Search

Given the differential risk of type 1 diabetes (T1D) in offspring of affected fathers versus affected mothers and our observation that T1D cases have differential DNA methylation near the imprinted DLGAP2 gene compared to controls, we examined whether methylation near DLGAP2 mediates the association between T1D family history and T1D risk. In a nested case–control study of 87 T1D cases and 87 controls from the Diabetes Autoimmunity Study in the Young, we conducted causal mediation analyses at 12 DLGAP2 region CpGs to decompose the effect of family history on T1D risk into indirect and direct effects. These effects were estimated from two regression models adjusted for the human leukocyte antigen DR3/4 genotype: a linear regression of family history on methylation (mediator model) and a logistic regression of family history and methylation on T1D (outcome model). For 8 of the 12 CpGs, we identified a significant interaction between T1D family history and methylation on T1D risk. Accounting for this interaction, we found that the increased risk of T1D for children with affected mothers compared to those with no family history was mediated through differences in methylation at two CpGs (cg27351978, cg00565786) in the DLGAP2 region, as demonstrated by a significant pure natural indirect effect (odds ratio (OR) = 1.98, 95% confidence interval (CI): 1.06–3.71) and nonsignificant total natural direct effect (OR = 1.65, 95% CI: 0.16–16.62) (for cg00565786). In contrast, the increased risk of T1D for children with an affected father or sibling was not explained by DNA methylation changes at these CpGs. Results were similar for cg27351978 and robust in sensitivity analyses. Lastly, we found that DNA methylation in the DLGAP2 region was associated (P < 0.05) with gene expression of nearby protein-coding genes DLGAP2, ARHGEF10, ZNF596, and ERICH1. Results indicate that the maternal protective effect conferred through exposure to T1D in utero may operate through changes to DNA methylation that have functional downstream consequences. [ABSTRACT FROM AUTHOR]

Objective: Mechanisms underlying the role of non‐human leukocyte antigen (HLA) genetic risk variants in type 1 diabetes (T1D) are poorly understood. We aimed to test the association between methylation and non‐HLA genetic risk. Methods: We conducted a methylation quantitative trait loci (mQTL) analysis in a nested case‐control study from the Dietary Autoimmunity Study in the Young. Controls (n = 83) were frequency‐matched to T1D cases (n = 83) based on age, race/ethnicity, and sample availability. We evaluated 13 non‐HLA genetic markers known be associated with T1D. Genome‐wide methylation profiling was performed on peripheral blood samples collected prior to T1D using the Illumina 450 K (discovery set) and infinium methylation EPIC beadchip (EPIC validation) platforms. Linear regression models, adjusting for age and sex, were used to test to each single nucleotide polymorphism (SNP) ‐probe combination. Logistic regression models were used to test the association between T1D and methylation levels among probes with a significant mQTL. A meta‐analysis was used to combine odds ratios from the two platforms. Results: We identified 10 SNP‐methylation probe pairs (false discovery rate (FDR) adjusted P <.05 and validation P <.05). Probes were associated with the GSDMB, C1QTNF6, IL27, and INS genes. The cg03366382 (OR: 1.9, meta‐P =.0495), cg21574853 (OR: 2.5, meta‐P =.0232), and cg25336198 (odds ratio: 6.6, meta‐P =.0081) probes were significantly associated with T1D. The three probes were located upstream from the INS transcription start site. Conclusions: We confirmed an association between DNA methylation and rs689 that has been identified in related studies. Measurements in our study preceded the onset of T1D suggesting methylation may have a role in the relationship between INS variation and T1D development. [ABSTRACT FROM AUTHOR]