Systemic interindividual DNA methylation variants in cattle share major hallmarks with those in humans

Abstract

Background: We recently identified ~ 10,000 correlated regions of systemic interindividual epigenetic variation (CoRSIVs) in the human genome. These methylation variants are amenable to population studies, as DNA methylation measurements in blood provide information on epigenetic regulation throughout the body. Moreover, establishment of DNA methylation at human CoRSIVs is labile to periconceptional influences such as nutrition. Here, we analyze publicly available whole-genome bisulfite sequencing data on multiple tissues of each of two Holstein cows to determine whether CoRSIVs exist in cattle.

Results: Focusing on genomic blocks with ≥ 5 CpGs and a systemic interindividual variation index of at least 20, our approach identifies 217 cattle CoRSIVs, a subset of which we independently validate by bisulfite pyrosequencing. Similar to human CoRSIVs, those in cattle are strongly associated with genetic variation. Also as in humans, we show that establishment of DNA methylation at cattle CoRSIVs is particularly sensitive to early embryonic environment, in the context of embryo culture during assisted reproduction.

Conclusions: Our data indicate that CoRSIVs exist in cattle, as in humans, suggesting these systemic epigenetic variants may be common to mammals in general. To the extent that individual epigenetic variation at cattle CoRSIVs affects phenotypic outcomes, assessment of CoRSIV methylation at birth may become an important tool for optimizing agriculturally important traits. Moreover, adjusting embryo culture conditions during assisted reproduction may provide opportunities to tailor agricultural outcomes by engineering CoRSIV methylation profiles.

Keywords: Bovine; CoRSIV; Cow; DoHAD; Systemic interindividual epigenetic variation.

Fig. 1

Screen for correlated regions of systemic interindividual variation (CoRSIVs) in cattle. A We analyzed WGBS data from three tissues representing the three embryonic germ layers from each of 2 cows (3842 and 3886; 6 methylomes total). B Unsupervised hierarchical clustering of average methylation groups the libraries by tissue (a random sample of 30,000 genomic bins is shown). C Illustration of our two-step approach: an example of a 300 bp block with concordant positive residuals is indicated by the red box. D A 200 bp CoRSIV (shaded region; 5 CpG sites) 1.8 kb upstream of the GALNTL5 transcription start site illustrates consistent interindividual difference in all three tissues. E Scatter plot of number of CpGs per CoRSIV vs. CoRSIV size for all 1,263 CoRSIVs initially identified (following masking of CpG-SNPs). Subsequent analyses and the rest of this paper focus on the 217 CoRSIVs, each of which contains at least 5 CpGs (right of dotted line). F Unlike genome-wide bins, hierarchical clustering of average methylation in the 217 CoRSIVs groups the libraries by individual

Fig. 2

Independent validation of systemic interindividual variation at CoRSIVs we identified. Each scatter plot shows the inter-tissue correlation in DNA methylation based on quantitative bisulfite pyrosequencing in each of 20 Holstein calves (10 male, 10 female). A CAB39L, B GIMAP7, C IL17F, D KCTD14. Inter-tissue correlations with R ≥ 0.71 (the cutoff for validation) are highlighted in red

Fig. 3

Cattle CoRSIVs are enriched at transcription start sites and associated with genetic variation. A Manhattan plot of systemic interindividual variation index (SIVI) for all genomic blocks with ≥ 5 CpGs. Blocks with SIVI ≥ 20 (dashed line) are considered CoRSIVs. B The overlap of CoRSIVs / controls with transcription start sites (TSS), transcription end site (TES), gene body, and intergenic regions. CoRSIVs are enriched at TSS. C The overlap with genetic variants in 1 kb windows centered on CoRSIVs / control regions. D CoRSIV-flanking regions (± 50 kb at 5 kb increments) show overlap with various classes of transposable elements, specifically LINE-1 and LTR elements. Column to the left of 0 KB mark shows direct overlap of CoRSIVs with transposable elements

Fig. 4

Establishment of DNA methylation at cattle CoRSIVs is particularly labile to periconceptional environment. A Graphic illustration of IVP/MOET experiments. B Overlap of differentially methylated regions (DMRs) between IVP and MOET offspring with CoRSIVs and control regions in three tissues: liver, muscle, and hypothalamus. C-E For the IVP vs. MOET experiments, volcano plots illustrating significant differences in average methylation between IVP and MOET at 217 CoRSIVs and matched control regions in (C) liver, (D) muscle, and (E) hypothalamus. Each point represents a CoRSIV or control region. P values are adjusted for multiple testing; significant p values are shown in pink