Structural mechanism of extranucleosomal DNA readout by the INO80 complex

The nucleosomal landscape of chromatin depends on the concerted action of chromatin remodelers. The INO80 remodeler specifically places nucleosomes at the boundary of gene regulatory elements, which is proposed to be the result of an ATP-dependent nucleosome sliding activity that is regulated by extranucleosomal DNA features. Here, we use cryo–electron microscopy and functional assays to reveal how INO80 binds and is regulated by extranucleosomal DNA. Structures of the regulatory A-module bound to DNA clarify the mechanism of linker DNA binding. The A-module is connected to the motor unit via an HSA/post-HSA lever element to chemomechanically couple the motor and linker DNA sensing. Two notable sites of curved DNA recognition by coordinated action of the four actin/actin-related proteins and the motor suggest how sliding by INO80 can be regulated by extranucleosomal DNA features. Last, the structures clarify the recruitment of YY1/Ies4 subunits and reveal deep architectural similarities between the regulatory modules of INO80 and SWI/SNF complexes.

Table S3.In vivo yeast mutants.

Fig S2 :
Fig S2: Examples of cryo-EM density map of the C. thermophilum A-module.(A) Structural model of C. thermophilum (Ct) A-module (left) without DNA or (right) with DNA bound.The protein subunits are color coded and annotated.(B) Cryo-EM density maps of nucleotides (ATPγS) bound to (left to right) Arp8, actin and Arp4 (surface cutoff: 2 Å) (62).(C) Detailed view of the Ies4-actin interface in the cryo-EM density map of the CtA-module.(D) Detailed view of the Ino80 HSA /Arp8 N interface in the cryo-EM density map of the CtA-module.(E) Detailed view of the Ies4/Arp8 N -DNA interface in the cryo-EM density map of the CtA-module bound to DNA.

Fig. S3 .
Fig. S3.The 2W-hairpin shows a conserved interaction mode with actin/ARPs.(A) Detailed view of the Ies4-actin interface in the S. cerevisiae A-module.The conserved tryptophan and proline residues are shown.(B) Detailed view of the DNGR-1-actin interface (PDB: 3J82).The conserved tryptophan and proline residues are shown.(C) Structure of the dysotrophin WWdomain in complex with a β-dystroglycan peptide (PDB: 1EG4).Conserved tryptophan and proline residues are shown.(D) Detailed view of the Rtt102-Arp9 interface in S. cerevisiae SWI/SNF A-module (PDB: 4I6M).The conserved tryptophan and proline residues are shown.

Fig. S4 .
Fig. S4.YY1 interacts with the H. sapiens Ino80 HSA .(A) Coomassie-stained SDS-PAGE gel of the purified H. sapiens (Hs) A-module (B) Cryo-EM reconstruction of HsA-module.The density is colored according to the underlying protein subunits.(C) Coomassie-stained SDS-PAGE gel showing the BS3 crosslinked INO80 complex.The red square comprising the high-molecular weight crosslinked species indicates the cut-out region for in gel digest used for mass spectrometry (MS) analysis.(D) Topological crosslink-MS scheme of HsINO80 A-module subunits, showing inter-protein links (green).

Fig. S5 .
Fig. S5.HSA surface residues are critical for INO80 function in budding yeast.(A) Ino80 variants are expressed at levels similar to the Ino80 WT protein.Yeast cells as in Fig. 2A expressing the indicated 2FLAG-tagged Ino80 variants under the control of the endogenous INO80 promoter were subjected to total protein analysis by western blotting using an anti-FLAG antibody.Pgk1 levels served as control.(B) Ino80 variants are expressed at levels similar to the Ino80 WT protein.Diploid yeast cells used for tetrad analysis in Fig. 2B expressing the indicated 2FLAGtagged Ino80 variants under the control of the endogenous INO80 promoter were subjected to total protein analysis by western blotting using an anti-FLAG antibody.Pgk1 levels served as control.(C)The Arp8∆N variant is expressed similar to Arp8 WT levels.Yeast cells as in Fig.2Cand D expressing the indicated 2FLAG-tagged Arp8 variants under the control of the endogenous ARP8 promoter were subjected to total protein analysis by western blotting using an anti-FLAG antibody.Pgk1 levels served as control.

Fig. S7 .
Fig. S7.Example of cryo-EM density map of the Ino80 motor (ADP•BeFx state).(A) Comparison of Ino80, ISWI and Snf2 interacting with nucleosomes and similarity of post-HSA and Auto-N.Ino80 binds the nucleosome at SHL-6 while Isw1 and Snf2 bind at SHL-2.(B) Detailed view of the cryo-EM density map of the Ino80 motor -bound DNA of the C. thermophilum C-module.The protein subunit is color coded and annotated (surface cutoff: 2 Å) (62).(C) Cryo-EM density map of ADP bound to the Ino80 motor (surface cutoff: 2 Å) (62).Note, the light BeFx moiety is not visible in the density map, in line with other studies (19).

Fig. S9 .
Fig. S9.Influence of A/T-rich DNA on CtINO80 ΔN nucleosome remodeling.(A) DNA curvature analysis of the 601-based 0N80 DNA template sequence and sequences with replaced A/T-rich DNA cassettes (https://github.com/cgohlke/dnacurve). (B) Sliding of different 0N80 nucleosomes by CtINO80 ΔN analyzed by native PAGE.(C) Raw data of ATPase assays in presence of different 0N80 nucleosomes.ATPase rates were determined for CtINO80 ΔN wild type (wt) along with nucleosome-stimulated rates.(D) Electrophoretic mobility shift assay of different 0N80 nucleosome substrates bound to CtINO80 ΔN analyzed by native page.

Fig. S11 .
Fig. S11.Cryo-EM data analysis of S. cerevisiae A-module (ADP-state).(A) Cryo-EM data processing workflow of S. cerevisiae A-module using Relion-3.0(58).A representative micrograph and the cryo-EM data processing workflow are shown.(B) Gold-standard Fourier shell correlation (FSC) curve of the final A-module reconstruction.The red line indicates the 0.143 cutoff criterion.

Fig
Fig. S12.Cryo-EM data analysis of S. cerevisiae A-module (ATPγS-state).(A) Good particles were selected through a series of focused 3D classifications and subsequently polished in Relion (58).The final 3D reconstructions were generated and the resolution values calculated by Relion independent half map FSC0.143 criterion.(B) The particles were picked ab initio and qualitatively filtered using WARP (60).(C) The ab initio 3D model without DNA bound was generated in cisTEM (61) and used as the 3D reference for DNA-bound datasets to avoid bias in DNA presence and conformation.(D) Gold-standard Fourier shell correlation (FSC) curve of the final A-module reconstruction (no mask applied).The red line indicates the 0.143 cutoff criterion.

Fig
Fig. S13.Cryo-EM data analysis of C. thermophilum A-module (ATPγS-state).(A) Good particles were selected through a series of focused 3D classifications and subsequently polished in Relion (58).The final 3D reconstructions were generated and the resolution values calculated by Relion independent half map FSC0.143 criterion.(B) The particles were picked ab initio and qualitatively filtered using WARP (60).(C) The ab initio 3D model without DNA bound was generated in cisTEM (61) and used as the 3D reference for DNA-bound datasets to avoid bias in DNA presence and conformation.(D) Gold-standard Fourier shell correlation (FSC) curve of the final A-module reconstruction (no mask applied).The red line indicates the 0.143 cutoff criterion.

Fig. S14 .
Fig. S14.Cryo-EM data analysis of C. thermophilum A-module and DNA (ADP•AlFx).The ab initio 3D model without DNA bound was generated in cisTEM (61) and used to avoid bias in DNA presence and conformation.(A) DNA-bound A-module species were isolated from DNAfree species, followed by discernment and isolation of A-modules with straight and curved DNA bound through a series of focused 3D classifications in Relion (58).The final 3D reconstructions were generated and the resolution values calculated by Relion independent half map FSC0.143 criterion.(B) The particles were picked ab initio and qualitatively filtered using WARP (60).(C) Gold-standard Fourier shell correlation (FSC) curves of the final A-module reconstructions (no mask applied).The red line indicates the 0.143 cutoff criterion.

Fig. S15 .
Fig. S15.Cryo-EM data analysis of C. thermophilum INO80 ΔN .(A) A small subpopulation with the C-and A-modules more coherently placed were isolated through a series of focused 3D classifications in Relion and the final 3D reconstruction was generated by Relion multibody 3D refinement (58).(B) The particles were picked ab initio and qualitatively filtered using WARP (60).(C) 2D class averages of isolated full INO80 complex (D) The ab initio 3D model was generated in cisTEM (61).(E) Angular distribution plot of the full INO80 complex 3D reconstruction.

Fig. S17 .
Fig. S17.Cryo-EM data analysis of H. sapiens A-module.(A) Cryo-EM data processing workflow of HsA-module using cryoSPARC v3.2.0 (57).Representative micrograph of HsAmodule and representative classes of a 2D classification of the particles used for the final A-module reconstruction after an ab initio model.(B) Gold-standard Fourier shell correlation (FSC) curve of the final A-module reconstruction.The red line indicates the 0.143 cutoff criterion.

Table S2 .
In vivo yeast strains.