Systemic interindividual epigenetic variation in humans is associated with transposable elements and under strong genetic control

Abstract

Background: Genetic variants can modulate phenotypic outcomes via epigenetic intermediates, for example at methylation quantitative trait loci (mQTL). We present the first large-scale assessment of mQTL at human genomic regions selected for interindividual variation in CpG methylation, which we call correlated regions of systemic interindividual variation (CoRSIVs). These can be assayed in blood DNA and do not reflect interindividual variation in cellular composition.

Results: We use target-capture bisulfite sequencing to assess DNA methylation at 4086 CoRSIVs in multiple tissues from each of 188 donors in the NIH Gene-Tissue Expression (GTEx) program. At CoRSIVs, DNA methylation in peripheral blood correlates with methylation and gene expression in internal organs. We also discover unprecedented mQTL at these regions. Genetic influences on CoRSIV methylation are extremely strong (median R²=0.76), cumulatively comprising over 70-fold more human mQTL than detected in the most powerful previous study. Moreover, mQTL beta coefficients at CoRSIVs are highly skewed (i.e., the major allele predicts higher methylation). Both surprising findings are independently validated in a cohort of 47 non-GTEx individuals. Genomic regions flanking CoRSIVs show long-range enrichments for LINE-1 and LTR transposable elements; the skewed beta coefficients may therefore reflect evolutionary selection of genetic variants that promote their methylation and silencing. Analyses of GWAS summary statistics show that mQTL polymorphisms at CoRSIVs are associated with metabolic and other classes of disease.

Conclusions: A focus on systemic interindividual epigenetic variants, clearly enhanced in mQTL content, should likewise benefit studies attempting to link human epigenetic variation to the risk of disease.

Keywords: CoRSIV; DNA methylation; DOHaD; Epigenetic epidemiology; Epigenome-wide association study; Genetics.

Fig. 1

Target-capture bisulfite sequencing in 807 GTEx samples confirms systemic interindividual epigenetic variation at CoRSIVs. A DNA samples were obtained from multiple tissues (representing the three embryonic germ layers) from each of 188 GTEx donors. B CoRSIV capture process using Agilent reagents. C Percentage of CoRSIVs for which target-capture bisulfite sequencing achieved various read depths; each point represents one of 807 samples. D Plots of read depth at two target regions illustrate specificity of targeting across all six tissues. The Y-axis scales are same for each region and indicated for thyroid. E Scatter plots between all possible tissue pairs illustrate high inter-tissue correlations at a CoRSIV within HPCAL1. F Heat map of inter-tissue correlations across 4086 CoRSIVs shows generally high correlation coefficients between all possible tissue pairs. G For the 232 tissue samples from 53 donors with data on at least 4 tissues (excluding cerebellum), unsupervised hierarchical clustering of methylation data at 2349 fully informative CoRSIVs groups perfectly by donor

Fig. 2

Genetic influences on CoRSIV methylation are strong and biased. A, B Representative plots of mQTL associations at individual CoRSIVs on chromosomes 1 and 2, respectively. Significant associations are shown for all SNVs within 1Mb of each CoRSIV; positive and negative beta coefficients are plotted in blue and red, respectively. The most significant SNV (Simes SNV) is circled. Insets show average CoRSIV methylation vs. Simes SNV genotype. C Distribution of distances between CoRSIVs and corresponding Simes SNVs. D For each of 4086 CoRSIVs, heat map depicts the number of tissues in which the Simes SNV falls within the same haplotype block, illustrating the largely systemic nature of mQTL at CoRSIVs. E Distribution of beta coefficients of significant Simes mQTL associations for the GoDMC blood mQTL data [42]. F Distribution of beta coefficients of significant Simes mQTL associations at 3723 CoRSIVs in blood DNA from 188 GTEx donors. G Distribution of beta coefficients of significant Simes mQTL associations across 2939 CoRSIVs in blood DNA from 47 newborns (USC). H Distribution of Simes mQTL R² (goodness of fit) for the GoDMC data. I Distribution of Simes mQTL R² at CoRSIVs (GTEx, blood). J Distribution of Simes mQTL R² at CoRSIVs (USC samples)

Fig. 3

Genic CoRSIV-flanking regions show long-range enrichments and depletions for specific classes of transposable elements. A Using 1 kb step sizes, each plot shows significant enrichments or depletions for CpG islands (CGI) and subclasses within each of 8 classes of transposable element within 50 kb of genic CoRSIVs. Compared to control regions, CoRSIV-flanking regions show long-range depletion of CpG islands and enrichment of specific classes of LINEs and LTRs. B Compared to CoRSIVs showing a positive mQTL beta coefficient, those with negative coefficients are depleted for CpG islands and show long-range depletion of specific LINE1s and all subclasses of Alus. C The strength of mQTL associations at CoRSIVs (R² in 4^th vs. 1^st quartile) is not associated with widespread differences in genomic content of transposable elements. D Compared to regions in which HM450 probes are located, CoRSIVs show short- and long-range enrichments for many subclasses of LINE1 and LTR retrotransposons

Fig. 4

CoRSIV mQTL SNVs are enriched for GWAS associations. A Within each of 8 disease/phenotype categories, the histogram shows the null distribution obtained by permutation testing for overlap of GWAS SNVs with SNVs randomly sampled within 1Mb of each CoRSIV. The red diamond shows the actual number of overlaps between CoRSIV mQTL SNVs and GWAS SNVs. Numbers of GWAS SNVs considered in each category are anthropometric: 8106, cancer: 3163, cardiovascular: 4816, hematological: 7461, immune: 5263, metabolic: 10,121, neurological: 14,741, and various: 14,573. B Statistical significance (Bonferroni-adjusted p-value) vs. fold enrichments for the analysis in A. Strong and statistically significant enrichments were found for all outcomes except cancer. C Statistical significance (Bonferroni-adjusted p-value) vs. fold enrichments for 8 metabolic traits and 4 cancer outcomes from the LDSC analysis confirms that the vicinity of CoRSIVs is enriched for heritability of metabolic traits