REVIEWS EPIGENETICS Chromatin accessibility and the regulatory epigenome Sandy L. Klemm1,4, Zohar Shipony1,4 and William J. Greenleaf1,2,3* Abstract | 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- binding factors Chromatin accessibility is the degree to which nuclear The accessible genome comprises ~2–3% of total Non- histone macromolecules macromolecules are able to physically contact chroma­ DNA sequence yet captures more than 90% of regions that bind either directly or tinized DNA and is determined by the occupancy and bound by TFs (the Encyclopedia of DNA elements indirectly to DNA. topological organization of nucleosomes as well as (ENCODE) project surveyed TFs for Tier 1 ENCODE chromatin- binding factors 13 Transcription factor other that occlude access to lines) . With the exception of a few TFs that are (TF). A non- histone protein that DNA. The nucleosome — a core structural element of enriched within either facultative or constitutive hetero­ directly binds to DNA. chromatin — consists of an octamer of histone proteins chromatin, the overwhelming majority of TFs surveyed encircled by ~147 bp of DNA1–8 (see Olins and Olins9 within the ENCODE project bind to open chromatin Architectural proteins for an excellent historical review). The composition and almost exclusively13. TFs dynamically compete with Proteins that have a structural role in organizing chromatin, post­ translational modification of nucleosomes reflect histones and other chromatin­binding proteins to mod­ 10 including linker and core distinct functional states and regulate chromatin acces­ ulate nucleosome occupancy and promote local access histone proteins, as well as sibility through a variety of mechanisms, such as altering to DNA13,23,24; in turn, the accessibility landscape of a insulator proteins. transcription factor (TF) binding through steric hin­ cell type modulates TF binding25–28. For multicellular drance10 and modulating nucleosome affinity for active systems, TFs have a broad range of functional roles, chromatin remodellers11. The topological organization providing dynamic regulation of transcription on short of nucleosomes across the genome is non-​uniform: timescales and establishing and maintaining persistent while densely arranged within facultative and constitu­ epigenetic canalization of cell types that share a common tive heterochromatin, histones are depleted at regulatory genome. Consequently, chromatin accessibility reflects 1Department of Genetics, loci, including within enhancers, insulators and tran­ both aggregate TF binding and the regulatory potential Stanford University, scribed gene bodies12,13. Internucleosomal DNA is often of a genetic locus. This perspective establishes an ana­ Stanford, CA, USA. bound by TFs, RNA polymerases or architectural proteins lytical foundation for tracing changes in accessibility to 2Department of Applied with linker histones, which facilitate higher­ order chro­ differential binding of transcriptional regulators that Physics, Stanford University, matin organization14–21 and regulate access to DNA. determine cellular state. Stanford, CA, USA. Nucleosome occupancy and linker histone occupancy are Recent technological advances have dramatically 3 Chan Zuckerberg BioHub, 22 San Francisco, CA, USA. variably dynamic across the genome, creating an acces­ broadened the application space of chromatin acces­ sibility continuum that ranges from closed chromatin sibility measurements by reducing biological material 4These authors contributed equally: Sandy L. Klemm, to highly dynamic, accessible or permissive chroma­ requirements to levels available clinically and by improv­ Zohar Shipony. tin (FIg. 1). This landscape of chromatin accessibility ing the discriminative capacity of these assays to iden­ *e- mail: [email protected] broadly reflects regulatory capacity — rather than a tify putative regulatory domains at both single­cell and 13,23,29–35 https://doi.org/10.1038/ static biophysical state — and is a critical determinant single­ molecule resolution . Our principal aims s41576-018-0089-8 of chromatin organization and function. in this Review are to provide an overview of recent NATURE REVIEWS | GENETICS REVIEWS s Dynamic Closed chromatin Permissive chromatin Open chromatin TF TF TF TF Pol II Fig. 1 | A continuum of accessibility states broadly reflects the distribution of chromatin dynamics across the genome. In contrast to closed chromatin, permissive chromatin is sufficiently dynamic for transcription factors to initiate sequence- specific accessibility remodelling and establish an open chromatin conformation (illustrated here for an active gene locus). Pol II, RNA polymerase II; TF, transcription factor. advances in population­ scale and single­ cell measure­ DNase­ sensitive chromatin across the genome using ments of chromatin accessibility, describe the principal short­ read sequencing (DNase I hypersensitive site biophysical determinants of accessibility and discuss sequencing (DNase­ seq))45,46 (Fig. 2a). Boyle et al.45 used the role of TFs in regulating accessibility at the nucleo­ a type II restriction enzyme to isolate and subsequently some length scale. We conclude by briefly discussing the barcode each DNase cut site (single cut), whereas functional consequences of chromatin accessibility and Hesselberth et al.46 applied strict size selection to enrich potential directions for future research. for sequenceable fragments arising from paired cleav­ age events within DHSs (double cut) (Fig. 2a). Although Measuring chromatin accessibility there is broad agreement between these sequencing Chromatin accessibility is almost universally measured approaches, the Boyle protocol may identify more by quantifying the susceptibility of chromatin to either accessible locations, whereas the Hesselberth protocol enzymatic methylation or cleavage of its constituent provides a simplified workflow and captures fewer frag­ DNA (Fig. 2). In principle, measurements of chromatin ments that originate within broadly inaccessible chro­ accessibility are dependent on the molecule for which matin (a higher signal­to ­noise ratio). Collectively, these access is being interrogated; however, remarkable con­ genome­ scale chromatin accessibility measurements servation of accessibility has been reported across a show that a minority of DHSs is found within promot­ diverse range of molecular probes29. In 1973, Hewish ers and transcription start site (TSS)­proximal regions, Nucleosome occupancy and colleagues used DNA endonucleases to fragment with over 80% of accessible regions resident within The fraction of time that a chromatin, showing that nucleosomes confer periodic distal enhancers13,43,45–47. particular sequence of DNA is 36 bound by the core histone hypersensitivity across the genome . This periodicity octamer. was probed with Southern blot hybridization, show­ ATAC-seq. Assay for transposase­ accessible chroma­ ing a canonical 100–200 bp phasing pattern among tin using sequencing (ATAC­ seq) uses a hyperactive Epigenetic canalization DNase hypersensitivity sites (DHSs) that is conserved Tn5 transposase to insert Illumina sequencing adap­ A set of persistent epigenetic 37,38 (Fig. 2b) features (alternatively, the across genomic loci. This and subsequent work tors into accessible chromatin regions . Similar 13,25,47 process of establishing this provided the earliest direct evidence for stereotypical to double­ cut DNase­ seq protocols , ATAC­ seq feature set) that molecularly nucleosome phasing. Similar techniques were used to selectively amplifies proximal double­ cleavage events defines a cell type and link chromatin remodelling with contemporaneous in accessible chromatin. ATAC­ seq measurements of comprises a continuum of transcriptional activation of the heat shock locus in accessibility are highly correlated with both double­ cut cellular states including cell 39 29,33 cycle phases and activation Drosophila melanogaster . Following the introduction (r > 0.8) and single­ cut (r > 0.75) DNase­ seq assays , 40 states. of PCR in 1985 (REF. ), a variety of quantitative methods although higher­ resolution analyses — such as for (described below) have been developed to measure site-​ TF footprinting48,49 — can reveal differences in sequence TF footprinting specific chromatin accessibility using endonucleases and bias. Owing to the high efficiency of Tn5­mediated High- resolution analysis of 41,42 chromatin accessibility data to ligation­ mediated PCR . adaptor ligation, highly complex ATAC­ seq librar­ 29,33 identify a local accessibility