Surrounded by the breathtaking Rocky Mountains, the DNA Methylation and Epigenomics meetings were held concurrently, with the opening address and a number of the sessions held jointly between the two meetings.

Adrian Bird (University of Edinburgh, UK) opened the meetings and discussed his work on DNA methylation – how DNA base composition plays an important role in determining the epigenome, and how DNA-binding proteins are attracted or repelled depending on methylation status. An example is DNA-binding protein MeCP2, which is highly abundant in the brain; mutations in the MECP2 gene cause the autistic spectrum disorder Rett syndrome. Adrian’s team has identified two critical regions of MeCP2 (the MBD and the NCoR-SMRT interaction domain) that determine the presence and severity of Rett Syndrome.

In the first joint session on Genome-Wide DNA methylation, a number of speakers discussed how epigenetics drives cell differentiation, and the role of chromatin marks and DNA methylation in this process:

Alexander Meissner (Harvard University, Broad Institute, USA) gave an overview of his laboratory’s studies on DNA methylation across development and disease. While most DNA methylation patterns in somatic tissues are static, and inherited through cell division in a highly regulated manner, in germ cells DNA methylation is much more dynamic. Alexander’s team knocked out DNA (cytosine-5)-methyltransferase 3A (DNMT3A) in differentiating embryonic stem cells (ESCs) and using WGBS showed that while the overall methylation pattern remained stable, distinct focal losses of methylation were observed at many sites, including genes such as Nanog and Foxa2. The cells, however, were still able to form tumours when injected into mice. Knockout of both DNMT3A and DNMT3B causes passage-dependent loss of DNA methylation, though at very slow rates. These models represent powerful tools to enhance the understanding of DNA methylation in human development and disease.

Ryan Lister (University of Western Australia) addressed the general assumption that DNA methylation of promoters silences gene expression and described his team’s interrogation of thousands of human promoters by genome-wide manipulation using a zinc finger-DNMT3A fusion protein.

In the Perturbations of DNA Methylation in Disease session, Margaret Goodell (Baylor College of Medicine, USA) discussed DNMT3A In normal and malignant hematopoiesis. DNMT3A mutations are associated with approximately 20 % of haematological malignancies, and research from Margaret’s laboratory has shown that hematopoietic stem cells (HSCs) from DNMT3A KO mice have significantly reduced differentiation potential. DNMT3A KO mice all develop haematological malignancies within approximately 400 days, with pathological characteristics similar to those in patients with haematological malignancies, suggesting that they represent a good model for human disease.

The HSCs studied had largely extended regions of low methylation (canyons) frequently containing transcription factors. The canyon borders were demarked by 5-hmC and become eroded in the absence of DNMT3A, suggesting that DNMT3A regulates canyon size. When TET2 (which converts 5-mC to 5-hmC) was knocked out, the spread of canyons was reduced. The balance between 5-mC and 5-hmC in the genome is a critical step for regulating gene expression to maintain cellular functions, and the action of Dnmt3a and Tet proteins at the same genomic sites may suggest that imbalance in either disrupts the broader regulatory mechanisms acting at these loci.

A number of very informative workshops were held between the two meetings:

ENCODE Datasets Tutorial

The goal of ENCODE (Encyclopedia of DNA Elements) is to build a comprehensive catalogue of functional elements in the human genome, and to provide high quality data and new technologies and analytical tools to for the scientific community. As speaker Bing Ren (University of California, San Diego, USA) put it: ‘Use the data, publish tomorrow, scoop us, but acknowledge us!’

Modeling DNA methylation

Fabian Muller (Max Planck Institute for Informatics, Germany) described the upgraded RnBeads software package, a comprehensive and user-friendly analysis pipeline for large-scale DNA methylation datasets. Max and his colleagues have used RnBeads to characterize methylomes for the BLUEPRINT and DEEP projects.

Yaping Lie (Broad Institute of MIT and Harvard University, USA) described meQTLFinder, a Bayesian model to detect meQTL, which greatly increases mdQTL detection power by using local chromatin states and long-range and chromatin interaction.

Roadmap Epigenomics Tutorial

This tutorial highlighted the resources generated by the NIH Roadmap Epigenomics Project. To date, the Roadmap Epigenomics Consortium has characterized the 111 human epigenomes, which have been profiled for histone modification patterns, DNA accessibility, DNA methylation and RNA expression. The 2,805 genome-wide datasets comprise 150.2 billion sequencing reads, equivalent to 3,174x coverage of the human genome. This map will be of broad use to the scientific and biomedical communities, for studies of genome interpretation, gene regulation, cellular differentiation, genome evolution, genetic variation and human disease.

Tissues and cell types profiled to date: