to develop novel therapeutic interventions. This was tested by evaluating the effect of OIR on the structure and function of the neuroretina, and by investigating the neuroprotective effect of erythropoietin in this model. The temporal and spatial distribution of retinal hypoxia in murine OIR was closely correlated with the distribution of HIF-α isoforms in the retina, with significant isoform-specificity in distinct cell populations. Synthetic siRNAs targeting HIF-1α and HIF-2α resulted in effective knockdown of RNA and protein in vitro with a high degree of isoform specificity. Efficient HIF-α protein knockdown in vivo was not achieved, however, despite optimisation of timing and transfection conditions. Retinal ischaemia in OIR was associated with significant dysfunction and focal degeneration of the inner neuroretina. Mild photoreceptor dysfunction was also detected. Erythropoietin deficiency in OIR resulted in an exaggerated retinal dysfunction that was reversed by exogenous supplementation. The results suggest that HIF-1α and HIF-2α have distinct roles in retinal oxygen sensing. RNAi-mediated HIF-α knockdown is effective in vitro but further development of delivery strategies is required to achieve effective knockdown in vivo. Murine OIR offers a valuable model of ischemic neurodegeneration in which to investigate endogenous responses and to evaluate novel therapeutic approaches. Endogenous expression of erythropoietin has a neuroprotective role in this model.

The experiments described in this thesis were designed in order to test the hypothesis that molecular chaperones are involved in the biogenesis of rhodopsin and may be used in the treatment of rhodopsin retinitis pigmentosa. Rhodopsin is the prototypical G-protein coupled receptor found at high concentration in the outer segments of rod photoreceptor cells. Rhodopsin is made up of the rod opsin apoprotein and 11-cis-retinal, the photoactive ligand. Rhodopsin initiates the phototransduction cascade under dim light conditions and mutations in its primary sequence have been linked to the neurodegenerative blinding disease, retinitis pigmentosa. Mutations such as P23H, cause the misfolding of the protein, resulting in its retention in the endoplasmic reticulum of heterologous expression systems and the inner segment of photoreceptor cells. Whilst selecting suitable modifiers, the subcellular compartments occupied by rhodopsin during its biogenesis and the chaperones resident in these, were considered. Calnexin is a central component of the quality control machinery in the endoplasmic reticulum. As calnexin has been widely documented to assist in the maturation of nascent glycoproteins, mouse embryonic fibroblast cells were used, which expressed a truncated version of calnexin, unable to bind client glycoproteins. The expression of rod opsin was compared in cells expressing truncated calnexin and in their wild-type counterparts, assessing the contribution of calnexin in the subcellular localization and biochemical profile of rod opsin. Calnexin was found to be dispensable for the maturation and folding of rod opsin. EDEM1, the ER-degradation enhancing mannosidase α-like 1 protein, has been shown to accelerate the degradation of misfolded glycoproteins, extracting these from futile folding attempts in the calnexin cycle. EDEM1 was found to enhance the degradation of P23H rod opsin and importantly, promoted the cell surface expression of any remaining P23H molecules which escaped degradation. The localization of EDEM1 in murine retina was determined to be within a subset of the inner segment and rhodopsin was found to form a physiological immune complex in porcine retina. The binding protein, BiP, associates with nascent proteins as these are translated and translocated in the ER lumen. A toxin that efficiently cleaves BiP in two fragments was used in order to probe the effects of BiP deletion on the biogenesis of wild-type and mutant rod opsin. Wild-type rod opsin was found retained in the endoplasmic reticulum in the absence of functional BiP and was misfolded as ubiquitin was recruited to the endoplasmic reticulum surface from a previous diffuse localization. Therefore BiP appears to be critical for maintaining rod opsin in a folding competent state. A chaperone on the cytoplasmic face of the endoplasmic reticulum, namely HSJ1b, has previously been shown to result in the stalling of wild-type and mutant rod opsin folding. We have investigated the effects of coexpressing CHIP, the carboxy-terminus of Hsp70-interacting protein. CHIP is an E3 ligase, which has been shown to present misfolded proteins to the proteasome for degradation, via an association with the Hsp70 machinery and HSJ1b initiates the process by stimulating ATP hydrolysis by Hsp70. In the presence of HSJ1b, expression of CHIP resulted in the degradation of rod opsin by the proteasome. Hence chaperones in the endoplasmic reticulum lumen and the cytoplasm can be used to manipulate mutant P23H rod opsin and may be used in the treatment of rhodopsin RP.