To investigate the role of the nucleosomal acidic patch on dimer exchange by SWR1C, a fluorescence-based assay was initially employed. Recombinant nucleosomes were assembled with octamers containing wildtype Saccharomyces cerevisiae H2A/H2B (AB) dimers and Xenopus laevis H3/H4 tetramers or with S. cerevisiae H2A/H2B dimers containing eight alanine substitutions for residues that comprise the nucleosomal acidic patch (AB-apm, see Method section for residue list). The DNA template for nucleosome reconstitutions was a 224-bp fragment containing a ‘601’ nucleosome positioning sequence and an asymmetric, 77 bp linker DNA to mimic the structure of a promoter-proximal nucleosome located next to a nucleosome-depleted region (77N0) (Lowary and Widom, 1998). These substrates contain a Cy3 fluorophore conjugated to the linker distal end of the nucleosomal DNA, and a Cy5 fluorophore attached to an engineered cysteine residue (H2A-119) within the histone H2A C-terminal domain. The Cy3 and Cy5 fluorophores are within an appropriate distance to function as an FRET pair, such that excitation of the Cy3 donor leads to efficient energy transfer to the Cy5 acceptor, as evidenced by the fluorescence emission peak at 670 nm. The dimer exchange activity of SWR1C is monitored by following the decrease in the 670 nm FRET signal due to eviction of the Cy5-labeled AB-Cy5 dimer (Figure 1A).

Reactions to measure dimer exchange by SWR1C were performed under single turnover conditions (excess enzyme to nucleosomal substrate) and contained free H2A.Z/H2B (ZB) dimers which act as an essential co-substrate. Addition of SWR1C to a wildtype nucleosome led to a rapid drop in FRET, showing the biphasic kinetics consistent with the sequential exchange of the two nucleosomal H2A/H2B (AB) dimers (Fan et al., 2022; Singh et al., 2019).  Incubation of SWR1C with a nucleosome that lacks an intact acidic patch (AB-apm/AB-apm nucleosome) showed little decrease in the FRET signal, indicating that an intact nucleosomal acidic patch is essential for SWR1C to catalyze H2A/H2B eviction (Figure 1B).

Since the FRET assay is limited to monitoring the eviction of nucleosomal H2A/H2B dimers, a gel-based assay that monitors deposition of H2A.Z was employed to confirm these findings (Luk et al., 2010). In this assay, an unlabeled, 77N0 nucleosome was incubated with SWR1C, ATP, and free H2A.Z/H2B dimers in which H2A.Z contains a 3xFLAG tag at its C-terminus. Reaction products are separated on native-PAGE, and formation of the heterotypic (AB/ZB) and homotypic (ZB/ZB) nucleosomal products are detected by their reduced gel migration due to the 3xFLAG tag on H2A.Z. Similar to the FRET-based assay, the SWR1C-catalyzed deposition of H2A.Z was largely complete by 30–60 min for the AB/AB nucleosomal substrate, but addition of SWR1C to the AB-apm/AB-apm nucleosome led to 10-fold reduction in H2A.Z deposition after 60 min (Figure 1—figure supplement 1A, B, D). Together, these assays indicate that SWR1C requires an intact nucleosomal acidic patch for H2A.Z deposition.

To further interrogate the role for each of the two nucleosomal acidic patches in dimer exchange by SWR1C, we performed dimer exchange reactions with asymmetrically assembled nucleosomal substrates. Such asymmetric nucleosomes allow the fluorescent labeling of only a single H2A/H2B dimer, such that we could monitor which dimer was targeted for replacement. We generated these substrates by leveraging the asymmetric features of the 601 nucleosome positioning sequence and the spontaneous incorporation of dimers into purified hexasomes (Levendosky and Bowman, 2019; Figure 1—figure supplement 2A). Hexasomes were generated by varying the ratio of X. laevis H3/H4 tetramers to yeast H2A/H2B heterodimers on 77N0 or 0N0 DNA that was labeled with Cy3 (Figure 1—figure supplement 2B). After hexasome purification, nucleosomes of different compositions were reconstituted by titration of the appropriate yeast heterodimer (Figure 1—figure supplement 2C, D). For nucleosomes with linker DNA, we list the asymmetric dimers in linker proximal, linker distal order throughout (e.g. AB/AB-apm for a nucleosome with a linker distal H2A/H2B dimer harboring the acidic patch alteration).

First, we monitored H2A.Z deposition with nucleosomes harboring an unlabeled H2A/H2B heterodimer with a disrupted acidic patch (AB-apm) assembled on the nucleosomal face opposite to the Cy5-labeled heterodimer that will be monitored for H2A.Z deposition (Figure 1—figure supplement 3G and H). Strikingly, H2A.Z deposition was crippled, irrespective of whether the linker distal or linker proximal H2A/H2B heterodimer was the target for exchange (Figure 1C, AB-apm/AB-Cy5 and AB-Cy5/AB-apm substrates). We then tested if the Cy5-labeled heterodimer that is targeted for replacement also requires an intact acidic patch (Figure 1—figure supplement 3I). In this case as well, H2A.Z deposition was not greater than the no enzyme control (Figure 1C, AB/AB-apm-Cy5 substrate). Thus, SWR1C requires that both faces of the nucleosome contain an intact acidic patch, irrespective of which H2A/H2B heterodimer is undergoing replacement.

Previously, we and others reported that SWR1C preferentially exchanges the linker distal H2A/H2B dimer, as monitored by either ensemble or single molecule FRET analyses (Fan et al., 2022; Poyton et al., 2022). The asymmetric nucleosome substrates, harboring only a single, labeled H2A/H2B dimer were leveraged to further investigate this preferential exchange. Consistent with previous results, a substrate with a labeled, linker distal heterodimer is exchanged at a ~threefold faster rate by SWR1C compared to the linker proximal dimer (Figure 1C, compare AB/AB-Cy5 to AB-Cy5/AB; Figure 1—figure supplement 3A, B). To rule out the possibility that this preferential exchange was influenced by the unlabeled H2A/H2B dimer on the opposing nucleosomal face, nucleosomes harboring an unlabeled ZB dimer were also reconstituted. In these cases as well, the linker distal H2A/H2B dimer was exchanged at a more rapid rate, compared to the linker proximal heterodimer (Figure 1C, compare ZB/AB-Cy5 to AB-Cy5/ZB; Figure 1—figure supplement 3D, E). Surprisingly, incorporation of the ZB heterodimer on the contralateral (opposite to the dimer being evicted) nucleosome face stimulated H2A.Z deposition, irrespective of linker orientation (Figure 1C; purple bars). The stimulatory effect of contralateral ZB heterodimer placement remained even after elimination of the linker DNA (Figure 1D, Figure 1—figure supplement 3C, F). Interestingly, these assays also confirmed that linker DNA has a stimulatory impact on the H2A.Z deposition reaction, with both linker distal and linker proximal eviction occurring at a faster rate than eviction from a nucleosome core particle (Ranjan et al., 2013). To test whether the stimulation by linker DNA was due to differences in nucleosome-binding affinity, we ensured the experiment was done under saturating enzyme levels by repeating it at a fourfold increase in the concentration of SWR1C. At this higher concentration of SWR1C, the rates were unchanged, and the stimulatory impact of linker DNA remained (Figure 2—figure supplement 1A). Thus, both linker DNA and nucleosomal ZB dimers appear to stimulate H2A.Z deposition.

The ATPase activity of SWR1C is stimulated by both its nucleosomal substrate and free ZB dimers (Luk et al., 2010).  Notably, the ATPase activity of SWR1C is not stimulated by a ZB/ZB nucleosome, implying that SWR1C can distinguish product from substrate (Luk et al., 2010).  To investigate the role of the nucleosomal acidic patch in the stimulation of SWR1C ATPase activity, SWR1C ATPase activity was measured in the presence or absence of different 77N0-Cy3 nucleosomes.  Consistent with previous results (Luk et al., 2010), SWR1C ATPase activity was stimulated by AB/AB nucleosomes as compared to the basal level. However, ATPase activity was not stimulated by either of the asymmetric acidic patch mutant nucleosomes (AB-apm/AB, AB/AB-apm) or by a nucleosome lacking both acidic patches (AB-apm/AB-apm), demonstrating the essential role of dual acidic patch engagement by SWR1C for stimulation of ATPase activity (Figure 2—figure supplement 1B).

The inability of APM nucleosomes to stimulate the ATPase activity of SWR1C could be due to a defect in binding or the inability to form an active complex post-binding. To explore whether disruption of each acidic patch might impact the nucleosome-binding affinity of SWR1C for nucleosomes, we leveraged an FP assay. Using a Cy3-labeled 77N0 nucleosome, we determined a Kd of 13.6 nM for an AB/AB nucleosome (Figure 2A), similar to prior studies (Fan et al., 2022). However, disruption of even a single nucleosomal acidic patch severely weakened SWR1C binding, such that a Kd was unable to be determined (Figure 2B–D). Given the range of SWR1C concentrations that could be used in this assay, we estimate that alteration of one or both nucleosomal acidic patches decreases binding affinity by at least 10-fold. These data indicate that the inability of APM nucleosomes to stimulate SWR1C ATPase activity is most likely due to a defect at the nucleosome-binding step (K_m), rather than K_cat, though an impact at this latter step cannot be eliminated as yet.

Recently, a potential arginine anchor domain, the archetypal motif for proteins interacting with the nucleosomal acidic patch, was identified within the N-terminal, acidic domain of the SWR1C subunit, Swc2. However, removal of this domain (Swc2-ZN) has only a modest impact on the dimer exchange activity of SWR1C, and these data, combined with the essential nature of the nucleosomal acidic patches for SWR1C activity, suggest that additional subunits must contribute to acidic patch recognition (Dai et al., 2021). The Swc5 subunit is one candidate, as loss of Swc5 eliminates H2A.Z deposition and nucleosome-stimulated ATPase activity (Wu et al., 2005; see Figure 3B, C). Furthermore, removal of the Swc5 subunit from SWR1C crippled nucleosome binding, as measured by FP (Figure 3D). Previous studies of Swc5 identified an acidic N-terminal domain that interacts preferentially with AB dimers, and a conserved C-terminal domain, termed BCNT, that is essential for dimer exchange by SWR1C (Sun and Luk, 2017; Huang et al., 2020). An alignment of Swc5 homologs revealed a region adjacent to the BCNT domain, containing multiple arginine residues in a conserved basic region (Figure 3A).

To investigate the role of this Swc5 arginine-rich domain, alanine substitutions were created in four basic residues (RRKR), and this derivative was recombinantly expressed and reconstituted into a SWR1C complex purified from a swc5 deletion strain (SWR1C^swc5∆). In addition, SWR1C was reconstituted with wildtype Swc5, a Swc5 derivative lacking the acidic N-terminal domain (Swc5^79–303), and a Swc5 derivative harboring alanine substitutions in the essential BCNT domain (Swc5^LDW-3A). In all cases, derivatives were incorporated into SWR1C complexes with equal efficiencies (Figure 3—figure supplement 1). Consistent with previous studies, reconstitution with wildtype Swc5 fully restored dimer exchange activity to SWR1C^swc5∆ (Figure 3B, red curve). In contrast, SWR1C harboring Swc5^LDW-3A had no detectable exchange activity, and the complex that lacks the N-terminal, acidic domain of Swc5 (Swc5^79–303) exhibited a modest defect, as expected from previous studies (Figure 3B; Sun and Luk, 2017). Strikingly, SWR1C that contained the Swc5^RRKR-4A derivative showed minimal activity in the FRET-based exchange assay (Figure 3B), and this complex exhibited only ~10% the activity of wildtype SWR1C in the gel-based H2A.Z deposition assay (Figure 1—figure supplement 1A, C, D). Dimer exchange activity for each SWR1C complex was also mirrored by their ATPase activity (Figure 3D). SWR1C ^swc5∆ lost nucleosome-stimulated ATPase activity, with or without free ZB dimers, as compared to wildtype. As expected, reconstitution of SWR1C ^swc5∆ with recombinant Swc5 rescued ATPase activity. Underscoring the importance of the arginine-rich domain, reconstitution of SWR1C ^swc5∆ with Swc5^RRKR-4A showed a large defect in nucleosome-stimulated ATPase activity, revealing its importance in both dimer exchange and ATPase activity. Additionally, we reinforced prior findings that the ATPase activity of the Swc5^LDW-3A complex was not stimulated by either nucleosomes or free ZB dimers, while the complex harboring Swc5^79–303 had wildtype levels of ATPase activity (Figure 2—figure supplement 1C).

To further investigate whether the arginine-rich domain within Swc5 binds to the nucleosomal acidic patch, in vitro binding assays were performed with nucleosome core particles (0N0) and recombinant Swc5 (Figure 4A–F). Increasing amounts of Swc5 were incubated with nucleosomes, and binding was visualized by native-PAGE. Addition of wildtype Swc5 to nucleosomes led to formation of discrete complexes with an apparent Kd of 125.7 nM (Figure 4A). Likewise, binding of Swc5^LDW-3A, which has a disrupted BCNT domain, bound to nucleosomes with nearly identical affinity, compared to wildtype Swc5 (Figure 4B). Importantly, high affinity binding of Swc5 to nucleosomes required an intact acidic patch, with binding reduced ~threefold when an APM nucleosome was used as the substrate (Figure 4C). Furthermore, the Swc5^RRKR-4A derivative also showed an ~threefold loss in affinity, with an apparent Kd of 379.1 nM (Figure 4D). Together these data demonstrate that Swc5 has nucleosome-binding activity, and the data are consistent with direct interactions between the Swc5 arginine-rich domain and the nucleosomal acidic patch.

A time-resolved FRET assay (TR-FRET) was employed (Figure 5A) as an additional solution approach to monitor the binding of Swc5 to nucleosomes (Wesley et al., 2022). In this assay, increasing concentrations of recombinant, 6His-tagged Swc5 were incubated with a 31N1 nucleosome. The 6His-Swc5 protein were labeled with a ULight alpha-6xHIS acceptor antibody, and biotinylated nucleosomal DNA was labeled with a Eu-streptavidin donor fluorophore (Figure 5A). The binding of Swc5 to the nucleosome was monitored by quantitative measurements of the time-resolved FRET signal between the ULight alpha acceptor and the Eu fluorophore. Binding curves for wildtype Swc5 yielded an apparent Kd of 133 ± 12 nM, a value that agrees well with the Kd measured by the nucleosome gel shift assay. Binding experiments performed in parallel with the Swc5^RRKR-4A derivative yielded a Kd of 593 ± 28 nM, confirming that the arginine-rich motif within Swc5 plays a key role in nucleosome recognition (Figure 5B).

While the gel shift and TR-FRET assays establish that Swc5 can bind to nucleosomes, data from these assays can only indirectly link the nucleosomal acidic patch to the Swc5 arginine-rich domain. To probe for more direct interactions, a fluorescence quenching assay was employed. These assays exploit nucleosomes harboring a site-specific, Oregon Green fluorophore whose emission is sensitive to the chemical environment whereby fluorescence is quenched by protein binding (McGinty et al., 2014; Winkler et al., 2012). Nucleosome core particles were reconstituted that contained an Oregon Green fluorophore covalently attached to either the histone H4 N-terminal domain (H4-tail) (Figure 6A, top) or to a residue directly adjacent to the nucleosomal acidic patch (Figure 6A, bottom). Titration of wildtype Swc5 led to the concentration-dependent quenching of the Oregon Green fluorophore positioned at the acidic patch, but little quenching was observed for the fluorophore located on the H4-tail (Figure 6B) demonstrating the sensitivity of this assay. Importantly, Swc5 did not quench the acidic patch probe when this substrate also harbored alanine substitutions within the acidic patch (Figure 6B). Examined in the same titration of concentrations, we found that the Swc5^LDW-3A derivative also showed specific quenching of the acidic patch probe (Figure 6C), but the Swc5^RRKR-4A derivative was defective for quenching the acidic patch probe (Figure 6D). Together these results are consistent with a direct interaction between the arginine-rich domain of Swc5 and the nucleosomal acidic patch, an interaction that is essential for dimer exchange by SWR1C.

We analyzed the structure of Swc5^79–303 in complex with a 147-bp nucleosome core particle using cryo-EM (Figure 7). We collected over 13,000 micrographs of the cross-linked sample and were able to easily resolve the nucleosome structure to 3 Å resolution (Figure 7—figure supplement 1L–O). On the other hand, the moiety corresponding to Swc5 displayed severe structural heterogeneity, hindering our ability to build an atomic model for Swc5. The protein appears to be flexible on the nucleosome, with continuous conformational dynamics apparent (Figure 7A–D, Figure 7—video 1). Nevertheless, reconstructions of particle subsets obtained by local 3D classification show that Swc5 forms multivalent interactions with the nucleosome via the nucleosome acidic patch, the H4 N-terminal histone tail, and the C-terminal helix of H2B (Figure 7E–G). In the majority of 3D reconstructions, we observe additional density at the nucleosome acidic patch attributable to an arginine residue, like in other complexes with arginine anchor interactions (Makde et al., 2010; McGinty and Tan, 2015). Unfortunately, the low local resolution precludes the determination of the precise Swc5 arginine residue involved in this interaction (Figure 7—figure supplement 1D–G). In contrast, the 3 Å resolution of the nucleosome core allows the identification of H2B-K123 as an interacting residue with Swc5. In most 3D classes, Swc5 forms either a partial or a complete bridge between H2B C-terminal helix including K123 and the N-terminal tail of H4.

Previous studies have shown that yeast lacking functional SWR1C grow poorly on media containing formamide (Kobor et al., 2004; Sun and Luk, 2017). To investigate the functional role of the Swc5 arginine-rich domain in vivo, growth assays were performed with isogenic swc5∆ strains that harbor low copy vectors that express different Swc5 derivatives (Figure 8). As expected, the swc5∆ strain grew well on rich media, but was impaired on formamide media (vector), while cells expressing wildtype Swc5 grew well on both media (SWC5). Consistent with a previous study, a strain expressing the derivative with a disrupted BCNT domain showed slow growth on formamide media (Swc5^LDW-3A), and the Swc5^79–303 derivative that lacks the acidic N-terminal domain showed a moderate growth defect (Sun and Luk, 2017). Importantly, a strain expressing the Swc5^RRKR-4A derivative had a severe growth defect on formamide media, consistent with an important role of the arginine-rich region for SWR1C function.

The yeast strain W1588-4C swr1::SWR1– 3xFLAG-P-kanMx-P htz1Δ::natMX (yEL190) was a gift from Ed Luk (SUNY Stony Brook). Deletion of SWC5 was constructed by standard gene replacement with a Hygromycin B cassette (pAG32) (Goldstein and McCusker, 1999), and used to make W1588-4C swr1::SWR1– 3xFLAG-P-kanMx-P htz1Δ::natMX swc5Δ::hphMX (CY2535). The swc5∆ strain MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0 YBR231c::kanMX4 (Y03371) used for spot assays was purchased from Euroscarf.

Spot assays were performed with the swc5∆ strain (Y03371) transformed with an empty CEN/ARS/URA3 vector pRS416, or ones containing WT and mutant SWC5. Cultures were diluted to an OD₆₀₀ of 1 and then serially diluted 1:10. A dilution (7 μl) was spotted onto synthetic complete media lacking uracil with and without 2% formamide. Plates were incubated for 3 days before imaging.

The vector (pQE80L) was used for recombinant Swc5 expression. The SWC5 gene and SWC5^79–303 were inserted in frame with the N-terminal 6x histidine tag through Gibson assembly (Gibson et al., 2009) after digesting the vector with BamHI and SphI. Once WT and truncated Swc5 were in frame, the alanine mutants [swc5 (RRKR-4A) and swc5 (LDW-3A)] were generated using QuickChange site-directed mutagenesis. The pRS416 CEN/ARS/URA3 vector was digested with XbaI and SacI to allow for insertion of the SWC5 gene with 500 bp upstream of the start site and 500 bp downstream of the stop. This region was initially PCR amplified from a W303 strain where the primers had homology to the XbaI and SacI cut sites of the vector. Gibson assembly was used to ligate the vector and insert (CP1566) (Gibson et al., 2009). Once the WT SWC5 gene was inserted, site-directed mutagenesis was used on CP1566 to make swc5 RRKR-4A (CP1579) and swc5 LDW-3A (CP1580). swc5 79–303 construct was generated using CP1566 to PCR amplify the 500 bp upstream stretch of DNA and start site of SWC5 and PCR amplifying another DNA fragment starting from Swc5 K79 to 500 bp downstream of the stop. Primers were designed so each PCR fragment had homology to each other and the pRS416 vector at XbaI and SphI so that the two fragments and the pRS416 vector could be ligated together by Gibson assembly. Site-directed mutagenesis was used to generate cysteine mutants used for Oregon Green labeling, as well as for single amino acid substitution used for assembly of acidic patch mutant (AB-apm) dimers – H2A-apm (H2A-E63A, H2A-E65A, H2A-D74A, H2A-D92A, H2A-D93A, H2A-E94A) and H2B-apm (H2B-E109A, H2B-E117A).

These experiments utilized a variety of nucleosomes, both in terms of histone variants, labels, and mutations. Broadly, all nucleosomes contained a tetramer of X. laevis H3 and H4, along with S. cerevisiae H2A/H2B heterodimers, with mutations or variant substitutions as appropriate, reconstituted on DNA containing the Widom 601 positioning sequence with or without linkers and/or fluorescent labels. We refer to the Widom 601 sequence as a 147-bp element, though structural studies have indicated that this sequence assembles a 145-bp nucleosome core particle (Vasudevan et al., 2010). We decided to use X. laevis H3/H4 tetramers due to slightly increased stability of nucleosomal product in solution. These histones show high sequence conservation with S. cerevisiae H3 and H4 and we have previously demonstrated there is no difference in SWR1C eviction activity on full yeast or hybrid yeast-xenopus nucleosomes (Singh et al., 2019). Furthermore, the H4 tail region that interacts with Swc5 in our cryo-EM models is virtually identical in the orthologs.

For experiments using nucleosomes with symmetrical dimers, hybrid octamers were generated as previously described and nucleosomes were reconstituted via salt dialysis deposition on a Widom 601 positioning sequence containing DNA template (Lowary and Widom, 1998; Luger et al., 1999b).

Symmetrical nucleosomes produced include AB/AB and AB-apm/AB-apm mutant nucleosomes used for Swc5 quenching and gel shift assays contained yH2B-S115C or xH4-Q27C labeled with Oregon Green 488 Malemide (Thermo Fisher) on a 0N0 DNA template (Lowary and Widom, 1998). Additionally, AB/AB and AB-apm/AB-apm containing nucleosomes in which all H2A dimers contain the label Cy5 were assembled on 77N0-Cy3 and 0N0-Cy3 templates.

Asymmetrical nucleosomes were produced by first reconstituting X. laevis H3/H4 tetramers as well as S. cerevisiae dimers of interest separately instead of generating octamers. Tetramers, a single dimer type, and template DNA were mixed in a 1.4:1.8:1.0 ratio followed by reconstitution via salt dialysis identically to regular nucleosome reconstitution. This ratio was found to reliably produce a mixture of hexasomes and nucleosomes, and in the case of hexasomes the dimer is known to be deposited on the ‘strong’ side of the Widom position sequence (Levendosky et al., 2016). For asymmetric nucleosomes containing a single dimer labeled with Cy5, the template DNA was varied such that the labeled dimer is deposited on the strong side of the 601 positioning sequence. Note that our previous study demonstrated that the rates of H2A.Z deposition by SWR1C are nearly identical for either the weak or the strong side of a 601 mononucleosome (Lowary and Widom, 1998). Hexasomes are then purified using a Bio-Rad MiniPrep Cell apparatus, a method by which proteins are collected via fraction collector after passage through a native-PAGE tube gel (60:1 Acrylamide/Bis), which is sufficient to separate hexasomes from nucleosomes (Levendosky et al., 2016; Levendosky and Bowman, 2019; Nodelman et al., 2020). After checking fractions on a native-PAGE gel (29:1 Acrylamide/Bis) stained with SYBR Gold, the hexasome containing fractions were pooled and concentrated, followed by buffer exchange into remodeling buffer (25 mM HEPES, pH 7.6, 0.2 mM EDTA, 5 mM MgCl₂, 70 mM KCl, 1 mM DTT) with 20% glycerol, aliquoting, and flash-freezing. The introduction of a S. cerevisiae heterodimer to hexasomes in solution leads to spontaneous incorporation and formation of a nucleosome, allowing for production of asymmetric nucleosomes with known orientation.

Six types of asymmetric nucleosomes for dimer exchange assays were produced in this manner containing linker-proximal or distal H2A-Cy5/H2B (AB-Cy5) heterodimers with placement of unlabeled AB, ZB, or AB-apm dimers on the contralateral side on a 77N0-Cy3 template DNA. Additionally, a linker distal AB-apm-Cy5 and linker proximal AB nucleosome, as well as AB/AB-Cy5 and ZB/AB-Cy5 core particles (0N0 template DNA) were produced.

For FP and ATPase studies, unlabeled dimers were used to produce hexasomes containing linker distal AB or AB-apm dimers on 77N0-Cy3. These hexasomes were then used to generate nucleosomes with dimers (listed here and throughout the manuscript in linker proximal/linker distal order) AB/AB, AB/AB-apm, AB-apm/AB, or AB-apm/AB-apm, all on the 77N0-Cy3 template.

FRET-based dimer exchanges assays were performed as previously described (Singh et al., 2019) using an ISS PC1 spectrofluorometer, a Tecan Infinite M1000 PRO microplate reader, or a Tecan Spark microplate reader. The nucleosome remodeling reactions (50–100 μl) were performed in remodeling buffer at room temperature with 30–60 nM SWR1C, 10 nM 77N0-Cy3 and at least one dimer-Cy5 containing nucleosome, 60–70 nM ZB dimers, and 1 mM ATP (to start reaction). Dimer exchange was observed by exciting the Cy3 fluorophore at 530 nm and monitoring the Cy5 emission 670. Data for nucleosomes with both dimers labeled were fit to the two-phase decay equation (Equation 1; where Y denotes RFU, Yi the initial RFU, Yp the final RFU, Pf the fast phase expressed as a fraction, Ps the slow phase expressed as a fraction, kf the rate constant of the fast phase, ks the rate constant of the slow phase, and t is time) while data for reactions with only one dimer labeled were fit to the one-phase decay equation (Equation 2; where Y denotes RFU, Yi the intial RFU, Yp the final RFU, k is the rate constant, and t is time).(1)Y=PfYi-Yp-kft+PsYi-Yp-kst+Yp(2)Y=Yi-Yp-kt+Yp

Gel-based dimer exchange assays used a modified protocol from a previous study (Mizuguchi et al., 2004; Ranjan et al., 2013). A 150-μl dimer exchange reaction contained 30 nM SWR1C, 10 nM 77N0 nucleosomes (WT or APM), and 70 nM H2A.Z^flag/H2B in remodeling buffer with 0.1 mg/ml BSA. The reaction was performed at room temperature and 20 μl of the reaction was taken at each time point (time 0 is before 1 mM ATP is added) and quenched with 1 μg of plasmid DNA to separate SWR1C and dimers from the nucleosome. Each time point was stored on ice until last sample was quenched. Each time point was loaded onto a 6% 0.5× TBE Native PAGE gel (29:1 Acrylamide/Bis) and electrophoresed for 90 min at 120 V. Gels were stained with SYBR gold and imaged on a GE Typhoon.

FP assays were performed as previously described (Baier and Peterson, 2022) using a CLARIOstar microplate reader. The affinity of SWR1C for various nucleosomal substrates was tested by mixing 10 nM 77N0-Cy3 nucleosomes with SWR1C serially diluted in concentration from approximately 1 μM to 1 nM in remodeling buffer. The dimensionless value of polarization was plotted for each concentration of SWR1C and the Morrison equation (Equation 3; where Bmax is maximum observed binding, [F] is the concentration of the labeled nucleosomes, [E] is the concentration of SWR1C, and Kd,app is the apparent dissociation constant) was used determine the binding affinity for substrates where binding was observed.(3)Y=Bmax[F]+[E]+Kd,app-[F]+[E]+Kd,app2-4[F][E]2[F]

ATPase activity was measured using a phosphate-binding protein reporter assay. For the phosphate sensor assay (Luk et al., 2010; Brune et al., 1994), 50 μl reactions were prepared with 8 nM SWR1C mixed with 0.5 μM phosphate sensor (fluorophore-conjugated phosphate-binding protein, Thermo Fisher) in remodeling buffer. Reactions containing nucleosome and ZB dimers had concentrations of 20 and 30 nM, respectively. Reactions were initiated by adding 100 μM ATP and monitored by the measuring the fluorescence of the MDCC fluorophore on a Tecan Spark by exciting at 430 nm and setting the emission filter to 450 nm. All reagents and the microplate were pre-treated with a ‘Pi Mop’ to remove free phosphate as previously described (Luk et al., 2010; Brune et al., 1994). FI was monitored over time and the linear range of intensity change was plotted to calculate the rate in reciprocal seconds.

Gel shifts with recombinant WT or Swc5 derivatives were performed in 15 μl reactions. Reactions contained varying amounts of Swc5 in a total volume of 5 μl of storage buffer (10% glycerol, 50 mM Tris pH 7.4, 50 mM NaCl), 0.25 μl 300 nM 0N0 nucleosomes, 6.15 μl water, 0.6 μl 50% glycerol, and 3 μl 5× binding buffer (125 mM HEPES pH 7.3, 250 mM NaCl, 25 mM MgCl₂, 0.05% Tween 20, 0.5 mg/ml BSA, 5 mM DTT). Reaction incubated in the dark at room temp for 30 min. Samples were loaded onto 6% native-PAGE gels (29:1 Acrylamide/Bis) and electrophoresed at 230 V for 1 hr. Gels were imaged on a GE typhoon imager and binding was calculated by measuring the disappearance of the free nucleosome band. GraphPad Prism 8 was used to generate binding curves.

Swc5 fluorescent quenching assays were assembled similar to the gel shift assays but in 40 μl reactions with 10 nM 0N0 Oregon Green nucleosomes (8 μl 5× binding buffer, 1.3 μl 300 nM nucleosomes, 11.7 μl water, 19 μl Swc5/storage buffer). Master mixes of each reaction incubated for 30 min at room temp in the dark before being loaded into 384-well flat black plate (PerkinElmer) and scanning each well 20 times on a Tecan Spark plate reader at 488 nm excitation and 530 nm emission. Quenching was calculated by taking the average relative fluorescence units (RFUs) of each Swc5 containing well and dividing RFU average of the nucleosome only wells. The normalized fluorescence was plotted using Prism 8.

Time-resolved FRET nucleosome-binding assays were performed as described by Wesley et al., 2022. Acceptor mixtures were prepared by mixing ULight alpha-6xHIS acceptor antibody (PerkinElmer) with 6xHIS-tagged Swc5 variants at a ratio of 1:20 and serially diluting across 13 concentrations in H66 buffer (20 mM HEPES pH 7.5, 66 mM NaCl, 5 mM MgCl2, 5 mM DTT, 5% glycerol, 0.01% NP-40, 0.01% CHAPS, and 100 µg/ml BSA). 2× donor mixtures were prepared by mixing 4 nM streptavidin-Eu (PerkinElmer) with or without 2 nM yeast nucleosomes containing 177 bp of Widom 601 DNA (31 + 145 + 1) with a 5′ biotin group on the 31 bp extension and incubating for 30 min at room temperature. Samples were prepared in 384-well plates by mixing 5 µl of 2× donor mixtures with 5 µL of acceptor mixtures at each dilution. Fluorescence signals were acquired at room temperature in a Victor Nivo multimode fluorescent plate reader (PerkinElmer) using an excitation filter at 320 nm and emission filters at 615 and 665 nm. Emission signals at 615 and 665 nm were measured simultaneously following a 100 μs delay. Kd values were determined from triplicate titrations of each Swc5 variant and are reported as means ± standard error of the mean.

Swc5^79–303 was mixed with 147 bp Widom 601 sequence-containing nucleosomes at a 2.5:1 molar ratio in H50 buffer (10 mM HEPES pH 7.5, 50 mM NaCl, 1 mM DTT(Dithiothreitol)). Precipitated material was removed by centrifugation at 16,000 × g for 10 min. The sample was chemically cross-linked using the GraFix method in a 10–30% glycerol gradient (Stark, 2010), prepared with a Gradient Master instrument (Biocomp) from top (H50 + 10% vol/vol glycerol) and bottom (H50 + 30% vol/vol glycerol, 0.15% vol/vol glutaraldehyde) solutions. The sample was applied on top of the gradient and centrifuged in a SW40Ti rotor (Beckman-Coulter) at 35,000 rpm, 4°C for 16 hr. Fractions were collected manually by pipetting from the top, and the reaction was quenched by adding Tris pH 7.5 to each fraction to 50 mM. Fractions were analyzed using a 6% native gel (29:1 Acrylamide/Bis) stained with ethidium bromide and negative-stain EM with uranyl formate to select for monodisperse particles of the complex. Selected fractions were dialyzed into H50 buffer supplemented with 0.1 mM PMSF and concentrated to ~1 mg/ml.

Swc5/nucleosome complex was prepared at 0.9 mg/ml. Quantifoil R1.2/1.3 holey carbon, 300 mesh, copper grids (Electron Microscopy Sciences) were glow discharged for 45 s at 15 mA with an easiGlow device (PELCO). After glow discharge, 3 µl of sample was pipetted onto the grid and Grade 595 filter paper (Ted Pella) was used to blot the grid for 3 s, at 4°C and 100% humidity. The grid was plunge-frozen in liquid ethane cooled to the temperature of liquid nitrogen with a Vitrobot Mark IV instrument (Thermo Scientific). A Krios G3i microscope (Thermo Scientific) operating at 300 kV was used in conjunction with a K3 direct electron detector camera (Gatan) to collect data in super-resolution mode at the Pacific Northwest Cryo-EM Center. 13,855 movies were collected at a nominal magnification of 29,000× with a physical pixel size of 0.788 Å (super-resolution, 0.394 Å). The total dose per exposure was 50 e⁻/Ų at a dose rate of ~18.5 e⁻/pix/s and each exposure was fractioned into 50 subframes. Data were recorded in the defocus range of −0.8 to −2.8 µm.

Data were processed with the cryoSPARC v4 suite (Figure 7—figure supplement 1; Punjani et al., 2017). Super-resolution movies were motion-corrected with ‘Patch Motion Correction’ and binned using 2× Fourier cropping. CTF(contrast transfer function) values were estimated with ‘Patch CTF Estimation’. Initial particles were picked with ‘Blob Picker’ from a subset of micrographs, then used for ab initio 3D reconstruction, and the best obtained model was used to generate templates for particle picking with ‘Template Picker’. Template-picked particles were refined into three classes using ‘Heterogenous Refinement’ with the previously generated ab initio reconstructions as initial volumes to remove junk particles. 1,846,495 nucleosomal particles were further filtered to remove low-quality particles using the ‘random-phase 3D classification’ strategy (Gong et al., 2016). Briefly, particles were sequentially subjected to multiple rounds of Heterogeneous Refinement against a low-pass filtered reconstruction (40, 30, 25, 20, 15, 10, and 8 Å low-pass) of the output volume from the previous round. This strategy yielded two classes of high-quality particles with the nucleosomal DNA either (1) completely wrapped around the core (585,573 particles, Non-uniform Refinement to 3.0 Å resolution), or (2) partially peeling off (392,348 particles, 3.1 Å). The Swc5 moiety is extremely noisy in these reconstructions due to apparent structural heterogeneity. We tried employing various methods including ‘3D Variability Analysis’ (Punjani and Fleet, 2021) to either further classify the particles into more homogenous subsets or to describe the intrinsic conformational dynamics, but with limited success (data not shown). We obtained the most informative results with cryoSPARC’s algorithm of ‘3D Classification’ without alignment. The 585 k particles of fully wrapped nucleosome were classified into 10 classes using a mask focused on the sides of the histone core in the regions with Swc5 moiety apparent. The obtained particle classes were refined to ~3.1 Å using ‘Non-uniform Refinement’ (Punjani et al., 2020). Modeling was performed with UCSF ChimeraX (Pettersen et al., 2021).