Structure of a RSC-nucleosome complex and insights into chromatin remodeling

Abstract

ATP-dependent chromatin-remodeling complexes, such as RSC, can reposition, evict or restructure nucleosomes. A structure of a RSC-nucleosome complex with a nucleosome determined by cryo-EM shows the nucleosome bound in a central RSC cavity. Extensive interaction of RSC with histones and DNA seems to destabilize the nucleosome and lead to an overall ATP-independent rearrangement of its structure. Nucleosomal DNA appears disordered and largely free to bulge out into solution as required for remodeling, but the structure of the RSC-nucleosome complex indicates that RSC is unlikely to displace the octamer from the nucleosome to which it is bound. Consideration of the RSC-nucleosome structure and published biochemical information suggests that ATP-dependent DNA translocation by RSC may result in the eviction of histone octamers from adjacent nucleosomes.

Figure 1

Cryo-EM RSC data and reconstruction of the RSC complex. (a) Raw images of individual RSC particles preserved in a thin layer of amorphous ice. Individual RSC particles are highlighted by yellow circles. Several different views of the complex are apparent. (b) Three-dimensional structure of RSC calculated from EM images of particles preserved in amorphous ice. Four views of the RSC reconstruction are shown at a threshold value corresponding to the total mass of the complex (~1.3 MDa). The scale bar represents 100 Å.

Figure 2

Statistical analysis of domain mobility in the RSC complex. (a) Multivariate statistical analysis was used to characterize the mobility of the bottom portion of the RSC structure (as seen in front views of the complex). Two-dimensional image classification was carried out focusing on pixels inside a mask surrounding the bottom domain (‘RSC views’, broken yellow contour). Class averages corresponding to ‘closed’ and ‘open’ conformations of the complex were obtained, as well as a third type of class average in which part of the bottom domain was undetectable (‘Missing’). Appreciable changes in the position and/or conformation of the domain are restricted to its left half (right, highlighted by the red rectangle). (b) Focused classification using a mask on either the front or back of side projections of the RSC complex revealed mobility of domains protruding forward and backward from the central RSC density. Contour plots help to show that features apparent in the three-dimensional reconstruction appear in slightly different positions in class averages generated by classification of images corresponding to a single projection of the RSC structure. Varying positions for flexible features evident in the three-dimensional RSC reconstruction are depicted in orange. Yellow contours highlight density that is consistently detected after classification of two-dimensional images but that is too mobile to appear in the three-dimensional structure at the threshold used for rendering.

Figure 3

Cryo-EM analysis of the RSC–nucleosome complex. (a) Three-dimensional reconstruction of the RSC–nucleosome complex. Two views of the complex (corresponding to views of the RSC structure shown in Figure 1) show a close correspondence to the structure of RSC alone. The threshold for the RSC–nucleosome reconstruction was set to match the size of the RSC density in the three-dimensional reconstruction of RSC. The only substantial difference between the RSC and RSC–nucleosome structures is the presence of additional density occupying the central cavity in the RSC–nucleosome reconstruction. (b) A difference map (surface represented in blue) obtained by subtracting the RSC reconstruction from that of the RSC–nucleosome complex shows the shape and location of extra density related to the presence of a nucleosome. A black, broken rectangle highlights the portion (depicted in Figure 4a) of the RSC–nucleosome structure that contains most of the nucleosome-related density.

Figure 4

Analysis of the density in the central cavity of the RSC–nucleosome reconstruction and comparison with the X-ray structure of the nucleosome. (a) A top view of the RSC–nucleosome complex (left) showing only a central slab (denoted by the broken rectangle in Figure 3b) that includes density resulting from nucleosome binding. Density presumably corresponding to the histone core (shown in solid blue) comes into close contact with RSC density (shown in yellow and gray) in several places. Non-RSC density adjacent to the central histone density is rendered as a semitransparent light blue surface. The distribution of density resulting from nucleosome binding is best appreciated after the removal (by difference mapping) of RSC density (right). (b) Comparison of presumed histone density with a low-resolution (25 Å) model of the histone core derived from its X-ray structure shows a close correspondence in shape and size. In these views, the histones are shown in space-filling representation: H2A, yellow; H2B, red; H3, blue; H4, green. See text for details. (c) The same slabbed top view of the RSC–nucleosome structure shown in a but with a model of the histones docked in place. RSC densities in close contact with the nucleosome are designated 1–3. The proximity of density 1 to the dyad suggests that it probably corresponds to the Sth1 ATPase subunit. Absence of DNA density around the dyad suggests that binding of Sth1 pulls DNA away from the histones (as indicated by the red arrow), which would explain the origin of a reported DNase I hypersensitivity site near the dyad generated by RSC binding in the absence of ATP. Possible changes in the arrangement of nucleosomal DNA are schematically represented by a black line, with regions where no DNA density is apparent in the RSC–nucleosome reconstruction shown with a broken line.