Epigenetic age signatures in bones

Abstract

Age prediction can help identify skeletal remains by limiting the search range for a missing person. Although age prediction methods based on odontology and anthropology are frequently used in the forensic field, DNA methylation is particularly promising age-predictive biomarker. In this study, we generated genome-wide DNA methylation profiles of bone samples from 32 identified skeletal remains with an age at death ranging from 31 to 96 years. Only 12 provided more than 800 K quality-filtered CpG methylation values using Illumina's Infinium MethylationEPIC BeadChip array. Methylation ages of the bone samples calculated using a recently developed skin & blood clock composed of 391 CpG sites were found to be very similar to their actual ages (MAD = 6.4 years). However, the low success rate in methylation profiling of bone DNA samples may prevent researchers from applying the array to this type of samples. Therefore, we selected a set of CpG sites that would enable age prediction based on only a few CpG sites in bone DNA samples. Nineteen age-associated CpG marker candidates were selected from 720 K quality-filtered CpG values of 21 male skeletal remain samples. Because age signatures for blood, such as markers on the ELOVL2, FHL2, KLF14 and TRIM59 genes, had showed strong age associations in 12 bone samples, we further tested the age association of the 5 well-known markers in a blood-based model and the 13 out of 19 CpG markers from the array of 21 bone samples with an independent set of 30 skeletal remain samples using SNaPshot multiplex based on single nucleotide primer extension. Four CpG sites on TMEM51, TRIM59, ELOVL2, and EPHA6 genes showed moderate or weak correlations between methylation and age, which suggests further investigation of these markers to predict the age of bones.