Extended Data Fig. 3: Cryo-EM processing pipeline and modelling of DNA-bound D2-I^3D.

(a) CryoEM processing pipeline for the DNA-D2-I^3D complex. A similar pipeline was used to process the D2-I^3D and ubD2-I^3D datasets. A crYOLO model was trained on a subset of micrographs and then used to pick all particles. An ab initio model was obtained after particle extraction and 2D classification. Several rounds of 3D refinement and classification, followed by CTF refinement and Bayesian polishing were performed next. The final consensus map was obtained after refinement of two selected classes (with the best density for the N-terminal region of FANCD2 and DNA) resulting from a 3D classification without image alignment. Individual maps of FANCD2 and FANCI were obtained by signal subtraction of the FANCI-DNA and the FANCD2-DNA density respectively, followed by a 3D refinement. Sharpened maps (B factor of -55 for FANCD2 and -60 for FANCI) were used for interpretation and modelling. (b) Modelling of FANCD2 and FANCI. Fit of the D2-I complex (shown as sticks) in the composite map (sharpened using B factor of -55). (c) Fit of the FANCD2 and FANCI models (shown as sticks) into the corresponding signal subtracted and focus-refined maps (sharpened using B factor of either -55 or -60). Representative densities of the C-terminal and the N-terminal regions of either FANCD2 or FANCI from the DNA-D2-I^3D (sharpened map) and the corresponding model regions (cartoon representation, side chains represented as sticks (oxygen atoms in red, nitrogen in blue) are shown. (d) A bent dsDNA was fitted into the corresponding DNA density (postprocessed composite map on the left and unsharpened map of DNA on the right). (e) Comparison of the FANCI and FANCD2 chains from the DNA-D2-I^3D model to the corresponding FANCI and FANCD2 chains from the G. gallus D2-I^WT (PDB 6TNG) and the ubiquitinated D2-I^WT (PDB 6TNF) complexes.