T cell affinity and autoimmunity : what is the origin of autoimmune T cells?

T cell receptor affinity for self-peptide/MHC ligand has a well established role in thymic selection. Above a certain affinity threshold (KD ² 6 ?M) a developing CD8+ T cell is negatively selected, while below this threshold it is positively selected, and no binding results in die by neglect. How TCR affinity contributes to the induction of autoimmune pathology remains poorly understood.
In the first part of my thesis, I examined the role of TCR affinity in peripheral T cells activation and the maintenance of peripheral tolerance. To do this, I make use of the RIP-OVA experimental autoimmune diabetes model. In RIP-OVA mice, OVA peptide (SIINFEKL) is under the control of the rat insulin promoter (RIP-OVA) and is expressed in both the pancreas and the thymus. Diabetes can be induced in RIP-OVA mice following adoptive transfer of OVA-specific CD8+ T cells (OT-I T cells) and immunization with OVA peptide and lipopolysaccharide (LPS). To determine the role of TCR affinity in diabetes induction, mice were immunized with OVA peptide variants that span the threshold between positive and negative selection. Only mice receiving high affinity peptides (above the threshold for inducing negative selection) were able to induce diabetes. In addition, higher affinity ligands induced stronger proliferation, activation marker upregulation and MAP kinase activation. These findings indicate that the affinity threshold established in the thymus is maintained in the periphery. In addition, we identified asymmetric division as a mechanistic link between TCR affinity and the induction of autoimmune pathology. Asymmetric division results in the generation of ÒproximalÓ daughter cells capable of undergoing sustained proliferation and differentiation into short-lived effector T cells. High-affinity TCR-pMHC interactions also promoted T cell upregulation of VLA-4, an integrin important for infiltration into target tissue. In contrast, T cells activated by below-threshold antigens underwent symmetric division, leading to abortive clonal expansion and failure to fully differentiate into tissue-infiltrating effector T cells.
In the second part of my thesis, I addressed the origin of self-reactive T cells. I was particularly interested to understand the impact of TCR affinity on 1) the efficiency of negative selection in the thymus; 2) the priming of a pathologic T cell response; and 3) the ability of a T cell to destroy self-antigen expressing target cells. To address these questions I made use of newly generated transgenic mouse lines expressing OVA-variant proteins with altered affinities for the OT-I TCR under the rat insulin promoter (RIP-variant mice).
To assess the efficiency of negative selection in RIP-variant mice, I crossed them with OT-I TCR transgenic mice. All double transgenic mice developed severe diabetes early after birth, suggesting that there were too many OT-I T cells in this system. As an alternative approach, I generated radiation bone marrow chimeras (further addressed below).
To assess the role of TCR affinity in the priming of a pathologic T cell response, I needed a model in which the affinity for priming is variable. I adoptively transferred OT-I T cells into RIP-variant transgenic mice followed with Listeria monocytogenes expressing OVA (Lm-OVA) (in the case of transfer into RIP-OVA mice) or variant peptide (depending on variant expression in host mice). Using this approach I found that T cells activated with threshold and below threshold ligands are poor inducers of effector T cell differentiation, and therefore comprise a low risk for autoimmunity.
To investigate the impact of TCR affinity for target tissue antigen on the induction of autoimmune pathology, I used a model in which priming efficiency is strong but the antigen affinity of the target tissue is variable. OT-I T cells were adoptively transferred into RIP-OVA variant mice and challenged with Lm-OVA. Diabetes was strongly induced in mice expressing both high affinity and threshold antigens in the pancreas. In contrast, mice expressing below threshold ligand remained diabetes free even after high numbers of OT-I T cells were transferred (107). These data indicate that threshold and below threshold target cells comprise a low risk for autoimmunity.
Since it was not possible to study T cell tolerance by breeding double transgenic mice, I generated mixed bone marrow chimeras. Lethally irradiated RIP-OVA and RIP-OVA variant mice were reconstituted with a mixture of OT-I and B6 bone marrow cells. These mice could be further used to determine the role of antigen affinity in negative selection, peripheral T cell priming and the ability of a T cell to lyse target cells. The results of these studies demonstrated that the efficiency of negative selection in vivo dramatically increases with above threshold self-antigens, such that fewer OT-I T cells survive negative selection in RIP-variant mice expressing high affinity antigen. By further challenging these mice with self-antigen/LPS, I was able to investigate whether OT-I T cells that survived negative selection were capable of inducing diabetes. Mice expressing antigens just above the negative selection threshold exhibited the highest risk of developing experimental autoimmune diabetes upon immunization with the corresponding peptide self-antigen. In contrast, mice expressing the cognate antigen for OT-I (OVA) (and the lowest of number of peripheral T cells) were completely free of diabetes, underscoring the importance of negative selection in preventing the accumulation of peripheral self-reactive T cells. In summary, these data suggest that just above the affinity threshold for negative selection, sufficient numbers of self-reactive T cells can escape deletion thereby constituting an increased risk for the development of autoimmunity.