A cell epigenotype specific model for the correction of brain cellular heterogeneity bias and its application to age, brain region and major depression

Abstract

Brain cellular heterogeneity may bias cell type specific DNA methylation patterns, influencing findings in psychiatric epigenetic studies. We performed fluorescence activated cell sorting (FACS) of neuronal nuclei and Illumina HM450 DNA methylation profiling in post mortem frontal cortex of 29 major depression and 29 matched controls. We identify genomic features and ontologies enriched for cell type specific epigenetic variation. Using the top cell epigenotype specific (CETS) marks, we generated a publically available R package, “CETS,” capable of quantifying neuronal proportions and generating in silico neuronal profiles from DNA methylation data. We demonstrate a significant overlap in major depression DNA methylation associations between FACS separated and CETS model generated neuronal profiles relative to bulk profiles. CETS derived neuronal proportions correlated significantly with age in the frontal cortex and cerebellum and accounted for epigenetic variation between brain regions. CETS based control of cellular heterogeneity will enable more robust hypothesis testing in the brain.

Distinct chromatin signatures of DNA hypomethylation in aging and cancer

“…However, the application of the SVA method of correction (and Houseman correction for blood) and the use of a pure‐cell dataset such as the glia dataset (Guintivano et al., 2013) allowed us to tackle this issue in two different ways. Additionally, the use of the blood validation dataset (Hannum et al., 2013) allowed us to verify the reliability of our workflow, as in terms of whole blood dmCpGs we obtained 89% concordance with previous studies using the same data (Fernández et al., 2015).…”

“…Kidney, skin, and glia tissue datasets were enlarged for control cases using additional samples from KIRP (TCGA), skin (Bormann et al., 2016), and glia (Guintivano et al., 2013), respectively. Tissues were chosen based on disease prevalence, control data availability, and previous literature analyses to include both novel and pre‐analyzed tissues.…”

“…Moreover, recent analyses mainly performed in mouse tissue have failed to confirm global hypomethylation during the aging process (Cole et al., 2017; Hahn et al., 2017; Masser et al., 2017) and, to date, no study has provided a back‐to‐back and systematic comparison of the epigenetic changes that occur in aging and cancer. To address this issue, here we have analyzed DNA methylation changes and their associated chromatin patterns in a total of more than 2,300 healthy and tumoral samples obtained from differentially aged individuals, using HumanMethylation450 BeadChip data generated by The Cancer Genome Atlas (TCGA) consortium and other datasets (Bormann et al., 2016; Guintivano, Aryee & Kaminsky, 2013; Hannum et al., 2013). Our results confirmed the relationship between DNA hypermethylation in aging and cancer, but they also revealed important differences in DNA hypomethylation changes in the two processes that might be important to understand the possible role of DNA methylation as a molecular link between decline related to aging and tumor development.…”

DNA methylation and healthy human aging

“…The authors of this study proposed that to be certain that specific sites are reliable, it is vital to correct for cell types across individuals, although some epigenetic clock sites have been shown to not associate with cellular composition (Talens et al ., 2012; Horvath, 2013; Jaffe & Irizarry, 2014). While this example refers specifically to blood, the same consideration is important for analysis of any tissue sample containing more than one cell type (Guintivano et al ., 2013; Montaño et al ., 2013). In the future, studies of age and DNA methylation will need to either include cell counts, use reference‐free cell admixture control, or, at the very least, check their age‐related hits against regions known to differ across cell types in order for results to be reliable.…”

Cell type-specific DNA methylation in neonatal cord tissue and cord blood: a 850K-reference panel and comparison of cell types

“…The first approach involves a direct measure of cell type proportions for each sample by cell-sorting, but this approach is extremely challenging to execute for large sample sizes, and impractical for frozen clinical samples. Second, and the most widely used approach requires the use of a reference panel assembled from DNA methylation profiles of the tissue’s constituent cell types [5,6]. When a reference panel is not available, the last approach is to use a reference-free method, but this approach is computationally intense and less accurate [7–11].…”