Extended Data Fig. 5: Endogenous TREX–mRNP complex cryo-EM image processing, reconstructions, and biochemistry of UAP56–ALYREF.

a. Three-dimensional image classification tree of endogenous TREX–mRNP complex cryo-EM data^27,58. The complete data set contained 840,469 TREX-mRNP particles, which were classified in multiple rounds of 3D classification (with regularization parameter T = 4 for all RELION classifications) and focused refinement in RELION^62,71. The best particles were used to extract symmetry related dimers, separately, yielding 415,848 dimer particles, which were further classified and refined in cryoSPARC⁶³. This yielded maps A (cyan), B (light green), and C (slate blue) (see Methods for details). The percentage of TREX–mRNP particles (black) or TREX dimer units (orange) contributing to each class are provided. The type of mask and overall resolution is indicated for each 3D refinement. b. Gold-standard Fourier shell correlation (FSC = 0.143) of the TREX–mRNA cryo-EM maps A, B, and C. c. Orientation distribution plots for all particles contributing to the TREX–mRNA cryo-EM maps A, B, and C, visualized in cryoSPARC⁶³. d. The composite TREX–mRNA cryo-EM density is shown from front and left side views (maps A, B, and C), and colored by local resolution as determined by cryoSPARC⁶³. e. The composite TREX–mRNA cryo-EM density (maps A, B, and C) is shown opposite of the refined TREX–mRNA coordinate model, which is shown as ribbons and colored as in Fig. 2d. f. Gallery of TREX–mRNA complex subunits THOC1, THOC5 (tRWD domain), and THOC6 are shown superimposed on their respective cryo-EM densities. Below each protein a representative segment of the protein is superimposed on the respective cryo-EM density. g. The TREX monomer A is mobile in the TREX–mRNA complex. Two densities obtained from 3D variability analysis (class 3 in grey and class 8 in green) are overlayed, revealing that monomer A can shift globally by ~25 Å. This mobility can explain why monomer A, and the associated UAP56 molecule, have a low local resolution. h. The TREX–mRNA map reveals density for the UAP56 RecA1 lobe, the ALYREF UBM, and putatively assigned mRNA, which were fitted as a single rigid body of a yeast Yra1–Sub2–RNA homology model (5SUP). The ALYREF UBM, which could be either N- or C-terminal, is visible at lower density threshold, and was modelled as the C-UBM based on its position in the yeast Yra1 (C-UBM)–Sub2–RNA crystal structure and an AlphaFold2 mulitmer model⁸⁴ of the ALYREF C-UBM bound to human UAP56. i. Mutation of human UAP56 residues at the ALYREF-UBM to UAP56 interface, supports the ALYREF-UBM density assignment. Top: Interface mutations are mapped onto the UAP56 coordinate model and labelled. Bottom: In vitro, a fluorescently labeled ALYREF C-UBM peptide binds to wildtype UAP56 but not mutated UAP56. This experiment was done once. For gel source data, see Supplementary Fig. 8. j. Comparison of human ALYREF-UBM–UAP56–RNA (this study) and yeast Yra1-UBM–UAP56–RNA–ATP-analog (5SUP)⁴⁰ structures. k. An RNA filter-binding assay suggests that the ALYREF RNA binding domains 1 and 2 (RBD1 and RBD2) might assist RNA delivery to UAP56, but not the isolated ALYREF_55–182 construct that forms EJC contacts (see Fig. 1, Extended Data Fig. 1). Left: Boundaries of protein constructs used for RNA affinity measurements using filter binding assays. Middle: Binding curves of the tested constructs. The plot shows mean values from n = 6 measurements, error bars indicate the standard deviation of each measurement, and solid or dotted lines show the fit of a “Specific binding with Hill-slope”-function to the data, with the Bmax constrained to 1 as implemented in GraphPad Prism (see Methods). Right: Measured dissociation constants (K_D) of the tested constructs as determined by the fits in the middle panel; spheres indicate the K_D determined form the fit and error bars indicate the 95% confidence interval determined from the fit. UAP56-RNA binding is not detectable with isolated UAP56 in absence of ATPγS, but does bind RNA with K_D of ~900 nM (95% confidence interval: 810–1,014 nM) in presence of 1 mM ATPγS. The ALYREF-RNA binding activity is contained in its RBD1 and RBD2 domains, but not in the WQHD motif or RRM domain. These experiments were done twice, with three technical replicates each.