Parallel profiling of DNA methylation and hydroxymethylation highlights neuropathology-associated epigenetic variation in Alzheimer’s disease

Smith et al. Clinical Epigenetics (2019) 11:52 https://doi.org/10.1186/s13148-019-0636-y RESEARCH Open Access Parallel profiling of DNA methylation and hydroxymethylation highlights neuropathology-associated epigenetic variation in Alzheimer’s disease Adam R. Smith1, Rebecca G. Smith1, Ehsan Pishva1,2, Eilis Hannon1, Janou A. Y. Roubroeks1,2, Joe Burrage1, Claire Troakes3, Safa Al-Sarraj3, Carolyn Sloan4, Jonathan Mill1, Daniel L. van den Hove2,5 and Katie Lunnon1* Abstract Background: Alzheimer’s disease is a progressive neurodegenerative disorder that is hypothesized to involve epigenetic dysfunction. Previous studies of DNA modifications in Alzheimer’s disease have been unable to distinguish between DNA methylation and DNA hydroxymethylation. DNA hydroxymethylation has been shown to be enriched in the human brain, although its role in Alzheimer’s disease has not yet been fully explored. Here, we utilize oxidative bisulfite conversion, in conjunction with the Illumina Infinium Human Methylation 450K microarray, to identify neuropathology-associated differential DNA methylation and DNA hydroxymethylation in the entorhinal cortex. Results: We identified one experiment-wide significant differentially methylated position residing in the WNT5B gene. Next, we investigated pathology-associated regions consisting of multiple adjacent loci. We identified one significant differentially hydroxymethylated region consisting of four probes spanning 104 bases in the FBXL16 gene. We also identified two significant differentially methylated regions: one consisting of two probes in a 93 base-pair region in the ANK1 gene and the other consisting of six probes in a 99-base pair region in the ARID5B gene. We also highlighted three regions that show alterations in unmodified cytosine: two probes in a 39-base pair region of ALLC, two probes in a 69-base pair region in JAG2, and the same six probes in ARID5B that were differentially methylated. Finally, we replicated significant ANK1 disease-associated hypermethylation and hypohydroxymethylation patterns across eight CpG sites in an extended 118-base pair region in an independent cohort using oxidative-bisulfite pyrosequencing. Conclusions: Our study represents the first epigenome-wide association study of both DNA methylation and hydroxymethylation in Alzheimer’s disease entorhinal cortex. We demonstrate that previous estimates of DNA hypermethylation in ANK1 in Alzheimer’s disease were underestimates as it is confounded by hypohydroxymethylation. Keywords: Alzheimer’s disease (AD), Brain, Ankyrin 1 (ANK1), DNA methylation (5mC), DNA hydroxymethylation (5hmC), Entorhinal cortex (EC), Epigenetics, Epigenome-wide association study (EWAS), Illumina Infinium Human Methylation 450K microarray (450K array) * Correspondence: 1 College of Medicine and Health, University of Exeter Medical School, Exeter University, RILD Building Level 4, Royal Devon and Exeter Hospital, Barrack Rd, Exeter EX2 5DW, UK Full list of author information is available at the end of the article © The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Smith et al. Clinical Epigenetics (2019) 11:52 Background Alzheimer’s disease (AD) affects approximately 35 million people worldwide [1]. It is a progressive neurodegenerative disease that leads to neuronal cell loss and results in severe cognitive decline. The characteristic hallmarks of AD include neurofibrillary tangles (NFTs) of hyperphosphorylated tau and amyloid beta (Aβ) plaques [2]. The amyloid precursor protein (APP) is normally processed by αand γ-secretase (non-amyloidogenic pathway), which forms a soluble, non-toxic Aβ fragment called P3. In AD, APP is cleaved by β- and γ-secretase (amyloidogenic pathway) resulting in the formation of a larger Aβ species, which aggregates to form the characteristic plaques associated with the disease. Although Mendelian inheritance of mutations in the APP gene and the PSEN1 and PSEN2 genes, which encode subunits of the γ-secretase enzyme, have been demonstrated in early-onset familial AD cases, these only account for ~ 5% of disease incidence [3]. The majority of AD cases are sporadic, occur late in life, and have, as yet, no defined etiology, with common single nucleotide polymorphisms (SNPs) accounting for only a third of disease risk [4]. Epigenetic processes mediate the reversible regulation of gene expression, occurring independently of DNA sequence variation and orchestrate a diverse range of important neurobiological processes in the brain. DNA methylation (5-methylcytosine—5mC) is the best characterized and most stable epigenetic modification [5], and recent epigenome-wide association studies (EWAS) have utilized the Illumina Infinium Human Methylation 450K microarray (450K array) to demonstrate robust and reproducible changes in DNA methylation at a number of loci in AD brain [6, 7], including the ANK1 gene. Although the focus of these studies has been on changes in DNA methylation in AD, the sodium bisulfite (BS) conversion approaches used cannot distinguish DNA methylation from DNA hydroxymethylation (5-hydroxymethylcytosine—5hmC). 5hmC has been previously identified at high levels in the developing [8] and adult brain [9], particularly in neurons [10], and as such may represent an important epigenetic mark to profile in the context of neurodegenerative diseases. Furthermore, levels of 5hmC potentially mask the true abundance of 5mC at specific loci in the genome, confounding existing EWAS analyses of AD. Recent studies have demonstrated that oxidative BS (OxBS) treatment enables the detection of 5hmC as thymine; therefore, by running matched BS- and OxBS-treated samples in parallel, it is possible to generate a quantitative measurement for total DNA modifications (BS data), DNA methylation (OxBS data), and, by proxy, DNA hydroxymethylation (BS data − OxBS data) and unmodified cytosine (uC) levels (1-BS data) [11]. Page 2 of 13 We have previously utilized this method in conjunction with Illumina 450K arrays to profile 5mC and 5hmC levels in parallel in post-mortem brain tissue from non-demented individuals to demonstrate brain region-specific differences in DNA modifications [11]. Here, we perform an EWAS of 5mC, 5hmC, and uC using the 450K array and brain tissue from 96 donors representing the spectrum of AD pathology defined by Braak staging, a standardized measure of NFT burden determined at autopsy [12], ranging (...truncated)