Sequence-dependent binding modes of INO80 control +1 nucleosome positioning

Abstract

Cellular self-organisation counteracts entropy at the expenditure of energy. In case of the first level of nuclear DNA organisation, this relates to regular nucleosome arrays and interspaced nucleosome-depleted regions (NDRs), for example at promoters or replication origins^1–5. The organisation of nucleosomes as building blocks of chromatin is orchestrated by the collective activities of ATP-dependent chromatin remodellers^6–9. Yet, how remodellers achieve positional specificity, in particular regarding the promoter-proximal +1 nucleosomes, remains unclear. Here, we show that the S. cerevisiae chromatin remodeller INO80 unexpectedly distinguishes DNA sequence asymmetry within the +1 nucleosome of the SWH1 gene through distinct inhibited and active nucleosome-binding modes. In structural and biochemical analyses of INO80 on nucleosomes with the endogenous sequence, we identified an inhibited binding mode where the entire INO80 remodelling unit flipped on the +1 nucleosome. INO80 adopted this remodelling-incompetent binding mode when facing the promoter, but a remodelling-competent mode when facing the gene body. This directional read-out of intra-nucleosomal DNA sequence asymmetry, together with extranucleosomal NDR sequence features, prevented nucleosome sliding into the NDR while permitting array formation over the gene. Our work shows how DNA features contribute to ATP-dependent self-organisation of promoter chromatin by INO80.

Eukaryotic nuclear DNA exists in a highly organised nucleoprotein complex termed chromatin. The basic building block of chromatin is the nucleosome, ~147 base pairs (bp) of DNA wrapped around histone protein octamers. A principle of chromatin organisation is the presence of nucleosome-depleted regions (NDRs) flanked by arrays of regularly spaced nucleosomes. The arrays are typically found across gene bodies, whereas NDRs demarcate regulatory regions such as promoters, enhancers and replication origins^1–5.

The nucleosomes flanking NDRs are designated +1 in the downstream and −1 in the upstream direction of transcription, and subsequent array nucleosomes are denoted +2, +3, … or −2, −3, …, respectively. In Saccharomyces cerevisiae (S. cerevisiae or Sc), +1 nucleosomes often occupy well-defined positions. They co-define the transcription start site (TSS), play key roles in transcription initiation and origin firing^10,11 and “phase” adjacent nucleosomal arrays^4,12–14. Such phased arrays repress cryptic transcription^15–18 and assist origin firing¹⁹.

The organisation of nucleosomes around NDRs and in genic arrays is a prime example of energy-driven cellular self-organisation and orchestrated, among others, by the collective activities of ATP-dependent chromatin remodellers^6–8,20,21. Remodellers couple ATP hydrolysis to nucleosome sliding, histone eviction and histone variant exchange. Remodeller-intrinsic mechanisms together with regulatory cues shape the nucleosomal landscape⁷. Notably, in some cases, just the DNA sequence alone may direct a remodeller to generate defined nucleosome positions.

The ScINO80 complex²² is the leading example for this sequence feature-linked mechanism as it can position +1 and −1 nucleosomes on its own across the yeast genome in vitro^6,23. INO80 is organised into a modular structure with a nucleosome-sliding C- and regulatory N- and A-modules^24–28 (Fig. 1A). Like its structural relative SWR1, INO80 is activated by the presence of long extranucleosomal DNA (>30 bp^29,30), which is bound by INO80’s A- and N-modules^31–35. However, the activation of INO80 by extranucleosomal DNA generates a mechanistic paradox regarding INO80’s +1 nucleosome positioning activity. NDRs are sufficiently long to accommodate INO80’s regulatory modules, and their often dA/dT-rich nature should in particular allow A-module binding^3,36. Thus, INO80 should slide nucleosomes into the NDR, rather than stop and position them at the boundary of the NDR. A possible key to resolving this paradox may be that not only DNA length, but also DNA sequence features, both intra- and extranucleosomal, affect nucleosome positioning^6,23,37, like intranucleosomal DNA rigidity in case of remodelling by INO80³⁸ or Chd1³⁹. In this study, we use reconstituted mononucleosomes and nucleosomal arrays around the +1 nucleosome of the S. cerevisiae SWH1 gene to uncover the interaction of INO80 with nucleosomal arrays assembled on native promoter-flanking DNA sequences and reveal the mechanism of DNA sequence-guided +1 nucleosome positioning by INO80.

Fig. 1: INO80 adopts different nucleosome-binding modes on promoter-proximal nucleosomal arrays.

A, ScINO80 modular organisation. ScIno80^ΔN truncation leaving only amino acids 462-1489 are indicated. HSA, Helicase SANT-associated; PTH, post-HSA. B, Choice of the natural array sequences from the SWH1 gene with indicated EcoRV site. NDR, nucleosome-depleted region. C, ATP hydrolysis rates of INO80^ΔN on the arrays and a control Widom 601 mononucleosome with 80 bp overhang of synthetic DNA (n = 3). D, Quantification of the fraction of EcoRV-digested array DNA. In the restriction enzyme accessibility (REA) assay, arrays were incubated with EcoRV and INO80^ΔN in absence or presence of ATP (n = 3). E, Cryo-EM density maps (low-pass filtered to 10 Å) with fitted models of different INO80 binding modes (R-state, I-state and O-state) on the nucleosomal arrays with the representative cartoons. The difference in open and closed Ino80 ATPase binding is visualised in the frames. SHL 6, ATPase binding site in the R-state, is marked in pink. SHL, superhelical location; XL, crosslinked.

INO80 adopts different nucleosome-binding modes on promoter-proximal nucleosomal arrays

We chose endogenous sequences based on the S. cerevisiae SWH1 gene, since synthetic Widom 601 sequences, used in most structural studies, do not lead to proper array formation in vivo^40,41. The SWH1 +1 nucleosome sequence combines strong intrinsic positioning capability by salt gradient dialysis (SGD) in vitro with being a bona fide well-positioned +1 nucleosome in vivo⁴². We tested three sequences (1-3) encompassing the +1 to +4 nucleosomes of SWH1 (Fig. 1B). Tetra-nucleosomal arrays assembled well on these sequences in SGD (Fig. S1A). The nucleosomal arrays entailed approximately 30 bp linker DNAs, which correspond to the minimal linker length for remodelling by INO80³⁰. All arrays induced robust ATPase activity in INO80^ΔN (lack of the N-module’s Nhp10 subunit had no effect on +1 positions in vivo⁴³, Fig. 1C, D). Array 1 showed modest remodelling, as expected from the short linkers and our restriction enzyme accessibility assay may not efficiently pick up local remodelling due to the natural cleavage site occurring within the +3 nucleosome. Array 2, with a linker extending into the gene body, showed moderately increased remodelling, probably caused by the stimulatory effect of additional extranucleosomal DNA on nucleosome sliding³². Array 3, extending into the promoter, showed similar remodelling as array 1, suggesting a lack of sliding toward the NDR (Fig. 1D, Fig. S1B). In summary, all sequences form homogeneous arrays and robustly stimulate INO80’s ATPase, with modest remodelling aimed at helping us to identify both sliding-poised and sliding-inhibited conformations.

We used array 1 for interrogating the interaction of INO80 bound to promoter-proximal chromatin structures. To reduce nucleosomal stacking, we introduced the H4K16Q mutation that mimics the “open” chromatin achieved with H4K16ac^44,45. INO80^ΔN slid a mononucleosome with H4K16Q as efficiently as with wild type H4 (Fig. S1C), consistent with prior observations that H4 tails do not per se affect INO80 activity³⁰. We recorded two cryo-EM datasets of INO80^ΔN bound to array 1 in the presence of ATP and BeF_x or AlF_x. In the ATP+BeF_x dataset, we observed the previously obtained canonical nucleosome-binding mode, with Ino80’s ATPase “motor” engaged with unwrapped entry DNA at SHL 6 and the Arp5/Ies6 “grip” bound to SHL 3, poised to pump DNA into the nucleosome core particle (NCP)^24,26. The A-module was either not visible in this mode or was folded back in recently found autoinhibited conformation³³, consistent with the effect of short linkers. We denote these modes as R-state (“remodelling”, extended A-module) and I-state (“inhibited”, folded back A-module) (Fig. 1E, Fig. S2).

Intriguingly, in the ATP+AlF_x dataset, we observed along R- and I-states a new nucleosome-binding mode that we called the O-state (“off-state”). The O-state was less stable, as its presence could be increased by mild glutaraldehyde crosslinking (Fig. S3). It strongly deviated from all other conformations observed for INO80 in that the Ino80 ATPase was not placed at SHL 6, but at SHL 1, and Arp5/Ies6 was bound to SHL 5 instead of SHL 3. This binding mode is different from the SHL 2-bound conformations of other remodellers⁷, and notably, the two lobes of the Ino80 ATPase were open (Fig. 1E). This indicates a defunct ATPase, even in presence of ATP+AlF_x.

These results established that the arrays with endogenous sequence captured a range of different binding modes of INO80: either the Ino80 ATPase was bound at SHL 6 and poised for sliding (R- and I-states), or at SHL 1 and incapable of ATP hydrolysis and sliding (new O-state).

Promoter-facing INO80 binds the SWH1 +1 nucleosome in the off-state

In our structural analysis of arrays, we could not distinguish to which of the four nucleosomes INO80^ΔN was bound in which conformations. However, given INO80’s limited remodelling of arrays towards the NDR and the long-standing question of how INO80 determines +1 nucleosome positioning (Fig. 1B–D), we tested whether the O-state could be arising on the +1 nucleosome and lead to setting this position. We assembled a mononucleosome on DNA comprising the SWH1 +1 nucleosome and 80 bp upstream promoter sequence (denoted 80S0, Fig. 2A). This sequence assembled a homogeneous nucleosome as judged by native gel electrophoresis (Fig. 3C, 80S0 -ATP lane). The cryo-EM structure of 80S0 in complex with INO80^ΔN in presence of ATP+AlF_x (or BeF_x with similar results) revealed that most, if not all, nucleosome-bound INO80 was in the O-state (even without crosslinking), although the 80 bp extranucleosomal promoter would allow INO80 to adopt a promoter-facing R-state. Again, the Ino80 ATPase was in an open conformation, with N- and C-lobes widely separated even in presence of ATP/ADP+AlF_x/BeF_x (Fig. 2A). Only the N-lobe was bound at SHL 1, while the C-lobe remained distant through its tight interaction with the Rvb1/2 subcomplex.

Fig. 2: INO80 distinguishes SWH1 +1 nucleosome asymmetry through different binding modes.

A, Cryo-EM density map (low-pass filtered to 10 Å) with fitted model. Promoter-facing INO80 binds the SWH1 +1 80S0 nucleosome in the O-state. SHL 6 is marked in pink. B, Cryo-EM density map (low-pass filtered to 10 Å) with fitted model. Gene body-facing INO80 binds the SWH1 +1 80S80 nucleosome in the R-state. C, Nucleosomal sequence mapping pipeline (see Suppl. Method). D, Cryo-EM density map of a free 80S80 nucleosome. E, Z-scores of 80S80 forward strand correlated with both nucleic acid chains (I and J) in the cryo-EM map of the free 80S80 nucleosome. F, Cryo-EM density focused map of an INO80-bound 80S80 nucleosome. G, Z-scores of 80S80 forward strand correlated with both nucleic acid chains (K and L) in the cryo-EM map of the INO80 bound 80S80 nucleosome.

Fig. 3: INO80 slides the highly asymmetric SWH1 +1 nucleosome away from, but not into, the promoter.

A, Model of an INO80-bound 80S80 nucleosome with mapped DNA sequence. dA, dT, dG, and dC nucleotides are coloured red, blue, green, and yellow, respectively. B, Architecture of nucleosomal substrates assembled on the following sequences: natural SWH1, reverse complement SWH1 +1 nucleosome, and Widom 601 replacing SWH1 +1 nucleosome, with extranucleosomal DNA reaching either into the promoter or gene body sequences. 50P and 50G demarcate the DNA sequences used for measuring A-module binding (see Fig. S8A). C, Representative sliding assay native gels for the above-mentioned nucleosomal substrates (see B). D, Remodelling progress of each nucleosomal substrate. In each panel, the upper dashed line fit shows the total remodelling products (all but end-positioned nucleosomes, marked with *), the lower solid line fit shows the emergence of the centrally positioned nucleosomes (marked with **). E, ATP hydrolysis rates of the nucleosomal substrates as in panel B. F, Composite profiles for all in vivo +1 nucleosomes (black trace) and for all nucleosomes positioned by INO80 in vitro (orange trace) show a similar propeller twist gradient. Vertical dashed line marks the dyad, vertical solid lines mark the nucleosome borders (+/− 73 bp from the dyad).

In summary, the SWH1 +1 nucleosome sequence strongly suppressed an “into the NDR”-sliding R-state in favour of a new O-state with a defunct Ino80 ATPase.

INO80 distinguishes SWH1 +1 nucleosome asymmetry through different binding modes

To test whether the SWH1 +1 nucleosome prevents formation of active INO80 states per se, or only in the direction of the promoter, we tested an 80S80 nucleosome, encompassing the SWH1 +1 nucleosome sequence and 80 bp flanking DNA from both the promoter and gene body sides. Here, the INO80 A-module could bind either of the extranucleosomal DNAs. Structure determination revealed that 80S80-bound full-length INO80 (INO80^fl) was now in the canonical active R-state with N- and C-lobes properly closed around SHL 6 and Arp5/Ies6 bound to SHL 3 (Fig. 2B).

These data indicated that the SWH1 +1 nucleosome does not intrinsically prevent formation of an active sliding state of INO80 but may do so specifically in the promoter-facing orientation. To test this, we sought to determine whether the R-state on 80S80 is specifically arising in the gene-body facing orientation. We developed an approach to score DNA sequences in cryo-EM maps based on map-model correlations in a sliding-window approach (Suppl. Method, Fig. 2C). Analysis of a free 80S80 nucleosome (Fig. 2D) revealed two high Z-score matches corresponding to the same sequence in opposite orientations (Fig. 2E). This is due to a cryo-EM processing drawback: the two pseudo-symmetric nucleosome orientations often cannot be distinguished, which results in an averaged map with both orientations superimposed. Importantly, our approach identified the previously mapped SWH1 +1 position⁴² with single-base resolution and could even deconvolute two superimposed DNA sequences.

To reveal whether INO80 bound to 80S80 facing the promoter, the gene-body, or either, we derived a 2.75 Å map of the INO80-bound 80S80 nucleosome using focused refinement (Fig. 2F). In this case, we got only a single high Z-score corresponding to the gene-body facing R-state (Fig. 2G). Altogether, these data revealed that the binding mode of INO80 on the SWH1 +1 nucleosome was strikingly dependent on the nucleosome orientation, with promoter-facing orientations promoting an O-state and gene body-facing orientations promoting an R-state. Furthermore, of the two possible binding conformations, the gene body-facing R-state predominated, which concurs with the O-state being less stable.

INO80 slides the highly asymmetric SWH1 +1 nucleosome away from, but not into, the promoter

To validate the structural observations, we performed sliding assays with the SWH1 +1 nucleosome, with sliding directionality enforced via the presence of extranucleosomal DNA only on the promoter or the gene body side (80S0 versus 0S80, Fig. 3B). INO80^ΔN could robustly slide 0S80 into the gene body sequence (Fig. 3C, D) but not 80S0 into the promoter sequence (Fig. 3C, D). Furthermore, the ATPase rate was high on 0S80, similar as on a Widom 601 0N80 mononucleosome (Fig. 1C, W601 0N80), but low on 80S0 (Fig. 3E). Altogether, these data are consistent with the structural findings that INO80 binds 0S80 in a sliding-competent R-state, while 80S0 enforces a sliding- and ATP hydrolysis-inhibited O-state.

To dissect the roles of nucleosome orientation versus flanking extranucleosomal DNA, we analysed INO80^ΔN sliding and ATP hydrolysis on nucleosome variants comprising modified intranucleosomal DNA segments (Fig. 3B). We inverted the SWH1 +1 nucleosome sequence with respect to extranucleosomal promoter and gene body DNAs (S_inv) and also swapped it with a Widom 601 sequence (W). Intriguingly, 80S_inv0 showed reasonable remodelling into the promoter region, in contrast to 80S0 (Fig. 3C, D), indicating that intranucleosomal DNA features of SWH1 +1 nucleosome strongly dictate the INO80 remodelling directionality. It is of note that 80S_inv0 displayed a higher amount of intermediate nucleosome species (Fig. 3C). Such longer-lived intermediates could be the result of the natural A-tracts in SWH1 promoter DNA consistent with sliding experiments on engineered extranucleosomal A-tracts³⁶ (Fig. 3D). 0S_inv80 displayed substantially reduced but detectable remodelling, while 80W0 even showed arrested sliding and low ATP hydrolysis rate. Thus, gene body extranucleosomal DNA can, to a low amount, rescue the repressed upstream sliding of SWH1 +1, while SWH1 promoter DNA strongly inhibits sliding of a Widom 601 nucleosome (Fig. 3C, D). Thus, overall sliding capability results from both extra- and intranucleosomal features, with a possible context-dependent domination of one or the other.

Our data show that the SWH1 +1 nucleosome is highly asymmetric with respect to INO80 binding and sliding competence. A potential intranucleosomal feature promoting this striking asymmetry is an A-tract at the NDR-facing SHL 6/7 (Fig. 3A). A-tracts feature prominently in DNA shape prediction plots, e.g., as low propeller twist, and their increased rigidity can impair remodelling speed⁴². In the case of SWH1 +1 nucleosome, the A-tract at SHL 6/7 might prevent the Ino80 ATPase from gripping DNA to adopt a stable R-state. Consistently, we found an enrichment of long A-tracts on the NDR-facing side compared with the gene body-facing side in +1 nucleosomes in S. cerevisiae (Fig. S4). More generally, a skewed asymmetry with respect to propeller twist is not only found for the sequence of the SWH1 +1 nucleosome but typical for all in vivo +1 nucleosome sequences and for the sequences in all nucleosomes that are positioned by INO80 on its own in a genome-wide nucleosome positioning assay (Figs. 3F, S5–7).

INO80 integrates nucleosomal and NDR sequence features into sliding arrest

The strong intranucleosomal contribution to the asymmetry of SWH1 +1 nucleosome sliding by INO80 raised the question of which role the A-module plays in the promoter-facing O-state. Binding of >30-40 bp linker DNA by the A-module is a prerequisite for robust INO80 activation^30,32,33,46. Thus, we first tested whether A-module can bind the SWH1 NDR DNA and whether this potential extranucleosomal DNA binding can occur in the O-state. Promoter DNA showed a 3-fold reduced binding compared to gene body DNA (Fig. S8A, Table S1). The reduced but still substantial affinity may contribute to the moderate inhibitory effect in promoting sliding of SWH1 +1 nucleosome to the central position (0S80 vs 80S_inv0 in Fig. 3C, D), but may not stop sliding into the NDR simply by dissociation of the A-module. Likewise, robust sliding of 80S_inv0 suggests that the A-module can bind the NDR at least in the R-state when sliding a nucleosome from the gene body towards the promoter.

In prior work, we investigated various Widom 601-derived sequences to study the effect of intranucleosomal DNA rigidity on sliding by INO80³⁸. One of the studied sequences (denoted M2_rev) assembled into multiple nucleosomal species with different extranucleosomal DNA lengths. Cryo-EM analysis of this population in complex with INO80^fl uncovered an intriguing compatibility between upstream extranucleosomal DNA A-module binding and the O-state. In the M2_rev dataset, we find the R-state (not shown) as major species along with the O-state as a minor species (Fig. 4A). Notably, the nucleosome of the O-state showed ~20–25 bp of extranucleosomal DNA on one side and ~50–55 bp of entry DNA on the other side potentially stabilising the overall complex including the A-module (Fig. S8B).

Fig. 4: INO80 integrates nucleosomal and NDR sequence features into sliding arrest.

A, Cryo-EM density map (EMD-45377) with fitted model of R-state with an open ATPase (model is based on the AlphaFold 3 predictions and rigid-body fitted to the volume) and cryo-EM density map (low-pass filtered to 10 Å) with fitted model of the INO80 complex bound to the M2_rev nucleosome in O-state. Comparing R- to O-state shows that they are topologically related through a flip of the C-module on the nucleosome. The A-module can remain attached on extranucleosomal DNA, with the N-lobe of the Ino80 ATPase as pivot point. B, Depiction of N-lobe pivoting from R- to O-state. N-lobe location in R-state is marked yellow. C, Cartoon depiction of the interplay of all observed binding modes. Upon arrival at the +1 nucleosome site, intrinsic entry promoter DNA features (for example, A-tracts; marked dark orange) at the +1 site destabilize Ino80 ATPase engagement at SHL 6. This destabilization promotes flipping of the C-module into an inhibited O-state, while the A-module may remain attached to extranucleosomal DNA. Upon encountering the neighbouring nucleosome, shortening linker DNA sensing causes A-module flip and formation of an autoinhibited I-state, also restricting movement in an alternative way.

Comparison of the R- and O-states with A-modules engaged with extranucleosomal DNA now shows they are topologically related by a flip of the C-module across the nucleosome, rather than by differential binding of the A-module to extranucleosomal DNA. The N-lobe could act as the pivot point and change orientation between the nucleosomal gyres (Fig. 4 A, B; Movie S1). The C-module flip is facilitated by two alternative connections of the A-module to the N-lobe regulatory site via the post-HSA helix^33,47,48, enabling the A-module to remain bound to extranucleosomal DNA in both R- and O-state. The compatibility with the O-state resolves the paradox of how A-module interaction with the NDR does not automatically lead to INO80 sliding the +1 nucleosome into the NDR.

Discussion

Our results suggest a mechanism which explains how INO80 can position a +1 nucleosome by merely reading the DNA sequence and without the involvement of DNA-bound barrier factors. INO80 may stochastically slide a nucleosome from the gene body toward the unoccupied +1 nucleosome site, with the A-module engaging DNA ahead of the sliding direction. Upon arrival at the promoter, the R-state could be destabilized by DNA features, causing the INO80 C-module to flip into the sliding-incapable O-state and thereby arrest further upstream sliding of the +1 nucleosome. Our results are consistent with previous findings showing that, in vivo, in the absence of INO80, +1 nucleosome occupancy was observed further downstream in the gene body^20,49,50.

Given sufficient intragenic linker DNA, the O-state may also serve as an intermediate toward a gene body-facing R-state. Such a reversal is plausible given INO80’s spin-rotating capacity^51,52 and evidence that remodellers can swap nucleosome sides without fully detaching^53,54. The combination of R-, O-, and I-states³³ is well suited to explaining the formation of phased arrays, starting with the DNA-encoded kinetic trap at +1 and followed by subsequent spacing of +2, +3, and downstream nucleosomes (Fig. S8C). Here, the O- and I-states play distinct roles. The O-state could form in response to extra-/intranucleosomal DNA features, whereas the I-state could be induced in response to extranucleosomal DNA length, in particular sensing linker DNA shortening. A role of the A-module in setting the linker DNA length is consistent with altered linker lengths upon HSA helix mutations in the A-module²⁹.

We and others previously noted that remodelling by INO80 depends on DNA sequence features, such as DNA shape, electrostatic potential, and rigidity^{6,23,36,37,55}. This is also reflected, for example, in the composite propeller twist profile of all nucleosomes positioned by INO80 in vitro across the genome, which becomes asymmetric if sequences are sorted according to their shape profiles (Figs. 3F, S5–6). Intriguingly, the asymmetric propeller twist gradient in this profile is similar to that in the composite profile of all in vivo +1 nucleosome sequences sorted according to the direction of transcription²³ (Fig. 3F), i.e., by a shape-independent sorting. Therefore, this similarity reflects the co-evolution of gene-oriented +1 nucleosome sequences and INO80’s sequence-dependent sliding directionality, so that INO80 does not slide nucleosomes into the promoter NDR but can generate +1 nucleosome positions, even on its own in vitro⁶. Our results now provide a mechanistic framework in which asymmetric nucleosomal DNA features can lead to orientation-dependent binding modes in R- and O-state, thereby driving directional sliding.

The S. cerevisiae SWH1 +1 nucleosome enabled the discovery of this general mechanism, even though SWH1 may represent an extreme case as a strict directional barrier for INO80, maybe because the propeller twist minimum is caused by an A-tract right at SHL 6/7 where the Ino80 ATPase needs to bind (Fig. S7). We found such sequence setup for some more in vivo +1 nucleosomes (Fig. S4), but skewed propeller twist is also generated by other sequences. More generally, nucleosomes positioned by INO80 in vitro or of other in vivo +1 nucleosomes (Fig. 3F) show a shallower slope of propeller twist and the propeller twist minimum is more upstream in the extranucleosomal linker/NDR. This may reflect that other sequences do not strictly stop sliding in one direction, but their similarly asymmetric DNA features still introduce a sliding direction bias, which amounts to a nucleosome positioning barrier²⁹. In these cases, epigenetic marks, such as H3 tail modifications, and histone variants, together with barrier proteins, could provide additional layers of regulation and assist stable +1 nucleosome positioning in vivo.

In summary, our results provide a mechanism by which DNA features can directly regulate the binding mode of a core remodeller unit, leading to directional sliding and +1 nucleosome positioning. This provides a general model for how nucleosome organisation, with accurately positioned and phased nucleosomal arrays at promoters, can emerge from the readout of genome sequence information by remodellers.

Materials and Methods

Expression and purification of S. cerevisiae INO80^fl, INO80^ΔN and INO80 A-module complexes from insect cells

Recombinant S. cerevisiae INO80^fl complex was generated as described previously⁵⁶, using two pFBDM vectors. One vector contained the coding sequences of the INO80 C-module subunits (C-terminally 2xFLAG tagged Ino80, Rvb1, Rvb2, Ies6, Arp5) and a second vector contained the remaining coding sequences for the subunits of the A- and N-modules (Actin, Arp4, Arp8, Taf14, Ies2, Ies4, Ies1, Ies3, Ies5, and Nhp10). INO80^ΔN, a complex in which truncation of the Ino80 N-terminus (amino acids 1-461) causes a loss of the Nhp10 module, was produced analogously using one vector containing the coding sequences of C-terminally 2xFLAG tagged Ino80(Δ1-461), Rvb1, Rvb2, Ies6, and Arp5 and a second vector containing the coding sequences for Actin, Arp4, Arp8, Taf14, Ies2, Ies4⁵⁷. INO80 A-module, a complex in which C- and N-modules are truncated, was produced analogously using one vector containing the coding sequence of C-terminally 2xFlag tagged Ino80(330-598), and a second vector containing the coding sequences for Actin, Arp4, Arp8, Taf14, Ies2, Ies4⁵⁷. Bacmids were generated using E. coli DH10 MultiBac cells. From each bacmid, baculoviruses were generated in Spodoptera frugiperda (Sf21) insect cells. Each baculovirus (1:200) was transferred to 1 l of Trichoplusia ni High Five culture for co-infection. Cells were cultured for 60 h at 27°C and harvested by centrifugation at 4°C.

For purification of the INO80^fl and INO80^ΔN complexes, cells were resuspended in lysis buffer containing 50 mM Tris HCl pH 8.0, 500 mM NaCl, 10% glycerol, 1 mM DTT, pepstatin A (0.28 μg/ml), PMSF (0.17 mg/ml), and benzamidine (0.33 mg/ml) and disrupted by sonication (3 × 1 min; duty cycle, 50%; and output control, 5 for INO80^fl and INO80^ΔN. 4 x 45 s; duty cycle, 50%; and output control, 5 for INO80 A-module). The lysate was cleared by centrifugation at 34,500×g for 45 min at 4°C. The supernatant was incubated with 3 ml of ANTI-FLAG M2 Affinity Gel for 1 h and washed with 50 ml of Wash 1 buffer (25 mM HEPES pH 8.0, 500 mM KCl, 10% glycerol, 0.05% IGEPAL CA630, 4 mM MgCl₂, and 0.25 mM DTT), 50 ml of Wash 2 buffer (25 mM HEPES pH 8.0, 200 mM KCl, 10% glycerol, 0.05% IGEPAL CA630, 4 mM MgCl₂, and 0.25 mM DTT), and 10 ml of buffer A (25 mM HEPES pH 8.0, 150 mM KCl, 2 mM MgCl₂, and 1 mM DTT). The protein was eluted from the matrix by incubation with 5 ml of buffer A supplemented with 0.2 mg/mL FLAG peptide in four incubation steps of 15 min each. The elution fractions were loaded onto a Mono Q 5/50 GL column and eluted by a linear salt gradient (150 mM KCl to 1 M KCl), resulting in highly pure INO80 complex. Aliquots were flash-frozen in liquid nitrogen in the presence of 20% of glycerol and stored at −80°C.

Mononucleosome and nucleosomal array preparation

H. sapiens (Hs) histone H4 mutant H4K16Q were cloned into pET-21b vector and X. laevis (Xl) histones (H2A, H2B, H3, and H4) were cloned into pET-3a vector. Histones were purified as previously described⁵⁸. In brief, histones were expressed in E. coli BL 21 (DE3) cells (Novagen) for 3 h at 37°C after induction at OD600 0.6 with 1 mM IPTG. Pelleted cells were disrupted by resuspension in wash buffer (50 mM Tris-HCl pH 8.0, 100 mM NaCl, 1 mM DTT) supplemented with 1 mg/ml lysozyme (Carl Roth, 8259), 1× protease inhibitor and 250 U benzonase (Merck Millipore, E1014) via three rounds of sonication. Inclusion bodies were washed two times with wash buffer supplemented with 1% Triton X-100 followed by two washes with standard wash buffer. Washed inclusion bodies were incubated with 1 ml DMSO for 30 min at room temperature followed by homogenization in resuspension buffer (7 M GdmCl, 20 mM sodium acetate pH 5.2, 1 mM EDTA) and 1 h incubation at room temperature. The supernatant was dialysed two times for 1.5 h against SAU 50 buffer (8 M urea; 20 mM sodium acetate, pH 5.2, 50 mM NaCl; 1 mM EDTA, 10 mM lysine). As first step of purification histones were separated by cation exchange chromatography (GE Healthcare HiTrap S HP) by applying a salt gradient. Histone-containing fractions were dialysed three times against refolding buffer (15 mM Tris-HCl, pH 8.0) of which one step needs to be over night to ensure proper refolding. Correct folded histones were finally separated by anion exchange chromatography (GE Healthcare HiTrap Q HP) using a salt gradient. Histone containing fractions were shock frozen in liquid nitrogen and lyophilized. WT HsHistones were purchased from The Histone Source, Colorado State University.

Histones were resuspended in unfolding buffer (7 M guanidinium chloride, 25 mM Tris-HCl pH 7.5, 1 mM DTT). HsH2A and HsH2B were mixed in a 1:1 ratio for HsH2A-H2B dimers; HsH3.3 and HsH4 were mixed in a 1:1 ratio for HsH3.3-H4 tetramers (both for WT and H4 variants), and XlH2A, XlH2B, XlH3 and XlH4 were mixed in a 1.2:1.2:1:1 ratio for Xl octamers. Histones were dialysed against 2 changes of 1 L of refolding buffer (2 M NaCl, 25 mM Tris-HCl pH 7.5, 1 mM DTT) for 1.5 h at 4°C, then against 1 change of 1L of refolding buffer for 16 h at 4°C, and finally against 1 change of 1L of refolding buffer for 1.5 h at 4°C. Histone dimers, tetramers or octamers were purified by size exclusion chromatography using a Superdex 200 16/60 column (Cytiva). After concentrating in centrifugal filters (Amicon Ultra, 10 MW cutoff), histone dimers, tetramers or octamers were stored in 50% glycerol at −20°C.

DNA for nucleosome reconstitution was amplified by polymerase chain reaction (PCR) and purified by anion-exchange chromatography using a 1 mL HiTrap^™ DEAE Sepharose^™ Fast Flow column (Cytiva). The DNA was subsequently precipitated by ethanol precipitation and resuspended in H₂O.

For reconstituting nucleosomes, DNA was mixed at a 1.1-fold molar excess with the histone Xl octamer (80S80 nucleosomes) and with Hs dimer and Hs tetramer (for the rest) in a buffer containing 25 mM HEPES pH 7.5, 0.25 mM DTT, and 2 M NaCl. The NaCl concentration was gradually decreased to 50 mM over 16 h at 4°C by salt gradient dialysis. After this, nucleosomes were purified by anion exchange chromatography using a 1 ml Resource Q column, and fractions containing nucleosomes were pooled and dialysed to 25 mM Tris-HCl pH 7.5, 0.25 mM DTT and 50 mM NaCl, concentrated to 1 mg/ml (Amicon Ultra, 30 MW cutoff), and stored at 4°C or aliquoted and flash-frozen in liquid nitrogen in the presence of 10% of glycerol and stored at −80°C.

DNA for array reconstitution was amplified by polymerase chain reaction (PCR) and purified using NucleoSpin Gel and PCR Clean-up, Maxi kit for gel extraction and PCR clean up (Macherey-Nagel, 740610.20).

For reconstituting arrays, DNA was mixed at a 0.83-fold molar excess with the histone dimer and tetramer in presence of recombinant albumin in a HEN2000 buffer containing 10 mM HEPES, pH 8; 0.1 mM EDTA; 2 M NaCl. The NaCl concentration was gradually decreased to 50 mM over 16 h at 4°C by salt gradient dialysis. After this, arrays were dialysed to HEN50 buffer (10 mM HEPES, pH 8; 0.1 mM EDTA; 50 mM NaCl), concentrated using Amicon Ultra, 30 MW cutoff, and stored at 4°C.

MNase array quality control

Correct nucleosomal array reconstitution was assessed by performing a MNase digest. Here, 1000 ng of array was digested with 20 U MNase (NEB, M0247S) using provided reaction buffer. The reaction was stopped using STOP-buffer (50 mM Tris pH 7.5, 4% SDS, 100 mM EDTA pH 8.0) at different incubation time points (1’ and 5’). Afterward samples were incubated with Proteinase K (Thermo Fisher Scientific, EO0491) for 1 h at 50°C and digested DNA was isolated using ethanol precipitation in the presence of glycogen (Roche, 10901393001). The isolated DNA fragments were separated using 1.5% (Biozym, 840004) agarose gel electrophoresis with 1×TAE (40 mM Tris pH 8, 1 mM EDTA, 20 mM acetic acid) as running buffer (120 V, 1.5 h) and stained with GelRed (Biotium, 41003). Gels were imaged using a GE Healthcare Typhoon FLA9000 imager.

Restriction enzyme accessibility assays

50 nM of INO80^ΔΝ and 12.5 nM arrays (assembled with WT Hs octamer) were mixed in a reaction buffer containing 40 mM Tris pH 7.5, 60 mM KCl, 5 mM MgCl₂, and 2 U/μL EcoRV. Each mixture was divided in two after a 2-minute preincubation at 30°C: the control without ATP and the reaction that was started with the addition of 1 mM ATP. Samples were quenched at various time-points with equal volumes of 20 mM Tris pH 7.5, 2% SDS, 70 mM EDTA and 20% glycerol and incubated with Proteinase K for 30 minutes at 50°C to digest all proteins.

The DNA fragments were then isolated by ethanol precipitation using glycogen (Roche, 10901393001). The isolated DNA (cut and uncut fragments) for each timepoint were resolved using 1.5% (Biozym, 840004) agarose gel electrophoresis with 1×TAE as running buffer (120 V, 1.5 h) and stained with GelRed (Biotium, 41003). Gels were imaged using a GE Healthcare Typhoon FLA9000 imager and were evaluated using ImageJ. After background subtraction, the total intensities of the cut and uncut band peaks were evaluated and the percentages of the fraction cut for each sample were plotted with Prism (GraphPad Software).

NADH-coupled ATPase assay

The ATP hydrolysis rate of INO80^ΔN was determined using an NADH-coupled ATPase assay. Reactions were carried out in a total volume of 50 μL containing 30 nM INO80^ΔN in assay buffer composed of 25 mM HEPES (pH 8.0), 50 mM KCl, 1 mM DTT, 2 mM MgCl₂, and 0.1 mg/mL BSA. The reaction mixture was supplemented with 0.5 mM phosphoenolpyruvate, 1 mM ATP, 0.1 mM NADH, 25 U/mL lactate dehydrogenase, and pyruvate kinase. Assays were performed at 25°C in non-binding black 384-well plates, and NADH consumption was monitored fluorometrically over one hour using a Tecan Infinite M100 plate reader (excitation 340 nm, emission 460 nm). Where indicated, ATPase activity was measured in the presence of 200 nM nucleosome or 50 nM of tetranucleosomal array. All reactions were performed in triplicate. ATP turnover rates were determined from the maximal initial linear rates of the NADH depletion curves and corrected by subtraction of a buffer blank.

Nucleosome sliding assays

Nucleosomes with 6’FAM–labelled extranucleosomal DNA were used for monitoring the sliding activity of purified recombinant INO80^ΔN complex. Nucleosome (150 nM) was incubated with 50 nM INO80^ΔN complex in sliding buffer [25 mM HEPES pH 8.0, 60 mM KCl, 7% glycerol, recombinant albumin (0.10 mg/ml), 0.25 mM DTT] at 25°C. The sliding reaction was started with the addition of 1 mM ATP and 2 mM MgCl₂ and stopped at indicated time points by addition of Lambda DNA (0.2 mg/ml). Nucleosome species were separated by native polyacrylamide gel electrophoresis (PAGE) on a 3 to 12% acrylamide Bis-Tris gel (Invitrogen) and visualised using the Typhoon imaging system (GE Healthcare). Experiments were performed in triplicate.

Nucleosome sliding assays were quantified using ImageJ by measuring band intensities corresponding to unremodelled nucleosomes (end-positioned; lowest band), sliding intermediates, and fully remodelled nucleosomes (centrally positioned; uppermost band). The fraction of centrally positioned nucleosomes was calculated as the intensity of the fully remodelled species expressed as a percentage of the total signal. The remodelled fraction was defined as the combined intensity of sliding intermediates and fully remodeled nucleosomes relative to the total intensity. Kinetic data of the remodelled fraction were fitted using a one-phase exponential decay model: y(t) = (y₀ – p)e^−Kt + p, the centrally positioned nucleosome data was fitted with y(t) = y_inf(1 – e^−Kt) for visualisation purposes.

Fluorescence anisotropy

Increasing concentrations of INO80 A-module (0, 8, 16, 32, 64, 128, 256 and 512 nM final) were prepared in assay buffer (25 mM HEPES, pH 8.0, 120 mM KCl, 2 mM MgCl₂, 2% glycerol, 0.01% Triton X-100, 1 mM DTT) and mixed at a 1:1 (v/v) ratio with 50-bp 6-FAM-labelled DNA (50P or 50G; Table Sxxx) to yield a final DNA concentration of 5 nM in a total volume of 20 μl (Greiner flat-bottom black 384-well plates). Samples were incubated for 20 min at room temperature prior to measurement of fluorescence anisotropy (λ_ex = 470 nm, λ_em = 520 nm) using a TECAN Infinite M1000 plate reader. All measurements were performed in triplicate.

For data analysis, background signal from the 0 nM protein control was subtracted from all data points. The resulting datasets were analysed using Prism (GraphPad Software) and fitted with a non-linear, non-cooperative 1:1 binding model $(y(t)=A_f-\frac{A_f-A_b}{x(K_d+x)}),$ where y denotes anisotropy, A_f and A_b represent the anisotropy of free and bound ligand, respectively, K_d is the dissociation constant, and x corresponds to receptor concentration. Apparent dissociation constants were derived from the fitted parameters.

Vitrification for cryo-EM

For INO80^ΔN-array NCP complexes, 600 nM INO80 and 150 nM of arrays were mixed in cryo-EM buffer (20 mM HEPES, pH 8.0; 60 mM KCl; 2 mM MgCl₂; and 1 mM DTT). 1 mM ATP was added to the mixture and the sample was incubated at 25 degrees for 10 min, then 1.5 mM of the respective nucleotide (AlF_x or BeF_x) was added. For INO80^fl-80N0 M2_rev nucleosome, INO80^fl-80S80 nucleosome and INO80^ΔN-80S0 nucleosome complexes; 500 nM INO80 and 500 nM nucleosome, 150 nM INO80 and 165 nM nucleosome, 600 nM INO80 and 600 nM nucleosome respectively were mixed in cryo-EM buffer. Then 1 mM ADP-AlF_x (M2_rev, 80S80) or 1.5 mM ADP-AlF_x (80S0) was added.

After the addition of the respective nucleotide, samples were incubated at 25 degrees for 10 min. In case of crosslinking (for AlF_x XL dataset), 0.1% glutaraldehyde was added and sample was incubated on ice for 10 min. Octyl-β-glucoside was added (0.045%), and 4.5 μl was applied onto a glow discharged Quantifoil R2/1 + 2 nm C Cu200 grid (arrays, 80S80, 80S0) or onto a glow discharged Quantifoil R2/1 Cu200 grid (M2_rev). The sample was vitrified in liquid ethane using an EM GP plunge freezer (Leica; 10°C and 90% humidity; 2.2 s blot; carbon grids were pre-blotted for 15 s).

Cryo-EM data collection

For INO80^ΔN-array NCP, INO80^fl-80S80 nucleosome and INO80^ΔN-80S0 nucleosome complexes, movies of particles embedded in vitreous ice were collected at liquid nitrogen temperature using a FEI Titan Krios G3 transmission electron microscope (300 kV, Thermo Fisher Scientific) equipped with a Selectris X imagining filter (slit width 10 eV) with a Falcon4 direct electron detector. 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 −0.5 to −2.6 μm. The total electron dosage of each movie was 40 e/Ų, fractionated into 40 movie frames.

For INO80^fl-80N0 M2_rev nucleosome complexes, movies of particles embedded in vitreous ice were collected at liquid nitrogen temperature using an FEI Titan Krios G3 transmission electron microscope (300 kV) equipped with a GIF quantum energy filter (slit width 20 eV) and a Gatan K2 Summit direct electron detector. The movies were recorded in counting mode using EPU acquisition software (Thermo Fisher Scientific) at ×130,000 magnification with a pixel size of 1.059 Å/pixel and nominal defocus range of −1.1 to −2.9 μm. The total electron dosage of each movie was 41.04 e/Ų, fractionated into 40 movie frames.

Cryo-EM data processing

For all datasets, movie frames were motion-corrected using MotionCor2⁵⁹. All subsequent processing steps were performed in cryoSPARC (v5.0.6 and former versions)⁶⁰, and the resolutions reported here are calculated based on the gold-standard Fourier shell correlation criterion (FSC = 0.143) with 3DFSC plots using Remote 3DFSC Processing Server⁶¹ or with cryoSPARC⁶⁰. CTF parameters of each dataset were determined using patch CTF estimation (multi) in cryoSPARC⁶⁰. The data collection and refinement statistics for each dataset are summarized in the cryo-EM validation tables 2–5. Figures and composite maps were generated in UCSF ChimeraX 1.5 or 1.6.145.32⁶².

For the INO80^ΔN-array complex AlF_x XL dataset, 40195 micrographs were collected. The processing scheme is depicted in Fig. S9. Initial particle picking was done using Blob picker (2621k particles). Particles were subjected to 2D classification, ab initio reconstruction, and heterogeneous refinement (1069k particles). Particles of INO80-NCP complexes, were then extracted with a box size of 600 px and a pixel size of 0.727 Å/pixel, and used for 3D-classification, sorting the particles in O- (259k particles) and R-states (134k particles). Particles of both states were further 3D-classified and particles with clearly defined features were used for homogeneous and/or non-uniform refinement of R-state⁶³. The final resolution for INO80-NCP complex in O-state overall and local refinements was 2.91 and 3.49 Å, respectively⁶¹. The final resolution for INO80-NCP complex in R-state overall and local refinements was 2.68 and 2.98 Å, respectively⁶¹.

For the INO80^ΔN-array complex BeF_x dataset, 31372 micrographs were collected. The processing scheme is depicted in Fig. S10. Initial particle picking was done using Blob picker (1332k particles). Particles were subjected to 2D classification, ab initio reconstruction, and heterogeneous refinement (198k particles). Particles of INO80-NCP complexes were used for template picking⁶⁴, 2D-, and 3D-classification, ab initio reconstruction and heterogeneous refinement. Particles were extracted with a box size of 600 px and a pixel size of 0.727 Å/pixel. 708k particles were used for homogeneous refinement of R-state. The final resolution overall and local refinements for INO80-NCP in the R-state was 2.52 and 2.43 Å, respectively⁶⁰. Particles of INO80-NCP in the R-state were further 3D-classified and particles of classes with the A-module density were then re-extraction with a box size of 800 px and refined. The final resolution for INO80-NCP complex in the I-state both overall and local refinements was 3 Å ⁶⁰.

For the INO80^ΔN-80S0 dataset, 16603 micrographs were collected. The processing scheme is depicted in Fig. S11. Initial particle picking was done using Blob picker (442k particles). Particles were subjected to 2D classification, ab initio reconstruction, and heterogeneous refinement (198k particles). Further 2D-, and 3D-classification, and homogeneous refinement were applied to sort the particles leading to the sharpest reconstruction. 150k particles were then extracted with a box size of 600 px and a pixel size of 0.727 Å/pixel and used for homogeneous refinement of O-state. The final resolution for INO80^ΔN-80S0 complex in the O-state overall and local refinements was 2.68 and 3.25 Å, respectively⁶⁰.

For the INO80^fl-80S80 nucleosome dataset 12000 micrographs were collected. Initial particle picking was done using Blob picker (9249k particles). For INO80^fl-80S80 nucleosome complex (the processing scheme is depicted in Fig. S12), particles were subjected to 2D classification, ab initio reconstruction, and heterogeneous refinement (226k particles). Particles with clearly defined features, were used for as input for a Topaz train job⁶⁵. After four rounds of Topaz, particles were extracted with a box size of 504 px and a pixel size of 0.727 Å/pixel. The particles were subjected to 2D classification and heterogeneous refinement. The class that showed the most defined features was selected (172k particles) and used for ab-initio reconstruction and non-uniform refinement⁶³ and further heterogeneous refinement into three classes. One class with 127k particles was used for homogeneous and local refinements. The final resolution of the last homogeneous refinement and the last local refinements were 2.89 and 2.75 Å, respectively⁶¹.

For 80S80 nucleosome (the processing scheme is depicted in Fig. S13), particles were subjected to 2D classification, and particles with clearly defined features, were used for as input for a Topaz train job⁶⁵. After four rounds of Topaz, particles were extracted with a box size of 360 px and a pixel size of 0.727 Å/pixel. The particles were subjected to 2D classification and heterogeneous refinement. The class that showed the most defined features was selected (650k particles) and used for further heterogeneous refinement into four classes. One class with 450 classes was used for non-uniform refinement⁶³, 3D classification, and a extra round of heterogeneous refinement. Three classes with 82k, 121k and 112k particles were selected and used for non-uniform refinement⁶³, first separately and then combined. The combined particles were used for 1 round of 3D classification with 2 classes. One class with 205k particles was selected and used for local refinement. The final resolution of the last local refinement was 2.81 Å ⁶¹.

For the INO80^fl-80N0 M2_rev nucleosome dataset, 14722 micrographs were collected. The processing scheme is depicted in Fig. S14. Initial particle picking was done using Blob picker (4739k particles). Particles were subjected to 2D classification, ab initio reconstruction, and heterogeneous refinement (355k particles). Particles with clearly defined features, were used for two rounds of Topaz⁶⁵, 2D-, and 3D-classification, ab initio reconstruction and heterogeneous refinement. 55k particles were used for non-uniform refinement⁶³ and further 2D classification and ab initio reconstructions, followed by heterogeneous refinement into two classes. One class with 25k particles was used for 3D variability analysis. 14k particles were used for non-uniform and local refinements⁶³, 3D flexibility analysis⁶⁶ and final non-uniform refinement with a box size of 512 px and a pixel size of 1.059 Å/pixel. The final resolution of the INO80^fl-80N0 M2_rev nucleosome complex was 4.37 Å ⁶¹.

Model building and refinement

Model templates were taken from AlphaFold2 34265844 model predictions for all S. cerevisiae protein chains, and from the cryo-EM INO80 Core structure for the H. sapiens histones (6FML)⁶⁷, and from the cryo-EM Nucleosome structure for the X. laevis histones (6WZ5)⁶⁸, respectively. The DNA model template was also taken from the cryo-EM Nucleosome structure for X. laevis histones (6WZ5)⁶⁸ and mutated to the S. cerevisiae natural promoter sequence as described in the Suppl. Method “Mapping Nucleosomal DNA”.

All model templates were manually placed into the corresponding composite cryo-EM maps, followed by rigid-body refinement with ChimeraX⁶². The initial models were then energy-minimised using ISOLDE⁶⁹ and manually inspected and corrected with Coot⁷⁰.

Reciprocal space refinement was done at 2.8 Å resolution using jelly-body restraints with Servalcat⁷¹ against the calculated structure factors from the composite cryo-EM maps, resulting in R=0.254, <FSC>=0.871, rmsd bonds=0.012 Å, rmsd angles=1.61° for the INO80 core bound to array NCP in O-state, and in R=0.262, <FSC>=0.844, rmsd bonds=0.011 Å, rmsd angles=1.68° for the INO80 core bound to 80S80 SWH1 +1 NCP in R-state.

Genome-wide chromatin remodelling assay

S. cerevisiae genomic plasmid library⁷² chromatin was reconstituted with embryonic D. melanogaster histones by salt-gradient dialysis (SGD) under low assembly degree conditions and remodelled by INO80 as described⁵⁶ with the following modifications. DTT (2 mM) was used as reducing agent instead of β-mercaptoethanol during SGD. Remodelling reactions were done with 600 ng whole-genome plasmid library DNA reconstituted as SGD, 20 nM recombinant S. cerevisiae WT INO80^fl, 200 U KpnI (only replicates 1 and 2, Fig. S6) in a total volume of 50 μl for 2 h in the presence of ATP. For replicates 2 and 3, the remodelling reaction buffer did not contain ammonium sulphate.

DNA shape analysis

Genome-wide nucleosome dyad positions were called from paired-end Illumina (Next Seq 1000 for replicates 1 and 2, NextSeq 2000 for replicate 3) MNase-seq data. MNase-seq read pairs were mapped using Bowtie2 (2.5.4) to the yeast reference genome (SacCer3, R64-1-1), filtered for nucleosomal size fragments (130 to 180 bp) and trimmed to 50 bp around the inferred dyad (calculated from paired-end fragment midpoints). Read counts were normalized by subsampling each group of samples to the number of reads in the sample with the lowest sequencing depth (target_n).

Nucleosome peak positions in the aggregated MNase-seq data were identified on dyad-centred coverages using the NucleR (2.42.0) peak-calling algorithm (Bioconductor). Only the top 3% of dyad positions ranked by peak height (97% threshold) were used as high-confidence nucleosome dyads. These positions were stored as .rds or .bed files for downstream analysis.

A matrix of DNA shape features was computed around each dyad using the DNAshapeR (1.38.0) package within a symmetrical window of ±160 bp (halfwin = 160L). For direct comparison with⁵⁶, we focused on the propeller twist (ProT) feature and generated dyad-aligned composite plots. These composite plots gave approximately symmetrical profiles. We reasoned that the symmetry could stem from the overlay of randomly oriented asymmetric profiles and sought to disentangle sequence orientations by DNA shape-based clustering (Fig. S5). Each dyad-centred ProT profile was Z-score-transformed row-wise such that each dyad profile had mean 0 and standard deviation 1. Only the profiles were retained where at least 95% of positions had finite Z-scores. The ProT matrix was used to compute a Euclidean distance matrix, which was clustered via hierarchical clustering with Ward.D2 linkage yielding a dendrogram based on ProT patterns of nucleosomal sequences. At the level of k = 2 clusters, we got two distinct clusters of nucleosomes. The right-side-heavy cluster, which showed the lowest ProT in the 0 to +160 bp region, appeared as mirror image of the left-side-heavy cluster profile and was flipped and merged with the left-side-heavy cluster to generate the final composite profile (plots were smoothed with 41 bp rolling mean for visualization – zoo/rollmean (1.8.14)). The ProT profile was also calculated for the sequence of the SWH1 in vivo +1 nucleosome dyad ±160 bp.

Data availability

For the cryo-EM structure and model of INO80 core bound to SWH1 +1 NCP in R-state, the coordinates are deposited in the Protein Data Bank (PDB, https://www.wwpdb.org/) with the accession code 30II; the cryo-EM density maps are deposited in the Electron Microscopy Data Bank (EMDB, https://www.ebi.ac.uk/emdb/) with the accession codes EMD-57769 (overall refinement), EMD-57790 (nucleosome refinement) and EMD-57808 (composite map). For INO80 core bound to array NCP in O-state in the ATP+AlF_x crosslinked dataset, the coordinates are deposited in the PDB with the accession code 30HC; the corresponding cryo-EM density maps are deposited in the EMDB with the accession codes EMD-57741 (overall refinement), EMD-57755 (nucleosome refinement) and EMD-57767 (composite map). For INO80 core bound to array NCP in R-state from this dataset, the accession codes are EMD-57978 (composite map), EMD-57857 (overall refinement) and EMD-57858 (nucleosome refinement).

For INO80 core bound to array NCP in I-state and R-state in the BeF_x dataset, the composite maps are deposited as EMD-58979 and EMD-57982, respectively, along with EMD-58976 and EMD-57980 for overall refinements and EMD-57977 and EMD-57981 for nucleosome refinements. For INO80 core bound to SWH1 +1 NCP in O-state, the accession codes are EMD-57985 (composite), EMD-57983 (overall refinement), and EMD-57984 (nucleosome refinement). The cryo-EM structure of INO80 core and A-module bound to M2_rev nucleosome in O-state is deposited with the accession code EMD-58053.

All codes are available via our GitHub repository https://github.com/Hopfner-Lab/map2seq.

Plasmids and materials generated in this study are available upon request from the corresponding author (KarlPeter.Hopfner@lmu.de).