Identification of autosomal cis expression quantitative trait methylation (cis eQTMs) in children’s blood

Abstract

Background: The identification of expression quantitative trait methylation (eQTMs), defined as correlations between gene expression and DNA methylation levels, might help the biological interpretation of epigenome-wide association studies (EWAS). We aimed to identify autosomal cis-eQTMs in child blood, using data from 832 children of the Human Early Life Exposome (HELIX) project. Methods: Blood DNA methylation and gene expression were measured with the Illumina 450K and the Affymetrix HTA v2 arrays, respectively. The relationship between methylation levels and expression of nearby genes (transcription start site (TSS) within a window of 1 Mb) was assessed by fitting 13.6 M linear regressions adjusting for sex, age, and cohort. Results: We identified 63,831 autosomal cis-eQTMs, representing 35,228 unique CpGs and 11,071 unique transcript clusters (TCs, genes). 74.3% of these cis-eQTMs were located at <250 kb, 60.0% showed an inverse relationship and 23.9% had at least one genetic variant associated with the methylation and expression levels. They were enriched for active blood regulatory regions. Adjusting for cellular composition decreased the number of cis-eQTMs to 37.7%, suggesting that some of them were cell type-specific. The overlap of child blood cis-eQTMs with those described in adults was small, and child and adult shared cis-eQTMs tended to be proximal to the TSS, enriched for genetic variants and with lower cell type specificity. Only half of the cis-eQTMs could be captured through annotation to the closest gene. Conclusions: This catalogue of blood autosomal cis-eQTMs in children can help the biological interpretation of EWAS findings, and is publicly available at https://helixomics.isglobal.org/.

Integration of public DNA methylation and expression networks via eQTMs improves prediction of functional gene–gene associations

“…Only 16% (3,861 out of 23,862 eQTMs) of the Ruiz-Arenas et al 3 eQTMs were also significant in our study, but 97% (3,731 of the 3,861) had consistent effect direction. The small overlap of eQTMs between our study and Ruiz-Arenas et al 3…”

“…Ruiz-Arenas et al also observed. 3 We further identified eQTM associations for 2,243 additional genes compared to Bonder et al 4 and 5,872 additional genes compared to Ruiz-Arenas et al 3 As expected, the majority of the eQTM genes (n=4,176; 67%) were negatively correlated with the nearby CpG sites (Table 1). We tested whether this directionality was consistent among different tissues by replicating the discovered eQTMs in each of the three tissues separately and observed that over 98% of the eQTMs identified significantly in both the meta-analysis and in each tissue shared the same direction (Figure 1).…”

“…The explained variance for gene expression levels was always lower than 0.4 (Supplementary Figure 1). We compared these associations to the largest eQTM analysis published to date by Bonder et al 4 and a recent eQTM study in a child cohort by Ruiz-Arenas et al 3 . As shown in Supplementary Figure 2, 83% (4,269) of the Bonder et al 4 eQTMs were also found significant in our study, and 99% (4,246) of eQTMs were of the same direction.…”

“…One way to determine this is to study the association between DNA methylation and gene expression levels (i.e. expression quantitative trait methylation; eQTMs 3,4 ). Because most eQTM studies focus on a single tissue, it is often unclear whether the eQTMs identified in one tissue are consistently present in other tissues as well, and only a few eQTM studies have compared eQTMs across tissues 5 or different cell types 6 .…”

Epigenome-Wide Meta-Analysis of Prenatal Maternal Stressful Life Events and Newborn DNA Methylation

“…We did conduct an analysis of functionality using the Human Early Life Exposome expression quantitative trait (eQTM) loci database 55 however, none of the ve CpG sites from the primary or secondary meta-analyses were identi ed as an eQTM. Further functional analyses are necessary to better de ne the relationship between DNA methylation of our sign cant loci and gene expression.…”