Genome-Wide DNA Methylation Differences Between Late-Onset Alzheimer's Disease and Cognitively Normal Controls in Human Frontal Cortex

Evidence supports a role for epigenetic mechanisms in the pathogenesis of late-onset Alzheimer's disease (LOAD), but little has been done on a genome-wide scale to identify potential sites involved in disease. This study investigates human postmortem frontal cortex genome-wide DNA methylation profiles between 12 LOAD and 12 cognitively normal age- and gender-matched subjects. Quantitative DNA methylation is determined at 27,578 CpG sites spanning 14,475 genes via the Illumina Infinium HumanMethylation27 BeadArray. Data are analyzed using parallel linear models adjusting for age and gender with empirical Bayes standard error methods. Gene-specific technical and functional validation is performed on an additional 13 matched pair samples, encompassing a wider age range. Analysis reveals 948 CpG sites representing 918 unique genes as potentially associated with LOAD disease status pending confirmation in additional study populations. Across these 948 sites the subtle mean methylation difference between cases and controls is 2.9%. The CpG site with a minimum false discovery rate located in the promoter of the gene Transmembrane Protein 59 (TMEM59) is 7.3% hypomethylated in cases. Methylation at this site is functionally associated with tissue RNA and protein levels of the TMEM59 gene product. The TMEM59 gene identified from our discovery approach was recently implicated in amyloid-beta protein precursor post-translational processing, supporting a role for epigenetic change in LOAD pathology. This study demonstrates widespread, modest discordant DNA methylation in LOAD-diseased tissue independent from DNA methylation changes with age. Identification of epigenetic biomarkers of LOAD risk may allow for the development of novel diagnostic and therapeutic targets.