Ran-dependent nuclear export mediators: a structural perspective

Abstract

Nuclear export is an essential eukaryotic activity. It proceeds through nuclear pore complexes (NPCs) and is mediated by soluble receptors that shuttle between nucleus and cytoplasm. RanGTPase-dependent export mediators (exportins) constitute the largest class of these carriers and are functionally highly versatile. All of these exportins load their substrates in response to RanGTP binding in the nucleus and traverse NPCs as ternary RanGTP-exportin-cargo complexes to the cytoplasm, where GTP hydrolysis leads to export complex disassembly. The different exportins vary greatly in their substrate range. Recent structural studies of both protein- and RNA-specific exporters have illuminated how exportins bind their cargoes, how Ran triggers cargo loading and how export complexes are disassembled in the cytoplasm. Here, we review the current state of knowledge and highlight emerging principles as well as prevailing questions.

Figure 1

(A) Schematic overview of the nuclear export cycle. See main text for details. (B) The guanine nucleotide switch of Ran. Structure of Ran bound to GDP (left panel; Scheffzek et al, 1995; Partridge and Schwartz, 2009; PDB-ID 3GJ0) or GTP (middle and right panels; Vetter et al, 1999b; PDB-ID 1RRP). In the right panel, the structure of GTP-bound Ran is rotated to visualize the positively charged ‘back’ of Ran (with the twelve basic side chains represented by orange spheres). The central six-stranded β-sheet of Ran's G domain is shown in dark green; peripheral α-helices and loops are depicted in light green. Nucleotide binding is mainly accomplished by the G domain's loops (light green or colour coded), which contain several sequence motifs or residues that are highly conserved among guanine nucleotide-binding proteins (GNBPs; Vetter and Wittinghofer, 2001). The nucleotide is grey (sticks), the Mg²⁺ ion is shown as a sphere (magenta). When Ran switches between its nucleotide states, the so-called switch regions (switch I, human Ran³⁰⁻⁴⁷: red; switch II, Ran⁶⁵⁻⁸⁰: cyan; switch III, Ran^177−216: yellow) undergo drastic conformational rearrangements. There are also changes in the positioning and coordination of the Mg²⁺ ion. Upon nucleotide exchange (from GDP to GTP), switch I relocates. Helix α1b melts completely and gets partially absorbed in α1a, while β2E dissolves. The conformational changes associated with switch II are more subtle, but functionally important: nucleotide switching positions Ran's catalytic glutamine side chain (not depicted) in closer proximity to the GTP γ-phosphate and thereby prepares Ran for GTPase activation (see main text). Switch III consists of a long linker, followed by an α-helix (α6) and the highly acidic C-terminal DEDDDL motif. In RanGDP, the linker is close to the G domain and even contacts switch I, thus stabilizing the GDP state of Ran. Helix α6 packs against the ‘back’ of Ran, while the DEDDDL motif probably contacts the so-called ‘basic patch’ (blue, Ran^139−142). Upon nucleotide exchange to GTP, switch I would clash into the switch III linker, triggering displacement of switch III. The resulting ‘unlocking’ of the Ran's back is crucial for transport receptor binding (see main text, Figure 2C). Switch III of RanGTP is shown in the RanBD-bound conformation; in the absence of a RanBD, it is disordered. (C) The RanGTP sensor of Importin β. S. cerevisiae Importin β (Kap95p) is shown in complex with RanGTP (Lee et al, 2005; PDB-ID 2BKU). Impβ HEAT repeat helices are depicted as grey cylinders. HEAT repeats that interact with Ran are labelled and coloured in orange. In all contact maps, hydrophobic (distance ⩽4 Å) and polar contacts (distance ⩽3.8 Å) have been considered. HEAT 8 (in Ran-binding region 2) contains an acidic loop insertion that contacts Ran's basic back. Encircled numbers mark the distinct regions of Impβ's RanGTP sensor. Ran is shown in tube representation, coloured as in (B).

Figure 2

Ligand binding by exportins. The exportin complexes are drawn to scale and in identical orientation with respect to Ran. HEAT repeat numbering is according to the original references. See main text for more information. (A, C, E, G) Exportin–RanGTP complexes are shown as described in Figure 1C. (B, D, F, H) Left: Exportins are depicted as in Figure 1C, but Ran has been omitted. Instead, the respective cargo molecules are shown in blue (cartoon or surface representation). HEAT repeats and interrepeat loops contacting the cargo are labelled and shown in orange. Cargo regions that are contacted by Ran in the export complex are coloured in green. Right: Structures are rotated to visualize details.

Figure 3

(A–E) Comparison of the cytosolic and nuclear states of structurally characterized export mediators. The indicated transport receptors are shown in a surface representation, gradient-coloured from dark blue (N-terminus) to light blue (C-terminus). Left panels show the NTRs in their (Ran-free) cytosolic states, whereas right panels depict their nuclear (RanGTP-bound) forms. The orientations are the same with respect to HEATs 2–7. RanGTP is shown in green and the indicated transport cargoes are depicted in orange. The acidic loop of CRM1 is shown in magenta. Red arrows sketch the most apparent global conformational changes that occur upon RanGTP binding. See main text for details.

Figure 4

Binding of divergent NESs by CRM1. CRM1 is unique in that it is the only exportin that can achieve cargo specificity by recognizing very short linear and transplantable peptide signals that can be amazingly divergent—so-called nuclear export signals (NESs). The figure illustrates how CRM1 accomplishes multispecific NES recognition and shows the allowed sequence space of NES-active peptides. See main text for details. (A) Panels depict the surface of the CRM1 NES-binding site (‘hydrophobic cleft’). Hydrophilic areas are shown in blue, hydrophobic regions in grey and Cys528 (the target residue of leptomycin B) in yellow. The indicated NES peptides (orange cartoon representation) bind to the hydrophobic cleft mainly via their Φ residues (Φ⁰; Φ¹–Φ⁴; colour coded). In spite of their different Φ spacing patterns (see insets), the depicted NESs dock equivalent Φ side chains into virtually identical pockets within the hydrophobic cleft by assuming different backbone conformations. (B) NES sequence alignment based on the structures shown in (A). Colour coding is as in (A). (C) Consensus of NES peptides. Based on (A) and biochemical experiments (Güttler et al, 2010, and references therein), PKI-class and Rev-class NESs can be distinguished. At Φ⁰–Φ², they differ drastically in the preferred Φ spacing and their amino-acid preferences at the Φ positions. In PKI-type NESs not all five Φ residues are essential for NES activity—each individual Φ residue is dispensable provided that the remaining four Φ positions are sufficiently strong. Judged from this multispecificity of the hydrophobic cleft, the number of NES-active peptides is enormous.

Figure 5

Cytoplasmic disassembly of RanGTP–CRM1–NES complexes. CRM1 helices are shown as dark grey cylinders. The hydrophobic cleft (HEATs 11 and 12; dark blue) and the acidic loop (magenta) are indicated. RanGTP is depicted in green; functionally important motifs are coloured as specified. GTP (black) and the Mg²⁺ ion (magenta) are shown for orientation. Numbers in circles denote those B-helices of CRM1 that contact the acidic loop (black circles) or RanBP1 (white circle). (A) Upper: RanGTP–CRM1 interaction in the mammalian RanGTP–CRM1–SPN1 complex (PDB-ID 3GJX; Monecke et al, 2009). The acidic loop extends over HEAT repeats 10–15 and also interacts with Ran loops, including switch I. (Residues involved in this switch I interaction are shown as sticks.) Lower: The hydrophobic cleft of CRM1 is shown as in Figure 4A. The cleft is open and can thus interact with cargo. Cys528 is highlighted for orientation. (B) Upper: Overview of an export complex disassembly intermediate—the CRM1–RanGTP–RanBP1 complex from S. cerevisiae (PDB-ID 3M1I; Koyama and Matsuura, 2010). CRM1 and Ran are depicted as in (A), RanBP1 is shown in a surface representation (light grey). Note that RanBP1 and RanGTP interact in a tight molecular embrace with Ran's C-terminal switch III wrapping around RanBP1. RanBP1 also interacts with CRM1 HEAT 15 as well as the acidic loop. RanBP1 binding triggers major conformational changes in CRM1, mainly in the acidic loop, but also in the arrangement of HEAT repeats. Collectively, these changes close the hydrophobic cleft (lower panel) and thereby release the NES cargo. The loss of all major interactions between the acidic loop and Ran destabilizes the CRM1–RanGTP interaction. (C) Overlay of the RanBP1–RanGTP–RanGAP structure (Seewald et al, 2002) with the CRM1–RanGTP–RanBP1 complex (B). The structures were aligned via their RanBP1–RanGTP modules. From this overlay, only the CRM1–RanGTP–RanBP1 complex and RanGAP (transparent surface representation, light blue) are shown. Note the severe clashes of RanGAP with CRM1.