Putative Causal Variants Are Enriched in Annotated Functional Regions From Six Bovine Tissues

Abstract

Genetic variants which affect complex traits (causal variants) are thought to be found in functional regions of the genome. Identifying causal variants would be useful for predicting complex trait phenotypes in dairy cows, however, functional regions are poorly annotated in the bovine genome. Functional regions can be identified on a genome-wide scale by assaying for post-translational modifications to histone proteins (histone modifications) and proteins interacting with the genome (e.g., transcription factors) using a method called Chromatin immunoprecipitation followed by sequencing (ChIP-seq). In this study ChIP-seq was performed to find functional regions in the bovine genome by assaying for four histone modifications (H3K4Me1, H3K4Me3, H3K27ac, and H3K27Me3) and one transcription factor (CTCF) in 6 tissues (heart, kidney, liver, lung, mammary and spleen) from 2 to 3 lactating dairy cows. Eighty-six ChIP-seq samples were generated in this study, identifying millions of functional regions in the bovine genome. Combinations of histone modifications and CTCF were found using ChromHMM and annotated by comparing with active and inactive genes across the genome. Functional marks differed between tissues highlighting areas which might be particularly important to tissue-specific regulation. Supporting the cis-regulatory role of functional regions, the read counts in some ChIP peaks correlated with nearby gene expression. The functional regions identified in this study were enriched for putative causal variants as seen in other species. Interestingly, regions which correlated with gene expression were particularly enriched for potential causal variants. This supports the hypothesis that complex traits are regulated by variants that alter gene expression. This study provides one of the largest ChIP-seq annotation resources in cattle including, for the first time, in the mammary gland of lactating cows. By linking regulatory regions to expression QTL and trait QTL we demonstrate a new strategy for identifying causal variants in cattle.

Gene expression and RNA splicing explain large proportions of the heritability for complex traits in cattle

“…If ‫ܴܦܨ‬ ஐ >= 0.05 for an omics feature, no trans e/sQTL were selected. Also, those e/sQTL under at least two ChIP-seq peaks identified from multiple studies 6,42,43 were used in the Bayesian analysis described below.…”

Differentially CTCF-Binding Sites in Cattle Rumen Tissue during Weaning

“…These results are in accordance with a previous study in bovine rumen tissue from our group, where the CTFC peaks identified for BW and AW represented approximately 0.6% of the cattle genome (bosTau8) [ 8 ]. Another study in cattle identified CTCF peaks in the liver, lung, mammary gland, kidney, heart, and spleen in lactating Holstein cows, totaling an average of 456,881 CTCF peaks for all samples and tissues (with a minimum number of ~60,000 peaks and a maximum number of 790,000 peaks per sample), representing ~7% of the cattle genome (bosTau8), showing a higher number of peaks compared to our study probably due to the higher number of samples and number of reads [ 34 ]. However, their samples presented a great variability in the number of CTCF peaks depending on the tissue and replicate [ 34 ].…”

“…Although large numbers of regulatory elements have been identified in cattle recently, many more regulatory non-coding elements are expected to be identified in additional tissues and conditions. A few studies have used CTCF ChIP-seq to identify functional regions in animals, such as mouse [ 29 , 30 , 31 ], sheep [ 32 , 33 ], and cattle [ 34 ]. A recent study in lactating dairy cows identified millions of functional regions in the cattle genome using ChIP-seq data from six tissues (heart, kidney, liver, lung, mammary, and spleen), including CTCF-binding sites and differential binding sites between tissues [ 18 , 34 ].…”

“…A few studies have used CTCF ChIP-seq to identify functional regions in animals, such as mouse [ 29 , 30 , 31 ], sheep [ 32 , 33 ], and cattle [ 34 ]. A recent study in lactating dairy cows identified millions of functional regions in the cattle genome using ChIP-seq data from six tissues (heart, kidney, liver, lung, mammary, and spleen), including CTCF-binding sites and differential binding sites between tissues [ 18 , 34 ]. In a recent report from our group [ 8 ], different epigenetic marks in the cattle genome were identified, including CTCF, resulting in the first in vivo global map of regulatory elements and chromatin states in rumen tissue during weaning.…”

Genetic score omics regression and multi-trait meta-analysis detect widespread cis-regulatory effects shaping bovine complex traits

“…These pieces of evidence support that MTAO can be used to identify omics features, i.e., regulatory elements, of high importance to complex traits. Apart from gene expression and splicing, there are many other types of quantitative omics features such as the height of ChIP-seq peaks 38 and allele-specific imbalance 27 which can also be analysed by MTAO. Moreover, MTAO is summary-data based so it can be applied to any results from GSOR or conventional TWAS, as long as there are beta and standard error estimates for the association between the omics feature and the phenotype.…”