Circulating Unmethylated CHTOP and INS DNA Fragments Provide Evidence of Possible Islet Cell Death in Youth with Obesity and Diabetes

Abstract

Identification of islet β cell death prior to the onset of type 1 diabetes (T1D) or type 2 diabetes (T2D) might allow for interventions to protect β cells and reduce diabetes risk. Circulating unmethylated DNA fragments arising from the human INS gene have been proposed as biomarkers of β cell death, but this gene alone may not be sufficiently specific to report β cell death. To identify new candidate genes whose CpG sites may show greater specificity for β cells, we performed unbiased DNA methylation analysis using the Infinium HumanMethylation 450 array on 64 human islet preparations and 27 non-islet human tissues. For verification of array results, bisulfite DNA sequencing of human β cells and 11 non-β cell tissues was performed on 5 of the top 10 CpG sites that were found to be differentially methylated. We identified the CHTOP gene as a candidate whose CpGs show a greater frequency of unmethylation in human islets. A digital PCR strategy was used to determine the methylation pattern of CHTOP and INS CpG sites in primary human tissues. Although both INS and CHTOP contained unmethylated CpG sites in non-islet tissues, they occurred in a non-overlapping pattern. Based on Naive Bayes classifier analysis, the two genes together report 100% specificity for islet damage. Digital PCR was then performed on cell-free DNA from serum from human subjects. Compared to healthy controls (N = 10), differentially methylated CHTOP and INS levels were higher in youth with new onset T1D (N = 43) and, unexpectedly, in healthy autoantibody-negative youth who have first-degree relatives with T1D (N = 23). When tested in lean (N = 32) and obese (N = 118) youth, increased levels of unmethylated INS and CHTOP were observed in obese individuals. Our data suggest that concurrent measurement of circulating unmethylated INS and CHTOP has the potential to detect islet death in youth at risk for both T1D and T2D. Our data also support the use of multiple parameters to increase the confidence of detecting islet damage in individuals at risk for developing diabetes.