Department of Biology, Duke University, North Carolina, United States
Christopher M Vockley
Center for Genomic and Computational Biology, Duke University Medical School, North Carolina, United States; Department of Biostatistics and Bioinformatics, Duke University Medical School, North Carolina, United States
Rachel A Johnston
Department of Evolutionary Anthropology, Duke University, North Carolina, United States
Christina A Del Carpio
Department of Evolutionary Anthropology, Duke University, North Carolina, United States
Luis B Barreiro
Department of Pediatrics, Sainte-Justine Hospital Research Centre, University of Montreal, Montreal, Canada
Center for Genomic and Computational Biology, Duke University Medical School, North Carolina, United States; Department of Biostatistics and Bioinformatics, Duke University Medical School, North Carolina, United States; Program in Computational Biology and Bioinformatics, Duke University, North Carolina, United States
Department of Biology, Duke University, North Carolina, United States; Department of Evolutionary Anthropology, Duke University, North Carolina, United States; Institute of Primate Research, National Museums of Kenya, Nairobi, Kenya; Duke University Population Research Institute, Duke University, North Carolina, United States
Changes in DNA methylation are involved in development, disease, and the response to environmental conditions. However, not all regulatory elements are functionally methylation-dependent (MD). Here, we report a method, mSTARR-seq, that assesses the causal effects of DNA methylation on regulatory activity at hundreds of thousands of fragments (millions of CpG sites) simultaneously. Using mSTARR-seq, we identify thousands of MD regulatory elements in the human genome. MD activity is partially predictable using sequence and chromatin state information, and distinct transcription factors are associated with higher activity in unmethylated versus methylated DNA. Further, pioneer TFs linked to higher activity in the methylated state appear to drive demethylation of experimentally methylated sites. MD regulatory elements also predict methylation-gene expression relationships across individuals, where they are 1.6x enriched among sites with strong negative correlations. mSTARR-seq thus provides a map of MD regulatory activity in the human genome and facilitates interpretation of differential methylation studies.
