Genome-wide overlap in the binding location and function of chromatin-remodeling proteins

The transient, sequential targeting of multiple remodeler complexes to nucleosomes by specific DNA-bound factors results in the opening (right) and closing (left) of chromatin.

The transient, sequential targeting of multiple remodeler complexes to nucleosomes by specific DNA-bound factors results in the opening (right) and closing (left) of chromatin.

A single strand of DNA can stretch several meters. Yet dozens of these strands, which can be one-tenth as thin as a human hair, need to fit into the cell’s nucleus. To pack those strands into such a small space, DNA tightly winds itself around histone proteins, forming nucleosomes that are strung together into complexes called chromatin. Beyond efficiently packaging DNA, chromatin also regulates how and when DNA is used. The condensed coiling of the genome makes it inaccessible to proteins such as RNA polymerases and transcription factors that control the expression of specific genes. For DNA to become accessible local chromatin regions need to be “opened” up. This process is called chromatin remodeling, and involves the ATP-dependent removal, ejection, or restructuring of nucleosomes by large, multiprotein enzymes.

In addition to its role in regulating gene expression, chromatin remodeling plays a role in apoptosis, DNA replication, and repair. Abnormalities in chromatin remodeling proteins are often associated with human diseases, including cancer. Targeting chromatin-remodeling pathways has become a significant line of research for the treatment of several cancers.

Whereas regions of accessible chromatin have been mapped genome wide, little is known about the distribution of the various remodeler proteins, the roles they play in the chromatin reorganization process, and how those proteins interact. Gordon Hager, Ph.D., Chief of CCR’s Laboratory of Receptor Biology and Gene Expression, and colleagues have developed cell lines with dominant negative mutations in three remodelers—Brg1, Chd4, and Snf2h—in order to map the proteins’ location and investigate their genome-wide function.

Using chromatin immunoprecipitation sequencing (ChIP-seq)—a technique for determining the precise binding site of any protein—the researchers mapped the genome-wide locations of Brg1, Chd4, and Snf2h in mouse mammary epithelial cells. The remodelers were largely localized in open chromatin regions. The number of total genome-wide binding sites occupied by Brg1 was 38,896, Chd4 was 37,525, and Snf2h was 46,614. For all three proteins, about 60 percent of binding sites were located in promoters or gene bodies (introns and exons), with about 40 percent located in the regions between genes.

Pairwise comparisons of the distribution patterns of the proteins along the genome revealed significant overlap, contrary to previous studies. At least half of the sites for each remodeler were shared by the other two remodelers (59 percent, 62 percent, and 50 percent for Brg1, Chd4, and Snf2h, respectively). However, those comparisons gave no indication of whether or not the remodelers occupied the sites concurrently. Sequential ChIP analysis (re-ChIP) suggested that binding of remodeler proteins to sites of co-localization occurs sequentially, ruling out the possibility that the proteins interacted directly.

The researchers identified a handful of transcription factors (TF)—HEB, AML1, TEF, AP-1, and CTCF—associated with the remodeler protein binding sites. AP-1—a TF that controls differentiation, proliferation, and apoptosis—was found to be most associated with Brg1, while CTCF—an important regulator of the 3D structure of chromatin—was the transcription factor most closely associated with Snf2h and Chd4.

To determine the function of Brg1, Chd4, and Snf2h, Hager’s team created dominant-negative mutants for each protein, then used DNaseI-sequencing to analyze changes in DNase I hypersensitivity (DHS)—an indication that DNA has been made accessible via chromatin “opening.” While all three remodelers were found to be capable of opening and closing chromatin, Chd4 was most responsible for chromatin closing events. Because a majority of DNase I hypersensitive sites (DHSs) were found to be bound by all three remodelers, which in some cases worked synergistically and in other cases antagonistically, Hager and colleagues proposed a model in which chromatin remodelers and transcription factors facilitate chromatin’s sequential cycling through a complex series of states. However, the low temporal resolution of ChIP-seq and DNase I-sequencing makes it difficult to track dynamic changes in these states. According to the researchers, characterizing the exact sequence of these changes may be possible in the future through in vitro reconstitution of chromatin remodeling processes.

In summary, Hager and colleagues discovered considerable overlap in the genome-wide distribution and function of three chromatin-remodeling proteins. Their findings indicate a level of dynamic complexity not previously described for the function of mammalian regulatory circuits.

Summary Posted: Sun, 12/01/2013


Morris SA, Baek S, Sung M-H, John S, Wiench M, Johnson TA, Schiltz RL, Hager GL. Overlapping chromatin-remodeling systems collaborate genome wide at dynamic chromatin transitions. Nature Structural & Molecular Biology. December 8, 2013 PubMed Link