Extended Data Fig. 8: Flexibility and anchoring of the enzymatic modules.

a, Two-dimensional classes of SAGA side views revealing different positions of the enzymatic HAT and DUB modules. Bottom, approximate positions of the modules are highlighted in yellow for the HAT module and in blue for the DUB module. The HAT module seems more diffuse and flexible than the DUB module, whose position fluctuates less. b, Docking of the DUB crystal structure (PDB: 6AQR) into the 15 Å resolution cryo-EM map reconstructed from a small class of particles that probably share a similar position for the DUB. c, Domain organization of the Sgf73 subunit. d, The Sgf73 linker tethers DUB to the central module. The linker (S73L) connecting the N-terminal end (ND), which is part of the DUB, and the anchoring ___domain (S73A) embedded in the central module, includes a SCA7 ___domain but additionally includes roughly 90 amino acids predicted to be unstructured. All residues that cross-link³⁶ to the Sgf73 linker can be mapped on the surface of SAGA. These residues are depicted with light-blue spheres. A putative path for the 164-residue-long linker is delineated (dashed) according to the cross-linking sites and scarce traces of the linker found in the cryo-EM map. The top insert shows the density that forms contact between DUB and the central module. This contact point is located next to the Spt20 SEP and Taf5 LisH domains and is the only trace of the flexible Sgf73 linker in our maps. e, Two helical domains (ochre) anchor the HAT module at the surface of the Taf6 HEAT repeats (red). f, These helical domains are assigned to Ada3. Domain-deletion analysis identified two domains in Ada3 and one in Ada2 that serve to dock HAT on SAGA and whose deletion results in the release of intact HAT from SAGA³⁶. Fit between the maps and secondary structure predictions (PSIPRED 3.3) or 3D structure of the Ada3 and Ada2 domains is presented in addition to available cross-linking data between these domains and the Taf6 HEAT repeats.