Recognition and remodelling of nucleosomes and hexasomes by the human INO80 complex
Priyanka Aggarwal, Manmohan Sharma, Stephan Woike, Franziska Kunert, Annika Brem, Manuela Moldt, Karl-Peter Hopfner
TL;DR
This study reveals how the human INO80 complex slides and repositions nucleosomes and hexasomes by sensing DNA unwrapping and binding to specific regions.
Contribution
The study shows that human INO80 slides hexasomes as efficiently as nucleosomes and provides structural insights into its mechanism.
Findings
Human INO80 slides hexasomes as efficiently as H2A and H2A.Z nucleosomes.
INO80 binds to the entry point of extranucleosomal DNA and spin-rotates around the nucleosome core.
Acidic patch binding by IES2 differentiates (sub)nucleosomal species and affects nucleosome sliding.
Abstract
The ATP-dependent INO80 chromatin remodeller slides and repositions nucleosomes to shape and maintain chromatin around gene regulatory elements and replication origins. Recent work uncovered capabilities of yeast and fungal INO80 to bind and slide hexasomes, but whether this is a universal feature is unknown. Here, we show that human INO80 also slides hexasomes as efficiently as H2A and H2A.Z nucleosomes. By determining a variety of structures of human INO80 bound to canonical and H2A.Z nucleosomes as well as hexasomes, we reveal a predominantly topological sensing of nucleosomal species with at least three positions depending on entry DNA unwrapping. INO80 spin-rotates around the nucleosomal core particle as a function of entry DNA unwrapping. Different degrees of unwrapped entry DNA lead to two different nucleosomal and one hexasomal locations of INO80, determined by binding of the…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5| Primer | Sequence |
|---|---|
| IES6 (G135A, K136A, K137A) | Fwd 5′ AGCCAACCGTACCTGGAAGTACCTG 3′ |
| Rev 5′ GCGGCAGCCACAGCACCACCGTG 3′ | |
| RuvBL2 (D214A) | Fwd 5′ CGTGCTCGTGCGTACGACGCTATG 3′ |
| Rev 5′ GGTGAAGCTACGACCCAG 3′ | |
| IES2 (R179A, R181A) | Fwd 5′ AGCGGCTCTGCTGCAGAAGGCTC 3′ |
| Rev 5′ TGCGCAGCGGTCAGCAGACGCTC 3′ |
| Nucleosomal/Hexasomal substrate(s) | Sequence |
|---|---|
| 50N50 | 5′CTTCACACCGAGTTCATCCCTTATGTGATGGACCCTATACGCGGCCGCC |
| 0H80 0H80-Cy5.5 0N80-FAM 0N80-Cy5.5 | 5′ |
| 0N400H40-FAM | 5′ |
| H2A.Z Nucleosome | State NZ-7INO80-H2A.Z Nucleosome | State N-6 (Nucleosome only) | State N-6 INO80-Nucleosome | State N-7 (Nucleosome only) | State N-7 INO80-Nucleosome | Hexasome | State H-3 INO80-Hexasome | INO80apo-Nucleosome (C-and A-module) | |
|---|---|---|---|---|---|---|---|---|---|
| (PDB – 9GE4) | (PDB -9GCG) | (PDB -9GF6) | (PDB -9GEV) | (PDB -9GFM) | (PDB -9GFB) | (PDB -9GEL) | (PDB -9GE5) | ||
| (EMDB -51289) | (EMDB -51229) | (EMDB -51310) | (EMDB -51307) | (EMDB -51313) | (EMDB -51312) | (EMDB -51294) | (EMDB -51290) | (EMDB-53574) | |
| Data collection and processing | |||||||||
| Magnification | 130, 000 | 130, 000 | 130, 000 | 130, 000 | 130, 000 | 130, 000 | 165, 000 | 165, 000 | 130, 000 |
| Voltage (kV) | 300 | 300 | 300 | 300 | 300 | 300 | 300 | 300 | 300 |
| Electron exposure (e–/Å2) | 40 | 40 | 41.16 | 41.16 | 41.16 | 41.16 | 44.82 | 44.82 | 40 |
| Defocus range (μm) | -1.1 to -2.9 | -1.1 to -2.9 | -1.1 to -2.9 | -1.1 to -2.9 | -1.1 to -2.9 | -1.1 to -2.9 | -1.1 to -2.6 | -1.1 to -2.6 | -1.1 to -2.9 |
| Pixel size (Å) | 1.049 | 1.049 | 1.049 | 1.049 | 1.049 | 1.049 | 0.727 | 0.727 | 1.049 |
| Symmetry imposed | C1 | C1 | C1 | C1 | C1 | C1 | C1 | C1 | C1 |
| Final particle images (no.) | 94, 892 | 94, 892 | 64, 537 | 64, 537 | 82, 269 | 82, 269 | 34, 444 | 34, 444 | 16, 786 |
| Map resolution (Å) | 3.6 | 3.5 | 4.1 | 3.7 | 3.8 | 3.6 | 6.9 | 3.6 | 9.1 |
| FSC threshold | 0.143 | 0.143 | 0.143 | 0.143 | 0.143 | 0.143 | 0.143 | 0.143 | 0.143 |
| Refinement | |||||||||
| Initial model used (PDB code) | 2CV5 | 7ZI4 | 2CV5 | 7ZI4 | 2CV5 | 7ZI4 | 2CV5 | 7ZI4 | |
| Ligands | - | 7 ADP, 1 ATP, 1 Zn | - | 6 ADP, 1 ATP | - | 6 ADP, 1 ATP, 1 Zn | - | 7 ADP, 1 ATP, 1 Zn | |
| Model resolution (Å) | 3.74 | 3.61 | 3.94 | 3.63 | 4.08 | 3.93 | 4.98 | 4.05 | |
| FSC threshold | 0.5 | 0.5 | 0.5 | 0.5 | 0.5 | 0.5 | 0.5 | 0.5 | |
| Map sharpening | 85.8 | 72.5 | 135.9 | 110.4 | 128.9 | 105.8 | 285 | 92 | |
| Model composition | |||||||||
| Protein residues | 748 | 4673 | 798 | 4723 | 810 | 4735 | 544 | 4469 | |
| Nucleotides | 304 | 304 | 270 | 270 | 278 | 278 | 226 | 226 | |
|
| |||||||||
| Protein | 100.6 | 116.2 | 176.4 | 170.9 | 196.9 | 121.4 | 335.3 | 212.9 | |
| Nucleotide | 175.4 | 175.4 | 273.5 | 273.5 | 277.1 | 277.1 | 346.5 | 346.5 | |
| Ligand | – | 80.1 | – | 124.6 | – | 75.6 | - | 171.5 | |
| R.m.s. deviations | |||||||||
| Bond lengths (Å) | 0.012 | 0.012 | 0.012 | 0.015 | 0.012 | 0.012 | 0.013 | 0.015 | |
| Bond angles (°) | 2.022 | 1.990 | 1.958 | 1.970 | 1.921 | 1.944 | 2.053 | 1.959 | |
| Validation | |||||||||
| MolProbity score | 2.18 | 1.83 | 0.90 | 1.11 | 1.02 | 0.95 | 1.06 | 1.07 | |
| Clashscore | 5.62 | 6.34 | 0.19 | 0.71 | 0.45 | 0.32 | 0.07 | 0.64 | |
| Rotamer Outliers (%) | 3.06 | 1.56 | 1.06 | 1.14 | 1.19 | 0.87 | 1.55 | 1.15 | |
| CaBLAM Outliers (%) | 5.03 | 2.79 | 1.45 | 2.13 | 1.42 | 2.43 | 2.31 | 2.34 | |
| Ramachandran plot | 95.50 | 95.11 | 95.71 | 95.21 | |||||
| Favored (%) | 91.53 | 95.25 | 4.11 | 4.16 | 3.79 | 4.15 | 94.36 | 95.56 | |
| Allowed (%) | 7.65 | 4.25 | 0.39 | 0.73 | 0.51 | 0.64 | 5.26 | 3.87 | |
| Outliers (%) | 0.82 | 0.50 | 0.38 | 0.57 | |||||
| Model-Map scores | |||||||||
| CC (Mask) | 0.86 | 0.85 | 0.88 | 0.79 | 0.88 | 0.78 | 0.68 | 0.81 | |
| CC (Peaks) | 0.73 | 0.81 | 0.78 | 0.68 | 0.79 | 0.71 | 0.44 | 0.67 | |
| CC (Volume) | 0.85 | 0.84 | 0.87 | 0.78 | 0.87 | 0.77 | 0.69 | 0.80 | |
| INO80- Hexasome (C- and A-module) | INO80-H2A.Z Nucleosome (C- and A-module) | |
|---|---|---|
| (EMDB-53579) | (EMDB-53580) | |
| Data collection and processing | ||
| Magnification | 165, 000 | 130, 000 |
| Voltage (kV) | 300 | 300 |
| Electron exposure (e–/Å2) | 40 | 40 |
| Defocus range (μm) | -1.1 to -2.6 | -1.1 to -2.9 |
| Pixel size (Å) | 0.727 | 1.049 |
| Symmetry imposed | C1 | C1 |
| Final particle images (no.) | 22, 112 | 11, 540 |
| Map resolution (Å) | 8.9 | 11.6 |
| FSC threshold | 0.143 | 0.143 |
| Refinement | ||
| Initial model used (PDB code) | ||
| Ligands | ||
| Model resolution (Å) FSC threshold | ||
| Map sharpening | ||
| Model composition | ||
| Protein residues | ||
| Nucleotides | ||
|
| ||
| Protein | ||
| Nucleotide | ||
| Ligand | ||
| R.m.s. deviations | ||
| Bond lengths (Å) | ||
| Bond angles (°) | ||
| Validation | ||
| MolProbity score | ||
| Clashscore | ||
| Rotamer Outliers (%) | ||
| CaBLAM Outliers (%) | ||
| Ramachandran plot | ||
| Favored (%) | ||
| Allowed (%) | ||
| Outliers (%) | ||
| Model-Map scores | ||
| CC (Mask) | ||
| CC (Peaks) | ||
| CC (Volume) |
- —Deutsche Forschungsgemeinschaft10.13039/501100001659
- —European Research Council10.13039/100010663
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsGenomics and Chromatin Dynamics · DNA Repair Mechanisms · Epigenetics and DNA Methylation
Introduction
Chromosomal DNA resides in the nucleus in the form of chromatin, a dynamic and variable complex of DNA and associated proteins, primarily histones. The major building blocks of chromatin are nucleosomes, which consist of ~146 base pairs (bp) of DNA wrapped around a histone octamer in ~1.7 left-handed superhelical turns [1]. The histone octamer consists of two copies each of the canonical histones H2A, H2B, H3, and H4, or their respective variants and isoforms [2, 3]. Nucleosomes package and protect the DNA, however they are also important carriers of epigenetic information, which is encoded in their local chromosomal position, histone composition, and chemical modification [4]. The position and epigenetic states of nucleosomes help govern transcription and gene regulation, mark chromosomal loci, and help conduct other DNA-associated processes such as replication, recombination, and repair [5, 6].
The nucleosomal landscape along DNA is shaped by the collective action of ATP-dependent chromatin remodellers along with numerous other factors [5, 6]. Chromatin remodellers use the energy of ATP hydrolysis to slide, position, or edit (i.e. exchange histone variants) nucleosomes, or evict histones altogether. They are grouped into, but not limited to, four main families, denoted INO80, SWI/SNF, ISWI, and CHD. Chromatin remodellers display diverse domain and subunit architectures but share an ATP-dependent motor domain, which belongs to the Snf2 family among superfamily 2 helicases/translocases. Using cycles of ATP binding and hydrolysis, Snf2 ATPases translocate DNA and/or alter local DNA shape, which serves as the underlying chemo-mechanical activity governing diverse remodelling reactions [5, 6].
The INO80 family consists of INO80 and SWR1 type complexes [7, 8]. Both remodellers are involved in shaping chromatin, particularly around nucleosome-depleted or nucleosome-free regions (NDRs/NFRs), such as promoters and origins of replication. They can sense extended stretches of extranucleosomal entry DNA. Despite this shared ability, they catalyse distinct nucleosome reconfigurations: SWR1 primarily edits nucleosomes by exchanging histone variants, whereas INO80 mainly repositions nucleosomes along the DNA [8, 9]. INO80 can slide nucleosomes and place them at the boundaries of NDRs in vitro. INO80 can also generate nucleosomal arrays outwards from these NDRs [10, 11]. SWR1, on the other hand, incorporates the H2A variant H2A.Z to replace canonical H2A in nucleosomes [12–14]. In humans, the function of yeast SWR1 is carried out by SRCAP and TIP60 complexes [15, 16]. H2A.Z is enriched at promoters, enhancers, and origins of replication and has pleiotropic-, context-, and species-dependent functions including roles in transcription regulation, DNA repair, and others [17]. Histone exchange activity has also been reported for yeast INO80 in vitro, with a preference to exchange H2A.Z to H2A [18]. In vivo, INO80 regulates the genome-wide distribution of H2A.Z and evicts H2A.Z from promoters during transcriptional activation. This activity specifically targets unacetylated H2A.Z, whose removal is important for maintaining genome integrity, but still its physiological role is not yet fully understood [19].
Yeast and fungal INO80 can also slide subnucleosomal particles, such as hexasomes [20–22]. Hexasomes are characterized by loss of one histone H2A/H2B dimer. They are proposed to arise when RNA polymerase II transcribes through the nucleosome [23]. Depending on the orientation, they either block or allow transcription by RNAPII in vitro. Although our understanding of hexasomes’ roles or their presence in various genomic processes in vivo remains limited, they are thought to arise as transient intermediates or products during many nucleosomal processes and remodelling reactions [24]. This underscores the importance of understanding how remodellers interact with the subnucleosomal particles and non-standard nucleosome variants.
Previous biochemical studies have revealed that the INO80 complex possesses a modular structure with >15 distinct subunits [25, 26]. The largest polypeptide, Ino80 functions as scaffold on which the other modules or subunits assemble [26]. The N-terminal ‘N-module’ (human subunits NFRKB, TFPT/Amida, MCRS1, UCHL5, INO80D, INO80E assembled at the N-terminal region of INO80) is very divergent in evolution and its function is not well understood, since it is dispensable for in vitro remodelling activity [25, 27]. The central ‘A-module’ contains nuclear actin along with actin-related proteins ACTB (Actin), ACTL6A (ARP4), and ACTR8 (ARP8) and the zinc finger containing transcriptional repressor YY1 and includes an HSA (helicase-SANT-associated) domain situated in the middle of the Ino80 polypeptide chain [28]. The A-module binds extranucleosomal entry DNA and helps couple the motor activity to nucleosome sliding. INO80’s C-terminal region harbours the Snf2 domain (Ino80^motor^) and forms the nucleosome core particle mobilizing module along with nucleosome binding subunits ARP5 (actin-related protein 5), /IES6 (ino eighty subunit) subunit (INO80C), IES2 (INO80B), and the assembly chaperone heterohexamer AAA^+^ ATPases RuvBL1/RuvBL2 [28–30].
While structures of members from all major families have been visualized bound to canonical nucleosomes mostly binding at superhelical location (SHL) +2 position [13, 31, 32], considerably less is known how remodellers interact with non-canonical nucleosomes containing histone variants or subnucleosomal particles. Recently, yeast and fungal INO80 complexes and SWRI bound to hexasomes have been visualized by cryogenic electron microscopy (cryo-EM) [20, 21, 33]. Yeast and fungal INO80 complexes bind hexasomes at SHL−2/−3 position in a spin-rotated orientation compared to nucleosome binding at SHL−6 position, with the H2A/H2B dimer on the DNA-entry side absent [20, 21]. Lack of acidic patch (negatively charged surface region on the nucleosome, located mainly on histones H2A and H2B) interaction due to the missing H2A/H2B dimer and ~46 base pairs of unwrapped entry-side DNA led to a different binding mode through formation of new interfaces between yeast and fungal INO80 and the hexasome in comparison with nucleosome interactions [20, 21]. However, this mode of subnucleosomal particle binding does not appear to be a general feature of chromatin remodellers. The CHD1 remodeller requires an H2A/H2B at the DNA entry side of the nucleosome and slides hexasomes in the opposite way than observed for INO80 in vitro [34]. Furthermore, since some nucleosome recognition features, in particular the ARP5/ACTR5 insertion element [30] (‘grappler’) that binds the exposed histone surface and directly engages the proximal H3/H4 face of the hexasome (or the H2A/H2B acidic patch in nucleosomes) in yeast and fungal INO80 is missing in human INO80, it is yet unclear to what extent sliding of non-canonical nucleosomes is even present in the mammalian system.
Here, we show that the human INO80 ∆N complex (C- and A-module, lacking the N-terminal module) can slide hexasomes as efficiently as nucleosomes, similar to its yeast and fungal orthologs [20, 21]. This demonstrates an evolutionary conserved apparent indifference to nucleosomal and hexasomal species with respect to basal sliding activity. The structures of the human INO80 complex bound to hexasomes, H2A.Z-nucleosomes, and canonical nucleosomes in different states show a high capability for spin-rotation relative to the nucleosomal dyad, where essentially the location of entry DNA and its interaction with the Ino80^motor^ domain determine the placement of the remodeller.
Materials and methods
Expression and purification of human ∆N INO80 complex
The open reading frames of subunits of Homo sapiens (Hs) INO80 (C-module comprising RuvBL1, RuvBL2, ARP5, IES2, IES6, and Ino80^267-1552^ having a 6× histidine tag at the N-terminus and 2× streptavidin tag at the C-terminus) and A-module (ARP8, Actin, ARP4, and YY1) were optimized for insect cell expression and ordered from GeneArt (Thermo Fisher Scientific). The gene cassettes for C- and A-modules were assembled separately on two pBIG2ab vectors employing the biGBac cloning system [35]. The baculoviruses generation was done in Sf21 insect cells (Spodoptera frugiperda; Thermo Fisher Scientific, #11497013), after which the complexes were recombinantly expressed in High Five insect cells (Trichoplusia ni; Invitrogen, #B85502) by adding the two viruses at a ratio of 1:100 (volume virus:medium) to 3 l of insect cell media. Cells were cultured for 60 h at 27°C and harvested by centrifugation at 4°C.
The complex was purified by sequential Ni-NTA affinity, Strep-tag affinity, and HiTrap Q (1 ml) chromatography according to the previously published protocol [36]. Briefly, the cell pellet was resuspended in lysis buffer [50 mM Tris, pH 8.0, 500 mM NaCl, 1 mM tris(2-carboxyethyl)phosphine (TCEP), 2 mM benzamidine–HCl, 10% (v/v) glycerol] supplemented with 10 µl BaseMuncher nuclease (Abcam, ab270049) and one ethylenediaminetetraacetic acid-free protease inhibitor tablet (Sigma–Aldrich). The lysate was gently sonicated for 2 min (duty cycle, 50%; output control, 5). After sonication, the lysate was clarified by centrifugation at 30 500 × g for 1 h at 4°C, and then filtered through a 0.22-µm Millipore filter. The lysate was loaded onto a pre-equilibrated 5 ml HisTrap FF column (Cytiva) using a peristaltic pump. The column was washed with 10 column volumes of His-buffer A [50 mM Tris, pH 8.0, 250 mM NaCl, 1 mM TCEP, 10% (v/v) glycerol], and the complex was eluted with 5 column volumes of His-buffer B (His-buffer A with 250 mM imidazole). The eluted complex was immediately loaded onto pre-equilibrated Strep-Tactin^®^ Sepharose^®^ resin (IBA Lifesciences, 3 ml slurry) and incubated for 25 min, gentle rolling at 4°C.
The resin was then washed with Strep-buffer A [50 mM Tris, pH 8.0, 200 mM NaCl, 1 mM TCEP, 10% (v/v) glycerol]. After washing, the complex was eluted iteratively in Strep-elution buffer (Strep-buffer A supplemented with 5 mM desthiobiotin, Sigma–Aldrich) over a period of 60 min. The eluate was subsequently loaded onto a HiTrap Heparin HP column (Cytiva), and the protein was eluted using a linear salt gradient from 100% HiTrap-Q buffer A [50 mM Tris, pH 8.0, 1 mM TCEP, 10% (v/v) glycerol] to 100% HiTrap-Q buffer B [50 mM Tris, pH 8.0, 1 M NaCl, 1 mM TCEP, 10% (v/v) glycerol] over 12 column volumes. The fractions eluted, were analysed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis and those with the highest purity were pooled together. The protein complex was concentrated to 5 mg/ml using centrifugal filters (Centricon, 50 kDa cutoff, Millipore) and either used for vitrification on the same day or dialysed in storage buffer [50 mM Tris, pH 8.0, 250 mM NaCl, 10% (v/v) glycerol, 1 mM TCEP] before being flash-frozen in liquid nitrogen for the in vitro biochemical assays.
INO80 mutants were generated by site-directed mutagenesis polymerase chain reaction (PCR) and were expressed and purified using the same protocol as the wild-type protein. List of primers used for preparation of INO80 mutants are listed in Table 1.
Preparation of nucleosomes and hexasomes
Canonical human histones were obtained from the histone source at Colorado State University, USA, and unfolded in a buffer containing 7 M guanidinium chloride, 20 mM Tris (pH 7.5), and 1 mM dithiothreitol (DTT) at room temperature for 30 min with continuous rotation. Following this, histone dimers, tetramers, and octamers were refolded according to methods described in previous publications [37]. The histone H2A mutant (E61A/E64A/D72A/D90A) was also expressed and purified as previously reported [37]. Widom 601-DNA [38] with varying linker lengths was amplified via PCR and purified using anion exchange chromatography. After overnight dialysis into ultrapure water, the purified DNA was concentrated under vacuum and stored at −20°C.
The details of the primers used are provided in Table 2. Nucleosomes were reconstituted by mixing the histone octamer and DNA in a 1:1 ratio at 2 M NaCl, followed by gradual reduction of the salt concentration to 50 mM over 16 h at 4°C with continuous stirring. Hexasomes were prepared similarly by mixing dimer, tetramer, and DNA in a 1:1:1 ratio [21]. Both nucleosomes and hexasomes were purified using a 1 ml Source Q 4.6/100 Column (Cytiva). Fractions containing the purified samples were pooled, dialysed to 50 mM NaCl, concentrated using a 10-kDa cutoff Centricon (Millipore), and stored at 4°C.
Nucleosome and hexasome sliding assay
The sliding activity of INO80 on nucleosomes and hexasomes was measured as previously reported [21, 28, 30]. Briefly, 6-carboxyfluorescein-labelled 0N80 nucleosomes and Cy5.5-labelled 0H80 hexasomes were used to determine the remodelling activity of INO80 and mutants. For the sliding assay, INO80 or its mutant at a concentration of 75 nM was mixed with 50 nM nucleosomes or hexasomes in a buffer containing 25 mM HEPES (pH 8.0), 60 mM KCl, 7% (v/v) glycerol, 0.1 mg/ml bovine serum albumin (BSA), 2 mM MgCl_2_, and 0.25 mM DTT. The samples were pre-incubated at 37°C for 2 min, and the sliding reaction was initiated by adding ATP to a final concentration of 1 mM. Samples were collected at various time points, including before ATP addition, at 30 s, t 1, 2, 5, 10, 15, 30, and 60 min. Reactions were stopped by adding Lambda DNA (0.2 mg/ml; NEB). For loading, 1 µl of 50% (v/v) glycerol was added to each sample, and 6 µl of each sample was loaded onto a 3%–12% acrylamide Bis–Tris native gel (Invitrogen). Gels were run at 110 V for 90 min. Visualization was performed using the Typhoon imaging system (GE Healthcare).
For hexasomes, the gels were run for a longer period—first at 110 V for 90 min, followed by an additional 30 min at 150 V to achieve better separation of translocated versus non-translocated hexasomes. Visualization was performed using the Typhoon imaging system (GE Healthcare). All gels were then quantified using the ImageJ version 1.53k software [39].
To account for minor variations in sample loading and the presence of small amounts of nucleosomes in hexasome samples (and vice versa), a correction was applied. The percentage of the substrate’s band intensity (S%) was calculated at each time point relative to the combined intensity of the substrate, intermediate products, and final product. The S% value of the -ATP samples was used as a baseline, representing 100% non-remodelled substrate, and this baseline was then used to calculate the fraction of remodelled substrates. Graphs were generated using GraphPad Prism version 9.5.1 (Dotmatics) by fitting the data to a single-phase exponential decay model described by the following equation:
\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} \begin{eqnarray*} y = ({{y}_0} - p){{\rm e}^{-k}}{{\rm obs}^{\rm t}} + { p}, \end{eqnarray*}\end{document}where y0 is the initial fraction product, kobs is the observed rate constant, and p is the fraction product at the plateau.
ATPase assay
The ATPase activity of INO80 and mutants was determined using a NADH-coupled ATPase assay [28]. INO80 or mutants (75 nM) were incubated in an assay buffer containing 25 mM HEPES (pH 8.0), 50 mM KCl, 1 mM DTT, 2 mM MgCl_2_, and 0.1 mg/ml BSA, along with 0.5 mM phosphoenolpyruvate, 1 mM ATP, 0.1 mM NADH, 25 U/ml lactate dehydrogenase, and pyruvate kinase (Sigma–Aldrich) at 37°C in a final volume of 50 μl. The decrease in NADH concentration was monitored over 90 min using non-binding, 384-well black plates (Greiner Bio-One). Fluorescence was measured at an excitation wavelength of 340 nm and an emission wavelength of 460 nm using a Tecan Infinite M100 plate reader (Tecan). The ATPase rate was determined in the presence of 50 nM nucleosomes or hexasomes, with ATP turnover calculated using the maximal initial linear rates, corrected for a buffer blank, using the following equation:
\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} \begin{eqnarray*} &&{\rm APTase\ rate}\ {\left[{\rm ATP\ {complex}^{-1}}\ {\rm min^{-1}}\right]} \\&&\quad = \frac{\rm Slope}{\rm NADH\ slope\ (\rm cINO80\ (\mu \rm {moles}))} * 60, \end{eqnarray*}\end{document}where cINO80 is the concentration of human INO80 in micromoles, and NADH slope is the slope of the calibration curve calculated by carrying out titration of varying ADP concentrations ranging from 0 to 100 µM in 10 µM increments to constant NADH concentration. All the experiments were performed in triplicates.
Competition assay
Nucleosome and hexasome translocation were simultaneously compared within the same reaction. For this, a mixture of 50 nM 6-carboxyfluorescein (6-FAM)-labelled 0N80 nucleosomes and 50 nM Cy5.5-labelled 0H40 hexasomes was incubated with 50 nM of either wild-type or mutant INO80. The competition assays were performed in the same manner as the individual sliding assays. Gels were run for 1.5 h at 100 V, followed by an additional 30 min at 150 V, and then scanned for both 6-FAM and Cy5.5 signals. Visualization was performed using the Typhoon imaging system (GE Healthcare). In parallel, the translocation of H2A.Z-containing 0N80 Cy5.5-labelled nucleosomes was compared to canonical nucleosomes under similar conditions, with gels run for 1.5 h at 100 V. Quantification of the gels was carried out as described in the above section.
Vitrification of the INO80-nucleosome/hexasome complex
For cryo-EM sample preparation, the purified INO80 complex (final concentration 800 nM) was mixed with either 0N40 nucleosomes or 0H80 hexasomes (final concentration 500 nM). The mixture was dialysed in 2 l of dialysis buffer (20 mM HEPES, pH 8.0, 50 mM NaCl, and 0.25 mM TCEP) for 2 h using a 10-kDa cutoff Slide-a-lyzer Dialysis Tube (Thermo Fisher Scientific).
Three millimolars ADP and 5 mM MgCl_2_ were prepared at 10× concentration in a 2× buffer containing 60 mM HEPES (pH 8.0), 100 mM NaCl, 8 mM MgCl_2_, and 1 mM DTT. Three millimolars BeF_2_ and 15 mM NaF were also prepared at 10× concentration in the same buffer. The dialysed INO80–nucleosome complex or INO80–hexasome complex was supplemented with ADP–MgCl_2_ and BeF_2_–NaF stock solutions to reach a final 1× concentration. This complex was incubated on ice for 30 min, after which β-Octyl glucoside (Roth, Germany) was added to a final concentration of 0.05%. For vitrification, 4.5 μl of the sample was applied to glow-discharged Quantifoil R2/1 200 mesh copper grids, followed by a 2.2-s blot time using a Leica EM GP (Leica; 10°C, 90% humidity).
Electron microscopy and data collection
Movies of INO80–nuclesosome embedded in vitreous ice were captured at liquid nitrogen temperature using a Titan Krios G3 transmission electron microscope (Thermo Fisher Scientific). This microscope was equipped with a K2 Summit direct electron detector (Gatan) and a BioQuantum LS Imaging Filter (Gatan). The movies were recorded in counting mode using EPU acquisition software (Thermo Fisher Scientific) at a magnification of 130 000×, resulting in a pixel size of 1.059 Å/pixel, with a nominal defocus range of −1.1 to −2.9 μm. Each movie received a total electron dose of ~40 to 46 e^−^/Å^2^, distributed over 40 frames, with an exposure time of 250 ms per frame. Movies of INO80–hexasome particles embedded in vitreous solution were collected at liquid nitrogen temperature using a Titan Krios G3 transmission electron microscope (Thermo Fisher Scientific) equipped with a Falcon 4 Direct Detection Camera and BioQuantum LS Imaging Filter (Gatan) operated at 300 kV acceleration voltage. The movies were recorded in counting mode using EPU acquisition software (Thermo Fisher Scientific) at 165 000× magnification with a pixel size of 0.727 Å/pixel and nominal defocus range of −1.1 to −2.6 μm. The total electron dosage of each movie was 40 e^−^/Å^2^, fractionated into 40 movie frames and exposure time of 3.19 s per frame.
Cryo-EM data processing for INO80 complex
Movie frames were motion-corrected using MotionCor2 v1.4.5 [40]. All subsequent cryo-EM data processing steps were performed using cryoSPARC v3.3.1 [41], with resolutions calculated using the gold-standard Fourier shell correlation criterion (FSC = 0.143). CTF parameters were estimated using patch CTF estimation (multi). The same data processing workflow was uniformly applied to all datasets used in this study, and data collection and refinement statistics are summarized in Table 3 .
Cryo-EM data processing for INO80–hexasome complex
For the INO80–hexasome complex, initial particle picking was conducted using the blob picker (157 133 59 particles). Selected particles were subjected to 2D classification. Defined 2D classes were chosen and particles were used as input for a Topaz training job [42, 43]. The Topaz model was used for particle picking across 39 388 micrographs, yielding 981 468 particles, extracted with a box size of 440 pixels and a pixel size of 0.727 Å. Following 2D classification and ab initio reconstruction, the most well-defined classes were selected and subjected to heterogeneous refinement. The class with the best-defined features, comprising 34 444 particles, was chosen for further refinement, resulting in a final resolution of 3.6 Å for the INO80–hexasome complex after non-uniform refinement [44].
To obtain a focused map around the hexasome, the volume was imported into ChimeraX 1.7.1 [45]. The ‘segment map’ function in ChimeraX was used for splitting the volume. The first volume was generated for the hexasome, and the second volume was generated for RuvBL1/2, Ino80^motor^, ARP5/IES6 subunits, and IES2 (aa 208–342). The masks for hexasome (lowpass filter: 12, dilation radius: 15) as well as the RuvBL1/2, Ino80^motor^, ARP5/IES6 subunits, and IES2 (aa 208–342) (lowpass filter: 12, dilation radius: 0) were then generated using these volumes in cryoSPARC v3.3.1 [41]. Particle subtraction was thereafter done using the mask generated for RuvBL1/2, Ino80^motor^, ARP5/IES6 subunits, and IES2 (aa 208–342). These subtracted particles were used against the mask generated for the hexasome to do the local refinement in cryoSPARC v3.3.1 using the pose/shift Gaussian during alignment (standard deviation of prior over rotation: 10, standard deviation of prior over shifts: 5). Thus, a focused map was produced for the hexasome at a resolution of 6.9 Å. The local resolution maps, angular distribution plots, and FSC curves are shown in Supplementary Fig. S1.
In the INO80–hexasome complex, certain 2D classes exhibited fuzzy map densities attached to the INO80–hexasome complex. These 2D classes (58 969 particles) were selected and re-extracted from the micrographs with a larger box size of 800 pixels. After another round of 2D classification, these particles underwent ab initio reconstruction followed by heterogeneous refinement in two classes. The resulting volume containing 26 073 particles, showing clear density for both the C- and A-module, and was further subjected to 3D classification with three classes. This classification further sorted the particles, isolating a subset of 22 112 particles that displayed density for both the C- and A-module together. Subsequent non-uniform refinement produced a low-resolution map at 8.95 Å. The map density for the C-module appeared clearer compared to the A-module due to the flexible nature of the HSA domain. To enhance the quality of the flexible A-module, particles were further processed using 3D-flex refinement [46] with default parameters. The final map obtained from this 3D-flex refinement was then used for docking the previously determined C-module structure and the AlphaFold3 [47] predicted A-module structure into the corresponding map densities. The overall processing scheme is illustrated in Supplementary Fig. S1.
Cryo-EM data processing for INO80–nucleosome complex
The initial processing for INO80–nucleosome dataset was processed similarly as mentioned above. After topaz training [42, 43], the resulting Topaz model was then applied as a template for particle picking across 19 528 micrographs, yielding 546 516 particles, extracted with a box size of 440 pixels and a pixel size of 1.046 Å. After selecting 2D classes with defined features, a round of ab initio reconstruction with five classes was performed. Classes with most well-defined features were chosen for heterogeneous refinement into three classes, followed by 3D classification, resulting in two distinct classes, named state N-6 and state N-7. Both classes were further refined using non-uniform refinement [44], achieving final resolutions of 3.7 Å for state N-6 and 3.6 Å for state N-7.
To obtain focused maps around the nucleosome and IES2 (aa 140–189) in both states N-6 and N-7, the volumes were imported into ChimeraX 1.7.1 [45]. The ‘segment map’ function in ChimeraX1.7.1 [45] was used to split the volumes for both states (N-6 and N-7). The first volume was generated for the nucleosome and IES2 (aa 140–189), and the second volume was generated for RuvBL1/2, Ino80^motor^, ARP5/IES6 subunits, and IES2 (aa 208–342). The masks for nucleosome and IES2 (aa 140–189) (lowpass filter: 12, dilation radius: 15) as well as the RuvBL1/2, Ino80^motor^, ARP5/IES6 subunits, and IES2 (aa 208–342) (lowpass filter: 12, dilation radius: 0) were then generated from these volumes in cryoSPARC v3.3.1.
Particle subtraction was performed using the mask generated for RuvBL1/2, Ino80^motor^, ARP5/IES6 subunits, and IES2 (aa 208–342). These subtracted particles were used against the nucleosome and IES2 (aa 140–189) mask to do a local refinement in cryoSPARC v3.3.1 [41] using the pose/shift Gaussian priors (rotation prior SD: 10; shift prior SD: 5). The focused maps were refined to resolutions of 4.1 Å for state N-6 and 3.8 Å for state N-7. The local resolution maps, angular distribution plots, and FSC curves are shown in Supplementary Fig. S2.
Cryo-EM data processing for INO80 H2A.Z–nucleosome complex
The INO80 H2A.Z–nucleosome complex was processed similarly as mentioned above. Briefly, from a total of 36 316 micrographs, initial processing steps described above were performed. Particle picking was carried out using a blob picker, followed by particle extraction from micrographs with a box size of 400 pixels and a pixel size of 1.046 Å, yielding 299 459 particles. These particles underwent multiple rounds of 2D classification. After final 2D classification, 5076 particles from well-defined classes were selected and used as input for training a Topaz model [42, 43]. This trained Topaz model was subsequently employed for particle picking across all 36 316 micrographs, resulting in 3 108 522 particles, extracted with a box size of 440 pixels. After additional rounds of 2D classification, particles displaying clearly defined features were selected. These particles underwent ab initio reconstruction into six classes. The best-defined classes were selected and subjected to heterogeneous refinement into six classes, followed by 3D classification, which yielded two similar classes. These two classes, comprising 94 892 particles, were combined and further refined using non-uniform refinement [44], achieving a final resolution of 3.5 Å.
To obtain a focused map around the H2A.Z–nucleosome, the volume was imported into ChimeraX 1.7.1 [45]. The ‘segment map’ function in ChimeraX was used for splitting the volume. The first volume was generated for the nucleosome, and the second volume was generated for RuvBL1/2, Ino80^motor^, ARP5/IES6 subunits, and IES2 (aa 208–342). The masks for nucleosome (lowpass filter: 12, dilation radius: 15) as well as the RuvBL1/2, Ino80^motor^, ARP5/IES6 subunits, and IES2 (aa 208–342) (lowpass filter: 12, dilation radius: 0) were then generated using these volumes in cryoSPARC v3.3.1 [41]. Particle subtraction was performed using the mask generated for RuvBL1/2, Ino80^motor^, ARP5/IES6 subunits, and IES2 (aa 208–342). These subtracted particles were used against the nucleosome and IES2 (aa 140–189) mask to do a local refinement in cryoSPARC v3.3.1 [41] using the pose/shift Gaussian priors (rotation prior SD: 10; shift prior SD: 5). Thus, a focused map was produced for the nucleosome at a resolution of 3.6 Å. The local resolution maps, angular distribution plots, and FSC curves are shown in Supplementary Fig. S3.
Additionally, similar to observations in the INO80–hexasome complex, some particles displaying fuzzy density in the 2D classification were identified. These particles were selected and re-extracted with a larger box size of 800 pixels, followed by additional rounds of 2D classification. The well-defined 2D classes obtained were selected and used to train another Topaz model [42, 43]. This trained model was then applied for particle picking across the entire set of 36 316 micrographs, yielding 179 796 particles, extracted with a box size of 800 pixels. After multiple rounds of 2D classification, particles exhibiting clear features were chosen. These selected particles underwent ab initio reconstruction into two classes, and the class with the most clearly defined features was subjected to heterogeneous refinement into two classes. The better defined class containing 11 540 particles was further refined using non-uniform refinement, resulting in a low-resolution map of 11.6 Å. Similar to the INO80–hexasome complex, the A-module map quality was poorer compared to the C module due to inherent flexibility. To improve the quality of the flexible A-module map, the particles were further processed using 3D-flex refinement [46] using default parameters. The final map obtained from this 3D-flex refinement was subsequently used to dock the previously determined C-module structure along with the AlphaFold3 [47] predicted structure of the A-module into their corresponding map densities. The processing scheme is illustrated in Supplementary Fig. S3.
Cryo-EM data processing for INO80apo–nucleosome complex
For the INO80–nucleosome complex without nucleotide, a total of 30 012 micrographs were collected, and initial processing steps were performed as described above. Following multiple rounds of 2D classification, particles with well-defined features were selected and used to train a new Topaz model [42, 43]. This trained model was then applied to the full dataset, yielding 605 639 particles, which were extracted with a box size of 440 pixels.
As in previous datasets, several 2D classes showed fuzzy densities, suggestive of partially resolved regions. These particles were processed using a similar workflow. Heterogeneous refinement and subsequent 3D classification yielded a reconstruction with clearly defined C- and A-module densities. This reconstruction was further refined using non-uniform refinement [44], resulting in a final resolution of 9.1 Å from 16 786 particles.
Due to the intrinsic flexibility of the A-module, its density remained suboptimal, prompting the use of 3D-flex refinement [46] to improve map quality. The final 3D reconstructions for both classes were used to dock the previously determined C-module structure and the AlphaFold3 [47] predicted A-module structure into their respective densities. The processing workflow is summarized in Supplementary Fig. S4.
Model building and refinement
The model for the INO80 C-module–nucleosome complex was initially constructed for state N-7 and then used as the basis for the model building of state N-6, INO80–H2A.Z containing 50N50 nucleosomes, and the INO80–hexasome complex state H-3. These models were built into B-factor sharpened maps derived from focused refinements. The starting models were obtained from PDB entries 7ZI4 (for the C-module) [29] and 2CV5 [48] (for the nucleosome). These models were first rigid-body fitted using ChimeraX version 1.7.1 [45], and then model rebuilding was performed using COOT version 0.9.8.93 [49]. For State N-7, DNA was modelled beyond SHL−3 and fitted into the density using German-McClure restraints in COOT. The N-terminal region of IES2 (aa 140–189), in both states N-6 and N-7, was modelled by rigidly fitting the AlphaFold [50] predicted structure of IES2 (AF-Q9C086-F1-v4) into the corresponding cryo-EM densities, and refined using COOT. The model building and refinement were carried out iteratively using interactive molecular dynamics flexible fitting in ISOLDE v1.4 [51]. Reciprocal space refinement, using jelly-body restraints, was performed with REFMAC5 [52] against maximum-likelihood weighted structure factors calculated from cryo-EM half-maps. Additional model building was conducted in COOT against the maximum-likelihood estimate of the expected true map calculated by REFMAC5. Final model corrections were made using ISOLDE against the same REFMAC5 map, followed by a concluding round of reciprocal space refinement with jelly-body restraints in REFMAC5.
All structural figures were prepared using UCSF ChimeraX 1.7.1 [45].
Results
Human INO80 can slide hexasomes and nucleosomes
We recombinantly reconstituted the Hs ∆N INO80 complex, consisting of an evolutionarily conserved C-(Ino80^motor^, ARP5, IES6, IES2, RuvBL1, RuvBL2) and A-module (ARP8, Actin, ARP4, YY1) (Supplementary Fig. S5a–c). This complex follows prior work [29], however with the addition of the transcriptional regulator YY1. In the presence of ATP, the human INO80 complex demonstrated robust nucleosome sliding activity, similar to what has been previously observed with Hs, fungal, and yeast INO80 complexes [20, 29, 30]. We next tested hexasome sliding activity of human INO80 and found robust sliding capability (Fig. 1a–d). We used 80 bp of extranucleosomal DNA for nucleosome and 40 bp extrahexasomal DNA for hexasome substrates, to minimize influences of linker DNA lengths on sliding rates caused by additional DNA unwrapping from histone core in the case of the hexasome. Our results show that human INO80, like yeast and fungal INO80, can efficiently slide hexasomes [20, 21]. Notably, we do not see a strong preference for either hexasomes or nucleosomes if sufficient extrahexasomal DNA (~46 bp) is available (Supplementary Fig. S5d).
Structure of INO80 bound to hexasome. (a) Cartoon depiction of a canonical nucleosome (PDB:2CV5) [48] and the hexasome from this study. The hexasome was reconstituted with Widom 601 DNA sequence [38] (80 bp linker on one side), spanning a 101 bp footprint and lacking the H2A–H2B dimer. (b) Cartoon illustration of a 0H80 nucleosome and a 0H80 hexasome. The entry-side H2A–H2B dimer is shown in dark grey, H3–H4 in light grey, the 601 DNA in black, the flanking DNA in red, and the unwrapped DNA in green. (c) Evaluation of the sliding activity of INO80 on 0N80 nucleosome and 0H80 hexasome substrates is shown. Band intensities of remodelled and unremodelled nucleosome/hexasome species were quantified, and the fraction of remodelled nucleosomes/hexasomes was plotted against time. Data points were fitted using an exponential equation (see the ‘Materials and methods’ section). Mean values and individual data points (n = 3, technical replicates) are shown. (d) Native PAGE analysis of sliding of 0N80 nucleosomes and 0H80 hexasomes by INO80 is shown. (e) Cryo-EM structure of the INO80 C-module bound to the hexasome. INO80 binds on the proximal side of the hexasome that lacks the H2A–H2B dimer (state H-3). The dashed boxes in orange and purple highlight the interaction of the INO80 C-module with the nucleosome, whereas the box in green highlights the interaction of RuvBL1 with the Ino80motor. (f) The aa residue D214 in chain D of RuvBL2 is located in proximity (within 5 Å) to residues in the α2 helix of histone H3, positioning it to make a potential interaction. (g) Additionally, aa residue K165 in chain B of RuvBL1 interacts with the phosphate backbone of the unwrapped DNA near SHL−2.5. (h) The N-terminal of chain B of RuvBL1 forms an additional strand on the edge of a β-sheet of the C-terminal lobe of the Ino80motor.
Structure of human INO80 bound to hexasomes
To understand how INO80 recognizes hexasomes, we determined structures in complex with hexasome (0H80) using cryo-EM. The structures were determined in the presence of the ATP analog ADP–BeF_3_ without any chemical crosslinking.
We determined the structure of INO80–0H80 at 3.6 Å, with a focused map of the hexasome at 6.9 Å. The map revealed the INO80 C-module, while the A-module was not visible due to flexibility. The map density for the RuvBL1/2, the ARP5/IES6 subunit, the Ino80^motor^, and the IES2 subunit was well resolved, with clear density for ADP visible in all six subunits of the RuvBL1/2 (Supplementary Fig. S6). The INO80 C-module engages the ‘proximal face’ of the hexasome, the side where the H2A/H2B dimer is missing. Due to the missing H2A/H2B dimer, an additional ~46 bp of DNA were unwrapped from the histone core. The Ino80^motor^ domain binds DNA at the resulting new entry position of linker DNA, that is, SHL−3. Similar to yeast and fungal INO80, [20, 21] human INO80 adopts an ~145° ‘rigid body’ spin rotation around the ‘axle’ of the hexasome ‘wheel’ relative to its location on the nucleosome. As a consequence, the ARP5/IES6 subunits bind DNA on the opposite side of the hexasome near the dyad (SHL0) and SHL+1. We denote this INO80 binding state ‘H-3’ with the Ino80^motor^ domain located at hexasome SHL−3 (Fig. 1e).
Hexasome recognition is mostly governed through binding of hexasomal DNA, with the exception of one RuvBL2 aspartic acid residue (D214), which is in close proximity to the H3 loop 2. This interaction might stabilize this H3/H4 pair at the entry side of the DNA (Fig. 1f). The predominantly DNA-mediated hexasome binding contrasts yeast and fungal orthologs, where the Arp5 grappler insertion [30] (largely missing in human INO80) binds either the H2A/H2B acidic patch on the nucleosome or H3/H4 on the hexasome. DNA-mediated contacts are found at the ATPase (entry DNA), where the RuvBL1 K165 residue binds the phosphate backbone of hexasomal DNA near the SHL−2.5 position and may stabilize hexasomal DNA at the new linker DNA entry site (Fig. 1g). Interestingly, amino acid residues ^3^IEEV^6^ of the same RuvBL1 protomer form an additional β-strand attached to the β-sheet of C-lobe of Ino80^motor^ (Fig. 1h). This type of β-strand addition has not been observed in the fungal and yeast INO80 complexes and possibly rigid the interaction of the Ino80^motor^ with the RuvBL1/2 domain.
INO80 binds nucleosomes in different positions
We determined the structure of human INO80 bound to nucleosomes (0N40) and observed two distinct nucleosome binding states, denoted state N-7 and state N-6, from the same dataset (Fig. 2a and b). The C-module (Ino80^motor^, ARP5, IES6, IES2, RuvBL1, RuvBL2) displayed well-resolved map density, including ADP in each of the 6 RuvBL1/2 subunits in both the states (Supplementary Figs S7 and S8). In state N-7 (3.6 Å, with the nucleosome locally refined to 3.8 Å), the Ino80^motor^ binds nucleosomal DNA at the SHL−7 position, while ARP5/IES6 subunits are located at SHL−3 (Fig. 2a). This position is similar to the previously determined structure of human INO80 in complex with the 50N25 nucleosome [29]. Interestingly, in N-6 we observed a second state (3.7 Å, with the nucleosome locally refined to 4.1 Å), in which human INO80 underwent a spin rotation by one SHL. This rotation positioned the ATPase domain at SHL−6 and the ARP5/IES6 subunits at SHL−2 (Fig. 2b). This position is similar to nucleosome binding of yeast and fungal INO80 [20, 30]. Transitioning from state N-7 to N-6 involves detachment of DNA from H2A at SHL−6, leading to unwrapping of ~10 bp from the histone core (Fig. 2c). In addition, the H3 tail moves away from the Ino80^motor^ (Fig. 2d and e), and DNA contacts made by the proximal-side histone H3 αN and H2A loop 2 are broken, resulting in the partial exposure of the H2A/H2B dimer (Supplementary Fig. S9a and b).
INO80 binds nucleosome in different positions. Cryo-EM structures of INO80 C-module bound to nucleosomes in states (a) N-7 and (b) N-6. The boxes in orange highlight the interaction of the INO80 C-module with the nucleosome, whereas the boxes in light blue highlight the H3 tail interaction with nucleosomal DNA. (c) Structural comparison of nucleosome states N-7 (in grey) and N-6 (in light grey) relative to the canonical nucleosome (in white) (PDB:2CV5) [53]. The transition from state N-7 to state N-6 results in unwrapping of ~10 bp from the histone core. The histone core is aligned. (d) The N-terminal H3 tail of the nucleosome in state N-7 is positioned near the C-lobe of Ino80motor. Despite its proximity, no map density was observed to indicate a direct interaction between the H3 tail and the ATPase subunit. (e) In state N-6, due to the translocation of INO80, Ino80motor moves away from the histone core and the H3-αN helix. (f) No interactions were observed between histone core residues and RuvBL2 in state N-7. (g) In contrast to state N-7, where no interaction was observed with the histone core, state N-6 shows a close proximity for a potential salt bridge interaction between aa residues R71 of histone H2A (α2) and D214 of the RuvBL2 subunit of INO80. (h) Evaluation of the sliding activity of INO80 RuvBL2 D214A mutant on 0N80 nucleosome and 0H80 hexasome substrates is shown. Band intensities of remodelled and unremodelled nucleosome/hexasome species were quantified, and the fraction of remodelled nucleosomes/hexasomes was plotted against time. Data points were fitted using an exponential equation (see the ‘Materials and methods’ section). Mean values and individual data points (n = 3, technical replicates) are shown. (i) ATPase activity of INO80 RuvBL2 D214A mutant, with and without stimulation by nucleosomes/hexasomes. ATPase rates were calculated from the linear region of the raw data and corrected using a buffer blank. Mean values and individual data points (n = 3, technical replicates) are presented. (j–m) Comparison of INO80 structures bound to different substrates. Shown are the multiple states of the INO80–nucleosome complex [N-7, N-6, and INO80-H2A.Z centrally positioned (50N50) nucleosome (NZ-7), alongside the INO80–hexasome structure (H-3)]. SHL is defined by DNA translocation, with regions proximal to the DNA entry site designated as ‘−’ and those further along the DNA path designated as ‘+’. In state N-7, the Ino80motor binds to the nucleosome at SHL−7, and the ARP5/IES6 subunits bind at SHL−3. In state N-6, the Ino80motor and ARP5/IES6 shift to SHL−6 and SHL−2, respectively, with ARP5/IES6 rotating ~40° compared to state N-7. In state H-3, the Ino80motor binds to entry DNA at SHL−3, while ARP5/IES6 subunits bind near the dyad (SHL0 position). Notably, INO80 interacts with the hexasome from the side lacking the H2A/H2B dimer, undergoing an ~145° rotation relative to state N-7. Finally, in the centrally positioned nucleosome (50N50), INO80 binds similarly to state N-7.
RuvBL2 residue D214 forms a salt bridge with the H2A α2 residue R71 in N-6 but not N-7 (Fig. 2f and g). We hypothesize that D214 might help stabilize the histone core in both hexasomes and nucleosomes when DNA is unwrapped and that the N-6 state is the ‘active’ state on nucleosomes. Consistently, INO80^RuvBL2_D214A^ showed a moderately decreased sliding activity for both nucleosomes and hexasomes compared to the wild type (Fig. 2h and Supplementary Fig. S9c). Hereby, the ATPase activity of INO80 was either unaffected (hexasome), or even slightly increased (nucleosome) compared to wild type (Fig. 2i), showing that the reduction in the sliding activity is not due to the compromised ATP hydrolysis rate.
Given the distinct modes of nucleosome binding by INO80, we also investigated how human INO80 binds a nucleosome containing the H2A variant H2A.Z. H2A.Z possesses additional acidic residues particularly in the acidic patch region [14] (Supplementary Fig. S10a and b). As previously observed, we also found that H2A.Z-containing nucleosomes are preferred substrates over canonical nucleosomes [27] (Supplementary Fig. S10c). We determined the structure of human INO80 bound to a centrally positioned 50N50 H2A.Z-containing nucleosome using cryo-EM in the presence of ADP–BeF_3_ (state NZ-7). The overall resolution achieved was 3.5 Å, with a focused map resolution around the nucleosome of 3.6 Å. The C-module displayed well-resolved map density for all the subunits, and ADP was clearly resolved in each of the six RuvBL1/2 subunits (Supplementary Fig. S11). Despite the extended acidic patch on the H2A.Z-containing nucleosome, the binding mode of INO80 was similar to that of canonical nucleosomes in state N-7 (Supplementary Fig. S10d).
Comparing H-3, N-6, and N-7 (H2A- and H2A.Z-nucleosomes) states shows that human INO80 moves in a rigid body manner without substantial changes in the C-module structure (Fig. 2j–m). Since the structures for states N-7, N-6, H-3, and the nucleosome H2A.Z were all done in the presence of ADP–BeF_3_, the different nucleosome interaction modes are not a consequence of the nucleotide binding to the Ino80^motor^. The lack of histone contacts in conjunction with the spin rotation relative to the nucleosomal dyad suggests that a defining feature of nucleosome/hexasome binding by human INO80 in respect to the point of entry DNA, a topological feature (see the ‘Discussion’ section). In summary, we reveal an intrinsic flexibility in nucleosome binding of human INO80 that also resolves the discrepancy in nucleosome recognition between human and fungal INO80 in earlier work [30].
Both ARP5 and IES6 nucleosomal DNA binding are required for robust sliding
To further understand the remodelling of nucleosomes and hexasomes, we investigated the role of the ARP5/IES6 module. The DNA-binding domain (DBD) of ARP5 [30] interacts with nucleosomal DNA in the minor groove at SHL−3 in state N-7 and SHL−2 in state N-6, while it interacts with the hexasomal DNA at SHL+1 in state H-3 (Fig. 3a). Previous studies with fungal INO80 have shown that the conserved DBD loop residues of ARP5 are essential for sliding activity of fungal INO80 on both nucleosomes and hexasomes [21, 30]. However, we observed that also a loop of the IES6 subunit (aa H129–K142), which contains multiple positively charged residues, is in close proximity to the DNA phosphate backbone in the major groove at SHL positions −4 in state N-7, at SHL−3 in state N-6, and SHL+2 in the case of state H-3 (Fig. 3a). Multiple sequence alignment [54] of this IES6 region (aa H129–K142) showed that the positively charged residues are partially conserved among species (Fig. 3b). To test the functional relevance of IES6 DNA binding, we introduced point mutations in the DNA binding loop of IES6 (G135A/K136A/K137A). These residues interact with the DNA phosphate backbone, and their mutation significantly affected the sliding activity of INO80 while still maintaining robust ATPase activity for both nucleosome and hexasome substrates (Fig. 3c and d, and Supplementary Fig. S12a). The interactions of IES6 with the histone core and nucleosomal/hexasomal DNA suggest that the IES6 loop region plays a regulatory role in human INO80 sliding.
Both ARP5 and IES6 nucleosomal DNA binding are required for robust sliding. (a) Zoomed-in view showing the ARP5/IES6 subunits positioned close to the nucleosomal/hexasomal DNA. ARP5 interacts with the phosphate backbone at the minor groove of DNA. This interaction occurs at SHL−3 in state N-7 and SHL−2 in state N-6. In state H-3, ARP5 interacts near the dyad. IES6 establishes additional contacts at the major groove of DNA, binding at SHL−4 in state N-7 and SHL−3 in state N-6 via its loop aa residues GKK (135–137). The IES6 loop aa residues GKK (135–137) interact with hexasomal DNA near the SHL+1 position. The INO80 ARP5/IES6 subunits show a similar interaction with the centrally positioned nucleosome as observed in state N-7. (b) Multiple sequence alignment [54] is shown for the loop residues of IES6 involved in interactions with nucleosomal and hexasomal DNA. H.s., Homo sapiens; S.c., Saccharomyces cerevisiae; D.m., Drosophila melanogaster; A.t., Arabidopsis thaliana; C.t., Chaetomium thermophilum; X.t., Xenopus tropicalis. The partially conserved aa residues are shown in the red box. (c) Evaluation of the sliding activity of INO80 IES6 (G135A, K136A, K137A) mutant is shown. Band intensities of remodelled and unremodelled nucleosome/hexasome species were quantified, and the fraction of remodelled nucleosomes/hexasomes was plotted against time. Data points were fitted using an exponential equation (see the ‘Materials and methods’ section). Mean values and individual data points (n = 3, technical replicates) are shown. (d) ATPase activity of INO80 IES6 (G135A, K136A, K137A) mutant, with and without stimulation by nucleosomes/hexasomes. ATPase rates were calculated from the linear region of the raw data and corrected using a buffer blank. Mean values and individual data points (n = 3, technical replicates) are shown.
Acidic patch binding is required for nucleosome but not hexasome sliding
The H2A/H2B acidic patch is the most commonly recognized epitope on the nucleosome and has been shown to regulate the activity of various chromatin remodellers [55]. Fungal ARP5 has an insertion domain (‘grappler’) that directly binds the acidic patch via a foot region [30]. Human ARP5 has a much smaller insertion domain that particularly lacks the acidic patch binding foot. However, like fungal and yeast INO80 [30], human INO80 also possesses the putative acidic patch binding by IES2. To examine the effect of the acidic patch on the INO80 chromatin remodeller, we mutated the H2A acidic patch residues (E61A/E64A/D72A/D90A) (Supplementary Fig. S12b), referred to as APM nucleosome or APM hexasome. The INO80 complex exhibited reduced sliding activity (Fig. 4f and Supplementary Fig. S12c) but retained robust ATPase activity with the APM nucleosome (Supplementary Fig. S12d). This data contrast findings for fungal INO80 [30], where mutations of the acidic patch residues completely abolished nucleosome sliding activity. Interestingly, the APM hexasome had no effect in the case of hexasomes, suggesting that IES2-mediated acidic patch contacts are important in the context of nucleosome remodelling but not hexasome remodelling (Fig. 4f and Supplementary Fig. S12c).
Roles of IES2: DNA unwrapping from the distal side and its engagement with the acidic patch. (a) The density map shows the modelled IES2 N-terminal helix (aa G140–S189) interacting with the distal side of the nucleosome in state N-7. (b) Comparison of IES2 interaction with nucleosomal DNA in states N-7 (in grey) and N-6 (in light grey) with that of the canonical nucleosome (in white) (PDB:2CV5) reveals that nucleosomal DNA in states N-7 and N-6 cannot follow the same path as in the canonical nucleosome. If it did, a steric clash would occur between the exit side nucleosomal DNA and the IES2 N-terminal helix. The histone core is aligned. (c) Multiple sequence alignment of IES2 subunit across species highlights the conserved arginine anchor residues (highlighted in light blue) and partially conserved arginine residues (highlighted in light green). S.c., Saccharomyces cerevisiae; S.p., Saccharomyces pombe; D.m., Drosophila melanogaster; H.s., Homo sapiens; A.t., Arabidopsis thaliana; C.t., Chaetomium thermophilum. (d) Interaction of conserved arginine residue (R179) of IES2 subunit with the acidic patch residues (E61, D90) of histone H2A. Evaluation of the sliding activity of (e) INO80 IES2 (R179A, R181A) mutant and (f) APM nucleosome/hexasome. Band intensities of remodelled and unremodelled nucleosome/hexasome species were quantified, and the fraction of remodelled nucleosomes/hexasomes was plotted against time. Data points were fitted using an exponential equation. Mean values and individual data points (n = 3, technical replicates) are shown.
IES2 functionally interacts with the acidic patch on nucleosomes but not hexasomes
In our INO80 structure with 0N40 nucleosomes, in both states N-6 and N-7, we were able to improve the resolution of the IES2 subunit compared to the previously published structure [29], revealing additional density at the distal side of the nucleosome. Based on the shape, chemical environment, and AlphaFold model [50] (AF-Q9C086-F1-v4), this density corresponds to the N-terminus of IES2 (G140–S189). We built the model for this region of IES2 and found that it forms multiple contacts with the distal side histones and nucleosomal DNA. The resulting model includes IES2 residues G140–S189 (region I), T236–N263 (region II), and P271–R342 (region III), which make contacts with the histone core as well as with the Ino80^motor^ and with the RuvBL1/2 subunits of the INO80 complex (Supplementary Fig. S13a). The N-terminal α-helix (G140–K158) of IES2 region I is positioned near the DNA exit site at SHL+5 in both states N-7 and N-6 (Fig. 4a). We observed no density for DNA beyond SHL+5 in either state. When aligning the canonical nucleosome with the map density, the IES2 α-helix (G140–K158) causes steric hindrance at the DNA exit, suggesting that this interaction needs partially unwrapped exit DNA in a chromatin environment (Fig. 4b). DNA unwrapping from both entry and exit sides has been reported in other chromatin remodellers like Chd1 and SWR1 [56], 57].
Region I also exhibits a highly conserved RxR arginine anchor motif, which binds the acidic patch (Fig. 4c and d). Mutating R179 and R181 to alanines resulted in moderately decreased sliding activity (Fig. 4e and Supplementary Fig. S13b) but increased the ATPase activity of INO80 in the nucleosome context three-fold (Supplementary Fig. S13c). A previous INO80 delta IES2 construct was deficient both in nucleosome sliding and ATPase activity [36]. Finally, no effect of the RxR mutation was observed on hexasome sliding activity or hexasome-stimulated ATPase activity (Supplementary Fig. S13c), consistent with a lack of density for IES2 at the hexasome distal histone surface including the acidic patch.
The region II (throttle helix) [30] interacts with the nucleosomal DNA near the major groove at different SHL positions in states N-6, N-7, and H-3 (Supplementary Fig. S13a). Region III of IES2 binds RuvBL2. Both regions essentially bind as previously established [29, 30]. While the moderate spin rotation between N-6 and N-7 still allow for IES2 to bind the distal acidic patch, the spin rotation on the hexasome and resulting position of the throttle helix relative to the acidic patch is not compatible with acidic patch binding of IES2, explaining why acidic patch mutants affect nucleosome but not hexasome sliding.
In summary, we can conclude that the IES2 acidic patch anchor RxR is critical to convert ATPase activity into sliding of nucleosomes, while it does not appear to play a role in mobilizinghexasomes.
Overall structure of the INO80 C- and A-module bound to nucleosome/hexasome
To define the overall architecture of the C- and A-module, we determined the structure of INO80 in the absence of nucleotide, as previously done for fungal INO80 under nucleotide-free conditions [28]. Two-dimensional classification revealed particles containing the INO80 C-module along with density corresponding to the A-module. To improve resolution, these particles were re-extracted using a box size of 800 pixels (see the ‘Materials and methods’ section).
Applying the same strategy to the other datasets (INO80-hexasome, INO80-nucleosome, and INO80-H2A.Z) revealed some 2D classes with A-module density in the INO80–hexasome and INO80–H2A.Z dataset (see the ‘Materials and methods’ section and Supplementary Figs S5 and S7*)*. However, no additional A-module density was detected in the INO80–nucleosome complex. To further improve the reconstruction, we performed 3D flexible refinement in cryoSPARC [46], which yielded an improved map for the A-module and enabled fitting of the AlphaFold3 [47] predicted A-module structure alongside the observed C-module.
In the hexasome dataset, the A-module binds to unwrapped nucleosomal DNA (Fig. 5a). In contrast, in the nucleosome dataset, it binds to extranucleosomal DNA (Fig. 5b). Thus, although engaging different DNA sequences, the A-module consistently targets topologically equivalent regions in the H-3 and N-6 states, recognizing extranucleosomal or extrahexasomal DNA, respectively.
Overall structure of the INO80 C- and A-module bound to nucleosome/hexasome. Maps generated by 3D flexible refinement [45] showing the INO80 C-module combined with the AlphaFold-predicted A-module structure fitted into the density for (a) INO80apo–nucleosome (C- and A-module) and (b) INO80–hexasome (C- and A-module). Map density for the post-HSA region is visible in the INO80apo–nucleosome map but absent in the INO80–hexasome reconstruction. (c) Representative 2D classes used for reconstruction of the C- and A-module complex, illustrating A-module binding to both the entry and exit sides of extranucleosomal DNA. (d) Map from 3D flexible refinement of the C- and A-module complex highlighting density at the entry and exit sites of extranucleosomal DNA. Post-HSA map density is not visible.
Importantly, in the nucleotide-free INO80 structures, we observed clear density for the post-HSA domain (Fig. 5b). This density is consistently absent in structures determined in the presence of ADP–BeF_3_ (Fig. 5a and d), suggesting that the post-HSA domain is a dynamic feature in INO80, whose conformation may be coupled to the nucleotide state of the Ino80^motor^, consistent with previous observations in fungal INO80 [28].
Notably, in the H2A.Z nucleosome dataset (with 50 bp flanking DNA on both sides, 50N50), the A-module appears to engage both the entry and exit DNA (Fig. 5c and d). A prior study [27] demonstrated that INO80 can sense both entry and exit DNA to position nucleosomes, a property that likely contributes to its ability to centrally position nucleosomes.
Discussion
Chromatin remodellers utilize ATP hydrolysis to alter nucleosome structure, composition, or position and regulate DNA accessibility. These enzymes are broadly classified into four main families: SWI/SNF, INO80, ISWI, and CHD. While SWI/SNF [58], ISWI [31], and CHD remodellers [59] typically bind SHL−2 on the nucleosome, INO80 exhibits a distinct binding mode, as biochemical and structural data have demonstrated from three different species that it binds at SHL−6/−7 on intact nucleosomes [20, 29, 30]—a position notably different from other remodellers
Recent work on yeast and fungal INO80 showed that they can mobilize not only nucleosomes but also hexasomes [20, 21]. The occurrence and role of hexasomes during various DNA-associated processes is a developing field [24, 60]. Hexasomes have been reported to be generated by RNA polymerase II during transcription and can either stall or accelerate elongation depending on whether the promoter-proximal or promoter-distal H2A–H2B dimer is lost [23, 61]. Moreover, loss of one H2A–H2B dimer creates an asymmetrically accessible substrate for DNA-binding factors and histone-modifying enzymes [62, 63]. Beyond transcription, hexasomes containing variants such as H2A.Z and H3.3 have been shown to facilitate access of base-excision repair glycosylases to sites of DNA damage, underscoring their versatile, emerging roles in genome maintenance [64].
Given the emerging role of hexasomes as potential substrates but also possible intermediates prompted us to investigate whether human INO80, like its fungal and yeast orthologs, can act on such subnucleosomal species. This is an important question since human INO80 lacks an important part of the ARP5 insertion domain that binds to H2A/H2B acidic patch on the nucleosome, or H3/H4 in the hexasome in case of yeast and fungal INO80 [21, 29]. Nevertheless, we find that human INO80 efficiently slides hexasomes. Compared to the nucleosome, the hexasome spin rotates relative to human INO80, resulting in the engagement of Ino80^motor^ at the SHL−2/−3 position. Human INO80 does not, per se, slide hexasomes more efficiently than nucleosomes; when sufficient linker DNA is present for A-module binding, INO80 slides both substrates similarly. This suggests that, for human INO80, hexasomes are not inherently preferred substrate over nucleosomes, but may become favourable if not preferred substrate, if sufficiently long linker DNA is exposed as a result of DNA unwrapping providing a binding site for the A-module.
Interestingly, we also observed that INO80 can spin-rotate on the nucleosome, with the Ino80^motor^ bound to either SHL−6 or SHL−7. Comparing both binding modes, the main difference is a partial unwrapping of nucleosomal DNA in the SHL−6 bound state of the ATPase (N-6). Thus, whether bound to nucleosomes or hexasomes, the Ino80^motor^ consistently engages DNA at the entry side of the nucleosomal/hexasomal DNA. This binding behaviour appears to be remarkably independent of nucleotide presence and nucleosome type. The DNA entry side of the nucleosome is characterized by increased dynamics and accessibility, reduced DNA curvature, and fewer DNA–histone interactions compared to the tightly wrapped core DNA [53, 65–67]. This suggests that human INO80 integrates two structural features, DNA curvature around the nucleosome core (ARP5/IES6) and emerging straight DNA at the entry DNA (Ino80^motor^). The relatively modest differences between N-6 and N-7 suggest that such states could be dynamically sampled in chromatin, possibly coinciding with nucleosomal unwrapping/wrapping. Since we visualized three spin-rotated binding modes of INO80 from the same species, we can rule out species-specific differences and conclude that the human INO80 (C-module) complex binds nucleosomes in a topology-driven manner. The topology-driven interaction may equip INO80 with the possibility to act on a range of nucleosomes and also enable regulation through nucleosome and linker DNA features that impact entry DNA paths and dynamics. In contrast, remodellers that bind nucleosomes at SHL−2, such as SWI/SNF, ISWI, SWR1, and CHD [5, 6] engage DNA that is already bent by histone interactions, making them less sensitive to dynamic DNA shape features. This difference highlights INO80’s unique adaptation to exploit nucleosomal DNA flexibility for its specialized roles.
We observed that RuvBL1/2 interacts with the proximal side of histone H2A and interacts with the phosphate backbone of the unwrapped nucleosomal or hexasomal DNA at this position. In Ct INO80, on the contrary, interactions between Rvb1/2 and the histone core are not observed (Supplementary Fig. S14a). Possibly, RuvBL1/2 might help compensate to some degree the lack of the histone binding Arp5 foot in fungal or yeast INO80. The IES2 subunit is a conserved component of INO80 remodellers, with the C-terminal part providing a wrapper that connects RuvBL1/2, the Ino80^motor^, and the nucleosome gyres [20, 21, 29, 30]. The less well-understood N-terminal part of human IES2 contains an RxR motif that binds the acidic patch on the distal side of the nucleosome along with a mostly unstructured region towards the N-terminus. Our data suggest differential binding of IES2 to different nucleosomal species. Acidic patch binding by IES2 is observed on both nucleosomal binding modes (N-6, N-7), but we did not observe this interaction when bound to hexasomes (H-3). In the case of the H2A.Z 50N50 nucleosome (NZ-7), we find residual density corresponding to the RxR motif (Supplementary Fig. S15a). This suggests intact but probably weaker binding of IES2 to the H2A.Z acidic patch compared to that of the H2A containing nucleosomes (Supplementary Fig. S15b).
The lack of hexasome interactions can be explained by the ~145° spin rotation of INO80 compared to nucleosomes, strongly ‘misaligning’ the acid patch orientation with the location of IES2’s throttle helix. The linker between the RxR motif and the throttle is likely too short to compensate the rotation, leading to loss of acidic patch binding. Our biochemical and structural results, suggesting a moderate role of IES2 acidic patch binding on nucleosome sliding but not on hexasome sliding is consistent with our structural work, and also with prior work using yeast INO80 and asymmetric nucleosomes [22]. There, it was shown that the proximal acidic patch is critical for remodelling activity, whereas mutation of the distal acidic patch had minimal effect. Since human INO80 does not have the proximal acidic patch binding Arp5 grappler foot of fungal INO80, IES2 is the currently the only identified acidic patch binding element of human ∆N INO80 complex. Thus, the ‘symmetric’ APM nucleosomes in our study are to some extend comparable to ‘asymmetric’ nucleosomes for yeast INO80 [22]. In the absence of a grappler foot, IES2’s distal acidic patch binding may also have a comparatively more important role in regulating sliding than in species where the Arp5 foot is present. Nevertheless, the acidic patch mutations affect nucleosome sliding to a greater degree that IES2 RXR mutations, which also displayed increased ATP hydrolysis rates. Perhaps autoinhibitory states [24] are affected by IES2 mutations, or APM more broadly affects sliding than through IES2 contacts, e.g. by altering nucleosome stability.
We find density for the N-terminal part of IES2 in place of unwrapped nucleosomal DNA at the exit side of the nucleosome. IES2 is the paralog to SWR complex protein 2 (Swc2) in SWR1 remodellers and shares throttle helix and RuvBL1/2 binding element [7, 57]. Like IES2, Swc2 also binds the distal side acidic patch even though SWR1 binds nucleosomes very differently from INO80 (SHL−2) [68]. Interestingly, yeast Swc2 has also been shown to partially unwrap DNA from the exit side of the nucleosome [57]. It also plays an important role in SWR1’s ability to swap sides on the nucleosome and has H2A.Z histone chaperone functions [69]. Thus, its functional equivalence to Swc2 as histone binding element extends even beyond the acidic patch binding. Our results suggest that IES2 might either help unwrap or stabilize unwrapped exit DNA, with a preference of H2A over H2A.Z. Such a role could prevent H2A/H2B dimer loss during remodelling. Alternatively, IES2 could prepare the exit side of the nucleosome to become the new entry side, if INO80 also has SWR1’s capability to swap sides on a nucleosome [70].
Using the symmetric centre positioned H2A.Z nucleosome (50N50), we find that the A-module might interact simultaneously with nucleosomal entry and exit DNA sides (Fig. 5c and d). Previous research demonstrated that CHD1 can also sense both the entry and exit sides of DNA [71]. This dual sensitivity allows for faster and more precise responses, which is critical for generating regularly spaced nucleosomes. In the case of INO80, it may function in conjunction with complex autoregulatory mechanisms [72] to sense proper linker DNA lengths in regulating sliding and spacing, but details need to be clarified in future work.
Altogether, the differential interactions of IES2 with H2A.Z– and H2A–nucleosomal species may have physiological consequences in regulating H2A.Z and H2A containing chromatin [19], consistent with a role of Swc2 in SWR1. H2A.Z contains three different residue changes compared to canonical H2A within 5Å from IES2 (S14b), which might explain the reduced binding [36] (Supplementary Fig. S14a). However, the use of centrally positioned 50N50 nucleosome rather than H2A.Z could lead to dual entry/exit DNA binding of the A-module and could impact on IES2 binding via exit DNA stability. In principle, the proximal acidic patch binding by IES2 and histone contacts at unwrapped exit DNA could enable INO80 to differentiate between different nucleosomal species or help retain the proximal H2A/H2B dimer during the sliding reaction in canonical nucleosomes. However, a functional interplay of exit DNA unwrapping and IES2 nucleosome binding, possibly relevant in exchange reactions, needs to be studied in the future. Likewise, posttranslational modifications could be interesting in the context of IES2’s role in recognizing different nucleosomal species in future work. For instance, H2BK120ub has recently been shown to occlude the acidic patch [73]. Since this modification is a hallmark of transcribed chromatin, it could fully allow hexasome sliding but reduce nucleosome sliding by human INO80.
In conclusion, our work has illuminated hexasome sliding as a universal capability of INO80. We have also provided high-resolution structures of the INO80 C-module in two distinct states (N-7 and N-6), altogether arguing for a predominantly topological recognition of nucleosomes driven by the degree of entry DNA unwrapping. Exit DNA interactions of the A-module and binding of histones at unwrapped exit DNA by IES2 suggests that INO80’s spectrum of nucleosome recognition and potential remodelling intermediates is far more complex than anticipated from previous work. Studies on more complex nucleosomal substrates as well as different nucleotide analogues beyond ADP–BeF_3_^−^ such as ATPγS, AMP–PNP, or ADP–AlF_4_^−^ will be valuable to capture additional conformational states of the ATPase motor and to further elucidate the dynamic mechanism of INO80 within the chromatin context.
Supplementary Material
gkag138_Supplemental_Files
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Yuan GC, Liu YJ, Dion MF et al. Genome-scale identification of nucleosome positions in S. cerevisiae. Science. 2005;309:626–30. 10.1126/science.1112178.15961632 · doi ↗ · pubmed ↗
- 2Arents G, Burlingame RW, Wangt B et al. The nucleosomal core histone octamer at 3.1 Å resolution : a tripartite protein assembly and a left-handed superhelix. 1991;88:10148–52. 10.1073/pnas.88.22.10148.PMC 528851946434 · doi ↗ · pubmed ↗
- 3Luger K, Mäder AW, Richmond RK et al. Crystal structure of the nucleosome core particle at 2.8 Å resolution. Nature. 1997;389:251–60. 10.1038/38444.9305837 · doi ↗ · pubmed ↗
- 4Lai WKM, Pugh BF. Understanding nucleosome dynamics and their links to gene expression and DNA replication. Nat Rev Mol Cell Biol. 2017;18:548–62. 10.1038/nrm.2017.47.28537572 PMC 5831138 · doi ↗ · pubmed ↗
- 5Clapier CR, Iwasa J, Cairns BR et al. Mechanisms of action and regulation of ATP-dependent chromatin-remodelling complexes. Nat Rev Mol Cell Biol. 18:407–22. 10.1038/nrm.2017.26.PMC 812795328512350 · doi ↗ · pubmed ↗
- 6Eustermann S, Patel AB, Hopfner K et al. Energy-driven genome regulation by ATP-dependent chromatin remodellers. Nat Rev Mol Cell Biol. 2024;25:309–32. 10.1038/s 41580-023-00683-y.38081975 PMC 12303036 · doi ↗ · pubmed ↗
- 7Willhoft O, Wigley DB. INO 80 and SWR 1 complexes : the non-identical twins of chromatin remodelling. Curr Opin Struct Biol. 2020; 2:50–8. 10.1016/j.sbi.2019.09.002.PMC 717146931838293 · doi ↗ · pubmed ↗
- 8Yen K, Vinayachandran V, Pugh BF. SWR-C and INO 80 chromatin remodelers recognize nucleosome-free regions near +1 nucleosomes. Cell. 2013;154:1246–56. 10.1016/j.cell.2013.08.043.24034248 PMC 4090706 · doi ↗ · pubmed ↗