Search

Analytical techniques have been developed to measure the very small [230ThP32Th] atomic ratios (3-8x10-6) typical of basaltic lavas using a single focussing mass spectrometer. Compared to conventional alpha-counting methods, the technique presented typically gives a 2-5 fold improvement in reproducibility of 230 Th/232Th ratios, and furthermore uses -5x less sample, has a simpler chemical prepartion and a data aquisition time -100x shorter. The above technique was used in investigating a suite of historic (< 1000a) basanites from La Palma, Canaries. The lavas have undergone significant post melt segregration processes, involving residence in deep (>4kbar) magma chambers, that are inferred to persist for at least 30ka. The basanites show geochemical variations related to differences in the degree of partial melting, that are estimated to be between 0.3-2%. Rare earth element (REE) modelling suggests clinopyroxene, rather than garnet control during melting, contrary to expectations from recent geophysical models (McKenzie & O'Nions, in press). The La Palma mantle source is inferred to be LREE enriched, K& Rb depleted, and shows many incompatible element ratios characteristic of 'himu' basalts (Weaver 1991). In contrast, the 206pb/204pb ratios of the La Palma basanites (19.6±0.05) are lower than the isotopically defined'high p' basalts (Zindler & Hart 1986) which suggests that the La Palma source is geochemically similar, but younger than a typical 'high t' source. A suite of young (< 10,000a) high MgO tholeiites from Iceland show order of magnitude variations in highly incompatible element concentrations that correlate with FeO contents. Variations within these lavas is primarily attributable to sampling of instantaneous melts from different depths within a dynamic melting column. However, additional correlations of radiogenic isotope ratios with major and trace element contents highlights mixing of various plume melts with some 20% mid-ocean ridge basalt, derived from melting of upper mantle material entrained during plume upwelling.

A panel of monoclonal anti-idiotype antibodies specific for cell surface immunoglobulin (Ig) of the L₂C leukaemia of guinea pigs has been investigated. Equilibrium constants were established for reaction of the antibodies with surface Ig in situ. They ranged between 10⁸ and 10⁹M⁻¹ for intact IgG antibodies, and 10' and 108M-1 for their Fab'γ fragments. Values at O^u and 37^u were similar. All antibodies were shown to bind bivalently at equilibrium but showed striking differences in their ability to clear Ig from the L_2C cell surface by antigenic modulation in vitro. Differences in the ability to induce antigenic modulation correlated with the readiness of different antibodies to cross-link neighbouring Ig molecules and evidence is presented for the existence of two distinct types of anti-idiotype antibody: 1) those which form intra-Ig bridges and do not favour modulation (monogamous binders), 2) those which form inter-Ig bridges and favour modulation (bigamous binders). Competitive binding assays showed that simultaneous access of more than one anti-idiotype antibody is generally not permitted. This approach therefore revealed no gross spatial separation of the target determinants of monogamously and bigamously binding antibodies. Fab' arms from two monoclonal antibodies have been used in the preparation of univalent antibodies (FabFc and FabIgG) where Fab' is joined to xenogeneic effector (IgG or Fc) via tandem thioether bonds. The effector capabilities (complement cytotoxicity and antibody dependent cellular cytotoxicity) of a number of chimeric univalent antibodies sharing the same monoclonal Fab' arm but with various effector moieties were investigated and shown to differ according to the species donating the Fc or IgG moiety. In general, those univalent antibodies which were efficient recruiters of effectors in vitro gave the best protection against L_2C leukaemia in therapeutic trials, though by 1) using univalent antibodies which were poor activators of complement and 2) running trials in congenitally C3 deficient guinea pigs, the role of complement in tumour combat would seem to be relatively unimportant. By preparing univalent antibodies which share the same effector region by carry Fab's of different affinity, the role of affinity in therapeutic efficacy was shown to be minimal, although the available range of affinities was narrow.

The aim of this project was to develop a new microdevice to allow precise control of cellcell interactions, and to apply it to measure T cell signalling in arrays of single cells in realtime after stimulation by live antigen presenting cells. The specific T cell responses which underlie effective immunotherapy of cancer are mediated via cell-cell interactions and the formation of immune synapses. However, methods to study these interactions are limited by lack of control over single cell pairing and analysis methods. This project describes a new approach to track the activation profile of cytotoxic T cells against tumour cells in a high-throughput manner. A simple yet effective mechanism to pair the single T cells with antigen presenting cells was optimized to study cell-cell contact in a time-controlled manner. The cell-trap array is an open system consisting of thousands of microwells cast in an agarose block, which is biocompatible and permeable to nutrients allowing functional observations on viable cells to be carried out over a number of hours. T cells can be singularly isolated in the wells via passive sedimentation and size exclusion, achieving up to 90% occupancy. Activation profiles of thousands of single CD8+ cells could be screened in the same field of view using calcium-chelating fluorophore dyes to measure the early stages of T cell signalling. Custom software was developed to process experimental data and allow rapid detection of individual cells with the strongest activation signals defined by several different criteria. This capability could be used in future to identify and characterise single T-cells that have especially high propensity to be activated by immunotherapy such as vaccination or checkpoint blockade, in preclinical animal models and biopsy samples from human cancer patients.

MHC class I molecules on the cell surface present endogenous peptide to the adaptive immune system. The mature complex is trimeric, consisting of an α-chain encoding membrane-distal peptide binding and membrane-bound α3 domains, an invariant β2m protein, and a peptide. The α-chain is highly polymorphic, with the six alleles expressed by an individual capable of presenting all potential antigens due to promiscuous peptide binding. In addition to presenting antigen, MHC class I has a role in selecting high-affinity peptides during complex maturation within the endoplasmic reticulum. There is a large pool of available peptides, but only highaffinity interactions are capable of eliciting an immune response. Whilst the structure of the mature complex has been well documented by X-ray crystallography, the process of peptide optimisation remains unclear. It is apparent that a conformational change within the peptide binding domain is required to exchange low-affinity for high-affinity antigen. Molecular dynamics simulations have been employed to model the dynamics of MHC class I but have not yet been verified by direct observation of such motion. This thesis investigates the dynamics of MHC class I using nuclear magnetic resonance (NMR) spectroscopy, focusing on the possible existence of alternative conformations within the peptide binding domain that permit peptide exchange. Protein production methods have been established for reliable generation of mature MHC class I, consisting of the murine H-2Db α-chain allele, human β2m, and the high affinity FAPGNYPAL peptide. Following assignment of backbone resonances, NMR analyses include residual dipolar coupling data revealing the potential for up to a 20° twist between the peptide binding and membrane-proximal domains. Heteronuclear relaxation data shows that MHC class I is inherently dynamic, experiencing multiple conformations in regions throughout the secondary structure. Mobility is largely located within the peptide binding domain and provides confirmation of predictions from molecular dynamics. This thesis confirms that MHC class I is capable of forming an intermediate conformation during peptide binding, during which it is peptide-receptive.

Ankylosing Spondylitis (AS) is an autoimmune condition of the sacro-iliac region at the base of the spine. The pathogenesis of the disease is unknown despite strong genetic links with HLA-B27 and ERAP1. HLA-B27 is present in ~95% of AS patients and polymorphisms within ERAP1 have been shown to distinguish AS cases from controls. ERAP1 is an ER-resident aminopeptidase that trims peptides for binding to MHC Class I molecules, including HLA-B27, in antigen presentation to CD8+ T cells. No specific peptides have been found to induce disease and the disease is unlikely to be CD8+ T cell driven, suggesting that changes in the whole peptide repertoire may instead be important for disease generation through induction of other cell types. Natural Killer cells also interact with MHC Class I molecules through Killer immunoglobulin like receptors and have been previously implicated in AS, with AS patients having higher percentages of natural killer cells in the peripheral blood than controls. Here I show that the sequence of the peptide bound to HLA-B27 can affect NK cell activation and begin to create a panel of amino acids at key positions that are activating or inhibitory. I have shown that ASassociated ERAP1 allotype combinations induce more NK cell activation than controls due to a reduction in cell surface FHC expression. Further to this I show how these changes in NK cell activation can alter the cytokine profile of the surrounding PBMC population. In addition I have looked into how different alleles of HLA-B27 are associated with disease and show that peptides bound to HLA-B*2709 can inhibit NK cell activation in an NK cell dependent manner, while peptide-B*2705 complexes bind NK cells in an ERAP1- independent manner. This work taken together provides some interesting possibilities to explain the events that lead to AS disease.

Monoclonal antibody (mAb) immunotherapy has proven effective in the treatment of haematological malignancies. Antibodies towards unique epitopes within the same antigen can engage different effector mechanisms to facilitate cell clearance. Understanding what drives the engagement of these mechanisms is important for the development of new therapeutics. This thesis investigated the role of the antigen, in particular the epitope bound by a mAb in defining the effector mechanisms engaged. Specifically, the role of distance between the epitope and the target cell membrane in relation to the engagement of the effector mechanisms; complement dependent cytotoxicity (CDC), antibody dependent cellular phagocytosis (ADCP) and antibody dependent cellular cytotoxicity (ADCC) were assessed. A panel of model antigens were generated; incorporating the same epitope for a clinically relevant mAb attached to the N-terminus of various CD137 constructs. Extracellular domains of CD137 were removed or added in order to change the distance of the clinically relevant epitope from the cell membrane. These constructs were transfected into CHO-S and A20 cells and tested in vitro to assess the engagement of the three aforementioned effector mechanisms. It was found that the engagement of CDC and ADCC was diminished when targeting the largest (8-domain) construct (therefore most distal from the membrane), whilst ADCP was impaired with the smallest (membrane-flush) construct and required the presence of at least one extracellular domain for activity. However, ADCP engagement was restored when the membrane-flush epitope was tethered to the membrane via a GPI anchor rather than a transmembrane peptide domain. These findings were confirmed using two separate epitopes targeted by rituximab and CAMPATH-1H antibodies. Finally, the therapeutic response when targeting either the membrane-flush or 8-domain constructs was investigated in vivo. Mice who received A20 cells expressing the membraneflush construct exhibited clearance of tumour in the spleen following antibody therapy, whilst those that received tumours expressing the 8-domain construct did not respond to therapy. Together the work in this thesis demonstrates how the effector mechanisms engaged by a mAb can be altered dependent on the position of the epitope in relation to the cell membrane.

The ability to generate potent anti-tumour CTL responses is fundamental to the rejection of many tumours. However many potent anti-tumour responses are suppressed by regulatory T cells. The removal of this Treg suppression in the CT26 tumour model uncovers potent cryptic anti-GSW11 CTL, which become immunodominant. I characterise the GSW11-specific response and compare these with responses against the anti-tumour AH1-specific T cells, a representative of the global T cell population in both Treg depleted and replete tumour challenged mice. I show that the majority of CD8+ T cells recruited to the tumour site in Treg replete mice are GSW11-specific. Interestingly, the vast majority of these GSW11-specific T cells are in a PD1+ clonally exhausted state and thus unable to clear the tumour. I also show that the percentage of clonally exhausted cells, and the levels of PD1 on GSW11-specific T cells are much greater compared to AH1-specific T cells, thus indicating preferential suppression among CTL populations. Furthermore, my data revealed this phenomenon was also present in the draining lymph node during T cell priming following CT26 tumour challenge. Investigation of differences in TCR Vß usage revealed expansion of particular Vß families over others in the absence of Treg. This work highlights the complex interplay between peripheral tolerance mechanisms and CTL, which selectively hinders protective anti-tumour immunity.

Autoimmune diseases (AID) are chronic inflammatory diseases (CID) mediated by selfreactive T and B cells and are generally the results of the breakdown of T cell tolerance to self-antigen and failure of peripheral regulatory mechanisms. In this thesis I studied different mechanisms controlling the development of CIDs. I investigated the initial events involved in the activation of self-reactive CD4+ T cells which mediate the destruction of the thyroid in a mouse model of spontaneous thyroiditis. TAZ10 transgenic mice express a human T cell receptor (TCR) specific for a cryptic epitope of thyroid peroxidise (TPO) generated upon endogenous processing by thyroid epithelial cells (TEC), and a naturally occurring antagonistic epitope presented by dendritic cells (DC) upon exogenous processing of TPO. I have characterized the function of myeloid derived suppressor cells (MDSCs) in TAZ10 mice. MDSCs accumulate in lymphoid and non-lymphoid organs of TAZ10 mice during acute phases of inflammation and their number decrease as inflammation is fading. Despite their strong inhibitory function on T cell function and proliferation, MDSCs fail to prevent the activation of self-reactive T cells. I showed that the manipulation of MDSCs generated DCs that efficiently promoted the activation of T cells from TAZ10 mice. By contrast, peripheral T cells from patients with rheumatoid arthritis (RA) and lupus had a high proliferative activity compared to controls. Further analysis revealed that RA patients had reduced amounts of inhibitory MDSCs in peripheral blood. I showed that in TAZ10 mice TEC upregulate MHC class II molecules and present the cryptic epitope to TAZ10 T cells inducing their activation. I have demonstrated that DCs are responsible for the spreading of the TPO cryptic epitope from the thyroid to draining lymphnodes (DLN) resulting in the strong activation of transgenic T cells from TAZ10 mice. By adoptive transfer experiments, I showed that the activation of naive TAZ10 T cells occurs within days both in the thyroid and draining lymph-nodes (DLN) and resulted in the destruction of the thyroid. Altogether, this work shows for the first time that in a model devoid of any environmental insults, the normal turnover of TEC is sufficient to induce the activation of self-reactive T cells and the development of AID. In this thesis, I have highlighted the potential role of tissue transglutaminse 2 (TG2) in the treatment of CIDs. TG2 contributes to the pathogenesis of celiac disease and I have showed that TG2 activity promotes inflammation in patients with cystic fibrosis (CF). Mutation of the cystic fibrosis transmembrane regulator gene (CFTR) in CF patients is associated with increased TG2 expression and activity. In CF, TG2 promoted the crosslinking of the antiinflammatory peroxisome proliferator-activated receptor (PPAR) into perinuclear agresomes. The functional sequestration of PPAR was leading to increased inflammation. The finding of this function of TG2 in CF was relevant in TAZ10 mice as in-vivo inhibition of TG2 downregulated common markers of inflammation.

Class I heavy chain (HC) must assemble with β-microglobulin (β2m) and acquire optimal peptide in order to be presented to cytotoxic T cells (CTLs). Calnexin is involved in the initial folding of class I HC and subsequent assembly with β2m. Incorporation of "empty" or suboptimally loaded class I molecules into the multimolecular loading complex is essential for them to acquire optimal peptides. The loading complex consists of several cofactors: TAP, tapasin, ERp57 and calreticulin. The precise role of calnexin and calreticulin in the regulated assembly and peptide loading and the significance of their physical interaction with other cofactors of the loading as well as preloading complex still remains unclear. Using mouse fibroblasts that lack calreticulin, I have studied the role of calreticulin in the assembly and loading of H2-Kb and H2-Db expressed in these cells. MHC class I molecules in calreticulin-deficient cells are able to assemble with β2m normally, but their subsequent loading with optimal, stabilising peptides is defective despite their ability to interact with the TAP complex. The "empty" or suboptimally loaded class I molecules exit the ER rapidly. Reflecting the loading defect, presentation of endogenously processed antigens by class I molecules in calreticulin-deficient cells is impaired. I have used a human calnexin-deficient cell line CEM.NKR to study assembly of class I in the absence of calnexin. The results demonstrate that contrary to current understanding, calnexin has an important role in class I HC assembly with 32- microglobulin. The role of heavy chain glycosylation in class I biogenesis is still controversial. My findings suggest asparagine (N)-linked glycosylation of human class I heavy chain at position 86 is optimal and any deviations from "normal" glycosylation results in poor loading with peptides and some defect in the assembly with β2m. Despite affecting the loading function, glycosylation did not have significant effect on presentation of a high affinity binding epitope to HLA-A*0201 specific CTLs. Finally, I show that co-operation from all domains of calreticulin is essential in order to generate a fully functional calreticulin. Interestingly, proline-rich (P) -domain of calreticulin downregulated expression of a number of cellular proteins including MHC class I HC, despite restoring the cytosolic calcium levels in calreticulindeficient cells. The effect of P-domain on class I expression was at the level of transcription.