Integrated analyses of 5 mC, 5hmC methylation and gene expression reveal pathology-associated AKT3 gene and potential biomarkers for Alzheimer's disease.

Aims: 5 mC methylation and hydroxymethylation (5hmC) are associated with Alzheimer's disease (AD). However, previous studies were limited by the absence of a 5hmC calculation. This study aims to find AD associated predictors and potential therapeutic chemicals using bioinformatics approach integrating 5 mC, 5hmC, and expression changes, and an AD mouse model.

Methods: Gene expression microarray and 5 mC and 5hmC sequencing datasets were downloaded from GEO repository. 142 AD and 52 normal entorhinal cortex specimens were enrolled. Data from oxidative bisulfite sequencing (oxBS)-treated samples, which represent only 5 mC, were used to calculate 5hmC level. Functional analyses, random forest supervised classification and methylation validation were applied. Potential chemicals were predicted by CMap. Morris water maze, Y maze and novel object recognition behavior tests were performed using FAD AD mice model. Cortex and hippocampus tissues were isolated for immunohistochemical staining.

Results: C1QTNF5, UBD, ZFP106, NEDD1, AKT3, and MBP genes involving 13 promoter CpG sites with 5mc, 5hmC methylation and expression difference were identified. AKT3 and MBP were down-regulated in both patients and mouse model. Three CpG sites in AKT3 and MBP showed significant methylation difference on validation. FAD AD mice showed recession in brain functions and lower AKT3 expression in both cortex and hippocampus. Ten chemicals were predicted as potential treatments for AD.

Conclusions: AKT3 and MBP may be associated with AD pathology and could serve as biomarkers. The ten predicted chemicals might offer new therapeutic approaches. Our findings could contribute to identifying novel markers and advancing the understanding of AD mechanisms.