Mechanism of RanGTP priming the release of H2A-H2B from Kap114 and Importin-9

Previously we showed that the nuclear import receptor Importin-9 wraps around the H2A-H2B core to chaperone and transport it from the cytoplasm to the nucleus (Padavannil et al. 2019). However, unlike most nuclear import systems where RanGTP dissociates cargoes from their importins, RanGTP binds stably to the Importin-9•H2A-H2B complex and formation of RanGTP•Importin-9•H2A-H2B facilitates H2A-H2B release to the assembling nucleosome (Padavannil et al. 2019). Here we show cryo-EM structures of Importin-9•RanGTP and of its yeast homolog Kap114, including Kap114•RanGTP, Kap114•H2A-H2B, and RanGTP•Kap114•H2AH2B. In combination with hydrogen-deuterium exchange analysis of Importin-9 complexes and nucleosome assembly assays, we explain how the conserved Kap114/Importin-9 importins bind H2A-H2B and RanGTP simultaneously and how the GTPase primes histone transfer to the nucleosome. We show that RanGTP binds to the N-terminal repeats of Kap114/Importin-9 as in Kap114/Importin-9•RanGTP, and H2A-H2B binds via its acidic patch to the Kap114/Importin-9 Cterminal repeats as in Kap114/Importin-9•H2A-H2B. RanGTP-binding in RanGTP•Kap114•H2AH2B changes Kap114/Importin-9 conformation such that it no longer contacts the surface of H2AH2B proximal to the H2A docking domain that drives nucleosome assembly, positioning it for transfer to the assembling nucleosome. The reduced affinity of RanGTP for Kap114/Importin-9 when H2A-H2B is bound may ensure release of H2A-H2B only at chromatin.

Other interactions involve the extended h8 loops and h12-15 of the Importins contacting helix α4  However, the importin conformations, especially at the N-terminal regions, are quite different when bound to the two ligands (Figure 2A-C). The RanGTP-bound Kap114 superhelix is wider and has a shorter pitch than the H2A-H2B-bound one ( Figure 2C). HEAT repeats alignment between the Kap114•H2A-H2B and Kap114•RanGTP structures revealed rigid body regions and hinges between them that describe the conformational differences ( Figure 2 -Supplement 7A).
Repeats h5-h13 are similar in the two structures (RMSD=0.515 Å for 378 of 430 Cα atoms aligned), suggesting a rigid body block. Meanwhile, high RMSDs when aligning consecutive repeats suggest that h4-h5, h13-h14 and h16-h17 are hinges about which groups of HEAT repeats rotate (Figure 2 -Supplement 7A). These groups of repeats are 1) h1-h4 that contact H2A-H2B and also RanGTP, 2) the invariant h5-h13 core and 3) the h17-h20 repeats that make no contact with RanGTP and are very flexible in Kap114•RanGTP but make extensive contacts with H2A-H2B. The hinges at h4-h5 facilitate movement of the N-terminal repeats (h1-h4) relative to the central core repeats (h5-h13) while the h13-h14 and h16-h17 hinges facilitate movement of the C-terminal repeats (h17-h20) relative to the central core repeats (

Structure of the RanGTP•Kap114•H2A-H2B ternary complex
It was unclear from the binary structures how the ternary complex is arranged when Kap114 binds RanGTP and H2A-H2B simultaneously. We assembled the RanGTP•Kap114•H2A-H2B complex for structure determination by cryo-EM. The data collected produced three classes of ternary complexes that differ in the orientations of Kap114 repeats h16-h20. We solved the structure of the most highly populated class 3 ( Figure 2D Table 1).
The RanGTP•Kap114•H2A-H2B structure ( Figure 2D) shows the N-terminal half of Kap114 superhelix (composed of h1-h15) binding RanGTP by adopting the same conformation (RMSD=0.625 Å, 587 of 701 Cα atoms aligned) and making the same interactions with the GTPase as the binary Kap114•RanGTP structure ( Figure 3A, B, Figure 3 -Supplement 2). Most of the residues in the Kap114 h1-h4 repeats that contact H2A-H2B in Kap114•H2A-H2B are still accessible but not in the same conformation and thus no longer contact the histone in the ternary of Kap114 in the binary Kap114•H2A-H2B complex for comparison with (C). In all panels, Kap114 is gray, RanGTP is green with Switch 1 in violet and Switch 2 in cyan, H2A is yellow, and H2B is red. Dashed lines represent interactions <4 Å (yellow) and long-range electrostatic interactions <8 Å (gray). structure ( Figure 3C, D). This 'displaced' surface of H2A-H2B makes a few contacts with the Kap114 h2-h3 loop and the RanGTP switches, a drastic decrease in interaction when compared to Kap114•H2A-H2B ( Figure 3C, D). This newly exposed surface of H2A-H2B is also proximal to the H2A C-terminal tail that docks onto H3-H4 in the nucleosome, although this tail is not visible in either structure.
The region C-terminal of h16 in Kap114 does not contact RanGTP but adopts the same conformation and maintains the same extensive contacts with H2A-H2B as in the Kap114/Imp9•H2A-H2B structures ( Figure 3E

Solution analysis of Imp9•H2A-H2B and Imp9•RanGTP binary complexes
To probe Imp9/Kap114 complexes in solution, we performed bottom-up HDX-MS focusing on Imp9 for which a ternary complex structure is not available (Figure 4-7, Figure 4
These regions correspond to the binding sites of H2A-H2B in Imp9 (Padavannil et al. 2019).
Addition of RanGTP similarly reduced D-uptake within the Imp9 N-terminal repeats at h1 and the h3 loop, as well as within the middle repeats at the h8 loop and h13 repeat ( Figure 4C . Beyond known interfaces, we also see reduced D-uptake with either H2A-H2B or RanGTP in the Imp9 h8 loop, h16-h17 repeats and the flexible/disordered h19 loop ( Figure 4B, D, 5A Panels 1-2, 6A Panels 1-2, Figure 4 -Supplement 1A Panels 1-2). The larger reduction in D-uptake for h16-h17 with H2A-H2B than RanGTP is notable as it is likely a hinge that positions the C-terminal repeats for histone-binding. We can also compare D-uptake of peptides from RanGTP or H2A-H2B alone to those in the binary complexes ( Figure 4B, D, Figure 7A Panel 1, Figure 4 -Supplement 2). For RanGTP, the greatest reduction in D-uptake upon addition of Imp9 occurred at Switch 1, followed by Switch 2, residues 93-120 (near Imp9 h3 loop), and residues 121-158 (near Imp9 loops of h7, h8, h13 and h14), all aligning with interfaces seen in the Imp9•RanGTP structure ( Figure 4D, Figure

Solution analysis of the RanGTP•Imp9•H2A-H2B ternary complex
Solution analysis of the Imp9 ternary complex also suggests an arrangement similar to that seen in the RanGTP•Kap114•H2A-H2B structure. Comparison of Imp9 with H2A-H2B and Imp9 with both RanGTP and H2A-H2B, showed that addition of RanGTP caused an increase in D-uptake in h4-h5, indicative of less H2A-H2B occupancy at the Imp9 N-terminal repeats ( Figure   5A Panels 3-4, B, C). Unlike the N-terminal repeats, addition of RanGTP did not alter D-uptake of the h18 loop or the start of the h18-19 loop, showing that H2A-H2B maintains binding at these Cterminal repeats in the presence of RanGTP ( Figure 6A Panels 3-4, B, C). We do, however, observe less D-uptake at the h19 a-helix and in the h19 loop, suggesting subtly weaker contacts at this point. The ternary complex also has less D-uptake at the h16-h17 hinge, consistent with RanGTP causing a rigid body movement of the C-terminal repeats similar to that seen in the Imp9•RanGTP ( Figure 6A, B, Figure 6 -Supplement 1). Changes in H2A-H2B D-uptake also occurred, albeit with smaller magnitude that for many peptides was just below our stringent significance cutoffs (Figure 4 -Supplement 2). Binding of RanGTP caused increased D-uptake of H2A α1 suggesting increased flexibility of the H2A C-terminal tail, while decreased D-uptake occurred in H2A and H2B close to the acidic patch that is bound by the Imp9 h18-h19 loop.
The presence of H2A-H2B also reduces the occupancy of RanGTP on Imp9, an observation not predicted by the Kap114 ternary structure. This is indicated by the reduction in D-uptake upon addition of RanGTP to Imp9 being greater in the binary than in the ternary ( Figure   5A Panels 2-3). This trend is seen at the RanGTP-binding site in h1-h3 ( Figure 5A, B). An identical trend is observed for the RanGTP peptides involved in binding Imp9, with less change in D-uptake occurring at Switch 1, Switch 2, and residues 93-120 and 121-158 ( Figure 7A, B). Importantly, it is only the occupancy of RanGTP that is altered and not the actual contacts as the pattern of Duptake remains the same between the binary and ternary complexes for both Imp9 and RanGTP peptides ( Figure 5A Panels 2-3, 7A). Based on the reduced change in D-uptake of peptides at the RanGTP-Imp9 interface, we can estimate an affinity of ~200 nM, which is greater than the reported single digit nanomolar affinities for several Importin and RanGTP pairs (Figure 7 -Supplement Table 7) (Hahn and Schlenstedt 2011;Kochert et al. 2018). To test this prediction, we used fluorescence polarization to measure the affinity of RanGTP for Kap114/Imp9 and Kap114/Imp9•H2A-H2B ( Figure 7C, D). We measure the affinities to be 0.8/1 nM and 270/150 nM, respectively ( Figure 7C, D). These clearly show that the presence of H2A-H2B compromises the affinity of RanGTP to Imp9/Kap114 150-to 330-fold. uptake between Imp9 alone and Imp9+H2A-H2B, Imp9+RanGTP, or RanGTP+Imp9+H2A-H2B in Panels 1, 2, and 3, respectively. Difference in D-uptake between Imp9+H2A-H2B and RanGTP+Imp9+H2A-H2B is shown in Panel 4. Boundaries and peptides from binding sites (black) and regions of conformational change (gray) are noted. Blue/Red coloring indicates a difference ≥5% with a p-value ≤0.01 in a Welch's ttest (n=3). (B) Interface between the N-terminal HEAT repeats of Imp9 and RanGTP with example D-uptake plots for Imp9 residues 26-50 and 118-134 (colored blue on structure). (C) Interface between the N-terminal HEAT repeats of Imp9 and H2A-H2B with example D-uptake plots for Imp9 residues 174-226 (colored blue on structure). For B and C, D-uptake plots show Imp9 alone (purple), Imp9+H2A-H2B (orange), Imp9+RanGTP (green), and RanGTP+Imp9+H2A-H2B (black). Error bars are ±2SD with n=3. The y-axis is 80% of the maximum theoretical D-uptake, assuming complete back exchange of the N-terminal residue. Figure 6. Fractional D-uptake differences in the C-terminal HEAT repeats of Imp9. (A) Difference in Duptake between Imp9 and Imp9+H2A-H2B, Imp9+RanGTP, or RanGTP+Imp9+H2A-H2B in Panel 1, 2, and 3, respectively. Difference in D-uptake between Imp9+H2A-H2B and RanGTP+Imp9+H2A-H2B shown in Panel 4. Boundaries and peptides from binding sites (black) and regions of conformational change (gray) are noted. Blue/Red coloring indicates a difference ≥5% with a p-value ≤0.01 in a Welch's t-test (n=3). (B) Region of conformational change in the C-terminal HEAT repeats of Imp9 upon binding H2A-H2B and/or RanGTP with example D-uptake plots for Imp9 residues 728-805. (C) Interface between the C-terminal HEAT repeats of Imp9 and H2A-H2B with example D-uptake plots for Imp9 residues 837-863, 881-888, 899-914, and 921-941 (colored blue on structure). For B and C, D-uptake plots show Imp9 alone (purple), Imp9+H2A-H2B (orange), Imp9+RanGTP (green), and RanGTP+Imp9+H2A-H2B (black). Error bars are ±2SD with n=3. The y-axis is 80% of the maximum theoretical D-uptake, assuming complete back exchange of the N-terminal residue.

The GTPase in RanGTP•Kap114•H2A-H2B primes transfer of H2A-H2B to nucleosome
We performed nucleosome assembly assays where we titrated H2A-H2B and Kap114 in the presence or absence of RanGTP into tetrasomes ( Figure   Left and right views are 180° rotations.

Discussion:
Kap114 behaves like Imp9 in all our structural, biophysical and biochemical analyses.
Their structures bound to H2A-H2B or RanGTP are very similar with only minor differences, such as no interaction between Kap114 and the H2B N-terminal tail compared to 3-5 tail residues seen bound to Imp9 (Padavannil et al. 2019).The lack of H2B tail contacts with Kap114 is consistent with previous findings that the histone tails are not important for Imp9 binding, and that the removal of the histone tails does not affect their nuclear import (Thiriet and Hayes 2001;Padavannil et al. 2019). Both Imp9 and Kap114 occlude the nucleosomal DNA-binding regions of H2A-H2B with their N-terminal HEAT repeats, and the nucleosomal histone and DNA-binding sites with their C-terminal HEAT repeats. Such extensive interfaces render Imp9 and Kap114 effective H2A-H2B chaperones that compete interactions between H2A-H2B and DNA.
However, the extensive importin-histone interactions make it difficult to release H2A-H2B in the nucleus. The mechanism of RanGTP-mediated H2A-H2B release in the nucleus is indeed unusual. Unlike other import cargos that are easily dissociated from their importins by RanGTP, the GTPase cannot release H2A-H2B from Kap114 or Imp9 (Mosammaparast, del Rosario, and Pemberton 2005;J. Baumhardt and Chook 2018;Padavannil et al. 2019). Instead, RanGTP binds to form a stable RanGTP•Kap114/Imp9•H2A-H2B complex. This ternary complex suggests that H2A-H2B may continue to be chaperoned by its importin in the nucleoplasm, in line with the generally accepted notion that there is almost no free H2A-H2B in the cell (Gunjan, Paik, and Verreault 2006;Hammond et al. 2017). Importantly, neither Kap114 nor Imp9 alone can deposit H2A-H2B onto an assembling nucleosome, and in fact they are effective competitors of histone-DNA interactions. The binding of RanGTP however causes a switch where now they release H2A-H2B to DNA and assemble nucleosomes. The stable RanGTP•Kap114/Imp9•H2A-H2B complex is thus key for H2A-H2B release in the nucleus. We showed that conformational changes in the ternary RanGTP•Kap114•H2A-H2B complex, validated by solution HDX analysis of RanGTP•Imp9•H2A-H2B, released contacts between H2A-H2B and the N-terminal HEAT repeats of the importins to make H2A-H2B available for nucleosomal interactions with DNA and H3-H4.
Although we have a cryo-EM structure of RanGTP•Kap114•H2A-H2B and it is clear how RanGTP confers accessibility or exposure of H2A-H2B for nucleosomal interactions, it is unclear if/how RanGTP might bind differently when it first encounters the binary Kap114/Imp9•H2A-H2B complex. The conformations of histone-versus Ran-bound importins are different, but there is almost no overlap in the importin residues that bind the two ligands. Most of the importin residues that contact RanGTP are accessible in Kap114/Imp9•H2A-H2B and perhaps some of them may engage the GTPase weakly in an initial encounter complex even when H2A-H2B is fully engaged with both the N-and C-terminal repeats of Kap114/Imp9 (Padavannil et al. 2019). Importin fluctuations that sample multiple super-helical conformations may then result in a more stable RanGTP-bound conformation that resembles the binary Kap114/Imp9•RanGTP structures with H2A-H2B released from the N-terminal importin repeats and engaged only with the C-terminal repeats. Alternatively, there may be no separate encounter complex. Kap114/Imp9 in the binary histone complex may sample many conformations and RanGTP may selectively bind and stabilize the conformation seen in the Kap114/Imp9•RanGTP structures. The need for importin conformational change in Kap114/Imp9•H2A-H2B to bind RanGTP is consistent with the >200fold decrease in affinity for RanGTP binding to Kap114/Imp9•H2A-H2B compared to Kap114/Imp9 alone. The decreased RanGTP affinity may be useful to limit RanGTP-binding and H2A-H2B release to locations with the highest RanGTP concentration such as at chromatin near RCC1 (Nemergut et al. 2001;Fried and Kutay 2003;Weis 2003).
Although we have revealed how RanGTP primes the release of H2A-H2B from Kap114/Imp9, many questions remain about this process in cells and about Kap114/Imp9 functions in the nucleus. It is unclear if there are additional factors in the nucleus that further modulate H2A-H2B release. Kap114 was reported to be sumoylated and the modification improves the ability of RanGTP to release cargoes Sua7 (also known as TFIIb) and TBP (Rothenbusch et al. 2012). The sumoylation site, mapped to residue K909, is located in the long acidic h19 loop that is disordered in all our Kap114 structures (h19 loop deletion in Imp9 does not decrease H2A-H2B affinity; (Padavannil et al. 2019). Sumoylation may regulate RanGTPmediated release of H2A-H2B from Kap114/Imp9 by means of intranuclear targeting. While H2A-H2B may be released directly to the assembling nucleosome, it could also be released to another H2A-H2B chaperone. Candidate chaperones include Nap1, FACT, or APLF (Chen et al. 2016;Liu et al. 2020;Corbeski et al. 2022).
It is possible that Kap114/Imp9 has roles in the nucleus in addition to H2A-H2B import and deposition into nucleosomes. Afterall, it can chaperone H2A-H2B in the nucleus when bound to RanGTP. We clearly showed that Kap114/Imp9•RanGTP cannot disassemble H2A-H2B from a nucleosome, however, it is not known if nuclear Kap114/Imp9, in the presence of RanGTP, may chaperone H2A-H2B after it is removed from the nucleosome by remodelers during replication, transcription or DNA repair (Avvakumov, Nourani, and Côté 2011;Keck and Pemberton 2012).
We have revealed how Kap114/Imp9 chaperones H2A-H2B in the cytoplasm, through the nuclear pore complex, and likely continues to do so in the nucleus in the form of the RanGTP•Kap114/Imp9•H2A-H2B complex. This knowledge is the foundation to understand how additional molecular players contribute to cytoplasmic H2A-H2B processing, nuclear import, and H2A-H2B deposition into the nucleosome. Studies by Pemberton and colleagues showed Kap114 associating with histone chaperone Nap1, which binds H2A-H2B in both the cytoplasm and the nucleus (Straube, Blackwell, and Pemberton 2010). Imp9 also co-purified with Nap1L1 and Nap1L4 in mammalian cells (Havugimana et al. 2012;Obri et al. 2014;Jiang et al. 2017

Constructs, Protein Expression and Purification
ScKap114 was cloned into two modified vectors: pGEX-4T3 (GE Healthcare) and pmalE (New England BioLabs). The pGEX-43T was modified to have a TEV cleavage site inserted between the GST tag and Kap114. The pmalE was modified to have a His-tag at the N-terminus of MBP and a TEV cleavage site after the MBP.

Pull-down Binding Assays
Pull-down binding assays were performed by immobilizing purified His6MBP or His6MBP-Kap114 on amylose resin (New England BioLab). Resin was stored in Binding Assay (BA) buffer containing 20 mM Tris-HCl pH 7.5, 150 mM NaCl, 10% (v/v) glycerol and 2 mM DTT forming a 50% amylose resin BA slurry. 100 µL of slurry with a total solution volume of 50 µL BA buffer was brought up to a total solution volume of 100 µL with a final concentration of 10 µM His6MBP or His6MBP-Kap114, 50 µM ScH2A-H2B, and/or 50 µM RanGTP. RanGTP was added after a 30 min pre-incubation period of the other components and equilibrated for another 30 min at room temperature. Amylose resin was pelleted at 16,000 RCF for 1 min at 4°C using an Eppendorf Centrifuge 5415 R and washed 3 times with 600 µL of BA buffer stored at 4°C with excess solution carefully aspirated. 100 µL of 2x Laemmli sample buffer was added, and BA samples were boiled for 5 min. 10 µL of sample was loaded onto a 12% SDS-PAGE gel. Gels were visualized using Coomassie stain.

Cryo-EM Data Collection
Cryo-EM data for Kap114•H2A-H2B and Kap114•RanGTP were collected at the Pacific Northwest Cryo-EM Center on a Titan Krios at 300 kV with a Gatan K3 detector in correlated double sampling super-resolution mode at a magnification of 81,000x corresponding to a pixel size of 0.5295 Å.
Each movie was recorded for a total of 58 frames over 3.475 s with an exposure rate of 15 electrons/pixel/s. Datasets were collected using SerialEM (Mastronarde 2005) software with a defocus range of -0.8 and -2.5 µm.
Cryo-EM data collection for Imp9•RanGTP and RanGTP•Kap114•H2A-H2B were collected at the UT Southwestern Cryo-Electron Microscopy Facility on a Titan Krios at 300 kV with a Gatan K3 detector in correlated double sampling super-resolution mode at a magnification of 105,000x corresponding to a pixel size of 0.415 Å. Each movie was recorded for a total of 60 frames over 5.4 s with an exposure rate of 7.8 electrons/pixel/s. The datasets were collected using SerialEM (Mastronarde, 2005) software with a defocus range of -0.9 and -2.4 µm.

Cryo-EM Data Processing
A total of 9,158 movies were collected for Kap114•H2A-H2B and 5,744 movies were collected for Kap114•RanGTP. Both datasets were processed using cryoSPARC (Punjani et al. 2017 with 259,220 particles -were individually subjected to Non-uniform Refinement. We analyzed the local resolution at a 0.5 FSC threshold and determined that Class 1 was the best for model building. 4,767 movies were collected for the RanGTP•Kap114•H2A-H2B complex. This dataset was also processed using cryoSPARC. The movies were subjected to Patch Motion Correction (binned twice) and Patch CTF Estimation. Generation of templates was performed as for the binary complexes and 3,401,608 particles were extracted from the Template Picks. After five rounds of 2D Classification, 604,835 particles were further processed into five Ab-Initio classes and five Heterogeneous Refinement classes. Two of the five classes contained the complex of interest and were subjected to another round of Ab-Initio and Heterogeneous Refinement which generated four new classes. From there, three classes were chosen Non-uniform Refinement: Class 1 with 3.50 Å resolution using 114,232 particles, Class 2 with 3.49 Å resolution using 95,715 particles, and Class 3 with 3.28 Å resolution using 134,915 particles. All maps had local resolution analyzed at a 0.5 FSC threshold, and Class 3 was the best one for model building.
4,920 movies were collected for Imp9•RanGTP and processed using cryoSPARC. A ½ F-crop factor was applied during motion correction followed by patch CTF estimation. A small set of ~20 frames was used to generate the initial template for particle picking. 2,474,170 particles were initially extracted from all the micrographs. The first round of 2D classification produced 904,566 particles that were subjected to another three rounds of 2D classification. 232,798 particles were included for Ab-Initio modeling followed by heterogeneous refinement. Non-uniform refinement was carried out to generate the final 3.7 Å resolution map.
Samples were thawed for 50 s immediately prior to injection into a Waters™ HDX manager in line with a SYNAPT G2-Si. In the HDX manager, samples were digested by Sus scrofa Pepsin A (Waters™ Enzymate BEH) at 15°C and the peptides trapped on a C4 pre-column (Waters™ Acquity UPLC Protein BEH C4) at 1°C using a flowrate of 100 μl/min for 3 min. The chromatography buffer was 0.1% (v/v) formic acid. Peptides were separated over a C18 column (Waters™ Acquity UPLC BEH) at 1°C and eluted with a linear 3-40% (v/v) acetonitrile gradient using a flowrate of 40 μl/min for 7 min. Samples were injected in a random order.
Mass spectrometry data were acquired using positive ion mode in either HDMS or HDMS E mode.
Peptide identification of water-only control samples was performed using data-independent acquisition in HDMS E mode. Peptide precursor and fragment data were collected via collision-induced dissociation at low (6 V) and high (ramping 22-44 V) energy. HDMS mode was used to collect low energy ion data for all deuterated samples. All samples were acquired in resolution mode. Capillary voltage was set to 2.4 kV for the sample sprayer. Desolvation gas was set to 650 L/h at 175°C. The source temperature was set to 80°C. Cone and nebulizer gas was flowed at 90 L/h and 6.5 bar, respectively. The sampling cone and source offset were both set to 30 V. Data were acquired at a scan time of 0.4 s with a range of 100-2000 m/z. Mass correction was done using [Glu1]-fibrinopeptide B as a reference. For ion mobility, the wave velocity was 700 ms −1 and the wave height was 40 V.
Raw data of Imp9, H2A-H2B, and RanGTP water-only controls were processed by PLGS (Waters™ Protein Lynx Global Server 3.0.3) using a database containing S. scrofa Pepsin A, ScRanGTP, XlH2A-H2B, and human Imp9. In PLGS, the minimum fragment ion matches per peptide was 3 and methionine oxidation was allowed. The low and elevated energy thresholds were 250 and 50 counts, respectively, and overall intensity threshold was 750 counts. DynamX 3.0 was used to search the deuterated samples for peptides with 0.3 products per amino acid and 1 consecutive product found in 2 out of 4-5 controls. Data were manually curated. Structural images were made using scripts from DynamX in PyMOL version 2.5 (Schrödinger, http://www.pymol.org/pymol); and heatmaps were made using in-house Python scripts.
To allow access to the HDX data from this study, the HDX data summary table (Figure 4 -Supplement Table 3) and the HDX state data tables (Figure 4 -Supplement Table 4, 5, 6) are included. Theoretical maximum D-uptake used in percent calculations was determined as follows: 0.96×(residues in peptide -1 for N-terminal residue -the number of prolines not at the N-terminal position). Back exchange was calculated using peptides from the Imp9, H2A-H2B, or RanGTP only sample that had plateaued (<2% difference in D-uptake at 10 3 and 10 4 s) and had a D-uptake >40%. The number reported in Figure 4 -Supplement Table 3 is (100 -the average % D-uptake for these peptides at 10 4 s). The number of peptides used in the calculation was 83, 16, 6, and 9 for Imp9, H2A, H2B, and RanGTP, respectively. The raw mass spectrometry data are available from ProteomeXchange via the PRIDE partner repository with identifier PXD037571 (Perez-Riverol et al. 2016;Deutsch et al. 2020;Perez-Riverol et al. 2022).

Fluorescent Polarization
Fluorescence polarization (FP) assays were performed in a 384-well format as previously described (J. M. Baumhardt et al. 2020;Wing, Fung, and Chook 2022). Kap114, Imp9 and H2A-  (Scheuermann et al. 2016) and fitted with a 1:1 binding model, using error surface projection method to calculate the 95% confidence intervals of the fitted data. Fitted data were exported and plotted in GUSSI (Brautigam 2015).