Causal mechanisms in type 1 diabetes

• Main Author: Rubio García, Arcadio
• Format: Dissertation
• Language: English
• Published: University of Oxford 2022
• Subjects: Autoimmunity; Genetic regulation; Molecular aspects

Type 1 diabetes (T1D) is a common autoimmune disorder characterized by a progressive destruction of pancreatic β cells, which impairs insulin secretion and leads to a variety of long-term complications. Incidence and prevalence are highly variable geographically and steadily increasing worldwide, with the latter approaching one percent in some regions of Scandinavia and Middle East. Many different risk factors contribute to this variability. My thesis suggests biological mechanisms for the two major genetic and environmental factors, respectively, which were previously unknown. The highest contribution to T1D genetic risk comes from the human leukocyte antigen (HLA) DR and DQ loci. In particular, DQB1 position 57, and DRB1 positions 13 and 71 are estimated to account for 15% and 12% of the overall disease risk. Using a large hierarchical model, I have demonstrated that the frequency of certain negatively charged sequences in the epitope binding CDR3 region of T cell receptor (TCR) β chains is associated with susceptibility to T1D measured as the odds ratio (OR) due to DR-DQ diplotypes. These negatively charged sequences match the motifs present in early anti-insulin infiltrates isolated from NOD mice, a T1D animal model, and were also increased in the repertoires of T1D patients I used for validation. The same sequence biases were present in T conventional and regulatory cell subsets, which suggests a simple mechanistic interpretation of HLA effects in terms of changes to the positive thymic selection threshold. Clonotypes with negative charges in the CDR3β region are more likely to form a stable complex with HLA molecules expressed in the surface of thymic epithelial cells for diplotypes with a higher T1D OR as these typically lack aspartic acid, a residue with a negatively charged side chain, at DQB1 position 57. The biggest T1D environmental risk factors correspond to changes in the gut microbiome associated with industrialization. In particular, children that eventually progress to seroconversion and T1D have an increased abundance of Firmicutes and a decreased abundance of Bifidobacterium. Using extreme value statistics, I have shown the existence of many proteins in the gut microbiome whose similarity to insulin B 9-25, the primary epitope in T1D, is statistically significant. For some sequences the degree of similarity to insulin B 9-25 exceeds that of paralogs insulin-like growth factor I and II. Crossreactivity in one of the only three CD4+ T cell clonotypes that have been isolated from the pancreas of T1D organ donors, and recognize insulin B in the 8-24 register, to a bacterial mimotope I prioritized was demonstrated experimentally using T cell hybridomas. Furthermore, elution assays showed equivalent processing and presentation properties in the protective DQ6 and susceptible DQ8 molecules for proinsulin and a mimotope protein. Finally, I have also developed a state-space model to measure transcriptional activity at non-coding regulatory regions. In conjunction with chromosome contact data, this permitted establishing support for the candidacy of certain genes and variants in regions identified by genome-wide association studies as causal for T1D, and where our understanding of disease pathways is still incomplete. Furthermore, I observed that transcriptional activity in regulatory regions correlated with fold changes in protein-coding target genes linked by chromosome contacts. Active regulatory regions also contained long and highly specific transcription factor binding site motif combinations, which may represent potential targets to modulate immune circuits. Taken together, my findings improve our understanding of T1D etiology and may enable the development of better diagnostic and prognostic biomarkers, as well as interventions such as probiotics, immunotherapies or nucleic acid drugs.