ABSTRACT
The SWI/SNF chromatin remodeling complex changes the positions where nucleosomes are bound to DNA, exchanges out histone dimers, and disassembles nucleosomes. All of these activities depend on ATP hydrolysis by the catalytic subunit Snf2, containing a DNA-dependent ATPase domain. Here we examine the role of another domain in Snf2 called SnAC (Snf2 ATP coupling) that was shown previously to regulate the ATPase activity of SWI/SNF. We have found that SnAC has another function besides regulation of ATPase activity that is even more critical for nucleosome remodeling by SWI/SNF. We have found that deletion of the SnAC domain strongly uncouples ATP hydrolysis from nucleosome movement. Deletion of SnAC does not adversely affect the rate, processivity, or pulling force of SWI/SNF to translocate along free DNA in an ATP-dependent manner. The uncoupling of ATP hydrolysis from nucleosome movement is shown to be due to loss of SnAC binding to the histone surface of nucleosomes. While the SnAC domain targets both the ATPase domain and histones, the SnAC domain as a histone anchor plays a more critical role in remodeling because it is required to convert DNA translocation into nucleosome movement.
INTRODUCTION
The packing of DNA into chromatin involves the wrapping of 147 bp of DNA around a histone octamer to form a nucleosome and then assembly of nucleosomes into higher-order structures. Chromatin makes genomic DNA less accessible to protein factors important for transcription, replication, repair, and recombination. Chromatin structure is dynamic due to remodeling factors, some of which are ATP dependent, that facilitate making discrete regions accessible to other factors. ATP-dependent chromatin remodelers are single or large multisubunit assemblies composed of 1 to 17 subunits ranging from several kilodaltons to over a megadalton in molecular mass (1, 2). Each has a catalytic subunit with a conserved ATPase domain related to that of ATP-dependent DNA helicases. In helicases, this domain couples ATP hydrolysis to DNA translocation and subsequent unwinding of double-stranded nucleic acid substrates by means of a translocase domain and a duplex destabilizing wedge domain (3–5). Unlike helicases, chromatin remodelers do not have nucleic acid unwinding activity, but have retained the translocase activity, which in turn repositions or disassembles nucleosomes (5, 6).
Nucleosome movement by SWI/SNF- and ISWI-type complexes requires the ATPase domain to translocate along nucleosomal DNA near the dyad axis (7–10). DNA gaps near superhelical location 2 (SHL2) of the nucleosome block movement without interfering with binding of the remodeler. DNA translocation this far inside nucleosomes is challenging because there is no easy path for the ATPase domain to initially move. As the ATPase domain begins to translocate, it encounters histone-DNA interactions in both directions and has to overcome multiple histone-DNA interactions while trying to pull DNA into nucleosomes. The force required to disrupt histone-DNA interactions, as shown by mechanically unwrapping nucleosomes, is ∼23 pN and is a substantial force opposing the ATPase domain (11). SWI/SNF, on the other hand, was able to move nucleosomes against the inherent resistance of the nucleosomes plus an opposing mechanical force of ∼12 pN, which suggests that these enzymes could have a pulling force of >30 pN, and yet the ATPase domain alone does not appear to have a pulling force of 12 or 30 pN on free DNA, and instead a mechanical force of only ∼1 pN is required to stall SWI/SNF and RSC translocation on free DNA. Presumably the intrinsic pulling force of the ATPase domain should be the same in either case. The ATPase domain may have greater tendency to “slip” on DNA than nucleosomes, and a sufficient anchor to DNA is missing to prevent it from slipping. Evidence for tethering to the substrate being critical comes from a minimal complex of Arp7, Arp9, and a Tet dimer fused to a truncated copy of the catalytic subunit of RSC. This minimal complex can generate pulling forces of up to ∼30 pN due to the tight binding of the Tet dimer to its recognition site (12) and is much higher than that observed for the native RSC complex. There are questions that remain for SWI/SNF and RSC as to the nature of the anchor in the native complex, whether DNA or histones are the targets, and what is the protein domain that serves as an anchor?
Many of the interactions of SWI/SNF with nucleosomes are not through nucleosomal DNA but with the face of the histone octamer not bound to DNA and are likely important for remodeling. The interactions of SWI/SNF with nucleosomes have been probed by site-directed cross-linking to specific sites in DNA and histones. DNA cross-linking has shown that Snf2, the catalytic subunit of SWI/SNF, is exclusively associated with nucleosomal DNA near SHL2 and other subunits are not seen elsewhere with extranucleosomal or nucleosomal DNA (8, 13). The only exception was the Snf6 subunit, which is colocalized with the transcription activator Gal4-VP16 to its binding site in the extranucleosomal DNA region. In contrast, SWI/SNF extensively interacts with the part of the histone octamer not bound by DNA in nucleosomes. Four subunits of SWI/SNF were cross-linked to segments of the H3-H4 tetramer and H2A-H2B dimer, with Snf2 being one of the subunits most readily cross-linked (13). It seems likely that the critical anchor for SWI/SNF is histones rather than DNA. Furthermore, DNA does not seem to be the likely anchor, given that SWI/SNF does not require extranucleosomal or nucleosomal DNA, except for that bound by the ATPase domain, to move nucleosomes.
A domain in SWI/SNF that interacts with histones was identified in this study by tethering to histones a proteolytic agent that cleaves peptide bonds with hydroxyl free radicals, similar to other studies using Fe-EDTA tethered to DNA or protein for mapping protein-DNA and protein-protein interactions (14, 15). Previously this approach has shown that the ATPase and HSA (helicase SANT-associated) domains of Snf2 associate with free DNA when SWI/SNF is bound and identified the specific orientation of the ATPase domain relative to DNA (8). We now find that the SnAC and ATPase domains are associated with the histone proteins when SWI/SNF is bound to nucleosomes. Our previous studies showed that the SnAC (Snf2 ATP coupling) domain of Snf2 in Saccharomyces cerevisiae SWI/SNF is required for the ATPase and remodeling activities of SWI/SNF but not for complex integrity, substrate recognition, or recruitment function (16).
We found that without SnAC as the anchor domain, ATP hydrolysis and DNA translocation activities are uncoupled from nucleosome mobilization by SWI/SNF. Deletion of the SnAC anchor domain diminished the interactions of SWI/SNF with the histone octamer but not those with nucleosomal DNA. DNA movement inside the nucleosome is highly sensitive to the loss of the SnAC, while translocation on free DNA is not in terms of either movement rate, processivity, or pulling force.
MATERIALS AND METHODS
Nucleosome reconstitution, gel shift assays for binding, and remodeling.Mononucleosomes were reconstituted with 5.2 μg of PCR-generated DNA from the p-159-1 plasmid, which had 29 and 59 bp of DNA flanking either side of the 601 nucleosome positioning sequence (NPS) (29N59) or with 8 μg of sonicated salmon sperm DNA, 100 fmol 32P-labeled 29N59 DNA, and 9 μg wild-type Xenopus laevis octamer at 37°C by a rapid salt dilution method (17, 18). The labeled DNA was generated by PCR with an oligonucleotide labeled using Optikinase (USB) and [γ-32P]ATP (6,000 Ci/mol).
SWI/SNF complexes were purified as described previously (13). SWI/SNF was prebound to 29N59 nucleosomes (2.5 nM) containing only PCR-generated DNA at a molar ratio of 3:1 (full-binding conditions) for 15 min at 30°C and an additional 15 min at 25°C for measurement of rates of nucleosome movement. ATP was added to a final concentration of 320 μM for ΔSnAC SWI/SNF or 4.4 μM for wild-type (WT) SWI/SNF for different times, stopped, and SWI/SNF was competed off by addition of γ-thio ATP and salmon sperm DNA to final concentrations of 4.5 mM and 0.45 mg/ml, respectively. The remodeled products were analyzed on 5% native polyacrylamide gels (acrylamide/bisacrylamide ratio of 60:1) at 200 V in 0.5× Tris-borate-EDTA.
ATPase assays.ATPase kinetic assays were performed under conditions identical to remodeling kinetics with 0.055 μCi [γ-32P]ATP in a 13.5-μl reaction volume.
Site-directed mapping.A unique cysteine was engineered at the H2A 45 position (H2A45) by site-directed mutagenesis and overexpressed and purified as described previously (19). The histone mutant was refolded into octamer with WT H2B, H3, and H4. The mutant octamer was reconstituted into 29N59 nucleosomes with 1 to 4 pmol of labeled DNA. In the nucleosome structure, H2A45 is in close proximity to one strand of nucleosomal DNA 37 to 39 bp from the dyad axis. After UV cross-linking, DNA scission is initiated at the cross-linked site by alkaline conditions. p-Azido phenacyl bromide (APB) (Fluka) was added to 60 μl of reconstituted nucleosomes to a final concentration of 400 μM and incubated at 25°C for 3 h for conjugation of H2A45. Remodeling time courses were performed with reaction volumes scaled up a factor of 5, stopped by addition of γ-thio ATP, and UV irradiated for 3 min at 312 nm. Samples were denatured with 0.1% SDS for 20 min at 70°C, and histone-DNA cross-linked samples were purified by phenol-chloroform (4:1) extraction. The interphase containing the conjugates was washed three times with 1% SDS–1 M Tris-HCl (pH 8.0) and precipitated with sodium acetate and ethanol. After washing of the pellet with 70% ethanol, it was dried and resuspended in cleavage buffer (2% SDS, 20 mM ammonium acetate, 0.1 mM EDTA). DNA was cleaved at the cross-linked site by incubation in 0.1 M NaOH for 45 min at 90°C. The cleaved sample was neutralized with 2 M HCl and ethanol precipitated. Samples were resuspended in formamide and resolved by 6.5% PAGE with 8 M urea and visualized by phosphorimaging.
DNA cross-linking.Probe synthesis was performed with p-azidophenacyl bromide (APB)-modified phosphorothioate oligonucleotides that were radiolabeled close to the modification site with Optikinase (USB) and [γ-32P]ATP (20, 21). A series of modified oligonucleotides (IDT DNA) with the phosphorothioate located between the third and fourth nucleotides from the 5′ end of the oligonucleotide were designed to scan the extranucleosomal (every 3 bp) and nucleosomal (every ∼5 bp) regions. Nucleosomes containing salmon sperm DNA and labeled probe were reconstituted and photo cross-linked with SWI/SNF recruited by Gal4-VP16 as described previously (20–22). Histone cross-linking was performed as described previously (13) using SWI/SNF purified from PSY2 and PSY3 (16). The intensity of the cross-linked band of interest was normalized to the H380 Snf2 signal in the absence of ATP to find the relative cross-linking efficiency.
Single-molecule DNA translocation assays.The DNA used in the single-molecule experiments was a linearized 2.88-kb plasmid that contained a single asymmetric BsaI site on the opposite side of the plasmid from two Gal4 binding sites separated by 27 bp from a 601 high-affinity nucleosome positioning sequence. The plasmid was linearized by BsaI and ligated to two short duplex DNA molecules, which contained the 4-bp single-strand overhang that was complementary to one of the two single-strand overhangs created by the BsaI restriction enzyme. The two short double-stranded DNA (dsDNA) molecules were labeled with either biotin or digoxigenin. The 4-bp overhang that is created by BsaI is asymmetric. This ensures that the ligated linearized plasmid contained a biotin and a digoxigenin at opposite ends. The biotin-labeled end of the DNA molecule was attached to a commercial superparamagnetic bead (Dyna M280; Invitrogen) previously functionalized with streptavidin protein. The digoxigenin-labeled end was bound to a functionalized glass surface coated with antidigoxigenin.
Our magnetic tweezer setup was similar to the one described by Strick et al. (23, 24). Experiments were carried out in lab-built flow cells placed on a stage above a 60× oil immersion lens mounted on an inverted microscope. The magnetic field used to manipulate the magnetic bead was generated by two magnets. The force on the DNA was determined from the thermal fluctuations of the tethered bead in the x-y plane and the distance between the bead and the glass surface via the equipartition theorem F = kBTl/⟨Δx2⟩, where F is applied force, kB is the Boltzmann constant, l is the end-to-end distance of the DNA molecule, T is temperature, and Δx is the horizontal distance which the bead is displaced from where the DNA is attached to the surface. Data acquisition was achieved using Labview image analysis software. The program tracks the position of the bead to determine the force being applied to the DNA and the distance between the bead and the glass surface. The DNA extension can be determined with an error of ∼10 nm. The force was measured with 5 to 10% accuracy. To eliminate the effects of microscope drift, differential tracking was performed by monitoring at the same time a second polystyrene bead glued to the surface. Data were acquired using 5 nM WT SWI/SNF or 5 nM SnAC mutant and 0.25 nM Gal4-VP16 protein. The reaction buffer was 20 mM HEPES (pH 7.8), 3 mM MgCl2, 0.08% NP-40, 0.2 mM phenylmethylsulfonyl fluoride (PMSF), 2 mM β-mercaptoethanol (BME), 53 mM NaCl, 0.02% Tween 20, 0.1 mg/ml bovine serum albumin (BSA).
Nucleosome reconstitutions for single-molecule nucleosome remodeling measurements.Nucleosomes were reconstituted by salt double dialysis (25, 26) in a 50-μl volume with 0.1 μg of 2.9-kbp biotin-digoxigenin-labeled DNA plasmid, 1.0 μg of 250 bp containing the 601 DNA high-affinity nucleosome positioning sequence, 2 μg of 147-bp low-affinity DNA, 2 M NaCl, 1 mM benzamidine hydrochloride, 5 mM Tris-HCl (pH 8.0), and 0.5 mM EDTA. Samples were titrated with different amounts of histone of octamer, ranging from 1 to 3 μg, to find conditions where most of the tethers only have “1” or “0” nucleosomes. The histone octamer was prepared as previously described (27). Nucleosomes were characterized on a 5% acrylamide gel. The sample containing 1.6 μg histone octamer was used for most of the experiments presented here. This sample showed that on an average, 20 to 30% of the tested molecules have one nucleosome.
Magnetic tweezers and single-molecule nucleosome remodeling measurements.The nucleosome sample was preincubated with streptavidin-coated magnetic beads (Dynabead M280; Invitrogen) at a ratio of 1:2 for 20 min in 20 μl of 0.5× Tris-EDTA (TE) with 0.1 mg/ml BSA for 20 min. The sample was injected into the flow cell (assembled with an antidigoxigenin-coated cover glass slide) at 2 μl/s and incubated for 10 min. Before the protein is added into the flow cell, the height of the DNA was recorded as a function of time for around 300 s and at constant force of 3 pN using a lab-built magnetic tweezer apparatus. Finally, 5 nM WT or ΔSnAC SWI/SNF protein and 1 mM ATP flowed in. Addition of 0.5 nM Gal4-VP16 protein increased the probability of observing an event in a 5- to 10-min period. The height of the DNA in the presence of the protein was recorded for ∼300 s. The buffer conditions were identical to those described for the DNA experiments.
We determined the number of nucleosomes within a single-nucleosome sample by detecting the number of steps in the tether extension as the force was increased from 3 to 26 pN. The force was initially increased to 3 pN over 20 s, was increased from 3 to 26 pN over 100 s, and then was held at 26 pN for 250 s. As an additional control, the force was reduced to 0.3 pN and the height of the DNA was again tracked for ∼300 s. This last control determines if the protein was active during the measurement and whether or not the sample contained one or zero nucleosomes.
A total of 27 remodeling events were analyzed independently. Based on the Brownian fluctuations, an 80- ± 10-bp threshold was set for determining the change in extension of a DNA molecule due to its interaction with SWI/SNF. Rates were estimated by analyzing each remodeling event. The shortening rates were fitted using straight lines with an average of 100 ± 50 bp/s. The average time of the remodeling event was estimated to be 20 ± 10 s.
Conjugation of Fe-BABE to nucleosomes and mapping of Fe-BABE-mediated cleavage of Snf2.The 601 nucleosome positioning sequence (NPS) DNA prepared by PCR was used for nucleosome assembly. The DNA has biotin incorporated at one 5′ end, and the NPS has 69 and 60 bp of flanking DNA with Gal4-binding site within the flanking 60-bp stretch (biotin-69-601-60). Conjugation of Fe-BABE [Fe(III) (S)-1-(p-bromoacetamido-benzyl)EDTA] to the solvent-accessible lysines in the nucleosome was done using 2-iminothiolane (2-IT) to attach a free sulfhydryl group to the accessible lysines, as previously reported with some minor modifications (28, 29). Briefly, nucleosomes (0.4 μM) were incubated at 37°C for an hour with 0.85 mM Fe-BABE and 0.425 mM 2-IT in buffer containing 10 mM MOPS (morpholinepropanesulfonic acid; pH 8), 1 mM EDTA, 5% glycerol, 0.1% NP-40, and 0.5 mM PMSF. The control nucleosomes were incubated with Fe-BABE alone. Excess Fe-BABE and 2-IT were removed using a Sephadex G-50 spin column, and conjugation of Fe-BABE to the nucleosomal histones was analyzed by Western blotting using antichelate CHA225 antibody (28, 29).
Mapping of Fe-BABE-mediated cleavage of Snf2 was performed as described previously (8). The cleavage products were analyzed by SDS-PAGE and immunoblotting using antihemagglutinin (anti-HA) antibody (Pierce, Rockford, IL) against the C-terminal HA tag of Snf2. Cleavage sites, and hence the nucleosome-interacting regions of the SWI/SNF subunits, were determined by using truncated fragments of the same protein as molecular weight standards (30).
Histone cross-linking and label transfer.Mononucleosomes (29N59) containing mutant histones with unique cysteine residues (H2A19, H2A89, H2B109, and H380) were assembled and analyzed on native PAGE with ethidium bromide staining. β-Mercaptoethanol (from histone octamer) was removed by passing the assemblies through a Sephadex G-25 spin column. Sixteen nanomoles of PEAS {[N-(2-pyridyldithio) ethyl]-4-azidosalicylamide}, a radioiodinatable, cleavable, and photoactivatible cross-linking reagent from Molecular Probes, in dimethyl sulfoxide was radiolabeled with 2.5 mCi 125I in a 90-μl reaction volume adjusted with 100 mM sodium phosphate buffer (pH 7.4). Modification was carried out in an Iodogen (Pierce) iodination tube for 3 min at room temperature and stopped with 1 μl 2.5 mM tyrosine and 1 μl 80 mM methionine. Twenty picomoles of nucleosomes was modified with a 20 molar excess of iodinated PEAS for 30 min on ice. Unconjugated PEAS-125I and free 125I were removed by passing the sample through a Sephadex G-25 spin column. The modified nucleosomes were dialyzed against final dilution buffer (10 mM Tris-HCl [pH 7.5], 1 mM EDTA, 0.01% NP-40, 5% glycerol). The unmodified and modified nucleosomes were run on 4% native gel and ethidium bromide stained. WT or mutant SWI/SNF complexes were bound to the modified nucleosomes and incubated at 30°C for 30 min, and the bound products were separated on a 4% native polyacrylamide gel (acrylamide/bisacrylamide ratio of 36:1) at 200 V in 0.5× Tris-borate-EDTA at 4°C. SWI/SNF bound to nucleosomes was cross-linked by UV irradiation. “Plus ATP” samples were first prebound at 30°C for 30 min, and then 800 μM ATP was added for 10 min before nucleosome remodeling was stopped by UV irradiation. Dithiothreitol (DTT) was added to a final concentration of 100 mM to reduce disulfide bonds and transfer the 125I label to the cross-linked target protein. The samples were loaded for 4 to 20% SDS-PAGE, and radiolabeled subunits were identified by phosphorimaging.
RESULTS
The ATPase and SnAC domains of Snf2 associate with the histone portion of the nucleosome.The interactions of SWI/SNF with the histone octamer were mapped by targeted proteolysis to identify the domain(s) that anchors SWI/SNF to the nucleosome and facilitate the creation of an effectively higher pulling force. Fe-BABE [Fe(III) (S)-1-(p-bromoacetamido-benzyl)EDTA], an Fe-EDTA derivative, was covalently linked to the lysine residues on the nucleosome surface, including histone tails using 2-iminothiolane (2-IT) (28, 29). Fe-BABE, developed by Claude Meares and colleagues (14, 15), can be used to create hydroxyl free radicals that cleave peptide binds. The extent of Fe-BABE conjugated to histones was assessed by immunoblotting using chelate-specific CHA255 antibody (Fig. 1B). Modification of the histones had no apparent effect on SWI/SNF binding to nucleosomes, as shown by gel shift assay (data not shown). Fe-BABE-modified nucleosomes were bound to WT SWI/SNF, and cleavage was initiated by adding H2O2 and ascorbate. The SWI/SNF used in these experiments had a hemagglutinin (HA) epitope tag at the C terminus of Snf2, and immunoblotting detected full-length and proteolytic fragments containing the C-terminal end of Snf2 (8). Accurate determination of the cleavage sites was done using C-terminal HA epitope-tagged Snf2 polypeptides prepared by in vitro translation for size markers. Cleavage sites were mapped within an ∼10-amino-acid region, as shown previously (30). The SnAC and ATPase domains were both found to be in close proximity to histones. Snf2 was cleaved by modified histones near amino acid residues 810, 1098, and 1342 (Fig. 1A, lanes 2, 3, 5, and 6, and C, black arrows). The main Snf2 cleavage site was at amino acid 810 and is located in the N-terminal lobe of the ATPase domain between motifs I and Ia. The second, relatively weaker cleavage site was at approximately amino acid 1098 and is in the C-terminal lobe close to motif IV (31). These two cleavage sites flank a region previously shown to be cross-linked to nucleosomal DNA 17 and 18 bp from the dyad axis (8) and are consistent with SWI/SNF binding to nucleosomes (Fig. 1D). The third cleavage site was at approximately amino acid 1342 located inside the SnAC domain. Cleavage at these three sites within Snf2 was dependent on modification of the accessible lysines, as seen when 2-IT is omitted (Fig. 1A, compare lanes 2 and 3 with 8 and 9). Also, no cleavage was observed when ascorbate and hydrogen peroxide were omitted (Fig. 1A, compare lanes 1 and 2 or lanes 4 and 5). These contacts were not changed upon remodeling, as shown by the addition of ATP and are consistent with these interactions being maintained during nucleosome movement (Fig. 1A, compare lanes 2 and 3 or lanes 5 and 6). The addition of Gal4-VP16 did not significantly alter the contacts of Snf2 with the nucleosomal histones, although cleavage around amino acid 1098 appeared to be slightly stronger (Fig. 1A, compare lanes 2 and 3 with lanes 5 and 6). The interactions of SWI/SNF with free DNA have been previously investigated in a similar manner by tethering Fe-BABE to DNA and are summarized in Fig. 1C (8). Cleavage at the SnAC domain is specific for nucleosomes and modified histones and is not observed when SWI/SNF binds to free DNA only. These data highlight that the likely target for the SnAC domain is histone proteins when SWI/SNF is bound to nucleosomes.
The SnAC domain associates with nucleosomal histones. (A) The interactions of Snf2 in the context of the whole SWI/SNF complex with nucleosomes were probed by Fe-BABE conjugated to solvent-accessible lysine residues in nucleosomes. Cleavage of Snf2 was detected by immunoblotting using an antibody against the C-terminal HA epitope of Snf2. Nucleosomes were modified with 2-iminothiolane (2-IT) and Fe-BABE (lanes 1 to 6) or with Fe-BABE alone (lanes 7 to 9) as a negative control. ATP (1 mM) was added in lanes 3 and 6, and the mixture was incubated before addition of ascorbate and hydrogen peroxide. (B) The successful coupling of Fe-BABE to nucleosomes was detected by immunoblotting using the antichelate antibody CHA225. Nucleosomes were incubated with Fe-BABE alone (lane 2) or Fe-BABE and 2-IT (lane 3). (C) A map of the Snf2 domains, including the ATPase domain with conserved helicase motifs is shown. The other domains include the QLQ and HSA (helicase-SANT-associated) domains on the N-terminal half and the SnAC (Snf2 ATP coupling), tandem AT hooks and the bromodomain (Bromo) in the C-terminal half of Snf2. The regions of Snf2 shown to contact free DNA (gray arrows) and the histone octamer (black arrows) by Fe-BABE-mediated proteolysis are shown. The thickness of the arrows is proportional to the frequency of cleavage. (D) A structural model of the ATPase domain bound to DNA based on the crystal structure of Rad54 is shown. The regions shown to be close to histones by cleavage or to nucleosomal DNA by cross-linking are highlighted in light gray.
Absence of the SnAC domain reduces binding of SWI/SNF to the open histone octamer face of nucleosomes.Another approach was used to find whether the interactions of SWI/SNF with the histone octamer face are perturbed when the SnAC domain is absent. Unique cysteines were engineered into four locations on the histone octamer surface (Fig. 2A) and were designed to not perturb the nucleosomes' structure or stability and their ability to be remodeled; they were previously used for site-directed histone photo cross-linking (13). The SnAC domain was found to be required for stable binding of Snf2 to the histone portion of nucleosomes in the absence of ATP as there was a 2- to 3-fold reduction in Snf2 cross-linking at all four positions when the SnAC domain was removed (Fig. 2B to E). The interactions of the Snf5 and Swp82 subunits of SWI/SNF were also reduced upon deletion of the SnAC domain (Fig. 2D and E). Remodeling decreased the extent of Snf2 and Snf5 cross-linking at residues 19 and 113 of H2A to levels similar to the reduction observed when the SnAC domain was removed (Fig. 2C and D, compare open bars to bars with the different shades of gray). Remodeling only weakly decreased or did not decrease at all Snf2, Snf5, and Swp82 cross-linking at residue 109 of H2B and residue 80 of H3. The SnAC domain as shown by histone cross-linking has a general role in establishing the interactions between SWI/SNF and the histone octamer face of nucleosome. Taken together, these data indicate that the SnAC domain interacts with the histone portion of nucleosomes and stabilizes SWI/SNF binding to histones.
Loss of SnAC reduces interactions of SWI/SNF with the open face of the histone octamer. (A) The locations of the four positions in the open face of the histone octamer inside nucleosomes used to probe for the interactions of SWI/SNF are shown. A side view and top view are displayed. (B) Modified nucleosomes were bound to ΔSnAC (lanes 3, 5, 7, and 9) and WT SWI/SNF (lanes 2, 4, 6 and 8) in the absence of ATP (B) or the presence of 800 μM ATP (gel not shown). After cross-linking, the radiolabel was transferred, and the subunits of SWI/SNF were resolved by 4 to 20% SDS-PAGE. Labeled proteins were detected by phosphorimaging. Lane 1 had photoreactive H380 nucleosomes, but no SWI/SNF was added before cross-linking. The relative amounts of Snf2 (C), Snf5 (D), and Swp82 (E) cross-linked (X-link) at the different histone positions for WT SWI/SNF (−ATP and +ATP) and ΔSnAC SWI/SNF (−ATP and +ATP) are shown. The cross-linking experiments were done in triplicates, and standard deviations are shown.
Snf2 remains in contact with nucleosomal DNA in the absence of the SnAC domain.A histone anchor domain in SWI/SNF may or may not influence the binding of the complex to nucleosomal DNA. Previously Snf2 was found to contact nucleosomal DNA at superhelical location 2 (SHL2) by site-directed DNA cross-linking and to translocate on DNA from SHL2 in a 3′-to-5′ direction (10, 13). Potential changes or loss of Snf2 interactions with nucleosomal DNA due to deletion of the SnAC domain were determined by site-directed DNA photo cross-linking. Every ∼5th bp in nucleosomal DNA corresponding to positions facing in or away from the histone octamer core and every 3rd bp in extranucleosomal DNA were scanned by DNA cross-linking. The precise location of the phosphate backbone with regard to its position near the histone octamer had been previously determined for the 601 nucleosome by hydroxyl radical footprinting (32). The photoreactive group was attached to the phosphate backbone and probed contacts in both the major and minor grooves of DNA (20, 22, 33). After UV irradiation, a radiolabel was transferred to the cross-linked subunit by enzymatic digestion of DNA and the subunit was identified by SDS-PAGE and phosphorimaging. WT or ΔSnAC SWI/SNF was recruited by Gal4-VP16 onto mononucleosome substrates. Snf2 was cross-linked most efficiently 17 bp from the dyad axis with WT SWI/SNF and corresponds to a position pointing in toward the histone octamer (Fig. 3A to C). Peptide mapping studies of the cross-linked Snf2 subunit suggest the ATPase C-terminal lobe (between motifs IVa and V) wedges between the DNA and histone surface at this location (8). The other two positions next to bp −17 face away from the histone octamer at bp −22 and −11 and are less efficiently cross-linked (Fig. 3B). Generally, Snf2 is cross-linked to ∼3 helical turns of DNA from bp −33 to 0 and is centered at the SHL2 position. Except for two positions toward the entry site at bp +52 and +63 on the side of nucleosomal DNA facing away from the histone octamer, Snf2 is not significantly cross-linked to DNA elsewhere in the nucleosome and is highly localized.
Loss of the SnAC domain does not reduce the interactions of Snf2 with nucleosomal DNA. (A) SWI/SNF was recruited to nucleosomes by Gal4-VP16 binding 27 bp from the entry site of the nucleosome, as described in Materials and Methods. The cross-linked proteins were analyzed by 4 to 20% SDS-PAGE and detected by phosphorimaging. A representative of several gels for WT (top panel, lanes 1 to 8) and ΔSnAC (bottom panel, lanes 9 to 16) SWI/SNF is shown for cross-linking with nucleosomal DNA at sites spanning from bp −17 to −54. Cross-linking of WT Snf2 at bp −17 (lanes 8 and 17) was used as the reference in all gels for normalization. (B and C) Mapping the interactions of Snf2 with DNA by DNA cross-linking was done at 27 positions along nucleosomal DNA and 10 along extranucleosomal DNA. The nucleosomal positions probed every helical turn facing either in (B) or away from (C) the histone octamer. Every 3rd bp was probed in extranucleosomal DNA (B). The amount of label transferred to the Snf2 subunit was quantified and normalized to the amount of label transferred to Snf2 when cross-linked at bp −17 using WT SWI/SNF (relative cross-linking). The numbering system is relative to the nucleosomal dyad axis as zero, with positions to the left being negative and those to the right being positive.
Snf2 is associated with nucleosomal DNA at the SHL2 position in ΔSnAC SWI/SNF, as seen by efficient DNA cross-linking of Snf2 at bp −17 and −33, comparable to WT SWI/SNF (Fig. 3B and C). There are, however, differences in the DNA cross-linking patterns, indicating that the manner in which Snf2 is bound is altered. The most notable difference is the additional positions to which Snf2 is efficiently cross-linked to DNA with ΔSnAC SWI/SNF. Only with ΔSnAC SWI/SNF was there relatively strong cross-linking of Snf2 inside at bp −38 and others on the exposed surface of nucleosomal DNA at bp 0 and +42 (Fig. 3B and C). These data suggest that the binding of Snf2 to nucleosome DNA in the nucleosome is broadened in the absence of the SnAC domain and potentially has greater flexibility to scan additional proximal DNA sites. Given that ΔSnAC SWI/SNF retains contact with nucleosomal DNA, it should be able to hydrolyze ATP and translocate on DNA, but it may not be able to mobilize nucleosomes without a histone anchor.
Lack of SnAC domain uncouples nucleosome mobilization from ATP hydrolysis and DNA translocation.We examined if loss of the SnAC domain caused an uncoupling of nucleosome mobilization from ATP hydrolysis. Previously, ΔSnAC SWI/SNF was shown to hydrolyze ATP 7- to 8-fold less than WT (16), and in order to better assess remodeling differences separate from that of ATP hydrolysis, a higher concentration of ATP was used with ΔSnAC SWI/SNF (320 μM) than with wild-type SWI/SNF (4 μM). Under these conditions, the rate of hydrolysis was ∼2 times faster for ΔSnAC SWI/SNF than wild-type SWI/SNF (Fig. 4A [1.2 versus 0.65 nM s−1 for ΔSnAC versus WT]). Nucleosome remodeling was followed by measuring changes in the electrophoretic mobility of nucleosomes caused by their shifting positions on DNA. Under these conditions favoring ΔSnAC SWI/SNF, the wild-type SWI/SNF remodeled nucleosomes about 120 times more efficiently (Fig. 4C and D [26 pM s−1 versus 0.21 pM s−1] and Table 1), as determined by electrophoretic mobility shift assay (EMSA). The approximate number of ATPs hydrolyzed per remodeling event by ΔSnAC SWI/SNF was on average 5,700, as determined from the rate of ATP hydrolysis (1.2 nM s−1) divided by the rate of remodeling (0.21 pM s−1), and in comparison, WT SWI/SNF hydrolyzed about 25 ATPs per remodeling event (0.65 nM ATP s−1 versus 0.026 nM nucleosome s−1). This implies that WT SWI/SNF was 270 times more efficient at converting ATP hydrolysis into nucleosome movement. Nucleosome remodeling was also assayed by restriction enzyme (RE) accessibility, which measures the relative rates of nucleosomal DNA cleavage due to site exposure (46). Cleavage of the HhaI site near the dyad axis was monitored over time with WT and ΔSnAC SWI/SNF under the same conditions mentioned before. Exposure of the HhaI site is due to formation of DNA bulges on the surface of the nucleosomes and/or nucleosome movement away from its original position. The extent of site exposure with the ΔSnAC complex was ∼64 times lower than with the WT complex (Table 1) (4.5 versus 0.07 pM DNA cut s−1 for WT versus ΔSnAC), which is comparable to the reduction in nucleosome movement observed by EMSA (data not shown).
The SnAC domain couples ATP hydrolysis of SWI/SNF to nucleosome movement. The rates of ATP hydrolysis of the WT (A) and ΔSnAC (B) SWI/SNF were measured under full-binding conditions with 6.4 nM 29N59 mononucleosomes (containing 29 and 59 bp of extranucleosomal DNA) and 20 nM SWI/SNF. The ATP concentrations used for WT and ΔSnAC SWI/SNF were 4.4 μM and 320 μM, respectively. The time scale was also different for WT SWI/SNF (0 to 10 min) versus that for ΔSnAC SWI/SNF (0 to 120 min). (C and D) The rate of nucleosome movement by WT (lanes 1 to 8) and ΔSnAC (lanes 9 to 17) SWI/SNF was monitored by a gel shift assay under the same conditions as in panels A and B. The amounts of nucleosomes moved versus time for WT and ΔSnAC SWI/SNF are compared.
Kinetic parameters of the ΔSnAC complex obtained from Michaelis-Menten analysis, remodeling, and ATPase assays
The uncoupling of ATP hydrolysis from nucleosome movement could be due to defects in DNA translocation such as that observed for mutations in motif V of the ATPase domain (34). Single DNA molecules with one end tethered to a glass slide and the other to a magnetic bead (Fig. 5A) were used to directly examine the rate of DNA translocation of both ΔSnAC and wild-type SWI/SNF, as previously reported for RSC (35). The height of the bead was recorded at different tension forces lower than 2 pN in the presence of ATP with and without SWI/SNF. DNA was observed to be shortened with SWI/SNF and ATP due to translocation of SWI/SNF and formation of DNA loops (Fig. 5B and C), as reported previously (36). The ATP concentration ranged from 0.2 μM to 10 mM, and the maximum rate of DNA translocation observed was at ≥1 mM ATP (data not shown). In the absence of protein, the DNA molecules undergo restricted Brownian fluctuations, as expected (data not shown). SWI/SNF was recruited to a unique location on the 2.88-kb DNA fragment by Gal4-VP16 binding to two adjacent sites in the middle of the DNA template (8, 13). The recruitment of SWI/SNF by Gal4-VP16 made it possible to observe more DNA translocation events mediated by SWI/SNF at higher tension forces, where thermal fluctuations could be further suppressed and allow for shorter translocation events to be resolved. The selective recruitment of SWI/SNF by Gal4-VP16 was demonstrated with DNA shortening by SWI/SNF being dependent on Gal4-VP16 when competitor DNA was present (data not shown). Recruitment of SWI/SNF by Gal4-VP16 did not increase the force required to stall translocation by SWI/SNF.
SnAC is not required for translocation on free DNA, but is on nucleosomes. (A) The magnetic tweezer geometry for detection of SWI/SNF translocation events along a single DNA molecule. (B) Time traces of the end-to-end distance of a 2.88-kb DNA molecule with and without 5 nM SWI/SNF. The gray points are the raw data, while the black curve is a 1-s rolling average of time series. The long spikes that correspond to DNA shortening caused by SWI/SNF (bottom time trace) are not present in the absence of the SWI/SNF complex (top time trace). (C) Two of the characteristic translocation events from the bottom time trace in panel B are shown. The translocation rates were determined by fitting individual events using a straight line (dashed line). We observed two distinct types of translocation events. The top time series shows an event where SWI/SNF shortens the DNA end-to-end distance, stalls, and then reverses directions. The bottom time series shows an event where SWI/SNF shortens the end-to-end distance and then reverses direction without any significant stall time. Both types of translocation events were previously reported for RSC. (D) Force dependence of DNA shortening measured at 1 mM ATP for WT and ΔSnAC SWI/SNF complexes. (E) Force dependence of the translocation rates for WT and ΔSnAC SWI/SNF. Data were taken at 1 mM ATP, in the presence of Gal4-VP16. (F) Example time traces of the end-to-end distance of a 2.88-kb DNA molecule containing 1 nucleosome, where the flow cell contains 5 nM WT or ΔSnAC SWI/SNF and 1 mM ATP. Changes in end-to-end distances were observed under a pulling force of 3 pN. A remodeling event is circled in the upper trace. Totals of 15 traces for WT SWI/SNF and 13 traces for ΔSnAC SWI/SNF were obtained with tethers containing a single nucleosome.
We repeated the same magnetic tweezer experiment with the ΔSnAC SWI/SNF and observed transient shortening of DNA length similar to that for wild-type SWI/SNF. Changes in DNA extension greater than the maximum height change induced by Brownian motion (determined in the absence of protein) were analyzed in order to compare the extent and rate of translocation of WT or ΔSnAC SWI/SNF on free DNA. The signal was smoothed with a rolling average of 1 s. The peaks with drops greater than peak drops due to Brownian fluctuations were identified with MatLab. Once a translocation event or spike was identified and the magnitude determined, the rate of translocation or velocity was fit using straight lines, as shown in Fig. 5C. We found that at 0.3 pN, the average loop size on DNA is 580 ± 20 bp for the WT and 550 ± 70 for the ΔSnAC mutant (Fig. 5D and Table 1). The levels of force dependence of DNA shortening for both WT and ΔSnAC SWI/SNF are very similar, as shown in Fig. 5D. At 0.3 pN, we found the rate of translocation is 600 ± 100 bp/s for the WT and 500 ± 100 for the ΔSnAC mutant (Fig. 5E and Table 1). Single-molecule measurements showed that mutants lacking the SnAC domain have the same average loop size and translocate at the same rate as WT along duplex DNA, and therefore the SnAC domain is not required for SWI/SNF translocation along duplex DNA.
The SnAC domain is essential for translocation on nucleosomal templates.A similar setup was used to examine the rate of nucleosome movement and DNA translocation in a nucleosomal context. The same DNA template as before containing a 601 nucleosome positioning sequence was used to reconstitute and position a single nucleosome on DNA. SWI/SNF translocation of nucleosomes and shortening of DNA tolerate significantly higher tension forces than translocation of SWI/SNF on free DNA, as shown previously (36). A single nucleosome was observed to be bound to each DNA as the change in end-to-end distances observed when adjusting the tension force from 0.3 to 26 pN corresponded to that expected for unwrapping one nucleosome from DNA (data not shown). Changes in DNA length were observed at 3 pN when WT SWI/SNF and ATP were added with shortening of the end-to-end distances, but no shortening was observed with ΔSnAC SWI/SNF (Fig. 5F). In 13 of 15 traces, there was clear evidence of DNA shortening with WT SWI/SNF, but in 13 traces with ΔSnAC SWI/SNF, no DNA shortening was observed. There was no inherent problem with these templates, as shown by WT and ΔSnAC SWI/SNF being able to equally translocate along the free DNA portion of the template when the tension force was reduced to 0.3 pN (data not shown). Single-molecule experiments show that in the absence of SnAC, SWI/SNF is unable to move DNA through nucleosomes creating DNA loops and thereby shortening the DNA tether.
In order to examine more carefully the molecular changes in DNA and histone interactions associated with nucleosome movement, we monitored the interactions of residue 45 of histone H2A with DNA before and during remodeling. Alanine 45 of histone H2A was changed to a cysteine and conjugated to an aryl azide to probe these interactions. After cross-linking to DNA, the DNA site is labile under alkaline conditions and the cleavage site mapped with base pair resolution. Residue 45 is close to nucleosomal DNA 37 and 39 bp from the dyad axis and in one of its two orientations is ∼15 bp from where the ATPase domain of Snf2 is normally bound (8, 19). SWI/SNF was bound to nucleosomes, and the position most proximal to the ATPase domain binding site was monitored. WT SWI/SNF moved nucleosomes away from the original position, as monitored by reduction of cutting at bp −37, with almost half having moved after 10 s and nearly completely moved after 160 s (Fig. 6A and B). Within 10 s, the cross-linked site was shifted to bp −87 and was 50 bp from its original position, placing the edge of the nucleosome 22 bp past the DNA end. Later after 40 s, a new cross-linked position was also seen at bp +72 that represents a 110-bp step to the other side with the longer extranucleosomal DNA, placing the nucleosome 49 bp off the other edge of DNA (Fig. 6B). ΔSnAC SWI/SNF did not significantly move DNA inside nucleosomes from its original position, and there doesn't appear to be any significant movement of nucleosomal DNA ∼15 bp away from where the ATPase domain is likely bound (Fig. 6C). Even limited movement of nucleosomal DNA is not observed without the SnAC domain of SWI/SNF. Given that Snf2 is still bound to nucleosomal DNA, hydrolyzes ATP, and can equally well translocate along DNA without the SnAC domain, it is very significant that nonetheless it is unable to move DNA inside the nucleosome ∼15 bp from where the ATPase domain is bound. These data highlight the critical importance of a histone anchor in SWI/SNF to facilitate in the initial movement of DNA inside nucleosomes.
ΔSnAC SWI/SNF cannot move nucleosomal DNA close to the binding site of the ATPase domain. (A) Nucleosome positions were mapped using site-directed cross-linking and alkali-mediated cleavage at different times during remodeling. The photo cross-linker was placed at residue 45 of histone H2A, and alkali-induced cleavage of DNA was monitored with the bottom strand radiolabeled. The original cut site was at bp −37 (black rectangle), and the DNA translocation site was 15 to 25 bp from the dyad axis (yellow rectangle). Remodeling was initiated by the addition of ATP, and changes in histone-DNA interactions were examined after 10 s (red), 40 s (green), and 160 s (purple). (B and C) Lane profile overlay analyses of the sequencing gels from site-directed mapping experiments are shown for WT (B) and ΔSnAC (C) complexes. The x and y axes denote base pair positions (with the dyad being zero, left of the dyad being negative, and right of the dyad being positive) and cross-linking/cleavage signal in arbitrary units (AU), respectively. The time-dependent decrease in cross-linking/cleavage signal at the original nucleosome position is shown in the left panel. The appearance of cross-linking/cleavage signal upon remodeling is shown at the −87 (middle panel) and +72 (right panel) positions, indicating 50 bp of movement toward the shorter linker and 110 bp of movement toward the longer linker, respectively. Note the lack of movement measured by either disappearance of cross-linking/cleavage at the original nucleosome position or the appearance of cross-linking/cleavage at the −87 or +72 positions in the case of the ΔSnAC complex (C).
DISCUSSION
Conversion of ATP hydrolysis and DNA translocation by SWI/SNF into nucleosome movement requires the SnAC domain to bind to histones.Deletion of the SnAC domain of Snf2 uncouples ATP hydrolysis and DNA translocation by SWI/SNF from nucleosome mobilization. The SnAC domain is evolutionarily conserved in all eukaryotic SWI/SNF complexes and is essential for the in vitro and in vivo activity of SWI/SNF (16). Although SnAC enhances ATP hydrolysis by SWI/SNF, its role is more pronounced for being required to convert ATP hydrolysis and DNA translocation into changes of nucleosome translational positions. The two roles of SnAC are distinguished from each other by using different concentrations of ATP with each in order to have equivalent rates of ATP hydrolysis with wild-type SWI/SNF and SWI/SNF lacking the SnAC domain. Even when ΔSnAC SWI/SNF hydrolyzed ATP two times faster than wild-type SWI/SNF, ΔSnAC SWI/SNF moved nucleosomes ∼120 times slower than wild-type SWI/SNF. To our knowledge, the extent and type of uncoupling shown for the SnAC domain have never been observed before for any ATP-dependent chromatin remodeler. Most uncoupling of chromatin remodelers occurs at a level generally only <5 times that of wild type and involves a reduction in DNA translocation compared to ATP hydrolysis. Previously, mutations in motif V of the ATPase domain of SWI/SNF have uncoupled ATP hydrolysis from DNA translocation (34, 37–39). These mutations are in a motif that is conserved in multiple DNA helicases, and in some of these instances, this motif has been shown to directly interact with DNA. Changes in motif V of Snf2 probably uncouple ATP hydrolysis from DNA translocation by causing the enzyme to hold less tightly to DNA. Loss of the SnAC domain, however, does not have this effect as ΔSnAC SWI/SNF translocates on single DNA molecules with rates and total distances traversed comparable to those of wild-type SWI/SNF. Besides mutations within the ATPase domain, a domain outside the ATPase domain known as the HSA domain when removed reduces remodeling efficiency about a factor of <2 (40). The HSA domain binds to the actin-related proteins Arp 7 and 9, which are shared between RSC and SWI/SNF. The HSA domain likely stimulates chromatin remodeling through the interaction of the Arp 7 and 9 subunits (41–43). The HSA domain also has a propensity to bind to free DNA, which may also enhance the remodeling activities of RSC and SWI/SNF (8). The SnAC domain, unlike the HSA domain, does not bind or recruit other subunits to the SWI/SNF complex or bind DNA like the HSA domain and seems to have a very different target.
The SnAC domain is a histone anchor required to create sufficient force to mobilize nucleosomes.SWI/SNF requires a stable anchor to create a sufficient pulling force to move nucleosomes when the ATPase is bound at SHL2. The strong uncoupling of ATP hydrolysis and DNA translocation from nucleosome movement when the SnAC domain is removed is characteristic of a crucial histone anchor. The SnAC domain satisfies several of the criteria expected for a histone anchor of SWI/SNF. First and foremost, SnAC appears to bind to the histone component of nucleosomes, as shown by an innovative protein footprinting technique using an artificial protease tethered to the histones in nucleosomes. The only other domain shown to interact in this manner is the ATPase domain. Consistent with this finding is the observation that when SnAC is missing, the interactions of SWI/SNF with the open histone octamer face of the nucleosome are reduced overall, as seen by site-directed cross-linking. These effects are specific to histones as Snf2 is still observed to bind well to nucleosomal DNA, as would be expected for a histone anchor domain that uncouples DNA translocation from moving nucleosomes. In support of our work, when the ATPase-helicase domains are switched between BRG1 and SNF2h, a region from residues 1250 to 1386 of BRG1, including the SnAC domain, was required in addition to the ATPase domain to retain the remodeling activity of any BRG1 chimeras (44). In another study, a region from amino acids 1223 to 1420 from BRG1 encompassing the SnAC domain (residues 1332 to 1390) was found to be essential for remodeling and histone H3 interactions (45). These data show the SnAC domain is important for regulation of SWI/SNF activity in both yeast and humans. The other criterion is that SnAC is exquisitely required for SWI/SNF to move nucleosomal DNA even short distances from its original positions when proximal to the bound ATPase domain. This was shown using a method to track changes in histone-DNA interactions with base pair resolution 37 bp from the dyad axis.
Our model for the SnAC domain is that it binds to the histone octamer and provides an anchor for the ATPase domain (Fig. 7). This anchor is required for the ATPase domain to gain traction and create a sufficient pulling force to move DNA inside nucleosomes. Just like a Tet dimer fused to the Sth1 catalytic subunit converts it from being able to create a pulling force of ∼1 pN to being able to create a pulling force of almost 30 pN (12), the SnAC domain as an anchor also can make the ATPase domain of Snf2 have a significantly higher pulling force. Thus, the pulling force of a chromatin remodeler may be more a function of the strength of its anchor to the nucleosomes than the strength of the ATPase domain alone. It will be important to find how other chromatin remodelers interact with nucleosomes and potentially have key anchors like SWI/SNF or not. Histone anchors like that observed for SWI/SNF may be more a function of those remodelers that disassemble nucleosomes than of those remodelers involved in changing nucleosome spacing.
Model for an anchor to histones being required for SWI/SNF to mobilize nucleosomes. The histone octamer is shown as a sphere with numbers indicating different superhelical positions and a black line representing DNA with positions having the corresponding number. The ATPase lobes are gray ovals (I and II), and the hinge region is shown as a black line connecting the two lobes. The dark gray region is the SnAC domain. (A) When SnAC is present, the ATPase domain is anchored to the histone octamer and pulls DNA through it. The ATPase domain when tethered has sufficient force to disrupt histone-DNA interactions toward the entry site and create DNA bulges on the surface of nucleosomes. (B) In the absence of SnAC as a histone anchor, the ATPase domain is not fixed and cannot create a sufficient pulling force to disrupt histone-DNA interactions.
ACKNOWLEDGMENTS
This work was supported by NIH grant GM 48413 to B.B., a Career Award from the Burroughs Wellcome Fund to M.G.P., NIH grant GM 083055 to M.G.P., and an American Heart Association postdoctoral fellowship (12POST9380003) to P.V.
We thank Nilanjana Chatterjee for help with the DNA photoaffinity labeling experiments, Claude Meares for providing CHA225 antibodies, Kimberly Dimauro for help with single-molecule experiments, and Robert Forties for calculating the end-to-end distance distribution of a worm-like chain.
FOOTNOTES
- Received 9 July 2012.
- Returned for modification 10 August 2012.
- Accepted 2 November 2012.
- Accepted manuscript posted online 12 November 2012.
- Copyright © 2013, American Society for Microbiology. All Rights Reserved.
