A nucleotide-driven switch regulates flanking DNA length sensing by a dimeric chromatin remodeler

Abstract

The ATP-dependent chromatin assembly factor (ACF) spaces nucleosomes to promote formation of silent chromatin. Two copies of its ATPase subunit SNF2h bind opposite sides of a nucleosome, but how these protomers avoid competition is unknown. SNF2h senses the length of DNA flanking a nucleosome via its HAND-SANT-SLIDE (HSS) domain, yet it is unclear how this interaction enhances remodeling. Using covalently connected SNF2h dimers we show that dimerization accelerates remodeling and that the HSS contributes to communication between protomers. We further identify a nucleotide-dependent conformational change in SNF2h. In one conformation the HSS binds flanking DNA, and in another conformation the HSS engages the nucleosome core. Based on these results, we propose a model in which DNA length sensing and translocation are performed by two distinct conformational states of SNF2h. Such separation of function suggests that these activities could be independently regulated to affect remodeling outcomes.

Figure 1. SNF2h dimers remodel faster than monomers and coordinate their activities via the HSS domain

(A) Top, domain architecture of wild type SNF2h. Bottom left, schematic of SpyCatcher method for covalent dimerization of SNF2h (Supplemental Experimental Procedures). Bottom right, schematic of dimeric SNF2h variants. (B) Gel shift nucleosome remodeling assay comparing wild type SNF2h to [wt]-[wt] dimer. 0/60 nucleosomes (40 nM) were remodeled with the indicated concentrations of each enzyme and stopped at various time points. (C) FRET nucleosome remodeling assay comparing wild type SNF2h to [wt]-[wt] dimer. Left, schematic of the assay setup (green donor, red acceptor) and representative kinetic traces. Right, maximum rate constants (k_max) from remodeling 7.5 nM FRET-labeled 0/60 nucleosomes with saturating amounts of each enzyme (400 nM wild type; 20 nM [wt]-[wt]). Bars are mean ± s.e.m. from three replicates. (D) Gel shift nucleosome remodeling as in (B), but with excess, saturating nucleosomes. Left, gel of 0/60 nucleosomes (660 nM) remodeled by indicated concentrations of wild type SNF2h or [wt]-[wt]. Right, initial rate of appearance of centered product. Bars are mean ± s.e.m. from four replicates. (E) Gel shift nucleosome remodeling as in (B), comparing SNF2h dimer variants from (A). (F) Comparison of maximal FRET remodeling rates of SNF2h dimer variants, as in (C). [wt]-[wt] is same as (C), re-plotted for comparison. Saturating concentrations of dimers: 20 nM [wt]-[WB]; 160 nM [ΔHSS]-[wt]; 80 nM [WB/ΔHSS]-[wt]. Bars and reported values are mean ± s.e.m. from three replicates. See also Figure S1.

Figure 2. Nucleotide state regulates binding of the HSS domain to flanking DNA or the nucleosome core

(A) Top, schematic of core and 30/30 nucleosomes. Bottom, representative binding curves showing the increase in fluorescence intensity from various concentrations of wild type SNF2h binding to 7.5 nM H4 A15C-Cy3 nucleosomes in the presence of ADP. Points are mean ± s.e.m. from three replicates, fit to a cooperative binding model. (B) Comparison of wild type SNF2h binding affinities for 30/30 (striped bars) and core (solid bars) nucleosomes in the presence of no nucleotide (apo), ADP or ADP-BeF_x. Fold changes indicate the decrease in affinity for core relative to 30/30 nucleosomes in a given nucleotide state. Error bars are s.e.m. **** p<0.0001; n.s. not significant. (C) Comparison of wild type SNF2h and SNF2h ΔHSS binding affinities for 30/30 (left, striped bars) or core (right, solid bars) nucleosomes in the presence of ADP. Fold changes and p-values reflect the decrease in affinity from deletion of the HSS domain. *** p<0.001; **** p<0.0001. (D) Same as (C), but in the presence of ADP-BeF_x. (E) Simple model for the proposed nucleotide-stabilized conformational states of SNF2h. In the ADP state (top) the HSS domain is extended, binding flanking DNA. In the ADP-BeF_x state (bottom) the HSS is retracted, helping bind the nucleosome core. For clarity, only one protomer is shown. See also Figure S2.

Figure 3. A nucleotide-driven conformational change involving the HSS domain

(A) Structural model of the SNF2h HSS domain (green) bound to a nucleosome (gray), indicating the locations of FRET probes used in (B–D). A homology model of HSS-DNA complex based on PDB:2Y9Z (Yamada et al., 2011) was generated using Phyre2 (Kelley and Sternberg, 2009) and manually docked onto the nucleosome PDB:1KX5 (Davey et al., 2002). Positions of Cy3 donor (green spheres) and Cy5 acceptor (red sphere) dyes are indicated. (B) FRET between SNF2h A930C-Cy5 (1 μM) and 0/20 Cy3 DNA(+20) nucleosomes (200 nM). Left, schematic of label positions. Middle, representative emission spectra in the presence of ADP (blue) or ADP-BeF_x (red), showing donor (~565 nm) and acceptor (~670 nm) fluorescence in response to donor excitation. Right, FRET Ratios in the presence of ADP (0.51 ± 0.004) or ADP-BeF_x (0.34 ± 0.005) given as mean ± s.e.m. from three replicates. See Figures S3E–S3F for FRET determination by two alternate methods. (C) FRET between SNF2h A930C-Cy5 (1 μM) and 0/20 H2A T11C-Cy3 nucleosomes (200 nM), in the presence of ADP (0.39 ± 0.003) or ADP-BeF_x (0.56 ± 0.003), shown as in (B). See Figures S3G-S3H for FRET determination by two alternate methods. (D) FRET between SNF2h A930C-Cy5 (1 μM) and 0/20 Cy3 DNA(−25) nucleosomes (200 nM), in the presence of ADP (0.40 ± 0.011) or ADP-BeF_x (0.49 ± 0.004), shown as in (B). **** p<0.0001, *** p<0.0005 See also Figure S3 and Supplemental Experimental Procedures.

Figure 4. Impairing DNA length sensitivity abolishes nucleosome centering

(A) Schematic showing the location of the NegC motif in the primary sequence of SNF2h, and the mutations in SNF2h mNegC. (B) Restriction enzyme accessibility assay comparing stimulation of remodeling activity by flanking DNA for SNF2h wild type and mNegC. Top, schematic of the assay workflow. Remodeling exposes a buried PstI restriction site, DNA products are separated on a native gel and fraction of cut DNA is quantified. Middle, kinetics of nucleosomal DNA cutting over time for wild type (left) and mNegC (right) remodeling core and 30/30 nucleosomes. Saturating concentrations of enzyme were used (300 nM and 900 nM for 30/30 and core nucleosomes, respectively). Points are the mean of three or four replicates, fit to a single exponential. Bottom, maximum rate constants (k_max) obtained from fits. Values are mean ± s.e.m. (C) Gel shift nucleosome remodeling assay comparing wild type SNF2h to mNegC. 0/60 nucleosomes (40 nM) were remodeled with the indicated concentrations of each enzyme and stopped at various time points. See also Figure S4.

Figure 5. A model for nucleosome remodeling by SNF2h

SNF2h initially engages nucleosomes in a non-productive ground state (top middle). Binding of the HSS domain to flanking DNA leads to productive engagement of the ATPase domain (top left and right) and ATP hydrolysis. In this DNA length-sensing state (purple box), the HSS domain binds flanking DNA, but the enzyme cannot translocate. ATP hydrolysis from the DNA length-sensing state induces a slow, rate-limiting conformational change to a translocation-competent state (vertical arrow). For clarity, this transition is only shown for one SNF2h protomer. In the translocation-competent state (orange box), ATP hydrolysis is coupled to translocation of DNA across the surface of the histone octamer. Translocation proceeds until strain on the nucleosome is released, resetting the enzyme to the length-sensing state. In this model, longer flanking DNA increases the likelihood of SNF2h undergoing a conformational change to the translocating state and, therefore, makes the overall reaction sensitive to flanking DNA length. The two SNF2h protomers take turns for binding to their respective flanking DNAs (top). This allows discrimination between long and short flanking DNA, and helps prevent the two motors from trying to move the nucleosome simultaneously in opposite directions. See also Figure S5.