Stefan Petrovic, Dipanjan Samanta, # Thibaud Perriches, # Christopher J. Bley,
Karsten Thierbach, Bonnie Brown, Si Nie, George W. Mobbs, Taylor A. Stevens, Xiaoyu Liu, Giovani Pinton Tomaleri, Lucas Schaus, André Hoelz*
(2022). Science, 376, eabm9798

Figures from the paper:

Abstract:

Nuclear pore complexes (NPCs) mediate the nucleocytoplasmic transport of macromolecules. Although the arrangement of the structured scaffold nucleoporins in the NPC’s symmetric core had been determined, their cohesion by multivalent unstructured linker nucleoporins remained elusive. Combining biochemical reconstitution, high‑resolution structure determination, docking into cryo-electron tomographic reconstructions, and physiological validation, we elucidated the architecture of the evolutionarily conserved linker‑scaffold, yielding a near-atomic composite structure of the human NPC’s ~64MDa symmetric core. Whereas linkers generally play a rigidifying role, the linker-scaffold of the NPC provides the plasticity and robustness necessary for the reversible constriction and dilation of its central transport channel and the emergence of lateral channels. Our results substantially advance the structural characterization of the NPC symmetric core, providing a basis for future functional studies.

Architecture of the linker-scaffold in the nuclear pore

Fig. 1. Outline of the symmetric core inner ring architecture. (A) Cross‑sectional schematic of the NPC architecture. (B) Domain structures of the Chaetomium thermophilum inner ring linker and scaffold nucleoporins (nups). Nic96 consists of linker (residues 1-390) and scaffold (residues 391-1112) regions. Nomenclature for nup homologs from C.thermophilum, Saccharomyces cerevisiae, and Homo sapiens. Multiple paralogs exist for some S.cerevisiae nups. (C, D) Schematic map of previously established linker-scaffold interactions in alternative, mutually exclusive inner ring complexes organized around Nup192 and Nup188 scaffold hubs (35). Black lines connecting colored bars indicate interactions between nup regions.

Fig. 2. Structural and biochemical analyses of the Nup192-Nic96 and Nup188-Nic96 interactions. (A) Domain structures of C.thermophilum Nup188, Nup192, and Nic96. (B) Cartoon representation of the 3.8Å C.thermophilum Nup192•Nic96R2 single particle cryo-EM structure. Inset regions are magnified to illustrate the molecular details of the Nup192‑Nic96 interaction. Red circles indicate residues involved in the Nup192-Nic96 interaction. (C, D) Summary of structure-guided (C) SUMO‑Nic96R2 and (D) Nup192 mutations’ effect on Nup192•SUMO-Nic96R2 complex formation, assayed by size‑exclusion chromatography (SEC); (+++) no effect, (++) weak effect, (+) moderate effect, (-) abolished binding. (E) Dissociation constants (KDs) determined by triplicate isothermal titration calorimetry (ITC) experiments, with the mean and associated standard error reported. (F) SEC coupled to multi‑angle light scattering (SEC‑MALS) interaction analyses of Nup192•SUMO‑Nic96R2 and interaction‑abolishing mutants. Measured molecular masses are indicated, with theoretical masses in parentheses. (G) Two views in cartoon representation of the 4.4Å C.thermophilum Nup188•Nic96R2 crystal structure phased and built with the 2.8Å Nup188NTD and 3.4Å Nup188Tail (PDB ID 5CWU) (28) crystal structures. Inset regions are magnified to illustrate the molecular details of the Nup188‑Nic96R2 interaction. Red circles indicate residues involved in Nup188-Nic96R2 binding. (H, I) Summary of structure‑guided (H) SUMO-Nic96R2 and (I) Nup188 mutations’ effect on Nup188•SUMO-Nic96R2 complex formation, assayed by SEC. (J) KDs determined by triplicate ITC experiments with the mean and associated standard error reported. (K) SEC-MALS analysis of Nup188•SUMO-Nic96R2 and interaction-abolishing mutants.

Fig. 3. Structural and biochemical analyses of the Nup192-Nup145N interaction. (A) Domain structures of C.thermophilum Nup192, Nic96, Nup53 and Nup145N; effect of each 5‑Ala substitution on Nup145N binding to Nup192, assessed by SEC, indicated by colored boxes above Nup145N primary sequence. (B) Summary of SEC binding analysis identifying the minimal Nup145NR1 (red) region sufficient for Nup192 binding; (+++) no effect, (++) weak effect, (+) moderate effect, and (-) abolished binding. (C) Cartoon representation of the 3.2Å C.thermophilum Nup192•Nic96R2• Nup53R1•Nup145NR1 single particle cryo-EM structure. Insets indicate regions magnified to illustrate molecular details of (D) the Nup192‑Nup53R1 and (E) the Nup192‑Nup145NR1 interactions. Red circles indicate residues involved in the Nup192‑Nup53R1 (41) and Nup192‑Nup145NR1 interactions. (F) Effect of Nup145N alanine substitutions (cyan squares) on Nup192 binding, assayed by SEC (left). Effect of structure‑guided Nup192 alanine substitutions on SUMO-Nup145NR1 binding, assayed by SEC (right). (G) KDs determined by triplicate ITC experiments, with the mean and associated standard error reported. (H) SEC-MALS analysis of Nup192•SUMO‑Nic96R2•Nup53•Nup145N and Nup192•SUMO‑Nic96R2•Nup53• SUMO‑Nup145NR1 complex formation and disruption by mutants. Measured molecular masses are indicated, with respective theoretical masses provided in parentheses.

Fig. 4. Structural and biochemical analyses of the Nup188-Nup145N interaction. (A) Domain structures of C.thermophilum Nup188, Nic96, and Nup145N; effect of each 5‑Ala substitution on Nup145N binding to Nup188NTD, assessed by SEC, indicated by colored boxes above Nup145N primary sequence. (B) Summary of SEC binding analysis identifying the minimal Nup145NR2 (red) region sufficient for binding to Nup188NTD; (+++) no effect, (++) weak effect, (+) moderate effect, and (-) abolished binding. (C) Cartoon representation of the 2.8Å C.thermophilum Nup188• Nic96R2•Nup145NR2 single particle cryo‑EM structure. Inset indicates region magnified to illustrate molecular details of the Nup188‑Nup145NR2 interaction. Red circles indicate residues involved in the Nup188‑Nup145NR2 interaction. (D) Effect of Nup145N alanine substitutions (cyan squares) on Nup188NTD binding, assayed by SEC (left). Effect of structure‑guided Nup188NTD alanine substitutions on SUMO-Nup145NR2 binding, assayed by SEC (right). (E) KDs determined by triplicate ITC experiments, with the mean and associated standard error reported. (F) SEC‑MALS analysis of Nup188•SUMO‑Nic96R2•Nup145N and Nup188•SUMO‑Nic96R2•SUMO‑Nup145NR2 complex formation and disruption by mutants. Measured molecular masses are indicated, with respective theoretical masses provided in parentheses.

Fig. 5. Functional in vivo dissection of the S.cerevisiae NPC linker-scaffold. (A) Cross‑sectional view of the S.cerevisiae NPC composite structure generated by docking linker‑scaffold structures into an ~25Å in situ sub-tomogram averaged cryo-ET map (EMD-10198) (36) (top). Schematic representation of a S.cerevisiae NPC spoke (bottom). (B) Domain structure of scNup116 variants; Gle2‑binding sequence (GLEBS); phenylalanine‑glycine repeats (FG); scNup192‑binding region (R1); scNup188‑binding region (R2); scNup157/170‑binding region (R3). (C) Viability analysis of a ten‑fold dilution series of a nup100Δnup116Δnup145Δ/NUP145C strain expressing scNup116 variants and subjected to 5-FOA selection for loss of rescuing wildtype plasmid. (D) Subcellular localization at permissive (30°C) and growth-challenging (37°C) temperatures of a representative subset of eGFP‑scNup116 variants in a nup100Δnup116Δnup145Δ/NUP145C-mCherry strain. (E) Representative images and quantitation (n>500) of subcellular localization of 60S pre-ribosomal export reporter scRpl25‑mCherry and poly(A)+ RNA at 30°C and 37°C, in the presence of a representative subset of scNup116 variants. (F) Domain structures of scNup192, scNup188, and scNic96 variants. (G) Viability analysis of a ten-fold dilution series of nup192Δ,  nup188Δpom34Δ, and nic96Δ S.cerevisiae strains expressing scNup192, scNup188, and scNic96 variants, respectively, and subjected to 5‑FOA selection for loss of rescuing wildtype plasmids. (H) Representative images and quantitation (n>500) of the nuclear scRpl25‑mCherry and poly(A)+ RNA retention in the presence of scNup188 and scNic96 variants at indicated growth‑challenging temperatures in nup188Δpom34Δ and nic96Δ S.cerevisiae strains, respectively. (I) Subcellular localization at permissive (30°C) and growth‑challenging (37°C) temperatures of the scCNT subunit scNup57‑eGFP in the presence of mCherry‑scNic96 variants. (J) Schematic model of CNT positioning in wildtype, scNic96R2 deletion, and glycine-serine (GS)‑linker replacement strains. Squares associated with variant labels are color-coded according to the nup binding partners targeted by the mutation. All experiments were performed in triplicate. Mean and associated standard error are reported for all quantitation. Scalebars are 5 μm.

Fig. 6. Evolutionary conservation of the human linker-scaffold network. (A) Domain structures of the human inner ring nups. NUP93 consists of linker (residues 1‑173) and scaffold (residues 174‑819) regions. (B) Schematic summary of the linker-scaffold interactions in complexes organized around the NUP188 and NUP205 scaffold hubs. Black lines connecting colored bars indicate interactions between nup regions. (C) Summary of SEC binding analysis identifying the minimal NUP53R2 region (red) sufficient for NUP93SOL binding; (+++) no effect, (++) weak effect, (+) moderate effect, and (-) abolished binding. (D) Effect of each 5‑Ala substitution on SUMO-NUP53N binding to NUP93SOL, assessed by SEC, indicated by colored boxes above NUP53 primary sequence. (E) Cartoon representations of the 2.0Å H.sapiens apo NUP93SOL, 3.4Å H.sapiens NUP93SOL•NUP53R2, 2.7Å C.thermophilum Nic96SOL•Nup53R2 (PDB ID 5HB3) (35) crystal structures and their superposition. An ~12° displacement of the C-terminal α‑helical solenoid, pivoted about the hinge loop, is observed between the apo NUP93SOL and NUP93SOL•NUP53R2 structures. (F) Schematic of the human NUP93SOL and C.thermophilum Nic96SOL fold architectures. (G, H) Summary of the effect of structure-guided mutations in (G) SUMO‑NUP53N and (H) NUP93SOL on NUP93SOL•SUMO-NUP53N complex formation, assayed by SEC. (I-K) Magnified views of regions indicated with insets in (E), comparing molecular details of NUP53R2/Nup53R2 binding sites. (L) SEC analyses of NUP93SOL•SUMO‑NUP53N complex formation and disruption by mutants. SDS‑PAGE gel strips of peak fractions visualized by Coomassie staining.

Fig. 7. Architecture of the human NPC symmetric core outer rings. Composite structure generated by quantitatively docking crystal and single particle cryo-EM structures into an ~12Å cryo‑ET map of the intact human NPC (EMD-14322) (47), viewed from the (A) cytoplasmic and (B) nuclear face. Nuclear envelope and docked structures are rendered in isosurface and cartoon representation, respectively. Insets indicate regions encompassing two spokes (top), 90°‑rotated and magnified (middle), and schematized (bottom). Cross‑spoke distances between the distal NUP205-bound NUP93R2 and distal NUP93SOL are indicated in red. Linker binding sites on scaffold nup surfaces are indicated by colored circles. Dashed transparent shapes indicate the absence of proximal NUP205 and NUP93 from the nuclear outer ring.

Fig. 8. Linker-scaffold architecture of the human NPC inner ring. Composite structure generated by quantitatively docking crystal and single particle cryo-EM structures into an ~12Å cryo‑ET map of the intact human NPC (EMD-14322) (47), viewed from (A) the cytoplasmic face, and (B) the central transport channel cross-section. Nuclear envelope and docked structures are rendered in isosurface and cartoon representation, respectively. (C-F) Starting from the nuclear envelope, successive layers reveal the architecture of three inner ring spokes of the human NPC. Corresponding schematics illustrate linker paths between binding sites on scaffold surfaces (colored circles). (C) NUP53RRM domains homodimerize between spokes to link cytoplasmic peripheral with nuclear equatorial, and conversely cytoplasmic equatorial with nuclear peripheral copies of NUP155. (D) NUP53RRM domains link cytoplasmic peripheral with nuclear equatorial, and conversely cytoplasmic equatorial with nuclear peripheral copies of NUP93SOL. (E) NUP205 and NUP188 bind to the equatorial and peripheral NUP93R2, respectively. NUP98 connects NUP205 and equatorial NUP155. NUP53 connects NUP205 and cross‑spoke peripheral NUP93SOL (F) CNT is recruited by NUP93R1 and positioned by NUP93R2 binding to NUP188 and NUP205. (G) Closeup views of inner ring modules assembled around NUP188 and NUP205 scaffold hubs, and their superposition. Dashed lines indicate unstructured linker nup segments and FG‑repeat regions.

Fig. S1. Crystallographic analysis of Nup192ΔHEAD-bound Nic96 fragments. (A) Domain structures of Nup192 and Nic96. Black lines indicate the co-crystallized fragments. (B) Isomesh representation contoured at 5 σ of anomalous difference Fourier maps calculated from X-ray diffraction data collected at selenium anomalous peak wavelengths for Nup192ΔHead•Nic96R2 (orange) and Nup192 ΔHead•Nic96R2 A289M (magenta) SeMet‑derivatized co-crystals and superposed with the cartoon representation of the Nup192ΔHead•Nic96187-301 structure, validating the sequence register of the Nic96R2 peptide. (C) Comparison of electron density maps derived from Nup192ΔHead•Nic96187-301 and Nup192ΔHead•Nic96R2 co-crystals demonstrating that Nic96 residues 187-239 are not resolved in the electron density.

Fig. S2. ITC analysis of the Nup192-Nic96R2 interaction. Representative baseline-corrected ITC experiments of (A) wildtype SUMO-Nic96R2 titrated against wildtype Nup192, (B) SUMO‑Nic96R2 FFF mutant titrated against wildtype Nup192, (C) wildtype SUMO-Nic96R2 titrated against Nup192 LAF mutant, and (D) wildtype SUMO-Nic96187-301 titrated against wildtype Nup192. The least-squares fit thermodynamic parameters for a single binding site model are shown in the inset boxes. Overall ITC experimental conditions and results of experiments conducted in triplicate are reported in Table S5.

Fig. S3. Single particle cryo-EM analysis of the Nup192•Nic96R2 complex. (A) Representative motion corrected cryo-electron micrograph. (B) 2D class averages representative of different particle orientations. (C) Summary of single particle data processing workflow with indicated number of micrographs or particles involved at each step. (D) Global mean and directional gold‑standard masked Fourier Shell Correlation (FSC) curves along the x, y and z axes estimated with 3DFSC (116), as well as the unmasked model-to-map FSC curve, with the 0.143 cutoff indicated. The 1 standard deviation of directional FSC resolutions is indicated for the global mean FSC (dashed lines), resulting in an anisotropic 3.5‑4.0 Å directional resolution range at the 0.143 cutoff. A histogram of resolutions evenly sampled from directional FSCs at the 0.143 cutoff illustrates the anisotropic spread of the data. (E) Three-dimensional angular distribution plot of particles. (F) Isosurface representation of the Nup192•Nic96R2 cryo-EM density (EMD‑24056) colored according to protein chain identity (left) or local resolution estimation (right). (G) Isomesh representation of representative sharpened cryo-electron microscopy (cryo-EM) density contoured at a 10 σ cutoff level and ball-and-stick representation of corresponding atomic coordinates.

Fig. S4. Comparison of crystal composite and single particle cryo-EM structures of Nup192•Nic96R2. (A) Cartoon representation of the crystal structures of Nup192 ΔHead•Nic96187-301 (blue and pale green) and the previously solved Nup192NTD structure (yellow, PDB ID 4KNH) (41), and their superposition are shown. The Nup192•Nic96R2 composite crystal structure was generated by aligning and merging residues in common between the two structures. (B) Cartoon representation of the Nup192•Nic96R2 composite crystal structure (blue and pale green) and the 3.8 Å Nup192•Nic96R2 single particle cryo-EM structure (light cyan and pink), shown from three points of view to illustrate the conformational difference between the two structures. Curved arrows indicate the displacement of the Tower domain induced by the conformational change.

Fig. S5. Perturbation of the Nup192-Nic96R2 interaction by Nic96 point mutants. SEC and SDS‑PAGE analysis corresponding to Fig. 2C. SEC profiles of wildtype Nup192 alone (cyan) and Nup192 preincubated with wildtype SUMO-Nic96R2 (dark blue) are shown. SEC profiles of SUMO‑Nic96R2 mutants preincubated with Nup192 are colored according to the measured effect: no effect (green, +++), weak effect (yellow, ++), moderate effect (orange, +), and abolished binding (red, -). The inset box shows a closeup view of elution profile shifts at the peak fractions. The gray bar indicates fractions that were resolved on SDS-PAGE gels and visualized by Coomassie staining. All SEC profiles were obtained using a Superdex 200 10/300 GL column. The results are summarized in the table (top right).

Fig. S6. Perturbation of the Nup192-Nic96R2 interaction by Nup192 point mutants. SEC and SDS-PAGE analysis corresponding to Fig. 2D. SEC profiles of wildtype Nup192 alone (cyan) and Nup192 preincubated with wildtype SUMO-Nic96R2 (dark blue) are shown. SEC profiles of Nup192 mutants preincubated with SUMO-Nic96R2 are colored according to the measured effect: no effect (green, +++) and abolished binding (red, -). The inset box shows a closeup view of elution profile shifts at the peak fractions. The gray bar indicates fractions that were resolved on SDS-PAGE gels and visualized by Coomassie staining. All SEC profiles were obtained using a Superdex 200 10/300 GL column. The results are summarized in the table (top left).

Fig. S7. SEC-MALS analysis of the Nup192-Nic96R2 interaction. SEC-MALS and SDS-PAGE analysis corresponding to Fig. 2F. SEC-MALS profiles of nups are shown individually (red and blue) and after their preincubation (green) for: (A) wildtype SUMO-Nic96R2 binding to wildtype Nup192, (B) the SUMO-Nic96R2 FFF mutant binding to wildtype Nup192, (C) wildtype SUMO‑Nic96R2 binding to the Nup192 LAF mutant. Measured molecular masses are indicated for the peak fractions and corresponding theoretical molecular masses are reported in parenthesis. Gray bars indicate fractions that were resolved on SDS-PAGE gels and visualized by Coomassie staining. All SEC profiles were obtained using a Superdex 200 Increase 10/300 GL column. All SEC-MALS results are summarized in Table S4.

Fig. S8. ITC analysis of the Nup188-Nic96R2 interaction. Representative baseline-corrected ITC experiments for (A) wildtype SUMO-Nic96R2 titrated against wildtype Nup188, (B) SUMO‑Nic96R2 FFF mutant titrated against wildtype Nup188, (C) wildtype SUMO-Nic96R2 titrated against Nup188 FLV mutant, and (D) wildtype SUMO-Nic96187-301 titrated against wildtype Nup188. The least-squares fit thermodynamic parameters for a single binding site model are shown in the inset boxes. Overall ITC experimental conditions and results of experiments conducted in triplicate are reported in Table S5.

Fig. S9. Crystallographic analysis of the Nup188•Nic96R2 complex. (A) Domain structures of Nup188 and Nic96. Black lines indicate the fragments used for crystallization. (B) Cartoon representation of the 2.8 Å crystal structure of Nup188NTD. (C) Cartoon representations of the 2.8 Å Nup188NTD (light cyan) and previously solved 3.4 Å Nup188Tail (teal; PDB ID 5CWU) (28) crystal structures (left) used as starting models for the 4.4 Å Nup188•Nic96R2 crystal structure (light purple and pale green), and their superposition (right). (D) Isomesh representation contoured at 5 σ of anomalous difference Fourier map (orange) calculated from X-ray diffraction data collected at the selenium anomalous peak wavelengths for a Nup188•Nic96R2 SeMet‑derivatized co-crystal and superposed with the cartoon representation of the Nup188•Nic96R2 structure, validating the sequence register of Nic96R2 and the previously unmodeled Nup188 region.

Fig. S10. Perturbation of the Nup188-Nic96R2 interaction by Nic96 point mutants. SEC and SDS‑PAGE analysis corresponding to Fig. 2H. SEC profiles of wildtype Nup188 alone (cyan) and Nup188 preincubated with wildtype SUMO-Nic96R2 (dark blue) are shown. SEC profiles of SUMO‑Nic96R2 mutants preincubated with Nup188 are colored according to the measured effect: no effect (green, +++), moderate effect (orange, +), and abolished binding (red, -). The inset box shows a closeup view of elution profile shifts at the peak fractions. The gray bar indicates fractions that were resolved on SDS-PAGE gels and visualized by Coomassie staining. All SEC profiles were obtained using a Superdex 200 10/300 GL column. The results are summarized in the table (top right).

Fig. S11. Perturbation of the Nup188-Nic96R2 interaction by Nup188 point mutants. SEC and SDS‑PAGE analysis corresponding to Fig. 2I. SEC profiles of wildtype Nup188 alone (cyan) and Nup188 preincubated with wildtype SUMO-Nic96R2 (dark blue) are shown. SEC profiles of Nup188 mutants preincubated with SUMO-Nic96R2 are colored according to the measured effect: no effect (green, +++), weak effect (yellow, ++), moderate effect (orange, +), and abolished binding (red, -). The inset box shows a closeup view of elution profile shifts at the peak fractions. The gray bar indicates fractions that were resolved on SDS-PAGE gels and visualized by Coomassie staining. All SEC profiles were obtained using a Superdex 200 10/300 GL column. The results are summarized in the table (top left).

Fig. S12. SEC-MALS analysis of the Nup188-Nic96R2 interaction. SEC-MALS and SDS‑PAGE analysis corresponding to Fig. 2K. SEC-MALS profiles of nups are shown individually (red and blue) and after their preincubation (green) for: (A) wildtype SUMO-Nic96R2 binding to wildtype Nup188, (B) the SUMO-Nic96R2 FFF mutant binding to wildtype Nup188, (C) wildtype SUMO‑Nic96R2 binding to the Nup188 FLV mutant. Measured molecular masses are indicated for the peak fractions and corresponding theoretical molecular masses are reported in parenthesis. Gray bars indicate fractions that were resolved on SDS-PAGE gels and visualized by Coomassie staining. All SEC profiles were obtained using a Superdex 200 Increase 10/300 GL column. All SEC-MALS results are summarized in Table S4.

Fig. S13. 5-Ala scanning mutagenesis of Nup145N sequence for residues necessary for Nup192 binding. SEC and SDS-PAGE analysis corresponding to Fig. 3A. (A) Domain structure of Nup145N. Mutations were introduced into the Nup145N construct indicated by the black line above the domain structure. The positions of the 5-Ala mutated residues are indicated above the primary Nup145N sequence with boxes colored according to the measured effect on Nup192 binding: no effect (green), moderate effect (yellow), and severe effect (red). Asterisks above boxes indicate SEC experiments for which representative SDS-PAGE gels are shown in (B). (B) SEC profiles of wildtype Nup192 alone (cyan) and Nup192 preincubated with wildtype Nup145N (dark blue) are shown. SEC profiles of Nup145N 5-Ala mutants preincubated with Nup192 are colored according to the measured effect, as in (A). The inset box shows a closeup view of elution profile shifts at the peak fractions. The gray bar indicates fractions that were resolved on SDS‑PAGE gels and visualized by Coomassie staining. All SEC experiments were analyzed by SDS-PAGE, but only experiments representative of the three observed effect levels are shown for brevity. All SEC profiles were obtained using a Superdex 200 10/300 GL column.

Fig. S14. Mapping of the minimal Nup145N region sufficient for Nup192 binding. SEC and SDS-PAGE analysis corresponding to Fig. 3B. (A) Domain structure of Nup145N with gray bars indicating truncation construct boundaries analyzed. The Nup145NR1 peptide is indicated by a red bar. The binding of SUMO-Nup145N peptides to Nup192, as assessed by SEC and SDS-PAGE, is summarized and colored according to the measured effect: no effect (green, +++), weak effect (yellow, ++), moderate effect (orange, +), and abolished binding (red, -). (B) SEC profile of wildtype Nup192 (dark blue) is shown as reference. SEC profiles of Nup192 preincubated with SUMO-Nup145N peptides are colored according to the measured effect, as in (A). Dashed lines indicate the peak elution volumes across the offset dimension. The gray bar indicates fractions that were resolved on SDS-PAGE gels and visualized by Coomassie staining. All SEC profiles were obtained using a Superdex 200 10/300 GL column. Asterisks indicate degradation products.

Fig. S15. Isothermal titration calorimetry analysis of Nup192-Nup145N binding. Representative baseline‑corrected ITC experiments for Nup145N, Nup145N KKRMYKLRKR, SUMO-Nup145NR1, and SUMO-Nup145NR1 KKRMYKLRKR titrated against a preformed complex of Nic96R2, Nup53, and either wildtype Nup192 or mutant Nup192 LIFH. The least-squares fit thermodynamic parameters for a single binding site model are shown in box beneath ITC plot. Overall ITC experimental conditions and results of experiments conducted in triplicate are reported in Table S5.

Fig. S16. Single particle cryo-EM analysis of the Nup192•Nic96R2•Nup145NR1•Nup53R1 complex. (A) Representative motion corrected cryo-electron micrograph. (B) 2D class averages representative of different particle orientations. (C) Summary of single particle data processing workflow with indicated number of micrographs or particles involved at each step. Dashed boxes indicate classes that were selected for subsequent steps in the workflow. (D) Global mean and directional gold-standard masked Fourier Shell Correlation (FSC) curves along the x, y and z axes estimated with 3DFSC (116), as well as the unmasked model-to-map FSC curve, with the 0.143 cutoff indicated. The 1 standard deviation of directional FSC resolutions is indicated for the global mean FSC (dashed lines), resulting in an anisotropic 3.1‑3.6 Å directional resolution range at the 0.143 cutoff. A histogram of resolutions evenly sampled from directional FSCs at the 0.143 cutoff illustrates the anisotropic spread of the data. (E) Three-dimensional angular distribution plot of particles. (F) Isosurface representation of the Nup192•Nic96R2•Nup145NR1•Nup53R1 cryo-EM density (EMD-24057) colored according to protein chain identity (left) or local resolution estimation (right). (G‑I) Isomesh representation of representative sharpened cryo-EM density contoured (G) at a 10σ cutoff level and stick representation of the corresponding Nup192 polypeptide chain of the Nup192•Nic96R2•Nup145NR1•Nup53R1 structure, (H) contoured at 6σ cutoff level around the Nup145NR1 peptide shown in stick representation (cyan), and (I) contoured at 7σ cutoff level around the Nup53R1 peptide shown in stick representation (magenta).

Fig. S17. Nup145NR1 and Nup53R1 binding to Nup192•Nic96R2 does not induce conformational changes. Cartoon representation of the 3.8 Å Nup192•Nic96R2 (pale cyan and pink) and the 3.2 Å Nup192•Nic96R2•Nup145NR1•Nup53R1 (blue, pale green, cyan, and magenta) single particle cryo-EM structures, and their superposition, viewed from three sides.

Fig. S18. Perturbation of the Nup192-Nup145N interaction by Nup145N point mutants. SEC and SDS‑PAGE analysis corresponding to Fig. 3F. (A) Domain structure of Nup145N. Mutations were introduced into the Nup145N construct indicated by the black line above the domain structure. Primary sequence containing the mutated residues is shown beneath the domain structure. (B-C) SEC profiles of wildtype Nup192 alone (blue) and Nup192 preincubated with wildtype Nup145N (dark blue) are shown. SEC profiles of Nup192 preincubated with Nup145N mutants are colored according to the measured effect and the results summarized in the adjacent table: no effect (green, +++), weak effect (yellow, ++), moderate effect (orange, +), and abolished binding (red, -). The inset boxes show a closeup view of elution profile shifts at the peak fractions. Asterisks indicate SEC experiments for which representative SDS-PAGE gels are shown in (D). Gray bars indicate fractions that were resolved on SDS-PAGE gels and visualized by Coomassie staining. All SEC profiles were obtained using a Superdex 200 10/300 GL column. (D) A selection of SDS-PAGE gels representative of the effect levels observed in (B-C) is shown for brevity.

Fig. S19. Reconstitution of the Nup192•SUMO-Nic96R2•Nup53 complex. SEC-MALS profiles of nups are shown individually (red and blue) and after their preincubation (green) for (A) the complexation of Nup192 and SUMO-Nic96R2, and (B) subsequent complexation of Nup53 with Nup192•SUMO-Nic96R2. Measured molecular masses are indicated for the peak fractions and corresponding theoretical molecular masses are reported in parenthesis. Gray bars indicate fractions that were resolved on SDS-PAGE gels and visualized by Coomassie staining. All SEC profiles were obtained using a Superdex 200 Increase 10/300 GL column. All SEC-MALS results are summarized in Table S4.

Fig. S20. SEC-MALS analysis of the Nup192-Nup145N interaction. SEC-MALS and SDS‑PAGE analysis corresponding to Fig. 3H. (A-F) SEC-MALS profiles of nups are shown individually (red and blue) and after their preincubation (green) for either wilt-type Nup192•SUMO‑Nic96R2•Nup53 or mutant Nup192 LIFH•SUMO-Nic96R2•Nup53 binding to either wildtype or KKRMYKLRKR mutant of Nup145N and SUMO-Nup145NR1. Measured molecular masses are indicated for the peak fractions and corresponding theoretical molecular masses are reported in parenthesis. Gray bars indicate fractions that were resolved on SDS-PAGE gels and visualized by Coomassie staining. All SEC profiles were obtained using a Superdex 200 Increase 10/300 GL column. All SEC-MALS results are summarized in Table S4.

Fig. S21. Perturbation of the Nup192-Nup145N interaction by Nup192 point mutants. (A‑F) SEC and SDS-PAGE analysis corresponding to Fig. 3F. SEC profiles are shown for nups or nup complexes individually (red and blue) and after their preincubation (green) for SUMO-Nup145NR1 binding to pre-assembled complexes of SUMO-Nic96R2, Nup53, and wildtype or mutant Nup192. Gray bars indicate fractions that were resolved on SDS-PAGE gels and visualized by Coomassie staining. All SEC profiles were obtained using a Superdex 200 10/300 GL column. Asterisks indicate SUMO contaminant.

Fig. S22. 5-Ala scanning mutagenesis of Nup145N sequence for residues necessary for Nup188NTD binding. SEC and SDS-PAGE analysis corresponding to Fig. 4A. (A) Domain structure of Nup145N. Mutations were introduced into the Nup145N construct indicated by the black line above the domain structure. The positions of the 5-Ala mutated residues are indicated above the primary Nup145N sequence with boxes colored according to the measured effect on Nup188NTD binding: no effect (green), moderate effect (yellow). Asterisks above boxes indicate SEC experiments for which representative SDS‑PAGE gels are shown in (B). (B) SEC profiles of wildtype Nup188NTD alone (light purple) and Nup188NTD preincubated with wildtype Nup145N (dark blue) are shown. SEC profiles of Nup145N 5-Ala mutants preincubated with Nup188NTD are colored according to the measured effect, as in (A). The inset box shows a closeup view of elution profile shifts at the peak fractions. The gray bar indicates fractions that were resolved on SDS‑PAGE gels and visualized by Coomassie staining. All SEC experiments were analyzed by SDS‑PAGE, but only experiments representative of the three observed effect levels are shown for brevity. All SEC profiles were obtained using a Superdex 200 10/300 GL columnt.

Fig. S23. Mapping of the minimal Nup145N region sufficient for Nup188NTD binding. SEC and SDS‑PAGE analysis corresponding to Fig. 4B. (A) Domain structure of Nup145N with gray bars indicating truncation construct boundaries analyzed. The Nup145NR2 peptide is indicated by a red bar. The binding of SUMO-Nup145N peptides to Nup188NTD, as assessed by SEC and SDS‑PAGE, is summarized and colored according to the measured effect: no effect (green, +++), weak effect (yellow, ++), moderate effect (orange, +), and abolished binding (red, -). (B) SEC profile of wildtype Nup188NTD (dark blue) is shown as reference. SEC profiles of Nup188NTD preincubated with SUMO-Nup145N peptides are offset for clarity and colored according to the measured effect, as in (A). Dashed lines indicate the peak elution volumes across the offset dimension. The gray bar indicates fractions that were resolved on SDS-PAGE gels and visualized by Coomassie staining. All SEC profiles were obtained using a Superdex 200 10/300 GL column. Asterisks indicate degradation products.

Fig. S24. Nup53 is not part of the Nup188 symmetric core complex. SEC-MALS profiles of nups are shown individually (red and blue) and after their preincubation (green) for (A) the complexation of Nup188•SUMO-Nic96R2 and Nup145N, and (B) subsequent binding experiment between Nup188•SUMO-Nic96R2•Nup145N and Nup53. Measured molecular masses are indicated for the peak fractions and corresponding theoretical molecular masses are reported in parenthesis. Gray bars indicate fractions that were resolved on SDS-PAGE gels and visualized by Coomassie staining. All SEC profiles were obtained using a Superdex 200 Increase 10/300 GL column. All SEC-MALS results are summarized in Table S4.

Fig. S25. Single particle cryo-EM analysis of the Nup188•Nic96R2•Nup145NR2 complex. (A) Representative motion corrected cryo-electron micrograph. (B) 2D class averages representative of different particle orientations. (C) Summary of single particle data processing workflow with indicated number of micrographs or particles involved at each step. Dashed boxes indicate classes that were selected for subsequent steps in the workflow. (D-I) Resolution and particle orientation estimates for (D‑F) the Nup188•Nic96R2•Nup145NR2 and (G‑I) Nup188•Nic96R2•Nup145NR2 reconstructions. (D, G) Global mean and directional gold-standard masked Fourier Shell Correlation (FSC) curves along the x, y and z axes estimated with 3DFSC (116), as well as the unmasked model-to-map FSC curves, with the 0.143 cutoff indicated. The 1 standard deviation of directional FSC resolutions is indicated for the global mean FSC (dashed lines), resulting in anisotropic (D) 2.3-2.5 Å and (G) 2.7-2.9 Å directional resolution ranges at the 0.143 cutoff. A histogram of resolutions evenly sampled from directional FSCs at the 0.143 cutoff illustrates the anisotropic spread of the data. (E, H) Three-dimensional angular distribution plot of particles. (F, I) Isosurface representation of the cryo-EM density colored according to protein chain identity (left) or local resolution estimation (right). (J) Isomesh representation of representative sharpened cryo‑EM density of the 2.4 Å Nup188•Nic96R2 map (EMD‑24058) contoured at a 15σ cutoff level and ball-and-stick representation of the corresponding Nup188 polypeptide chain of the atomic Nup188•Nic96R2 model (light purple). (K) Isomesh representation of representative sharpened cryo-EM density of the 2.8 Å Nup188•Nic96R2•Nup145NR2 map (EMD‑24059) contoured at an 8 σ cutoff level around the Nup145NR2 peptide shown in stick representation (cyan).

Fig. S26. Nup188•Nic96R2 crystal structure and single particle cryo-EM structures of Nup188•Nic96R2 and Nup188•Nic96R2•Nup145NR2 do not present major conformational differences. Cartoon representation of the 4.4 Å Nup188•Nic96R2 crystal structure (yellow and blue), the 2.4 Å Nup188•Nic96R2 (pale cyan and pink) and the 2.8 Å Nup188•Nic96R2•Nup145NR2 (light purple, pale green, cyan) single particle cryo-EM structures, and their superposition, viewed from three sides.

Fig. S27. Perturbation of the Nup188NTD-Nup145N interaction by Nup145N point mutants. SEC and SDS‑PAGE analysis corresponding to Fig. 4D. (A) Domain structure of Nup145N. Mutations were introduced into the Nup145N construct indicated by the black line above the domain structure. Primary sequence containing the mutated residues is shown beneath the domain structure. (B) SEC profiles of wildtype Nup188NTD alone (light purple) and Nup188NTD preincubated with wildtype Nup145N (dark blue) are shown. SEC profiles of Nup188NTD preincubated with Nup145N mutants are colored according to the measured effect and the results summarized in the adjacent table: no effect (green, +++), weak effect (yellow, ++). The inset boxes show a closeup view of elution profile shifts at the peak fractions. Asterisks indicate SEC experiments for which SDS-PAGE gels representative of the effects observed are shown. Gray bars indicate fractions that were resolved on SDS-PAGE gels and visualized by Coomassie staining. All SEC profiles were obtained using a Superdex 200 10/300 GL column.

Fig. S28. Isothermal titration calorimetry analysis of Nup188-Nup145N binding. Representative baseline-corrected ITC experiments for Nup145N, Nup145N EDSILF, SUMO‑Nup145NR2, and SUMO‑Nup145NR2 EDSILF titrated against a preformed complex of Nic96R2and either wildtype Nup188 or mutant Nup188 HHMI. The least-squares fit thermodynamic parameters for a single binding site model are shown in box beneath ITC plot. Overall ITC experimental conditions and results of experiments conducted in triplicate are reported in Table S5.

Fig. S29. Perturbation of the Nup188NTD-Nup145N interaction by Nup188NTD point mutants. (A‑E) SEC and SDS-PAGE analysis corresponding to Fig. 4D. SEC profiles are shown for nups individually (red and blue) and after their preincubation (green) for SUMO-Nup145NR2 binding to wildtype or mutant Nup188NTD. Gray bars indicate fractions that were resolved on SDS-PAGE gels and visualized by Coomassie staining. All SEC profiles were obtained using a Superdex 200 10/300 GL column.

Fig. S30. SEC-MALS analysis of the Nup188-Nup145N interaction. SEC-MALS and SDS‑PAGE analysis corresponding to Fig. 4F. (A‑F) SEC-MALS profiles of nups or nup complexes are shown individually (red and blue) and after their preincubation (green) for either wildtype Nup188•SUMO-Nic96R2 or mutant Nup188 LIFH•SUMO-Nic96R2 binding to either wildtype or EDSILF mutant of Nup145N and SUMO-Nup145NR2. Measured molecular masses are indicated for the peak fractions and corresponding theoretical molecular masses are reported in parenthesis. Gray bars indicate fractions that were resolved on SDS-PAGE gels and visualized by Coomassie staining. All SEC profiles were obtained using a Superdex 200 Increase 10/300 GL column. All SEC-MALS results are summarized in Table S4.

Fig. 31. Structural comparison of the Nup192 and Nup188 linker-scaffold complexes. (A) Cartoon representations of the Nup192•Nic96R2•Nup53R1•Nup145NR1 and Nup188•Nic96R2•Nup145NR2 single particle cryo-EM structures, and their superposition. The Nup192 and Nup188 Tower domains (red and light red) and the Nup188 SH3-like domain (gold) are highlighted. In boxes, isolated and magnified views of conformations of Nic96R2 bound to Nup192 and Nup188 are shown to illustrate the Nic96R2 structural polymorphism. The two Nic96R2 conformations were superposed based on a common α-helical region between residues 263-284. Residues found to be involved in Nic96R2 binding to Nup192 and Nup188 (red) and residue M263 (orange), uniquely identifiable in anomalous difference Fourier maps derived from SeMet‑derivatized Nup192ΔHEAD•Nic96R2 and Nup188•Nic96R2 crystals, are shown in ball‑and‑stick representation. (B) Schematic of the Nup192 and Nup188 fold architectures. Rounded rectangles and arrows represent α-helices and β-strands, respectively. (C) Schematic of the secondary structure that Nic96R2 assumes in binding Nup192 and Nup188. Green boxes and black lines above the primary sequence indicate α‑helices and loops, respectively. Orange and red circles indicate residues highlighted in (A).

Fig. S32. Docking of the Nup192 and Nup188 complex structures into an ~25 Å in situ cryo‑ET map of the Saccharomyces cerevisiae NPC. Resolution-matched simulated cryo-EM densities of side-chain resolution structures were quantitatively docked into the ~25 Å sub‑tomogram averaged cryo-electron tomography (cryo-ET) map of the in situ imaged S. cerevisiae NPC (EMD-10198) (36). The placement of unique solutions in a single spoke is shown (top left). Arrows and numbers indicate the accepted solutions and their corresponding rank. Closeup views of the accepted solutions are shown on the right, with the rank indicated on the top left corner of the box. Isosurface representation of the nuclear envelope and the NPC are colored in dark gray and white, respectively. The docked structures are displayed in cartoon representation. Rug plots (blue) and histograms (light blue) of Pearson correlation scores and derived Fisher z scores fit with a normalized Gaussian curve (black) from a global search with 1 million random initial placements are shown (bottom left and middle). Arrows and numbers indicate accepted solutions and their rank, respectively. A tabular summary of the accepted solutions fitting statistics, along with one-tailed p-values calculated from the Fisher z score distribution, is shown (bottom right). The quantitative docking analysis of (A) the Nup192•Nic96R2•Nup145NR1•Nup53R1 single particle cryo-EM structure, (B) the alternative‑conformation Nup192•Nic96R2 composite crystal structure with Nup145NR1 and Nup53R1 peptides included by superposition with the single particle cryo-EM structure, and (C) the Nup188•Nic96R2•Nup145NR1 single particle cryo-EM structure, collocates the Nup192 and Nup188 complexes to the equatorial and peripheral positions of the inner ring, respectively.

Fig. S33. Comparison of the Nup192 and Nup188 complex placement in situ cryo-ET map of the S. cerevisiae NPC. Question mark-shaped cryo-ET densities carved out from (A) cytoplasmic peripheral, (B) nuclear peripheral, (C) cytoplasmic equatorial, and (D) nuclear equatorial positions of the inner ring of the 25 Å sub-tomogram averaged cryo-ET map of the in situ imaged S. cerevisiae NPC (EMD-10198) (36), where either Nup192•Nic96R2•Nup145NR1•Nup53R1 or Nup188•Nic96R2•Nup145NR1 were quantitatively docked, as indicated in the boxes (top). To compare the goodness of fit by visual inspection, two isosurface represented views of the carved density with cartoon representations of the quantitatively docked composite crystal Nup192, single particle cryo‑EM Nup192, or single particle cryo‑EM Nup188 complex structures are shown. At the peripheral positions, a tentative placement of the two conformations of the Nup192 complex (single particle cryo‑EM and composite crystal structures, respectively) results in unexplained excess density that can be accounted for by the SH3-like domain of Nup188 if the Nup188 complex (single particle cryo-EM structure) is placed in the same position. Unlike the shorter Nup188 Tower, the longer Nup192 Tower exceeds the cryo‑ET density when tentatively placed at the peripheral positions. On the contrary, the longer Tower is accommodated by the cryo-ET density at the equatorial position, but the SH3-like domain of the tentatively placed Nup188 exceeds the cryo-ET density at the equatorial position. Overall, the Nup188 α-helical solenoid presents a narrower superhelical twist compared to Nup192, which leaves unexplained density if tentatively fit at the equatorial positions. Conversely, the larger superhelical twist of the Nup192 α-helical solenoid exceeds the cryo-ET density if tentatively fit at the peripheral positions.

Fig. S34. Composite structure of the S. cerevisiae NPC. Linker-scaffold structures were docked into an ~25 Å sub-tomogram averaged cryo-ET map of the in situ imaged S. cerevisiae NPC (EMD-10198) to obtain a composite structure of the S. cerevisiae NPC complete with surface-bound linker nup segments. The crystal composite Nup192•Nic96R2•Nup145NR1•Nup53R1 structure and the single particle cryo-EM Nup188•Nic96R2•Nup145NR2 structure were quantitatively docked (fig. S32). Crystal structures of linker-scaffold complexes Nup170•Nup145NR3•Nup53R3 (full-length crystal composite structures) (35) and CNT•Nic96R1 (PDB ID 5CWS) (28) were first aligned to the composite structure from (36) and then locally rigid‑body fit into the cryo-ET map. The structures of the coat nup complex (CNC) and the scNup82•scNup159•scNsp1 complex were carried over from the composite structure from (36). (A) View from above the cytoplasmic face and (B) a cross-sectional view from the central transport channel. Isosurface representations of the nuclear envelope and the S. cerevisiae NPC are colored in dark gray and white, respectively. The docked structures are displayed in cartoon representation.

Fig. S35. Linker topology analysis of the composite structure of the S. cerevisiae NPC. (A) Schematic representation of a S. cerevisiae NPC spoke layers with the positional nomenclature indicated for each nup. Inner ring positional acronyms: cytoplasmic equatorial (CE), cytoplasmic proximal (CP), cytoplasmic bridge (CB), nuclear equatorial (NE), nuclear proximal (NP), and nuclear bridge (NB). Outer ring positional acronyms (grey shading): cytoplasmic (C), nuclear (N), and Nup82•Nup159•Nsp1 complex proximal (CFX) and distal (CFD). (B) Systematic measurement of Euclidean distances (Å) between linker-scaffold regions at various positions. The cross-spoke (x-sp.) distances reported are of the shortest connection to a particular position on an adjacent spoke, either in the clockwise or counterclockwise direction from the reference spoke. The shortest Euclidean distances consistent with the most parsimonious architecture of the linker‑scaffold are indicated in blue and boldface type. Distances incompatible with the maximum physically possible length of the extended linker (red) and longer (orange) or shorter (green) than the calculated root-mean square (r.m.s) end-to-end length of a linear polymer chain according to the Gaussian chain model developed by Flory, with a characteristic ratio parameter of ~3, typical for unstructured polypeptide linkers (134, 135). (*) Because scNup53/59RRM are not resolved in the dilated in situ cryo-ET map of the S. cerevisiae NPC, we report distances between the R2 and R3 regions but expect that the RRM homo-dimerization would further constrain the connections by extending the scNup53/59 R2-RRM and RRM-R3 linkers in between the inner ring spokes. (†) Because of the length of the linker between APD/CTD domains and the rest of the scNup100/116/145N domains, the assignment of connections is tentative. (‡) Linker lengths correspond to the number of residues in the scNup116 linkers; scNup100 paralog linker lengths are of similar length.

Fig. S36. Shortest-distance analysis of linker network in the S. cerevisiae NPC. Spatial distances in the composite structure of the S. cerevisiae NPC were measured between two copies of linker nup segments that are adjacent in their primary sequence. (A‑K) In the box on the left, linker nup segments and the scaffold surface that they are bound to are shown in cartoon representation and highlighted in color. The remainder of nups are colored in white. The nuclear envelope is rendered as dark gray isosurface. The straight-line linkage between the two closest copies of linker nup segments adjacent in their primary sequence is indicated by a solid red line. The measured Euclidean distance is reported in red. On the right, a schematic representation of the scaffold nup binding sites (highlighted in color) and the connecting linker nup segment (red), illustrates the topology of the linkage within the context of an entire S. cerevisiae NPC spoke (remainder of nups shown in gray).

Fig. S37. Architecture of the S. cerevisiae NPC linker-scaffold. Composite structure of the S. cerevisiae NPC generated by docking linker-scaffold structures into an ~25 Å sub-tomogram averaged cryo‑electron tomography (cryo-ET) map of the in situ imaged S. cerevisiae NPC (EMD‑10198) (36), viewed from (A) above the cytoplasmic face, and (B) a cross-section of the central transport channel. The nuclear envelope is rendered as a gray isosurface and the docked structures are represented as cartoons. (C-F) Cartoon representation of a single spoke’s layered architecture and schematic illustrating linker paths. Linker binding sites on scaffold nup surfaces are schematically indicated with colored circles. (C) scNup145N simultaneously binds the coat nup complex (CNC) component scNup145C in the outer rings and scNup157/170 in the inner ring. (D) scNup53/59 links equatorial Nic96SOL with peripheral scNup157/170 and peripheral scNic96SOL with equatorial scNup157/170, across the spoke midplane. (E) scNup192 and scNup188 bind to equatorial and peripheral scNic96R2, respectively. scNup100/116 link scNup192 and the equatorial  scNup157/170.scNup53/59 links scNup192 and the peripheral scNic96SOL of adjacent spokes. (F) The scCNT is recruited by scNic96R1. scNup100/116 simultaneously bind to the cytoplasmic asymmetric scNup82•scNup159•scNsp1 complex and to scNup188 or scNup157/170 in the inner ring.

Fig. S38. Multispecies sequence alignment of Nup145N. Sequences from twelve diverse species, including three S. cerevisiae paralogs Nup100, Nup116, and Nup145N, were aligned and colored by sequence similarity according to the BLOSUM62 matrix from white (less than 55 % similarity), to yellow (55 % similarity), to red (100 % identity). Numbering below alignment is relative to the C. thermophilum sequence. Regions of the sequence are annotated above the alignment: FG/GLFG repeats (pale cyan), Gle2-binding sequence – GLEBS (purple), R1 – Nup192 binding region (blue), R2 – Nup188 binding region (light purple), R3 – Nup170 binding region (orange), autoproteolytic domain – APD (green). Residues found by mutational analysis to affect Nup192 binding (Fig. 3, figs. S15, S18 to S20), Nup188 binding (Fig. 4, figs. S27 to S30), or Nup170 binding (35) are indicated by red circles above the alignment.

Fig. S39. Generation of the nup100Δnup116Δnup145Δ S. cerevisiae knockout strain. (A) Domain structure of the S. cerevisiae gene products: scNup145 (and its autoproteolytic products scNup145N and Nup145C), the scNup145N paralogs scNup100 and scNup116, and the scNup116-scNup145C chimera construct. (B) Complementation of double knockout strains by co‑transformation with a combination of plasmids expressing scNup100, scNup116,  scNup145N‑scNup145C, or the scNup116-scNup145C chimera. The scNup116-scNup145C chimera is a single construct that complements all double knockout combinations. (C) Gene knockout strategy employed in the generation of the nup100Δnup116Δnup145Δ strain. The scNup116-scNup145C chimera complements sequential knockouts of NUP100, NUP116, and NUP145 by homologous recombination with selection marker cassettes. The inducible expression of Cre results in the efficient excision of the loxP-natNT2-loxP cassette, allowing for a second round of selection using the natNT2 cassette. (D) The nup100Δnup116Δnup145Δ strain is rescued by the scNup116-scNup145C chimera construct but presents a slow growth phenotype. Growth can be restored to wildtype rates by co-transforming the strain with plasmids expressing scNup145C and scNup116. (E) Domain structure of the S. cerevisiae scNup145N paralogs and artificial constructs. scNup116 is the only paralog that possesses a GLEBS motif. To test if the GLEBS motif is the sole determinant of viable nup100Δnup116Δnup145Δ complementation by plasmids expressing scNup145N paralogs, constructs with the scNup116 GLEBS motif sequence inserted into the scNup100 and scNup145N FG-repeat region, as well as a scNup145N construct in which the FG-repeat region is replaced with the scNup116 FG-GLEBS region, were generated. (F) Viability analysis of the nup100Δnup116Δnup145Δ strain carrying a plasmid expressing scNup145C, after transformation with indicated scNup145N paralog constructs and selection of ten-fold serial dilution series spotted onto either SDC-LEU and SDC+5-FOA plates (shuffling).

Fig. S40. In vivo analysis of scNup116 functional regions and validation of interactions with scaffold proteins. Supporting data for analysis presented in Fig. 5. (A) Domain structure of scNup116 variants introduced into the nup100Δnup116Δnup145Δ S. cerevisiae strain, with colored squares indicating the functional region targeted by each mutation: GLEBS (pale cyan); FG repeats (dark gray); R1m, Nup192 binding region (blue); R2m, Nup188 binding region (light purple); R3m, Nup170 binding region (orange); R1m+R2m (blue and light purple); R2m+R3m (light purple and orange); R1m+R3m (blue and orange); R1m+R2m+R3m (blue, light purple, and orange); CTD (green). (B) Viability analysis of the nup100Δnup116Δnup145Δ strains after transformation with indicated scNup116 variants and selection of ten-fold serial dilution series spotted onto either SDC-LEU and SDC+5-FOA plates (shuffling). Untagged or N-terminally tagged with 3×FLAG or eGFP were introduced in strains carrying plasmids expressing scNup145C, scNup145C-3×HA, and scNup145C-mCherry, respectively. (C) Western blot analysis of expression levels of 3×FLAG-tagged non-rescuing scNup116 variants in unshuffled nup100Δnup116Δnup145Δ strain transformants selected in SDC-LEU media. 3×FLAG-scNup116 variants, scNup145C-3×HA, and the endogenous hexokinase loading control were detected with anti-FLAG, anti-HA, and anti-hexokinase antibodies, respectively. (D) Growth analysis at different temperatures of ten-fold serial dilution series of the shuffled nup100Δnup116Δnup145Δ strains carrying indicated scNup116 variants, replicated for untagged and N-terminally 3×FLAG or eGFP‑tagged constructs. (E) Western blot analysis of expression levels of 3×FLAG-tagged scNup116 variants in shuffled nup100Δnup116Δnup145Δ strain transformants selected on SDC+5-FOA media and subsequently grown in YPD media at 30 °C. 3×FLAG-scNup116 variants, scNup145C-3×HA, and the endogenous hexokinase loading control were detected with anti‑FLAG, anti-HA, and anti-hexokinase antibodies, respectively. (F) Subcellular localization at different temperatures of eGFP-scNup116 variants in a shuffled nup100Δnup116Δnup145Δ strain carrying a plasmid expressing scNup145C-mCherry. (G) Subcellular localization of the 60S pre‑ribosomal export reporter scRpl25-mCherry and eGFP-scNup116 variants in a shuffled nup100Δnup116Δnup145Δ strain. (H) Poly(A)+ RNA localization detected by FISH in a shuffled nup100Δnup116Δnup145Δ strain carrying plasmids expressing scNup116 variants and grown at 37 °C for 4 hours. (I) Quantitation of the 60S pre-ribosomal and poly(A)+ RNA retention in the nucleus. All experiments were performed in triplicate and mean and its standard error are reported. All scalebars are 5 μm.

Fig. S41. Multispecies sequence alignment of Nup192. Sequences from twelve diverse species were aligned and colored by sequence similarity according to the BLOSUM62 matrix from white (less than 55 % similarity), to yellow (55 % similarity), to red (100 % identity). Numbering below alignment is relative to the C. thermophilum sequence. Secondary structure observed in the Nup192 structures is shown above the alignment: α-helices (red bars), β-sheets (blue bars), and unstructured regions (black lines). Disordered regions are indicated by gray dots. Residues found by mutational analysis to affect binding to Nup53 (magenta) (41), Nup145N (cyan; Fig. 3, figs. S15, S20, and S21), and Nic96 (pale green; Fig. 2, figs. S2, S6, and S7) are indicated by circles above the alignment.

Fig. S42. In vivo validation of scNup192 interactions with scNic96 and S. cerevisiae Nup145N paralogs. Supporting data for analysis presented in Fig. 5. (A) Domain structure of scNup192 variants introduced into the nup192Δ strain, with colored squares indicating whether the mutation affects the S. cerevisiae Nup145N paralog binding site (cyan), the scNic96 binding site (pale green), or both (cyan and pale green). (B) Viability analysis of the nup192Δ strain after transformation with indicated scNup192 variants and selection of ten-fold serial dilution series spotted onto either SDC-LEU and SDC+5-FOA plates (shuffling). Variants were either untagged or N-terminally tagged with 3×HA or eGFP. (C) Western blot analysis of expression levels of 3×HA-tagged non-rescuing scNup192 variants in unshuffled nup192Δ strain transformants selected in SDC-LEU media. 3×HA-scNup192 variants and the endogenous hexokinase loading control were detected with anti-HA and anti-hexokinase antibodies, respectively. Asterisk indicates nonspecific band detected by the anti-HA antibody. (D) Growth analysis at different temperatures of ten-fold serial dilution series of the shuffled nup192Δ strain carrying indicated scNup192 variants, replicated for untagged and N-terminally 3×HA- or eGFP-tagged constructs. (E) Western blot analysis of expression levels of 3×HA-tagged scNup192 variants in shuffled nup192Δ strain transformants selected on SDC+5-FOA media and subsequently grown in YPD media at 30 °C. Asterisk indicates nonspecific band detected by the anti-HA antibody. 3×HA‑scNup192 variants and the endogenous hexokinase loading control were detected with anti-HA and anti-hexokinase antibodies, respectively. (F) Subcellular localization of eGFP‑scNup192 variants (green) and wildtype mCherry-scNup192 (red) in an unshuffled nup192Δ strain. (G) Subcellular localization eGFP-scNup192 variants at permissive (30 °C) and growth-challenging (37 °C) temperatures in a shuffled nup192Δ strain. Scalebars are 5 μm.

Fig. S43. Multispecies sequence alignment of Nup188. Sequences from eleven diverse species were aligned and colored by sequence similarity according to the BLOSUM62 matrix from white (less than 55 % similarity), to yellow (55 % similarity), to red (100 % identity). Numbering below alignment is relative to the C. thermophilum sequence. Secondary structure observed in the Nup188 structures is shown above the alignment: α-helices (red bars), β-sheets (blue bars), and unstructured regions (black lines). Disordered regions are indicated by gray dots. Residues found by mutational analysis to affect binding to Nup145N (cyan; Fig. 4, figs. S28, S29, and S30), and Nic96 (pale green; Fig. 2, figs. S8, S11, and S12) are indicated by circles above the alignment.

Fig. S44. In vivo validation of scNup188 interactions with scNic96 and S. cerevisiae scNup145N paralogs. Supporting data for analysis presented in Fig. 5. (A) Domain structure of scNup188 variants introduced into the nup188Δ, nup188Δpom34Δ, and nup188Δpom152Δ strains, with colored squares indicating whether the mutation affects the scNup100/116 paralog binding site (cyan), the scNic96 binding site (pale green), or both (cyan and pale green). (B) Subcellular localization of GFP-scNup188 variants transformed into a nup188Δ strain. Scalebars are 5 μm. (C) Viability analysis of the nup188Δpom34Δ and nup188Δpom152Δ strains after transformation with indicated scNup188 variants and selection of ten-fold serial dilution series spotted onto either SDC-LEU and SDC+5-FOA plates (shuffling). Variants were either untagged or N-terminally tagged with 3×FLAG or eGFP. (D) Western blot analysis of expression levels of 3×FLAG-tagged non-rescuing scNup188 variants in unshuffled nup188Δpom34Δ and nup188Δpom152Δ strain transformants selected in SDC-LEU media. 3×FLAG-scNup188 variants and the endogenous hexokinase loading control were detected with anti-FLAG and anti-hexokinase antibodies, respectively. Asterisk indicates nonspecific band detected by the anti‑FLAG antibody. (E) Growth analysis at different temperatures of ten-fold serial dilution series of the shuffled nup188Δpom34Δ and nup188Δpom152Δ strains carrying indicated scNup188 variants, replicated for untagged and N-terminally 3×FLAG or eGFP-tagged constructs. (F) Western blot analysis of expression levels of 3×FLAG-tagged scNup188 variants in shuffled nup188Δpom34Δ and nup188Δpom152Δ transformants selected on SDC+5-FOA media and subsequently grown in YPD media at 30 °C. 3×FLAG-scNup188 variants and the endogenous hexokinase loading control were detected with anti-FLAG and anti-hexokinase antibodies, respectively. (G) Subcellular localization eGFP-scNup188 variants at permissive (30 °C) and growth-challenging (16 °C) temperatures.

Fig. S45. Multispecies sequence alignment of Nic96. Sequences from twelve diverse species were aligned and colored by sequence similarity according to the BLOSUM62 matrix from white (less than 55 % similarity), to yellow (55 % similarity), to red (100 % identity). Numbering below alignment is relative to the C. thermophilum sequence. Secondary structure observed in the Nic96 structures is shown above the alignment: α-helices (red bars), β-sheets (blue bars), and unstructured regions (black lines). Disordered regions are indicated by gray dots. Residues found by mutational analysis to affect binding to the CNT (red) (28), Nup192 (blue; Fig. 2C, figs. S2, S5, S6), Nup188 (blue; Fig. 2H, figs. S8, S10, S12), or Nup53 (magenta) (35) in the C. thermophilum system, are indicated by circles above the alignment. Residues found by mutational analysis to affect binding between NUP93 and NUP53 (bright pink; Fig. 6, fig. S57) in the H. sapiens system, are indicated by circles below the alignment.

Fig. S46. In vivo validation of the scNic96R2 region. Supporting data for analysis presented in Fig. 5. (A) Domain structure of scNic96 variants introduced into the nic96Δ strain. (B) Viability analysis of the nic96Δ strain after transformation with indicated scNic96 variants and selection of ten-fold serial dilution series spotted onto either SDC-LEU and SDC+5-FOA plates (shuffling). Variants were either untagged or N-terminally tagged with 3×FLAG or eGFP. (C) Western blot analysis of expression levels of 3×FLAG-tagged non-rescuing scNic96 variants in unshuffled nic96Δ strain transformants selected in SDC-LEU media. 3×FLAG-scNic96 variants and the endogenous hexokinase loading control were detected with anti-FLAG and anti-hexokinase antibodies, respectively. (D) Growth analysis at different temperatures of ten-fold serial dilution series of the shuffled nic96Δ strain carrying indicated scNic96 variants, replicated for untagged and N-terminally 3×FLAG or eGFP-tagged constructs. (E) Western blot analysis of expression levels of 3×FLAG-tagged scNic96 variants in shuffled nic96Δ strain transformants selected on SDC+5-FOA media and subsequently grown in YPD media at 30 °C. 3×FLAG-scNic96 variants and the endogenous hexokinase loading control were detected with anti-FLAG and anti-hexokinase antibodies, respectively. (F) Subcellular localization of shuffled eGFP-scNic96 variants in the nic96Δ strain at different temperatures. (G) Viability analysis of the nic96Δnup57‑eGFP strain after transformation with indicated mCherry-scNic96 variants and selection of ten-fold serial dilution series spotted onto either SDC-LEU and SDC+5-FOA plates (shuffling). Scalebars are 5 μm.

Fig. S47. Multispecies sequence alignment of Nup53 from species related to S. cerevisiae. Sequences from eight diverse species, including two S. cerevisiae paralogs Nup53 and Nup59, were aligned and colored by sequence similarity according to the BLOSUM62 matrix from white (less than 55 % similarity), to yellow (55 % similarity), to red (100 % identity). Numbering below alignment is relative to the S. cerevisiae sequence. Regions of the sequence are annotated above the alignment: R1 – Nup192 binding region (blue), R2 – Nic96 binding region (green), RNA recognition motif – RRM (magenta), R3 – Nup157/Nup170 binding region (orange) (143), amphipathic α-helix (red).

Fig. S48. Multispecies sequence alignment of Nup53 from species related to C. thermophilum. Sequences from seven diverse species were aligned and colored by sequence similarity according to the BLOSUM62 matrix from white (less than 55 % similarity), to yellow (55 % similarity), to red (100 % identity). Numbering below alignment is relative to the C. thermophilum sequence. Regions of the sequence are annotated above the alignment: R1 – Nup192 binding region (blue), R2 – Nic96 binding region (green), RNA recognition motif – RRM (magenta), R3 – Nup157/Nup170 binding region (orange), amphipathic α-helix (red). Residues found by mutational analysis to affect Nup192 binding (41) or Nup170 binding (35) are indicated by red circles above the alignment.

Fig. S49. Multispecies sequence alignment of Nup53 from species related to H. sapiens. Sequences from fifteen diverse species were aligned and colored by sequence similarity according to the BLOSUM62 matrix from white (less than 55 % similarity), to yellow (55 % similarity), to red (100 % identity). Numbering below alignment is relative to the H. sapiens sequence. Regions of the sequence are annotated above the alignment: R1 – NUP205 binding region (blue), R2 – NUP93 binding region (green), RNA recognition motif – RRM (magenta), R3 – NUP155 binding region, amphipathic α-helix (red). Residues found by mutational analysis to affect binding to NUP93SOL (Fig. 6 and fig. S57) are indicated by red circles above the alignment.

Fig. S50. NUP155NTD binds to a topologically conserved region of NUP53. (A) Domain structure of NUP53 with gray lines indicating truncation construct boundaries analyzed. (B‑D) SEC profiles are shown for nups individually (red and blue) and after their preincubation (green) for NUP155NTD binding to various NUP53 constructs. Gray bars indicate fractions that were resolved on SDS-PAGE gels and visualized by Coomassie staining. All SEC profiles were obtained using a Superdex 200 10/300 GL column.

Fig. S51. NUP93SOL binds to a topologically conserved region of NUP53. (A) Domain structure of NUP53 with gray lines indicating truncation construct boundaries analyzed. (B‑D) SEC profiles are shown for nups individually (red and blue) and after their preincubation (green) for NUP93SOL binding to various NUP53 constructs. Gray bars indicate fractions that were resolved on SDS-PAGE gels and visualized by Coomassie staining. All SEC profiles were obtained using a Superdex 200 10/300 GL column. Asterisks indicate degradation products or contamination.

Fig. S52. NUP93R1 binding to the CNT is disrupted by mutating evolutionarily conserved residues. (A) Sequence alignment of orthologous C. thermophilum Nic96R1 and human NUP93R1 regions shows the evolutionary conservation of the LLLL mutation that abolishes binding of C. thermophilum Nic96R1 to the CNT (28). Residues are colored according to a multispecies sequence alignment, from white (less than 55 % similarity), to yellow (55 % similarity), to red (100 % identity) using the BLOSUM62 substitution matrix. (B‑C) Domain structure of NUP93 with black lines indicating the analyzed construct boundaries, and a schematic of the suggested NUP93R1‑CNT binding mode and location of evolutionarily conserved LIL residues (right). The H. sapiens CNT is composed of NUP54, NUP58 and NUP62. SEC profiles are shown for nups or nup complexes individually (red and blue) and after their preincubation (green) for the CNT binding to (A) NUP93ΔR2-SOL and the mutant NUP93ΔR2-SOL LIL, and (B) NUP93R1 and the mutant NUP93R1 LIL. Gray bars indicate fractions that were resolved on SDS-PAGE gels and visualized by Coomassie staining. All SEC profiles were obtained using a Superdex 200 10/300 GL column.

Fig. S53. Human NUP188 and NUP205 linker-scaffold network. (A) Domain structure of NUP93, NUP53, and NUP98. (B) Pulldown analysis of NUP188 and NUP205 interactions with NUP93R2, NUP53, and NUP98. Biotinylated His6-Avi-SUMO (BA-SUMO) fused with linker NUP93R2, NUP53 and NUP98 fragments were loaded onto streptavidin-conjugated magnetic beads and used as baits to pull down 3×FLAG‑NUP188 or 3×FLAG‑NUP205 from lysate. Linker nup truncations and substitutions are indicated with black lines and red circles above the stylized domain structures, respectively. Load and pulldown fractions were resolved by SDS-PAGE and visualized by western blot. Baits were resolved by SDS-PAGE and visualized by Coomassie staining.

Fig. S54. Mapping of the minimal NUP53 region sufficient for NUP93SOL binding. SEC and SDS‑PAGE analysis corresponding to Fig. 6B. (A) Domain structure of NUP53 with gray bars indicating truncation construct boundaries analyzed. The NUP53R2 peptide is indicated by a red bar. The binding of NUP53 truncation constructs to NUP93SOL, as assessed by SEC and SDS‑PAGE, is summarized and colored according to the measured effect: no effect (green, +++), weak effect (yellow, ++), moderate effect (orange, +), and abolished binding (red, -). (B, C) SEC profiles are shown for nups individually (dark blue and cyan) and after their preincubation (colored according to the measured effect, as in (A). (D) SEC profile of wildtype NUP93SOL (dark blue) is shown as reference. SEC profiles of NUP93SOL preincubated with SUMO-NUP53 truncation constructs are colored according to the measured effect, as in (A). Dashed lines indicate the peak elution volumes across the offset dimension. The gray bar indicates fractions that were resolved on SDS-PAGE gels and visualized by Coomassie staining. All SEC profiles were obtained using a Superdex 200 10/300 GL column.

Fig. S55. 5-Ala scanning mutagenesis of NUP53 sequence for residues necessary for NUP93SOL binding. SEC and SDS-PAGE analysis corresponding to Fig. 6D. (A) Domain structure of NUP53. Mutations were introduced into the NUP53N construct indicated by the black line above the domain structure. The positions of the 5-Ala mutated residues are indicated above the primary NUP53 sequence with boxes colored according to the measured effect on Nup192 binding: no effect (green), moderate effect (yellow), and abolished binding (red). (B) SEC profiles of wildtype NUP93SOL alone (cyan) and NUP93SOL preincubated with wildtype SUMO-NUP53N (dark blue) are shown. SEC profiles of SUMO-NUP53N 5-Ala mutants preincubated with NUP93SOL are colored according to the measured effect, as in (A). The gray bar indicates fractions that were resolved on SDS-PAGE gels and visualized by Coomassie staining. All SEC profiles were obtained using a Superdex 200 10/300 GL column.

Fig. S56. Crystallographic analysis of apo NUP93SOL and NUP93SOL•NUP53R2. Cartoon representation of apo NUP93SOL and NUP93SOL•NUP53R2 crystal structures. Inset boxes indicate magnified views of the NUP53 binding site of (A) the 2.0 Å crystal structure of apo NUP93SOL (yellow), (B) the 3.4 Å co-crystal structure of the NUP93SOL•NUP53R2 complex (green and magenta), and (C) a structural superposition of the two structures. Red circles indicate residues found by mutational analysis to affect NUP93SOL binding to NUP53 (Fig. 6 and fig. S57). Black arrows indicate conformational changes in the loops adjacent to the binding site induced by the binding of the NUP53R2 peptide. A global 12° displacement respect to the and N-terminal U-bend α-helical solenoid, pivoted about a connecting hinge loop, is observed. (D) Anomalous difference Fourier map (orange) calculated from X-ray diffraction data collected at the selenium anomalous peak wavelengths for NUP93SOL•NUP53R2 I94M SeMet-derivatized co-crystals and superposed with the cartoon and stick representation of the NUP93SOL•NUP53R2 structure (green and magenta), validating the sequence assignment of the NUP53R2 peptide. (E) NUP53R2 shown in cartoon and ball-and-stick representation. Surface representation of NUP93SOL binding site (pale green) with residues found by mutational analysis to affect NUP53 binding (Fig. 6 and fig. S57) highlighted (orange).

Fig. S57. Perturbation of the NUP93SOL-NUP53 interaction by point mutants. SEC and SDS‑PAGE analysis corresponding to Fig. 6, G and H. SEC profiles of wildtype NUP93SOL alone (cyan) and NUP93SOL preincubated with wildtype SUMO-NUP53N (dark blue) are shown. SEC profiles of (A) SUMO-NUP53N mutants preincubated with NUP93SOL (B) NUP93SOL mutants preincubated with SUMO-NUP53N are colored according to the measured effect: no effect (green, +++), weak effect (yellow, ++), moderate effect (orange, +), and abolished binding (red, -). The gray bar indicates fractions that were resolved on SDS-PAGE gels and visualized by Coomassie staining. All SEC profiles were obtained using a Superdex 200 10/300 GL column. The results are summarized in the tables (top right).

Fig. S58. Workflow for the sequential quantitative docking of high-resolution structures into the ~12 Å cryo-ET map of the intact human NPC. Cartoon representations of the high resolution structures used as search models at each step are shown on the left. Global searches were performed in the full map, as well as a map limited to the inner ring. At each step, map density assigned to the search model was removed from the search regions of the subsequent step. Cross-sectional views of the full map and inner ring search region (white) and the nuclear envelope (dark gray) are shown in isosurface representation (middle). Isolated contours of the assigned densities are rendered according to the color of the placed search model (right). The density left over after all assigned density is removed from the full map and inner ring search regions, along with a colored composite of all the assigned density.

Fig. S59. Docking of the coat nup complex structures into the ~12 Å cryo-ET map of the intact human NPC. Resolution-matched simulated cryo-EM densities of side-chain resolution structures of the coat nup complex (CNC) were quantitatively docked into an ~12 Å sub-tomogram averaged cryo-ET map of the intact human NPC (EMD-14322) (47). The search model was adopted from (35) and was composed of the yeast CNC-hexamer (PDB ID 4XMM) (29), the NUP84•NUP133 hetero-dimer (PDB ID 3I4R) (24), NUP43 (PDB ID 4I79) (30), NUP37 (PDB ID 4FHM) (25), and NUP133NTD (PDB ID 1XKS) (15) crystal structures. (A) The placement of unique solutions in a single spoke is shown (top left). Arrows and numbers indicate the accepted solutions and their corresponding rank. Closeup views of the accepted solutions are shown on the right, with the rank indicated on the top left corner of the box. Isosurface representations of the nuclear envelope and the NPC are colored in dark gray and white, respectively. The docked structures are displayed as cartoons. Rug plots (blue) and histograms (light blue) of Pearson correlation scores and derived Fisher z scores fit with a normalized Gaussian curve (black) from a global search with 1 million random initial placements are shown (bottom left and middle, respectively). A tabular summary of the accepted solutions fitting statistics, along with one-tailed p-values calculated from the Fisher z score distribution, is shown (bottom right). (B) The NUP133NTD rigid-body fit to the cryo-ET map was further locally optimized at each position: cytoplasmic distal (red), cytoplasmic proximal (dark blue), nuclear distal (green), and nuclear proximal (orange). (C) Comparison of cartoon representations of the CNC docked at different positions, with NUP133NTD polypeptide chains colored as in (B). To account for the difference in diameter of the proximal and distal CNC rings, the NUP133NTD position and distance from the rest of the structured NUP133 (yellow) varies significantly between the distal and proximal positions.

Fig. S60. Docking of the single particle cryo-EM Nup192•Nic96R2•Nup145NR1•Nup53R1 structure into the ~12 Å cryo-ET map of the intact human NPC. Resolution-matched simulated cryo-EM densities of the 3.2 Å Nup192•Nic96R2•Nup145NR1•Nup53R1 single particle cryo-EM structure were quantitatively docked into (A) an ~12 Å sub-tomogram averaged cryo-ET map of the intact human NPC (EMD-14322) (47) and (B) a sub-segment of the map comprised solely of the inner ring. The placement of unique solutions in a single spoke is shown (top left). Arrows and numbers indicate the accepted solutions and their corresponding rank. Closeup views of the accepted solutions are shown on the right, with the rank indicated on the top left corner of the box. Isosurface representations of the nuclear envelope and the NPC are colored in dark gray and white, respectively. The docked structures are displayed as cartoons. Rug plots (blue) and histograms (light blue) of Pearson correlation scores and derived Fisher z scores fit with a normalized Gaussian curve (black) from a global search with 1 million random initial placements are shown (bottom left and middle). A tabular summary of the accepted solutions fitting statistics, along with one-tailed p-values calculated from the Fisher z score distribution, is shown (bottom right).

Fig. S61. Docking of the composite crystal Nup192•Nic96R2•Nup145NR1•Nup53R1 structure into the ~12 Å cryo-ET map of the intact human NPC. Resolution-matched simulated cryo-EM densities of the composite crystal Nup192•Nic96R2•Nup145NR1•Nup53R1 structure were quantitatively docked into (A) an ~12 Å sub-tomogram averaged cryo-ET map of the intact human NPC (EMD-14322) (47) and (B) a sub-segment of the map comprised solely of the inner ring. The placement of unique solutions in a single spoke is shown (top left). Arrows and numbers indicate the accepted solutions and their corresponding rank. Closeup views of the accepted solutions are shown on the right, with the rank indicated on the top left corner of the box. Isosurface representations of the nuclear envelope and the NPC are colored in dark gray and white, respectively. The docked structures are displayed as in cartoon representation. Rug plots (blue) and histograms (light blue) of Pearson correlation scores and derived Fisher z scores fit with a normalized Gaussian curve (black) from a global search with 1 million random initial placements are shown (bottom left and middle). A tabular summary of the accepted solutions fitting statistics, along with one-tailed p-values calculated from the Fisher z score distribution, is shown (bottom right). The crystallized conformation of Nup192 found in the composite crystal structure fits the cryo-ET map density with greater confidence than the conformation of Nup192 found in single particle cryo-EM structure (Fig. S60).

Fig. S62. Docking of the single particle cryo-EM Nup188•Nic96R2•Nup145NR2 structure into the ~12 Å cryo-ET map of the intact human NPC. Resolution-matched simulated cryo-EM densities of the 2.8 Å Nup188•Nic96R2• Nup145NR2 single particle cryo-EM structure were quantitatively docked into (A) the ~12 Å sub-tomogram averaged cryo-ET map of the intact human NPC (EMD-14322) (47) and (B) a sub-segment of the map comprised solely of the inner ring. The placement of unique solutions in a single spoke is shown (top left). Arrows and numbers indicate the accepted and tentative solutions and their corresponding rank. Asterisks indicate that solutions in the outer rings are tentative and shown for illustrative purpose, as quantitative docking in the ~12 Å cryo-ET map places the Nup192 complex with greater confidence into the outer ring positions (figs. S60 and S61). Closeup views of the accepted and tentative solutions are shown on the right, with the rank indicated on the top left corner of the box. Isosurface representations of the nuclear envelope and the NPC are colored in dark gray and white, respectively. The docked structures are displayed in cartoon representation. Rug plots (blue) and histograms (light blue) of Pearson correlation scores and derived Fisher z scores fit with a normalized Gaussian curve (black) from a global search with 1 million random initial placements are shown (bottom left and middle). A tabular summary of the solutions fitting statistics, along with one-tailed p-values calculated from the Fisher z score distribution, is shown (bottom right).

Fig. S63. Comparison of the Nup192 and Nup188 complex placement in the inner ring of the ~12 Å cryo-ET map of the intact human NPC. Question mark-shaped cryo-ET densities carved out from (A) cytoplasmic peripheral, (B) nuclear peripheral, (C) cytoplasmic equatorial, and (D) nuclear equatorial positions of the inner ring of an ~12 Å sub-tomogram averaged cryo‑ET map of the intact human NPC (EMD-14322) (47) where either Nup192•Nic96R2•Nup145NR1•Nup53R1 or Nup188•Nic96R2•Nup145NR2 complexes were quantitatively docked, as indicated in the boxes (top). To compare the goodness of fit by visual inspection, two isosurface representation views of the carved density with cartoons of the quantitatively docked composite crystal Nup192•Nic96R2•Nup145NR1•Nup53R1, single particle cryo-EM Nup192•Nic96R2•Nup145NR1•Nup53R1, or single particle cryo-EM Nup188•Nic96R2•Nup145NR2 structures are shown. At the peripheral positions, a tentative placement of the two conformations of the Nup192•Nic96R2•Nup145NR1•Nup53R1 complex (single particle cryo-EM and composite crystal structures, respectively) results in unexplained excess density that can be accounted for by the SH3-like domain of Nup188 if the Nup188•Nic96R2•Nup145NR2 complex (single particle cryo-EM structure) is placed in the same position. Unlike the shorter Nup188 Tower, the longer Nup192 Tower exceeds the cryo-ET density if tentatively placed at the peripheral positions. On the contrary, the longer Tower domain is accommodated by the cryo-ET density at the equatorial position, but the SH3-like domain of the tentatively placed Nup188 exceeds the cryo-ET density at the equatorial position. At the equatorial positions, the Nup192 Tower domain conformation found in the composite crystal Nup192 complex structure fits the cryo-ET density better than the Nup192 Tower domain conformation observed in the single particle cryo-EM Nup192 complex structure.

Fig. S64. Comparison of the Nup192 and Nup188 complex placement in the outer rings of the ~12 Å cryo-ET map of the intact human NPC. Question mark-shaped cryo-ET densities carved out from (A) cytoplasmic proximal, (B) cytoplasmic distal, and (C) nuclear distal positions of the outer rings of the ~12 Å sub-tomogram averaged cryo-ET map of the intact human NPC (EMD-14322) (47) where Nup192•Nic96R2•Nup145NR1•Nup53R1 complexes were quantitatively docked, as indicated in the boxes (top). To compare the goodness of fit by visual inspection, two isosurface representation views of the carved density with cartoons of the quantitatively docked composite crystal Nup192•Nic96R2•Nup145NR1• Nup53R1, single particle cryo-EM Nup192•Nic96R2•Nup145NR1•Nup53R1, or single particle cryo-EM Nup188•Nic96R2• Nup145NR2 structures are shown. The cryo-ET density is exceeded by the Nup188 SH3-like domain if the Nup188•Nic96R2•Nup145NR2 complex is tentatively placed at all three outer ring positions. On the other hand, the Nup188 α‑helical solenoid presents a narrower superhelical twist compared to Nup192, which leaves unexplained density in other parts of the cryo-ET density. At all outer ring positions, the Nup192 Tower conformation found in the composite crystal Nup192•Nic96R2•Nup145NR1•Nup53R1 complex structure fits the cryo-ET density better than the Nup192 Tower conformation observed in the single particle cryo-EM Nup192•Nic96R2•Nup145NR1•Nup53R1complex structure.

Fig. S65. Docking of the single particle cryo-EM Nup192•Nic96R2•Nup145NR1•Nup53R1 structure into an ~23 Å cryo-ET map of the intact human NPC. Resolution-matched simulated cryo-EM densities of the 3.2 Å Nup192•Nic96R2•Nup145NR1•Nup53R1 single particle cryo-EM structure were quantitatively docked into (A) the 23 Å sub-tomogram averaged cryo-ET map of the intact human NPC (EMD-3103) (38) and (B) a sub-segment of the map comprised solely of the inner ring. The placement of unique solutions in a single spoke is shown (top left). Arrows and numbers indicate the accepted solutions and their corresponding rank. Closeup views of the accepted solutions are shown on the right, with the rank indicated on the top left corner of the box. Isosurface representation of the nuclear envelope and the NPC are colored in dark gray and white, respectively. The docked structures are displayed as in cartoon representation. Rug plots (blue) and histograms (light blue) of Pearson correlation scores and derived Fisher z scores fit with a normalized Gaussian curve (black) from a global search with 1 million random initial placements are shown (bottom left and middle, respectively). A tabular summary of the accepted solutions fitting statistics, along with one-tailed p-values calculated from the Fisher z score distribution, is shown (bottom right). Top-scoring solutions were identified for four of the five positions identified by docking into the higher resolution ~12 Å cryo-ET map (Fig. S60). The cytoplasmic outer ring proximal position was ranked as the 8th highest scoring solution.

Fig. S66. Docking of the composite crystal Nup192•Nic96R2•Nup145NR1•Nup53R1 structure into an ~23 Å cryo-ET map of the intact human NPC. Resolution-matched simulated cryo-EM densities of the composite crystal Nup192•Nic96R2•Nup145NR1•Nup53R1 structure were quantitatively docked into (A) the 23 Å sub-tomogram averaged cryo-ET map of the intact human NPC (EMD-3103) (38) and (B) a sub-segment of the map comprised solely of the inner ring. The placement of unique solutions in a single spoke is shown (top left). Arrows and numbers indicate the accepted solutions and their corresponding rank. Closeup views of the accepted solutions are shown on the right, with the rank indicated on the top left corner of the box. Isosurface representations of the nuclear envelope and the NPC are colored in dark gray and white, respectively. The docked structures are displayed in cartoon representation. Rug plots (blue) and histograms (light blue) of Pearson correlation scores and derived Fisher z scores fit with a normalized Gaussian curve (black) from a global search with 1 million random initial placements are shown (bottom left and middle). A tabular summary of the accepted solutions fitting statistics, along with one-tailed p‑values calculated from the Fisher z score distribution, is shown (bottom right). Top-scoring solutions were identified for four of the five positions identified by docking into the higher resolution ~12 Å cryo-ET map (Fig. S61). The cytoplasmic outer ring proximal position was ranked as the 9th highest scoring solution.

Fig. S67. Docking of the single particle cryo-EM Nup188•Nic96R2•Nup145NR2 structure into an ~23 Å cryo-ET map of the intact human NPC. Resolution-matched simulated cryo-EM densities of the 2.8 Å Nup188•Nic96R2•Nup145NR2 single particle cryo-EM structure were quantitatively docked into (A) an ~23 Å sub-tomogram averaged cryo-ET map of the intact human NPC (EMD-3103) (38) and (B) a sub-segment of the map comprised solely of the inner ring. The placement of unique solutions in a single spoke is shown (top left). Arrows and numbers indicate the accepted solutions and their corresponding rank. Asterisks indicate that solutions in the outer rings are tentative and shown for illustrative purpose, as quantitative docking in the ~12 Å cryo‑ET map places the Nup192 complex with greater confidence into the outer ring positions (figs. S60 and S61). Closeup views of the accepted and tentative solutions are shown on the right, with the rank indicated on the top left corner of the box. Isosurface representations of the nuclear envelope and the NPC are colored in dark gray and white, respectively. The docked structures are displayed in cartoon representation. Rug plots (blue) and histograms (light blue) of Pearson correlation scores and derived Fisher z scores fit with a normalized Gaussian curve (black) from a global search with 1 million random initial placements are shown (bottom left and middle). A tabular summary of the accepted solutions fitting statistics, along with one-tailed p-values calculated from the Fisher z score distribution, is shown (bottom right). Top-scoring solutions were identified for four of the five positions identified by docking into the higher resolution ~12 Å cryo-ET map (fig. S62). The tentative cytoplasmic outer ring proximal position was ranked as the 11th highest scoring solution.

Fig. S68. Docking of the NUP93SOL•NUP53R2 co-crystal structure into the ~12 Å cryo-ET map of the intact human NPC. Resolution-matched simulated cryo-EM densities of the 3.4 Å NUP93SOL•NUP53R2 structure were quantitatively docked into (A) the ~12 Å sub-tomogram averaged cryo-ET map of the intact human NPC (EMD-14322) (47) from which cryo-ET density corresponding to hereto docked nups was subtracted (fig. S58) and (B) a sub-segment of the map comprised solely of the inner ring from which cryo-ET density corresponding to hereto docked nups was subtracted (fig. S58). The placement of unique solutions in a single spoke is shown (top left). Arrows and numbers indicate the accepted solutions and their corresponding rank. The asterisk indicates a copy of NUP93SOL•NUP53R2 that was placed manually into matching but weak density. The presence of a proximal copy of NUP93SOL is supported by the presence of a proximal copy of NUP205, which binds to NUP93R2. Closeup views of the accepted solutions are shown on the right, with the rank indicated on the top left corner of the box. Isosurface representations of the nuclear envelope and the NPC are colored in dark gray and white, respectively. The docked structures are displayed in cartoon representation. Rug plots (blue) and histograms (light blue) of Pearson correlation scores and derived Fisher z scores fit with a normalized Gaussian curve (black) from a global search with 1 million random initial placements are shown (bottom left and middle). A tabular summary of the solutions fitting statistics, along with one-tailed p-values calculated from the Fisher z score distribution, is shown (bottom right). Top scoring solutions in neither full map nor inner ring search regions did not include a nuclear equatorial copy of NUP93SOL•NUP53R2. However, its position could be inferred by the C2 symmetry relationship with the top-scoring cytoplasmic equatorial copy (ranked 3rd highest) and matched with the 36th highest scoring solution.

Fig. S69. Docking of the NUP93SOL•NUP53R1 co-crystal structure into an ~23 Å cryo-ET map of the intact human NPC. Resolution-matched simulated cryo-EM densities of the 3.4 Å NUP93SOL•NUP53R2 structure were quantitatively docked into (A) an ~23 Å sub-tomogram averaged cryo-ET map of the intact human NPC (EMD-3103) (38) and (B) a sub-segment of the map comprised solely of the inner ring. The placement of unique solutions in a single spoke is shown (top left). Arrows and numbers indicate the accepted solutions and their corresponding rank. Closeup views of the accepted solutions are shown on the right, with the rank indicated on the top left corner of the box. Isosurface representations of the nuclear envelope and the NPC are colored in dark gray and white, respectively. The docked structures are displayed in cartoon representation. Rug plots (blue) and histograms (light blue) of Pearson correlation scores and derived Fisher z scores fit with a normalized Gaussian curve (black) from a global search with 1 million random initial placements are shown (bottom left and middle, respectively). A tabular summary of the solutions fitting statistics, along with one-tailed p-values calculated from the Fisher z score distribution, is shown (bottom right). Top scoring solutions were identified for a single copy of NUP93SOL•NUP53R2 at the distal position in each one of the outer rings. No cryo‑ET density was observed for a proximal copy of NUP93SOL•NUP53R2 in the cytoplasmic outer ring. Top scoring solutions for all four copies of NUP93SOL•NUP53R2 in the inner ring were identified in the inner ring search region.

Fig. S70. Docking of the bridge composite crystal Nup170•Nup145N R3•Nup53R3 structure into the ~12 Å cryo-ET map of the intact human NPC. Resolution-matched simulated cryo-EM densities of the composite crystal Nup170•Nup145NR3•Nup53R3 conformation II structure (as identified by (35)) were quantitatively docked into (A) the ~12 Å sub-tomogram averaged cryo-ET map of the intact human NPC (EMD-14322) (47) from which cryo-ET density corresponding to hereto docked nups was subtracted (fig. S58) and (B) a sub-segment of the map comprised solely of the inner ring from which cryo-ET density corresponding to hereto docked nups was subtracted (fig. S58). The placement of unique solutions in a single spoke is shown (top left). Arrows and numbers indicate the accepted solutions and their corresponding rank. Closeup views of the accepted solutions are shown on the right, with the rank indicated on the top left corner of the box. Isosurface representations of the nuclear envelope and the NPC are colored in dark gray and white, respectively. The docked structures are displayed in cartoon representation. Rug plots (blue) and histograms (light blue) of Pearson correlation scores and derived Fisher z scores fit with a normalized Gaussian curve (black) from a global search with 1 million random initial placements are shown (bottom left and middle). A tabular summary of the solutions fitting statistics, along with one-tailed p-values calculated from the Fisher z score distribution, is shown (bottom right).

Fig. S71. Docking of the inner ring composite crystal Nup170•Nup145N R3•Nup53R3 structures into the ~12 Å cryo-ET map of the intact human NPC. (A) Resolution-matched simulated cryo-EM densities of the composite crystal Nup170•Nup145NR3•Nup53R3 conformation I structure (as identified by (35)) were quantitatively docked into a sub-segment of the ~12 Å sub‑tomogram averaged cryo‑ET map of the intact human NPC (EMD-14322) (47) comprised solely of the inner ring from which cryo-ET density corresponding to hereto docked nups was subtracted (fig. S58). The placement of unique solutions in a single spoke is shown (top left). Arrows and numbers indicate the accepted solutions and their corresponding rank. Closeup views of the accepted solutions are shown on the right, with the rank indicated on the top left corner of the box. Isosurface representations of the nuclear envelope and the NPC are colored in dark gray and white, respectively. The docked structures are displayed in cartoon representation. Rug plots (blue) and histograms (light blue) of Pearson correlation scores and derived Fisher z scores fit with a normalized Gaussian curve (black) from a global search with 1 million random initial placements are shown (bottom left and middle). A tabular summary of the solutions fitting statistics, along with one-tailed p-values calculated from the Fisher z score distribution, is shown (bottom right). The top-scoring solutions identified the two equatorial copies of Nup170•Nup145NR3•Nup53R3 in the inner ring. (B) Attempts of quantitative docking of all available Nup170•Nup145NR3•Nup53R3 conformations in the peripheral inner ring positions was unsuccessful. The composite crystal Nup170•Nup145NR3•Nup53R3 conformation III structure (as identified by (35)) was placed manually at the inner ring cytoplasmic peripheral (*) and nuclear peripheral (**) positions and the fit was locally optimized.

Fig. S72. Docking of the CNT•Nic96R1 co-crystal structure into the ~12 Å cryo-ET map of the intact human NPC. (A) Resolution-matched simulated cryo-EM densities of the CNT•Nic96R1 co-crystal structure (PDB ID 5CWS) (28) were quantitatively docked into a sub-segment of the ~12 Å sub-tomogram averaged cryo-ET map of the intact human NPC (EMD-14322) (47) comprised solely of the inner ring from which cryo-ET density corresponding to hereto docked nups was subtracted (fig. S58). The placement of unique solutions in a single spoke is shown (top left). Arrows and numbers indicate the accepted solutions and their corresponding rank. Closeup views of the accepted solutions are shown on the right, with the rank indicated on the top left corner of the box. Isosurface representations of the nuclear envelope and the NPC are colored in dark gray and white, respectively. The docked structures are displayed in cartoon representation. Rug plots (blue) and histograms (light blue) of Pearson correlation scores and derived Fisher z scores fit with a normalized Gaussian curve (black) from a global search with 1 million random initial placements are shown (bottom left and middle). A tabular summary of the solutions fitting statistics, along with one-tailed p-values calculated from the Fisher z score distribution, is shown (bottom right). The four inner ring positions were identified by the 2nd, 3rd, 6th, and 9th highest scoring solutions. (B) Two views of cartoon representations of the CNT•Nic96R1 co-crystal structures (red and pale green, respectively) docked in the isosurface representation of the ~12 Å cryo-ET map of the human NPC (white). Four globular extra cryo-ET densities (cyan, indicated by arrows) adjacent to the Nup57 α/β domain (left) are shown. NUP54, the metazoan ortholog of Nup57, possesses an additional ferrodoxin-like domain insert adjacent to the α/β domain. Ferrodoxin-like domains from Xenopus laevis xlNUP54 (PDB ID 5C2U, dark blue) (42) were manually fit into the extra densities and the fit locally optimized (middle). A representative pair of CNT•Nic96R1 and xlNUP54 structures fit in the cryo-ET density is shown to illustrate their relative positions (right).

Fig. S73. Composite model of a CNT•NUP93R1 inclusive of the NUP54 ferrodoxin-like domain. All structures are shown in cartoon representation. (A) The X. laevis crystal structures xlNUP54 (PDB ID 5C2U, dark blue) (42) and xlCNT (PDB ID 5C2U) (42), comprising the first two coiled coil segments CCS1 and CCS2 of xlNUP54 (light blue), xlNUP58 (brick), and xlNUP62 (salmon), were structurally aligned using overlapping residues 317-329 (inset box) to generate a X. laevis composite crystal structure. (B) The X. laevis composite crystal structure was superposed with the C. thermophilum CNT•Nic96R1 co‑crystal structure to generate a composite CNT•NUP93R1 model which includes a ferrodoxin-like domain and all coiled coil segments CCS1‑CCS3. (C) A top view of the composite CNT•NUP93R1 model structurally aligned with the docked CNT•Nic96R1 (PDB ID 5CWS) (28) co-crystal structure reveals that the ferrodoxin-like domains fit-optimized in the 12 Å cryo-ET map density (cyan) adopt varying degrees of displacement at the peripheral and equatorial positions, which is likely allowed by flexible hinge loops that connect the ferrodoxin-like domain to the rest of the NUP54 chain.

Fig. S74. Docking of the NUP53RRM and NUP98APD crystal structures into the ~12 Å cryo-ET map of the intact human NPC. Resolution-matched simulated cryo-EM densities of the (A) NUP53RRM homodimer (PDB ID 4LIR) and (B) NUP98APD (PDB ID 1KO6) (14) crystal structures were quantitatively docked into the ~12 Å sub-tomogram averaged cryo-ET map of the intact human NPC (EMD-14322) (47) from which cryo-ET density corresponding to hereto docked nups was subtracted (fig. S58). The placement of unique solutions in a single spoke is shown (top left). Arrows and numbers indicate the accepted solutions and their corresponding rank. Closeup views of the accepted solutions are shown on the right, with the rank indicated on the top left corner of the box. Isosurface representations of the nuclear envelope and the NPC are colored in dark gray and white, respectively. The docked structures are displayed in cartoon representation. Rug plots (blue) and histograms (light blue) of Pearson correlation scores and derived Fisher z scores fit with a normalized Gaussian curve (black) from a global search with 1 million random initial placements are shown (bottom left and middle). A tabular summary of the solutions fitting statistics, along with one-tailed p-values calculated from the Fisher z score distribution, is shown (bottom right). The 2nd‑10th highest scoring solutions for the NUP53RRM are placed at the same position as the 1st highest scoring solution, in slightly different orientations. The 2nd-7th highest scoring solutions for the NUP98APD are placed at the same position as the 1st highest scoring solution, in slightly different orientations.

Fig. S75. Docking of the X. leavis NPC cytoplasmic outer ring. (A) An anisotropic ~8 Å single particle cryo-EM composite map of the X. laevis cytoplasmic face of the NPC (light cyan; EMD‑0909) (133) was superposed to the ~12 Å cryo-ET map of the intact human NPC (wheat; EMD-14322) (47). (B) The superposition of the two maps placed the docked cytoplasmic outer ring nups into the X. laevis cytoplasmic face cryo-EM map. The fit of the structures was locally optimized. (C) Two views of isosurface representations of carved density and the rigid-body fit NUP93SOL•NUP53R2 crystal structure shown in cartoon representation (pale green and magenta), revealing correspondence between secondary structure and cryo-EM map features. (D) Two views of isosurface representations of proximal and distal question mark-shaped densities carved out from the X. laevis cytoplasmic face cryo-EM map and fit with the composite crystal Nup192•Nic96R2•Nup145NR1•Nup53R1 structure. Tentative placement of the Nup188•Nic96R2•Nup145NR2 single particle cryo-EM structure into the same densities indicates poor fit and confirms the assignment of NUP205 to both proximal and distal question mark-shaped densities of the cytoplasmic outer ring. (E, F) Closeup on regions indicated by inset boxes in (D). Isosurface of a long tubular cryo-EM density (pale green) running near-perpendicular to surrounding tubular cryo-EM density (white) is shown (top). Superposition of cryo-EM density (white) with the Nup192 (blue) and Nic96R2 (pale green), shown as cartoons, suggests that the long tubular density engulfed corresponds to the long NUP93R2 α-helix binding along NUP205 Tail domain (bottom).

Fig. S76. Unassigned density in an ~12 Å cryo-ET map of the intact human NPC. Isosurface representation of the ~12 Å sub-tomogram averaged cryo-ET map of the intact human NPC (EMD-14322) (47) with assigned cryo-ET density (white) and nuclear envelope (gray) rendered as isosurfaces. (A) Cross sectional views from the center of the transport channel (top) or of a lateral transect of a single spoke (middle and bottom) of the inner ring with unassigned density isosurfaces indicated (dark green). Structures that explain the assigned density are shown in cartoon representation (middle and bottom). (B) View of the cytoplasmic face isosurface with two clusters of unassigned density indicated (salmon and magenta). (C) View of the nuclear face isosurface with two clusters of unassigned density present on top (cyan) and sides (dark blue) of the nuclear outer ring. Schematics indicate the orientation of the viewer with respect to the NPC.

Fig. S77. Summary of novel additions to the symmetric core composite structure. (A) Novel high resolution structures of nups and nup complexes docked into intact human NPC symmetric core, shown as cartoons: Nup192•Nic96R2•Nup145NR1•Nup53R1 (blue, pale green, cyan, and magenta), Nup188•Nic96R2•Nup145NR2 (light purple, pale green, and cyan), NUP93SOL•NUP53R2 (pale green and magenta), and the NUP53RRM homodimer (magenta). (B) View from above the cytoplasmic face and (C) a cross-sectional view from the central transport channel of the symmetric core of the intact human NPC with novel structures shown in cartoon representation and colored according to (A). The remainder of the symmetric core nups are colored white. Isosurface representations of the nuclear envelope are colored in gray.

Fig. S78. Linker topology analysis of the composite structures of the human NPC. (A) Schematic representation of an NPC spoke layers with the positional nomenclature indicated for each nup. Inner ring positional acronyms: cytoplasmic equatorial (CE), cytoplasmic proximal (CP), cytoplasmic bridge (CB), nuclear equatorial (NE), nuclear proximal (NP), and nuclear bridge (NB). Outer ring positional acronyms (grey shading): cytoplasmic distal (CD), cytoplasmic proximal (CX), nuclear distal (ND), nuclear proximal (NX), and cytoplasmic nup filament complex (CFNC) (60). (B) Systematic measurement of Euclidean distances (Å) between linker-scaffold regions at various positions in the constricted/dilated composite structures of the human NPC, respectively. The cross-spoke (x-sp.) distances reported are of the shortest connection to a particular position on an adjacent spoke, either in the clockwise or counterclockwise direction from the reference spoke. The shortest Euclidean distances consistent with the most parsimonious architecture of the linker-scaffold are indicated in blue and boldface type. Distances incompatible with the maximum physically possible length of the extended linker (red) and longer (orange) or shorter (green) than the calculated r.m.s. end-to-end length of a linear polymer chain according to the Gaussian chain model developed by Flory, with a characteristic ratio parameter of ~3, typical for unstructured polypeptide linkers (134, 135). (*) Because NUP53RRM is not accurately resolved in the dilated in situ cryo-ET map of the human NPC, we do not report distances involving NUP53RRM but expect them to be larger than in the constricted NPC. (†) Because the NUP53RRM homodimer possesses 2-fold rotational symmetry and is anchored in place by the tethering connections of the shortest NUP53 RRM-R3 linker, the longer cross‑spoke connections would convert to the shorter within-spoke connections by a simple rotation of the NUP53RRM homodimer and are therefore not meaningful. (††) The distance between the cross-spoke related NUP205-bound NUP93R2 and NUP93SOL is ~15 Å shorter than the current estimate due to the additional ~240 residues elongating the C-terminal Tail region of the human NUP205 that, in comparison, are not present in the docked C. thermophilum Nup192 ortholog. (‡) Based on biochemical analysis of NUP93R1 and NUP93ΔR2-SOL constructs, the connecting linker length is likely shorter (fig. S52). (§) NUP98R3 binding to the cytoplasmic bridge NUP155 is possibly precluded by GLE1•NUP42 (60). (#) Bridge NUP155-bound NUP53R3 cannot connect to resolved inner ring RRM domains. The reported linker lengths and distances to outer ring NUP93SOL-bound NUP53R2 are agnostic of RRM domain location.

Fig. S79. Architecture of the outer rings of the NPC symmetric core. (A) Top and bottom views of the cytoplasmic and nuclear outer ring spoke protomers, and their structural superposition. NUP205 and NUP93 are shown in cartoon representation. CNCs are shown in surface representation and colored in gray, with the nups that interface with NUP205 and NUP93 highlighted in color. Circles by labels indicate if the nup is in a distal (dark gray) or proximal (light gray) position. The cytoplasmic spoke includes two copies of NUP205 (blue for distal and dark blue for proximal) and two copies of NUP93 (pale green for distal and dark green for proximal). The distal NUP205 (blue for cytoplasmic, light cyan for nuclear) and the distal NUP93 (pale green for cytoplasmic, pink for nuclear) are found at equivalent positions in the cytoplasmic and nuclear outer rings and form interfaces with both proximal (pale yellow) and distal (yellow) CNC. Schematics indicate linker connections: between distal NUP93SOL (pale green for cytoplasmic, pink for nuclear) and distal NUP205 of adjacent spokes (blue for cytoplasmic, light cyan for nuclear), between proximal NUP205 (dark blue) and proximal NUP93SOL (dark green) of the same cytoplasmic spoke, between NUP53 binding sites (magenta) on proximal NUP205 and distal NUP93SOL and distal NUP205 and proximal NUP93SOL. (B) Structural superposition of the distal and proximal nups of the cytoplasmic outer ring. NUP205 (blue for distal and dark blue for proximal) and NUP93 (pale green for distal and dark green for proximal) are shown in cartoon representation. CNCs are shown in surface representation and colored in gray, with the nups that interface with NUP205 and NUP93 highlighted (pale yellow for proximal and yellow for distal). (C) Schematic representations of top views of the cytoplasmic (left) and nuclear (right) outer rings. Gray octagons indicate the eight-fold symmetry. The CNC double rings (yellow for proximal and ochre for distal) assume a head-to-tail circular arrangement with eight spokes. A staple between spokes is provided by the distal NUP205 (blue) interacting with a distal NUP93 (pale green) from an adjacent spoke. Bridge NUP155 (orange) provide an interface between the outer rings and the inner ring (not shown). The cytoplasmic outer ring encompasses a proximal NUP205 (dark blue) and NUP93 (dark green) that interact within the same spoke. Linker nup binding sites are indicated with circles: NUP98 (cyan), NUP53 (magenta), and NUP93 (green). (D) Closeup views of the interfaces accommodating outer ring NUP205 and NUP93. Nups are shown as cartoons and the nuclear envelope as isosurface (dark gray). The nups interfacing with indicated NUP205 or NUP93 molecule are shown in color and the remainder of nups are indicated in gray.

Fig. S80. Shortest-distance analysis of linker network in the outer rings of the NPC. Spatial distances in the composite structure of human NPC were measured between two copies of linker nup segments that are adjacent in their primary sequence. (A‑I) In the box on the left, linker nup segments and the scaffold surface that they are bound to are shown in cartoon representation and highlighted in color. The remainder of nups are colored white. The nuclear envelope is rendered as dark gray isosurface. The straight-line linkage between the two closest copies of linker nup segments adjacent in their primary sequence is indicated by a solid red line. The measured Euclidean distance is reported in red. On the right, a schematic representation of the scaffold nup binding sites (highlighted in color) and the connecting linker nup segment (red), illustrates the topology of the linkage within the context of two outer ring NPC spokes, with the remainder of the nups shown in gray.

Fig. S81. Placement of the NUP53RRM homodimer between inner ring spokes. (A‑C) Different views of the interface between inner spokes of the human NPC are shown, as indicated by the inset box and schematic representation of the NPC (left). The composite NPC structures are shown as cartoons and the nuclear envelope as a dark gray isosurface. In the middle panel, globular cryo-ET densities, discontinuous with the rest of the ~12 Å human cryo-ET map, are shown as pink isosurfaces and fit with the docked NUP53RRM homodimer crystal structure (magenta). On the right, two spokes are delimited by a dashed line and one spoke is shown in gray. Shortest-distance connections with NUP53R2 and NUP53R3 peptides, bound to NUP93 and NUP155, respectively, are drawn as dashed lines. The NUP53R2 and NUP53R3 binding regions are N- and C-terminally adjacent to the NUP53RRM domain in the primary sequence. The NUP53RRM homodimer interfaces connect two inner ring spokes.

Fig. S82. Shortest-distance analysis of linker network in the inner ring of the NPC. Spatial distances in the composite structure of human NPC were measured between two copies of linker nup segments that are adjacent in their primary sequence. (A‑I) In the box on the left, linker nup segments and the scaffold surface that they are bound to are shown in cartoon representation and highlighted in color. The remainder of nups are colored white. The nuclear envelope is rendered as a dark gray isosurface. The straight-line linkage between the two closest copies of linker nup segments adjacent in their primary sequence is indicated by a solid red line. The measured distance is reported in red. On the right, a schematic representation of the scaffold nup binding sites (highlighted in color) and the connecting linker nup segment (red), illustrates the topology of the linkage within the context of three inner ring NPC spokes (remainder of nups shown in gray).

Fig. S83. Composite structure of the dilated in situ human NPC. (A) Comparison of the constricted human NPC imaged in purified HeLa cell nuclear envelopes and the dilated human NPC imaged in situ in SupT1-R5 cells. Outer and inner ring spoke subcomplexes from the composite NPC structure built into the constricted ~12 Å cryo-ET sub-tomogram averaged human NPC map (EMD-14322) (47) were manually docked and locally fit as rigid units in the dilated in situ ~37 Å cryo-ET sub-tomogram averaged human NPC map (EMD-11967) (45). The composite NPC structures are shown as cartoons superposed to isosurfaces of the constricted and dilated NPCs (grey). In transitioning between constricted and dilated states, the outer ring diameter of the human NPC increases by ~4%, whereas the diameter of the central transport channel increases by ~31%. (B) Closeup views the cytoplasmic outer ring and inner ring at the interface (dashed line) between two spokes in the constricted and dilated human NPCs. Measurements of the distance spanned by the NUP93 linker between NUP205 (blue) and NUP93SOL (pale green) of adjacent spokes in the dilated and constricted NPCs show little dilation between spokes in the outer rings. On the contrary, measurements of the distance spanned by the NUP53 linker between NUP205 (blue) and the peripheral NUP93 (pale green) of adjacent spokes of the inner ring, as well as the distance between two equivalent points in the CNT (red) layer closest to the central transport channel, suggest that the dilation of the NPC results in a gap between inner ring spokes. (C) Schematic representation of the NPC symmetric core cross-section. The translocation of inner nuclear membrane integral membrane proteins (INM-IMPs) is accommodated by gaps between the outer rings and the nuclear envelope and lateral channels between the spokes of the inner ring. (D) Schematic representation of the NPC symmetric core top view. The variable dimensions of the lateral gates impose a size limit to the passage of pore-facing INM-IMP’s soluble domains. Whereas small folded domains (left) can translocate through the lateral channel space, larger folded domains (middle) or nuclear localization sequence (NLS)-containing INM-IMPs actively transported by karyopehrins (Kapα/Kapβ-mediated classical transport is illustrated; right) are transported through the central transport channel, tethered by unstructured linkers that span the width of the inner ring.

Coordinates: