Thymocyte regulatory variant alters transcription factor binding and protects from type 1 diabetes in infants

Abstract

We recently mapped a genetic susceptibility locus on chromosome 6q22.33 for type 1 diabetes (T1D) diagnosed below the age of 7 years between the PTPRK and thymocyte-selection-associated (THEMIS) genes. As the thymus plays a central role in shaping the T cell repertoire, we aimed to identify the most likely causal genetic factors behind this association using thymocyte genomic data. In four thymocyte populations, we identified 253 DNA sequence motifs underlying histone modifications. The G insertion allele of rs138300818, associated with protection from diabetes, created thymocyte motifs for multiple histone modifications and thymocyte types. In a parallel approach to identifying variants that alter transcription factor binding motifs, the same variant disrupted a predicted motif for Rfx7, which is abundantly expressed in the thymus. Chromatin state and RNA sequencing data suggested strong transcription overlapping rs138300818 in fetal thymus, while expression quantitative trait locus and chromatin conformation data associate the insertion with lower THEMIS expression. Extending the analysis to other T1D loci further highlighted rs66733041 affecting the GATA3 transcription factor binding in the AFF3 locus. Taken together, our results support a role for thymic THEMIS gene expression and the rs138300818 variant in promoting the development of early-onset T1D.

Figure 1

Two parallel approaches were utilized to identify age-at-diagnosis (AAD) and type 1 diabetes (T1D) credible set SNPs that affect thymocyte regulatory motifs or transcription factor (TF) binding motifs. Genetic variants on chromosome 6q22.33 locus near the thymocyte selection associated gene, THEMIS, are associated with the AAD and a lower risk of early-onset type 1 diabetes. The 58 credible set SNPs on this locus were obtained from the original analysis by Inshaw et al.. THEMIS is responsible for the thymocyte selection in the thymus, a process thought to affect the development of T cell-related autoimmune diseases such as T1D. We sought DNA motifs located in active chromosome regions of thymocyte cells, defined as histone modification peaks and overlapping RNA expression, indicative of active enhancer regions. Resulting thymocyte motifs were integrated with the 58 credible set variants on the 6q22.33 locus, and 1165 other SNPs for T1D, to identify SNPs that disrupt DNA motifs identified in the thymocyte epigenetic data. As a parallel approach, we applied a deep-learning method to identify the credible set variants that alter TF binding motifs.

Figure 2

rs138300818 G insertion allele significantly increases sequence similarity with a thymocyte H3K27ac motif in CD4⁺αß cells. (a) Overlap between a thymocyte H3K27ac motif and rs138300818 alternative G allele (highlighted with star) and reference null allele flanking sequence. Bases conflicting with the motif are indicated with gray color. Panels (b) and (c): Sequence similarity score (b) and −log₁₀(p-value) distributions (c) for all credible set SNP—motif pairs where at least one of the SNP alleles (REF or ALT) forms a motif overlapping sequence (p < 0.05). rs138300818 is indicated as a red dot. (d) rs138300818 REF allele, but not ALT insertion allele, forms a RFX7 transcription factor binding site. The motifs are aligned with panel (a) sequence. (e) Score distribution for rs138300818 REF and ALT alleles against all available transcription factor binding sites from DeepBind database. (f) RFX7 gene expression is highest in brain (cerebellum, pituitary gland), and in thymus in FANTOM5 database. (g) RFX7 and THEMIS mRNA expression in early double negative [DN], DN (P/Q), double positive [DP] (P/Q), αβ entry, CD4⁺ , CD8⁺ , CD8AA, Treg, type 1/3 innate, and cycling T cells in scRNAseq of thymus. The DN (Q), DP (Q), and αβ entry T cells with the highest RFX7 expression are highlighted with larger point size.

Figure 3

rs138300818 G insertion allele—protecting from early-onset T1D—creates a thymocyte motif (a). At the same time, this motif disrupts the RFX5/7 binding motif which is present with the rs138300818 reference (null) allele (b). rs138300818 is located intronic in PTPRK. Non-coding RNA transcription was detected overlapping rs138300818 (grey vertical line) in all four studied thymocyte cell types (dark blue peaks). Roadmap Epigenomics chromatin state data for fetal thymus (PrimaryHMM) indicated strong transcription (state 4, green bar) overlapping the SNP flanking region (c). PCHiC data in fetal thymocytes and in naïve CD8 cells suggests that rs138300818 interacts with both PTPRK and THEMIS transcription start sites. The genetic association signal for early-onset T1D is colocalized with eQTL signal for THEMIS in human whole blood; rs138300818 reference allele—which forms the RFX7/5 binding motif—is associated with higher THEMIS expression (d).

Figure 4

Two T1D susceptibility SNPs have allele-specific thymocyte motifs. (a) The reference G insertion allele of rs142852921 on chromosome 7p15.2 (SKAP2) region matches a state 10 motif (distal active promoters) in CD3⁺CD4⁺CD8⁺ cells (REF p = 4.5 × 10 ^− 15, score 43.5; ALT p = 5.0 × 10 ^− 5, score −11.6). (b) In the AFF3 locus, rs66733041 reference allele (underlined sequence) forms a H3K4me3 motif in CD4⁺αβ cells (p = 1.4 × 10 ^− 11, score 31.1), as well as GATA3 and Aft3 binding motifs, all of which are disrupted by the 12 bp deletion allele (indicated with a star). For better match, the ALT sequence was converted to reverse complement format.