Structure of Importin13-Ubc9 complex: nuclear import and release of a key regulator of sumoylation

Abstract

Importin13 (Imp13) is an unusual β-karyopherin that is able to both import and export cargoes in and out of the nucleus. In the cytoplasm, Imp13 associates with different cargoes such as Mago-Y14 and Ubc9, and facilitates their import into the nucleus where RanGTP binding promotes the release of the cargo. In this study, we present the 2.8 Å resolution crystal structure of Imp13 in complex with the SUMO E2-conjugating enzyme, Ubc9. The structure shows an uncommon mode of cargo-karyopherin recognition with Ubc9 binding at the N-terminal portion of Imp13, occupying the entire RanGTP-binding site. Comparison of the Imp13-Ubc9 complex with Imp13-Mago-Y14 shows the remarkable plasticity of Imp13, whose conformation changes from a closed ring to an open superhelix when bound to the two different cargoes. The structure also shows that the binding mode is compatible with the sumoylated states of Ubc9. Indeed, we find that Imp13 is able to bind sumoylated Ubc9 in vitro and suppresses autosumoylation activity in the complex.

Figure 1

Structure of the Imp13–Ubc9 complex. (A) Cartoon view of the complex. Imp13 is shown in green with a colour gradient from grey (N-terminus) to green (C-terminus). Ubc9 is in purple. Secondary structure elements are labelled; (B) Cartoon view rotated 60° along the x axis. HEAT repeats are labelled from H1 to H20. These and all other protein structure figures were generated using PyMOL (http://www.pymol.org).

Figure 2

Details of the interactions between Imp13 and Ubc9. (A, B) show close up views of the interactions between Imp13 and Ubc9. In the upper left, the molecule is rendered as surface in a similar colour code and in a similar orientation as in Figure 1B. Interacting residues on Imp13 are in orange and interacting side-chains on Ubc9 are in pale blue for clarity. The views show the interaction between HEATs 1, 2 and 4 of Imp13 and the loop6 of Ubc9 (A), HEATs 7–9 of Imp13 and the loop1 of Ubc9, and between HEAT 6 and the helix α2 of Ubc9 (B). Residues of Imp13 and Ubc9 are labelled and HEAT repeats are in black (bold). Secondary structure elements of Ubc9 are labelled in purple. H bonds and charged interactions are shown as regular dotted lines and bold dotted lines, respectively. See also the corresponding electron density in Supplementary Figure S1. Residues mutated in 2C are highlighted with a red (Imp13) or blue (Ubc9) box. (C) Protein co-precipitations by GST pull-down assays. GST-tagged Imp13 wild type (wt) or mutant (labelled with a red box) was incubated with Ubc9 wt or mutant (blue box) in a buffer containing 50 mM NaCl. One-sixth of the sample was kept as input control (upper panel) and the rest was co-precipitated with glutathione sepharose beads (lower panel). Both input and pull-down samples were analyzed on Coomassie stained 15% SDS–PAGE. Lanes 1–4 show binding to GST as control. The far left lane was loaded with a molecular weight marker. The gels show some contamination between 35 and 25 KDa (marked with ^*), probably resulting from degradation of GSTImp13.

Figure 3

Conformational changes and binding mode of Imp13 in different complexes. (A–C) Conformational changes of Imp13 bound to (A) Ubc9, (B) RanGTP and (C) Mago-Y14. In the upper panel, the bound molecules are removed for clarity. Imp13 is represented as a cartoon with the same colour coding and view as in Figure 1A. The hinge point at HEAT 10 is represented as a dashed line in black. In the lower panel, Imp13 is represented as a ribbon trace while the bound molecules are shown as a cartoon. Ubc9 is in purple (A), Ran in yellow and GTP in black (B), Mago in blue, and Y14 in magenta (C). (D) Protein co-precipitation of GSTImp13 wt, of a GSTImp13 double mutant that was previously shown to impair the binding to Mago-Y14 (Bono et al, 2010) and of a GSTImp13 C-terminally truncated mutant (GSTImp13ΔC, H1–14). Both Imp13 K802E/R803E and GSTImp13ΔC mutants retain their ability to bind Ubc9 (lane 7, 10) and RanGTP (lane 9, 12). (E) On the left, competition assay of Ubc9 to Imp13–Mago-Y14 and on the right side of Mago-Y14 to Imp13–Ubc9. Both cargoes can displace each other and no concomitant binding is detected. In this competition experiment, 4 μg of GST, GST–Ubc9-Imp13 or GST–Mago-Y14-Imp13 were incubated on beads and competing amounts of Ubc9 (lanes 4 and 5) or of Mago-Y14 (lanes 7 and 8) were added (1 μg and 4 μg lanes 4, 7 and 5, 8, respectively).

Figure 4

Cargo release by RanGTP. (A) Cargo release by increasing amounts of RanGTP (lanes 4, 5, 9, 10, 12, 13) and of RanGTP + eIF1A (lanes 6, 7, 14, 15) is shown in a competition assay on Imp13–Ubc9 (lanes 4–7), Imp13ΔC (HEATs 1–14)–Ubc9 (lanes 9, and 10) and Imp13–Mago-Y14 (lanes 12–15). RanGTP promotes the full dissociation of Ubc9 bound to Imp13ΔC (lane 10). Lanes 1 and 2 show binding of RanGTP and eIF1A to GST as control. (B) Superposition of Imp13–Ubc9 and Imp13–RanGTP (PDB ID 2X19). Imp13–Ubc9 is shown in green with a similar view as in Figure 1 after a 180° rotation around the y axis, after optimal superposition of the N-termini. Imp13 in complex with RanGTP is shown in grey, Ran is shown in yellow (Supplementary Figure S4C). (C) Sausage-like representation of the conformational differences between Imp13 in complex with Ubc9, and Imp13 in complex with RanGTP (calculated according to (Cook et al, 2009)). The wider the section of the ribbon and the darker the color, the more variability exists between the two structures. The complex is in the same orientation as in panel 4B. Hinge regions are labelled.

Figure 5

Imp13 can bind Ubc9 in different sumoylation states. (A) Ubc9 structure superposition of Imp13–Ubc9 and Ubc9^*SUMO (PDB ID 2VRR). The view is similar as in Figure 1A. The catalytic cysteine (C93, as a yellow sphere) and the main SUMO-modified lysine (K14) on Ubc9 are labelled. (B) Detailed view of the superposition rotated 180° along the y axis. Marked with a circle are a loop and the very C-terminal portion of SUMO that might sterically clash with HEAT 8 and 10 of Imp13. K17 and R18 are shown, as well as K14, wherein the C-terminus of SUMO is covalently linked. (C) Pull-down analysis of the binding of Ubc9^*SUMO to Imp13 (Coomassie stained gel on the left side), revealed by western blot with an anti-Ubc9 (middle panel) and an anti-SUMO antibody (right panel). The arrow indicates the band corresponding to Ubc9^*SUMO at the expected molecular weight (lane 8) and specifically detected both by anti-Ubc9 (lane12) and anti-SUMO antibodies (lane 16) (Supplementary Figure S5B). In this experiment, SUMO is His-tagged and, therefore, in the gel shows a molecular weight higher than Ubc9. (D) In vitro sumoylation reaction in the presence of the purified E1 enzyme, SUMO and Ubc9 (wt or mutated) apo or in complex with Imp13 with or without RanGTP, and ATP. On lane 4, Ubc9 is not modified by SUMO when in complex with Imp13 whereas Ubc9 apo (lane 2) functions as SUMO substrate as shown by the appearance of a band of ∼35 KDa corresponding to Ubc9^*SUMO. A K17E/R18E Ubc9 double mutant is catalytically inactive in this condition (lane 3). The same mutant is unable to bind Imp13 (not shown and Figure 2C, lane 6). On lane 1 is the same reaction composition as in lane 2, without ATP as a negative control. (E) In vitro thioester formation reaction in the presence of the purified E1 enzyme (GSTAos–Uba), SUMO (HisSUMO in this reaction) and Ubc9 apo or in complex with Imp13 with or without ATP and in presence (reducing) or absence (non-reducing) of DTT. Ubc9 is modified by SUMO via formation of the thioester bond both in the apo form (lane 5) and when in complex with Imp13 (lane 6) as shown by the appearance of a band of ∼35 KDa corresponding to Ubc9∼SUMO. A similar reaction in reducing condition does not show the appearance of a band of the same molecular weight, confirming that the band in lanes 5 and 6 corresponds indeed to Ubc9∼SUMO. The three rightmost lanes of the gel correspond to a GST protein co-precipitation of the reactions performed in non-reducing conditions. On lane 9, it is clear that SUMO∼Ubc9 can bind to Imp13. GSTAos also precipitates on the beads but it is not bound to Ubc9 or SUMO∼Ubc9 as judged by the absence of the corresponding bands (lane 8). On lane 1 is the same reaction composition as in lane 2, without ATP and in reducing conditions, while on lanes 4 and 7 is the same reaction composition as in lane 5 and 8, without ATP and in non-reducing conditions as a negative controls. (F) SUMO conjugation of Ubc9 substrates E2-25K and RanGAP. The reaction was performed by incubating the purified E1 enzyme, SUMO (HisSUMO in this reaction) and Ubc9 apo or in complex with Imp13 in the presence either of E2-25K (lanes 1–3) or of RanGAP (lanes 4–6). Appearance of a weak band corresponding to SUMO conjugated E2-25K (∼45 KDa) in lane 3 and to RanGAP^*SUMO in lane 6, indicates that in presence of Imp13 the reaction is less efficient. In the positive controls, stronger bands appear in presence of Ubc9 apo (lanes 2 and 5). Lanes 1 and 4 are loaded with the same reactions in the absence of ATP as negative controls. On the leftmost lane of all gels the molecular weight markers were loaded.