How do remodelers mechanically shift DNA past the histone core?
An essential activity of most remodelers is an ability to reposition nucleosomes along DNA. What is the mechanism underlying this process? A number of distinct models have been proposed to explain nucleosome repositioning (Bowman, 2010; Clapier et al., 2017), but our recent results most strongly support a model that has been largely ignored by the remodeling community called twist diffusion (Yager & van Holde, 1985; Yager & van Holde, 2003; Winger et al., 2018). Twist diffusion results from the ATPase motor locally altering the geometry of nucleosomal DNA. The changes in geometry are coupled to the nucleotide-bound state of the remodeler: with ADP or the absence of nucleotide, DNA is slightly undertwisted, whereas a more canonical geometry is favored with ADP•BeF3- and transition state analogs (Winger et al., 2018). The initial changes in DNA twist allow the nucleosome to pull on and transiently store an extra base pair at the remodeler binding site. Returning DNA back to its canonical geometry on the nucleosome results in a corkscrew shift of nucleosomal DNA, which effectively transfers the extra base pair to another site further within the nucleosome (Bowman, 2019).
Several remodeler-nucleosome complexes captured by cryo-EM give strong support to the twist diffusion model (Farnung et al., 2017; Willhoft et al., 2018; Sundaramoorthy et al., 2018; Li et al., 2019; Yan et al., 2019; Chittori et al., 2019). DNA geometry was shown to be distorted (so-called twist defects) in two different ways. When bound to the ISWI and SWI/SNF ATPase motor in the nucleotide-free and ADP-bound states, only one strand of the DNA duplex was shifted, producing a half twist defect (Li et al., 2019; Yan et al., 2019; Chittori et al., 2019). A full twist defect, where the DNA was puckered by a full base pair addition, was captured in the SWR1-nucleosome complex bound to ADP•BeF3- (Willhoft et al., 2018).
 |
These data demonstrate that remodelers alter DNA geometry on the nucleosome, consistent with the idea that they shift DNA around the histone core through creation and elimination of twist defects. For twist defects to be converted into directional motion, the ATPase motors must act as Brownian ratchets. That is, the movement of DNA is powered by thermal motion, and the remodelers bias the direction of motion through the ATP binding and hydrolysis cycle. We hypothesize that there are two stages in particular where the remodelers behave as Brownian ratchets (Ren et al., 2019). We expect that one stage occurs when ATP binds. Prior to ATP binding, the ATPase favors a half twist defect. But the half twist defect must be converted to a full twist defect for twist diffusion to occur. We predict that ATP binding stabilizes a full twist defect, favoring a forward progression of the twist diffusion cycle.
We have proposed that a second step occurs upon ATP hydrolysis (Bowman, 2019; Bowman & Deindl, 2019). Since the twist defect state is metastable, it will spontaneously diffuse to another site on the nucleosome. When this happens, the DNA bound to the remodeler will return to a canonical conformation. Consistent with cryo-EM structures, we have shown that the canonical conformation is recognized by the ATPase motor in a state competent for ATP hydrolysis (Winger et al., 2018). Thus, after passage of a twist defect, ATP hydrolysis alters the grip of the ATPase, which blocks the original twist defect from easily returning.
Our current efforts are aimed at testing these hypotheses, and understanding the transitions between different conformational states. How is the twist defect stabilized by the ATPase motor? Why do some remodelers (Chd1, ISWI, SWI/SNF) shift DNA whereas others (SWR1) do not? Do non-remodeler ATPases with the same architecture use a similar twist-defect mechanism on non-nucleosomal substrates?
| back to Research Overview |