Physical access to DNA is a highly dynamic property of chromatin that plays an essential role in establishing and maintaining cellular identity. The organization of accessible chromatin across the genome reflects a network of permissible physical interactions through which enhancers, promoters, insulators and chromatin-binding factors cooperatively regulate gene expression. This landscape of accessibility changes dynamically in response to both external stimuli and developmental cues, and emerging evidence suggests that homeostatic maintenance of accessibility is itself dynamically regulated through a competitive interplay between chromatin-binding factors and nucleosomes. In this Review, we examine how the accessible genome is measured and explore the role of transcription factors in initiating accessibility remodelling; our goal is to illustrate how chromatin accessibility defines regulatory elements within the genome and how these epigenetic features are dynamically established to control gene expression.

Chromatin accessibility and the regulatory epigenome.

The process of identifying genomic sites that show statistical relationships to phenotypes holds great promise for human health and disease (Hindorff et al. 2009). However, our current inability to efficiently interpret noncoding variants impedes progress toward using personal genomes in medicine. Coordinated efforts to survey the noncoding genome have shown that sequences marked by DNA accessibility and certain histone modifications are enriched for variants that are statistically related to phenotypes (The ENCODE Project Consortium 2012; Roadmap Epigenomics Consortium et al. 2015). The first stages of a mechanistic hypothesis can now be assigned to variants that directly overlap these annotations (Fu et al. 2014; Kircher et al. 2014; Ritchie et al. 2014).

However, simply considering the overlap of a variant with annotations underutilizes these data; more can be extracted by understanding the DNA–protein interactions as a function of the underlying sequence. Proteins that recognize specific signals in the DNA influence its accessibility and histone modifications (Voss and Hager 2014). Given training data, models parameterized by machine learning can effectively predict protein binding, DNA accessibility, histone modifications, and DNA methylation from the sequence (Das et al. 2006; Arnold et al. 2013; Benveniste et al. 2014; Pinello et al. 2014; Lee et al. 2015; Setty and Leslie 2015; Whitaker et al. 2015). A trained model can then annotate the influence of every nucleotide (and variant) on these regulatory attributes. This upgrades previous approaches in two ways. First, variants can be studied at a finer resolution; researchers can prioritize variants predicted to drive the regulatory activity and devalue those predicted to be irrelevant bystanders. Second, rare variants that introduce a gain of function will often not overlap regulatory annotations in publicly available data. An accurate model for regulatory activity can predict the gain of function, allowing follow-up consideration of the site.

In recent years, artificial neural networks with many stacked layers have achieved breakthrough advances on benchmark data sets in image analysis (Krizhevsky et al. 2012) and natural language processing (Collobert et al. 2011). Rather than choose features manually or in a preprocessing step, convolutional neural networks (CNNs) adaptively learn them from the data during training. They apply nonlinear transformations to map input data to informative high-dimensional representations that trivialize classification or regression (Bengio et al. 2013). Early applications of CNNs to DNA sequence analysis surpass more established algorithms, such as support vector machines or random forests, at predicting protein binding and accessibility from DNA sequence (Alipanahi et al. 2015; Zhou and Troyanskaya 2015). More accurate models can more precisely dissect regulatory sequences, thus improving noncoding variant interpretation. However, to fully exploit the value of these models, it is essential that they are technically and conceptually accessible to the researchers who can take advantage of their potential.

Here, we introduce Basset, an open source package to apply deep CNNs to learn functional activities of DNA sequences. We used Basset to simultaneously predict the accessibility of DNA sequences in 164 cell types mapped by DNase-seq by the ENCODE Project Consortium and Roadmap Epigenomics Consortium (The ENCODE Project Consortium 2012; Roadmap Epigenomics Consortium et al. 2015). From these data sets, CNNs simultaneously learn the relevant sequence motifs and the regulatory logic with which they are combined to determine cell-specific DNA accessibility. We show that a model achieving this level of accuracy provides meaningful, nucleotide-precision measurements. Subsequently, we assign Genome-wide association study (GWAS) variants cell-type–specific scores that reflect the accessibility difference predicted by the model between the two alleles. These scores are highly predictive of the causal SNP among sets of linked variants. Importantly, Basset puts CNNs in the hands of the genome biology community, providing tools and strategies for researchers to train and analyze models on new data sets. In conjunction with genomic big data, Basset offers a promising future for understanding how the genome crafts phenotypes.

/pdf/basset-learning-the-regulatory-code-of-the-accessible-genome-3rjgomsu4d.pdf

Basset: learning the regulatory code of the accessible genome with deep convolutional neural networks

A subset of eukaryotic transcription factors possesses the remarkable ability to reprogram one type of cell into another. The transcription factors that reprogram cell fate are invariably those that are crucial for the initial cell programming in embryonic development. To elicit cell programming or reprogramming, transcription factors must be able to engage genes that are developmentally silenced and inappropriate for expression in the original cell. Developmentally silenced genes are typically embedded in "closed" chromatin that is covered by nucleosomes and not hypersensitive to nuclease probes such as DNase I. Biochemical and genomic studies have shown that transcription factors with the highest reprogramming activity often have the special ability to engage their target sites on nucleosomal DNA, thus behaving as "pioneer factors" to initiate events in closed chromatin. Other reprogramming factors appear dependent on pioneer factors for engaging nucleosomes and closed chromatin. However, certain genomic domains in which nucleosomes are occluded by higher-order chromatin structures, such as in heterochromatin, are resistant to pioneer factor binding. Understanding the means by which pioneer factors can engage closed chromatin and how heterochromatin can prevent such binding promises to advance our ability to reprogram cell fates at will and is the topic of this review.

/pdf/pioneer-transcription-factors-in-cell-reprogramming-1rrgyre0zt.pdf

Pioneer transcription factors in cell reprogramming

Absence of a simple code: how transcription factors read the genome

Eukaryotic transcription factors (TFs) are key determinants of gene activity, yet they bind only a fraction of their corresponding DNA sequence motifs in any given cell type. Chromatin has the potential to restrict accessibility of binding sites; however, in which context chromatin states are instructive for TF binding remains mainly unknown. To explore the contribution of DNA methylation to constrained TF binding, we mapped DNase-I-hypersensitive sites in murine stem cells in the presence and absence of DNA methylation. Methylation-restricted sites are enriched for TF motifs containing CpGs, especially for those of NRF1. In fact, the TF NRF1 occupies several thousand additional sites in the unmethylated genome, resulting in increased transcription. Restoring de novo methyltransferase activity initiates remethylation at these sites and outcompetes NRF1 binding. This suggests that binding of DNA-methylation-sensitive TFs relies on additional determinants to induce local hypomethylation. In support of this model, removal of neighbouring motifs in cis or of a TF in trans causes local hypermethylation and subsequent loss of NRF1 binding. This competition between DNA methylation and TFs in vivo reveals a case of cooperativity between TFs that acts indirectly via DNA methylation. Methylation removal by methylation-insensitive factors enables occupancy of methylation-sensitive factors, a principle that rationalizes hypomethylation of regulatory regions.

Competition between DNA methylation and transcription factors determines binding of NRF1

Here we describe Protein Interaction Quantitation (PIQ), a computational method that models the magnitude and shape of genome-wide DNase profiles to facilitate the identification of transcription factor (TF) binding sites. Through the use of machine learning techniques, PIQ identified binding sites for >700 TFs from one DNase-seq experiment with accuracy comparable to ChIP-seq for motif-associated TFs (median AUC=0.93 across 303 TFs). We applied PIQ to analyze DNase-seq data from mouse embryonic stem cells differentiating into pre-pancreatic and intestinal endoderm. We identified (n=120) and experimentally validated eight ‘pioneer’ TF families that dynamically open chromatin, enabling other TFs to bind to adjacent DNA. Four pioneer TF families only open chromatin in one direction from their motifs. Furthermore, we identified a class of ‘settler’ TFs whose genomic binding is principally governed by proximity to open chromatin. Our results support a model of hierarchical TF binding in which directional and non-directional pioneer activity shapes the chromatin landscape for population by settler TFs.

/pdf/discovery-of-directional-and-nondirectional-pioneer-41pzpp15cc.pdf

Discovery of directional and nondirectional pioneer transcription factors by modeling DNase profile magnitude and shape

Interactions between DNA and transcription factors (TFs) guide cellular function and development, yet the complexities of gene regulation are still far from being understood. Such understanding is limited by a paucity of techniques with which to probe DNA–protein interactions. We have devised magnetic protein immobilization on enhancer DNA (MagPIE), a simple, rapid, multi-parametric assay using flow cytometric immunofluorescence to reveal interactions among TFs, chromatin structure and DNA. In MagPIE, synthesized DNA is bound to magnetic beads, which are then incubated with nuclear lysate, permitting sequence-specific binding by TFs, histones and methylation by native lysate factors that can be optionally inhibited with small molecules. Lysate protein–DNA binding is monitored by flow cytometric immunofluorescence, which allows for accurate comparative measurement of TF-DNA affinity. Combinatorial fluorescent staining allows simultaneous analysis of sequence-specific TF-DNA interaction and chromatin modification. MagPIE provides a simple and robust method to analyze complex epigenetic interactions in vitro.

/pdf/a-multi-parametric-flow-cytometric-assay-to-analyze-dna-1ijlpj6nie.pdf

John Peter van Hoff

Papers

Discovery of directional and nondirectional pioneer transcription factors by modeling DNase profile magnitude and shape

Discovery of directional and nondirectional pioneer transcription factors by modeling DNase profile magnitude and shape

A multi-parametric flow cytometric assay to analyze DNA–protein interactions