The chances to develop Alzheimer's disease (AD) result from a combination of genetic and non-genetic risk factors , the latter likely being mediated by epigenetic mechanisms . In the past, genome-wide association studies (GWAS) have identified an important number of risk loci associated with AD pathology , but a causal relationship remains difficult to establish. In contrast, locus-specific or epigenome-wide association studies (EWAS) have revealed site-specific epigenetic alterations, which provide mechanistic insights for a particular risk gene but often lack the statistical power of GWAS . Here, combining both approaches, we report a previously unidentified association of the peptidase M20-domain-containing protein 1 (PM20D1) with AD. We find that PM20D1 is a methylation and expression quantitative trait locus coupled to an AD-risk associated haplotype, which displays enhancer-like characteristics and contacts the PM20D1 promoter via a haplotype-dependent, CCCTC-binding-factor-mediated chromatin loop. Furthermore, PM20D1 is increased following AD-related neurotoxic insults at symptomatic stages in the APP/PS1 mouse model of AD and in human patients with AD who are carriers of the non-risk haplotype. In line, genetically increasing or decreasing the expression of PM20D1 reduces and aggravates AD-related pathologies, respectively. These findings suggest that in a particular genetic background, PM20D1 contributes to neuroprotection against AD.

Thousands of regions in gametes have opposing methylation profiles that are largely resolved during the post-fertilization epigenetic reprogramming. However some specific sequences associated with imprinted loci survive this demethylation process. Here we present the data describing the fate of germline-derived methylation in humans. With the exception of a few known paternally methylated germline differentially methylated regions (DMRs) associated with known imprinted domains, we demonstrate that sperm-derived methylation is reprogrammed by the blastocyst stage of development. In contrast a large number of oocyte-derived methylation differences survive to the blastocyst stage and uniquely persist as transiently methylated DMRs only in the placenta. Furthermore, we demonstrate that this phenomenon is exclusive to primates, since no placenta-specific maternal methylation was observed in mouse. Utilizing single cell RNA-seq datasets from human preimplantation embryos we show that following embryonic genome activation the maternally methylated transient DMRs can orchestrate imprinted expression. However despite showing widespread imprinted expression of genes in placenta, allele-specific transcriptional profiling revealed that not all placenta-specific DMRs coordinate imprinted expression and that this maternal methylation may be absent in a minority of samples, suggestive of polymorphic imprinted methylation.

Our comprehensive list of bona fide-imprinted DMR probes will simplify and facilitate methylation profiling of individuals with imprinting disorders and is applicable to other diseases in which aberrant imprinting has been implicated, such as cancer and fetal growth.

It has been postulated that imprinting aberrations are common in tumors. To understand the role of imprinting in cancer, we have characterized copy-number and methylation in over 280 cancer cell lines and confirm our observations in primary tumors. Imprinted differentially methylated regions (DMRs) regulate parent-of-origin monoallelic expression of neighboring transcripts in cis. Unlike single-copy CpG islands that may be prone to hypermethylation, imprinted DMRs can either loose or gain methylation during tumorigenesis. Here, we show that methylation profiles at imprinted DMRs often not represent genuine epigenetic changes but simply the accumulation of underlying copy-number aberrations (CNAs), which is independent of the genome methylation state inferred from cancer susceptible loci. Our results reveal that CNAs also influence allelic expression as loci with copy-number neutral loss-of-heterozygosity or amplifications may be expressed from the appropriate parental chromosomes, which is indicative of maintained imprinting, although not observed as a single expression foci by RNA FISH.