Genome-wide quantification of the effects of DNA methylation on human gene regulation

Changes in DNA methylation affect development and disease, but not all regulatory elements act in a methylation-dependent (MD) manner. Here, we developed mSTARR-seq, a high-throughput approach that we used to quantify genome-wide MD regulatory activity across three human cell types. We identify thousands of MD regulatory elements, show that MD activity is predictable using sequence and chromatin state information, and identify specific transcription factors associated with higher activity in unmethylated or methylated states. MD regulatory elements exhibit greater cell type specificity than MD-independent elements, and explain variation in the strength of DNA methylation-gene expression correlations in vivo. Our findings thus provide a multi-tissue map of MD regulatory activity in the human genome, facilitating functional interpretation of the many emerging associations between methylation and trait variation.

DNA甲基化的改变会影响发育进程与疾病发生，但并非所有调控元件均以甲基化依赖型（methylation-dependent, MD）方式发挥功能。本研究开发了mSTARR-seq这一高通量研究方法，用于定量分析三种人类细胞类型全基因组范围内的甲基化依赖型调控活性。本研究鉴定了数千个甲基化依赖型调控元件，证实可通过序列与染色质状态信息预测甲基化依赖型调控活性，并筛选出与未甲基化或甲基化状态下更高活性相关的特定转录因子。相较于非甲基化依赖型调控元件，甲基化依赖型调控元件表现出更强的细胞类型特异性，且可解释体内DNA甲基化与基因表达相关性强度的变异情况。综上，本研究绘制了人类基因组中跨组织的甲基化依赖型调控活性图谱，可为日益增多的甲基化与性状变异相关关联的功能注释提供有力支撑。