Special Reviews

Crossing the Nuclear Envelope: Hierarchical Regulation of Nucleocytoplasmic Transport

See allHide authors and affiliations

Science  30 Nov 2007:
Vol. 318, Issue 5855, pp. 1412-1416
DOI: 10.1126/science.1142204


Transport of macromolecules between the nucleus and cytoplasm is a critical cellular process for eukaryotes, and the machinery that mediates nucleocytoplasmic exchange is subject to multiple levels of control. Regulation is achieved by modulating the expression or function of single cargoes, transport receptors, or the transport channel. Each of these mechanisms has increasingly broad impacts on transport patterns and capacity, and this hierarchy of control directly affects gene expression, signal transduction, development, and disease.

In eukaryotic cells, physical separation of the nuclear genomic material from the other intracellular compartments necessitates controlled nuclear entry and exit to carry out basic biological processes. The nucleocytoplasmic transport mechanism uses specific cellular factors and macromolecular complexes to regulate such bidirectional trafficking of RNA and protein cargoes. The timely translocation of specific cargo across the nuclear envelope is especially critical for correct cell division and gene expression. The importance of faithful nuclear import and export is underscored by the role transport plays in coordinating embryonic development. Furthermore, deregulation or hijacking of transport pathways is also central to various disease states. Here we highlight the multitiered strategies for regulating nuclear influx and efflux of proteins and RNAs and the impacts of dynamic intra- and extracellular cues.

Cellular Machinery for Nuclear Import and Export

Entrance into and exit from the nucleus occurs via the nuclear pore complexes (NPCs), >60 MD macromolecular structures that form channels spanning the double lipid bilayer of the nuclear envelope (Fig. 1, A and B) (1, 2). Each NPC is equipped to facilitate both import and export of proteins and RNAs (Fig. 1C) (3). The constituent proteins of the NPC, nucleoporins (Nups), serve precise roles in NPC assembly and function (4). Structural Nups contribute to the overall NPC architecture, and pore membrane proteins (Poms) anchor the NPC in the nuclear envelope. FG-Nups constitute a third class of Nups; they contain discrete domains with multiple phenylalanineglycine (FG), GLFG (L, leucine), or FxFG (x, any) repeat motifs, separated by charged or polar spacer sequences (5). The FG-Nups presumably line the central NPC channel and extend filaments on both the cytoplasmic and nucleoplasmic faces (Fig. 1B). Additionally, the unfolded nature of the FG domains might allow multiple topological positions in the NPC (68).

Fig. 1.

The NPC mediates bidirectional nucleocytoplasmic trafficking. (A) Scanning electron micrograph of NPC structures from the cytoplasmic (upper panel) and nucleoplasmic (lower panel) sides of a Xenopus laevis ooctye nuclear envelope. The sample was a prepared as described in (80). [Image courtesy of M. Goldberg, Durham University, and R. Stick, University of Bremen] (B) Receptor-mediated transport involves sequential steps. A transport receptor recognizes a signal-bearing cargo and forms a receptor-cargo complex (1). The receptor-cargo complex then docks on the near side of the NPC (2) before engaging in sequential, stochastic, low-affinity interactions with FG-Nups during translocation (3). At the far side of the NPC, the receptor-cargo complex is disassembled (4) to deliver the cargo. For Kapβ import, complex disassembly is initiated by nuclear binding of RanGTP to the receptor. For Kapβ nuclear export, RanGTP in the export receptor-cargo complex is hydrolyzed to RanGDP by the Ran GTPase activating protein (RanGAP), resulting in complex disassembly. The Ran guanine nucleotide exchange factor (RanGEF) is localized to the nucleus and generates high RanGTP concentrations, and RanGDP is imported into the nucleus by Ntf2 (not shown). (C) Individual NPCs are capable of bidirectional movement of both proteins and RNA. Immuno–electron microscopy of a single nuclear pore from a X. laevis oocyte is shown after microinjection with gold-labeled transport substrates. Nucleoplasmin-gold (120 to 220 Å) was injected into the cytoplasm (c) and tRNA-gold (20 to 50 Å) was injected into the nucleus (n). The nuclear membranes are highlighted by the dotted blue lines. [Reprinted from (3) with permission]

Movement of ions, metabolites, and other small molecules through the NPC occurs via passive diffusion, but the translocation of cargoes larger than ∼40 kD generally requires specific transport receptors (9, 10). The FG-Nups are considered to play an intimate role in generating this NPC permeability barrier, although roles for structural Nups in formation of the barrier have also been reported (11, 12). FG-Nups are essential components of the active translocation scaffold through the NPC (13). Transport receptors are thought to engage in multiple, stochastic, low-affinity interactions with FG-repeats during translocation. The series of interactions between an FG-Nup and a transport receptor during docking and translocation are energy-independent, with nucleotide hydrolysis apparently required only during terminal release and directionality steps (10).

The biophysical nature of the permeability barrier and the mechanism of translocation remain controversial and poorly understood (13). At the core of the debate is whether the NPC permeability barrier is physical or energetic, and recent studies suggest that either or both could be the case (1, 8, 1315). As a physical barrier, the FG-Nups would interact to form a gelatinous meshwork through which only small molecules could passively diffuse (14). Translocation through such a meshwork would be mediated by transport receptors that locally dissolve the barrier and permit passage of receptor-cargo complexes. As an energetic barrier, the NPC would form a repulsive gate, repelling non–FG-binding molecules (1, 8). Receptor binding to FG-repeats would effectively allow concentration of transport complexes at the NPC, overcoming the entropic barrier and increasing the probability of transport. Finally, a two-dimensional walk model has been proposed, with a selective barrier to passive diffusion formed by the hydrophilic spacer sequences between FG-repeats or other Nups (16). This would result in the organization of a continuous FG surface allowing transport receptor-cargo complexes to dock and pivot along neighboring FGs. The intrinsic translocation and barrier properties are critical to the regulatory role of the NPC in intracellular transport and signaling, and further biophysical and cellular studies will be required to differentiate between the proposed mechanisms.

Transport receptors themselves are central to the nuclear import and export steps of recognizing signal-bearing cargoes, interacting with the NPC, and delivering cargo to its destination compartment (Fig. 1B) (17). The largest group of transport receptors comprise structurally related members of the karyopherin-β/importin-β (Kapβ/Impβ) protein family (importins, exportins, or transportins). Most Kapβs bind cargoes directly, although Kapβ1 commonly uses adaptors to recognize cargo. Specifically, Kapα isoforms serve as adaptors between Kapβ1 and cargo (18). Directionality of transport for Kapβ family members is accomplished via the small guanosine triphosphatase (GTPase) Ran (Fig. 1B) (10) and might also be affected by Nup-Kapβ binding sites (13). Non-Kapβ transport receptors include the Ran guanosine diphosphate (RanGDP) import factor Ntf2 and the general mRNA export receptor—yMex67 in yeast, mNxf1/Tap in metazoans—that functions as a heterodimer with yMtr2 or mNxt1/p15, respectively (10). Some cargoes also require more than one transport receptor, such as the Saccharomyces cerevisiae 60S ribosomal subunit (19). Additionally, a limited number of cargoes undergo receptor-independent transport; for example, the Wnt signaling molecule β-catenin directly interacts with FG-repeats to mediate its own import (20).

It is increasingly clear that the regulation of nucleocytoplasmic transport encompasses multiple levels of control and involves all the key transport components: the NPC transport channel, the transport receptors, and the transport cargo itself. As described here, the use of multiple regulation strategies allows for hierarchical levels of impact on the overall transport process (Fig. 2). Regulation at the level of cargo usually leads to small-scale changes involving only that particular cargo, whereas modifications of transport receptor or adaptor function have an intermediate effect, potentially affecting all cargoes recognized by that receptor. Changes at the NPC level are more global and are positioned to influence multiple receptors and thus large numbers of cargo simultaneously.

Fig. 2.

Nucleocytoplasmic transport is regulated at the level of the NPC, transport receptors, and individual cargoes. The pyramid shows the hierarchy of levels used to regulate nucleocytoplasmic transport, with control at higher levels (right side) having broader impacts on trafficking. Each level is controlled by multiple mechanisms, with selected in vivo examples listed (front) and illustrated in Fig. 3.

Regulation at the Level of Individual Cargo

The signals displayed by cargo molecules and recognized by transport receptors are termed nuclear localization (import) sequences (NLSs) and nuclear export sequences (NESs) (10). These have classically been defined as primary amino acid motifs that are both necessary and sufficient for transport. However, it is now clear that such signals are composed of primary sequence and secondary or tertiary structural elements, and that they are present in both proteins and RNAs. The precise sequence and the substructural context of an NLS or NES defines its specificity for the various Kaps (21). Furthermore, the receptor-protein and receptor-RNA interactions that mediate cargo localization are governed by multiple posttranscriptional and posttranslational modifications (Fig. 2, cargo). This allows for specific transport regulation of individual cargo.

Control of nuclear localization by transport signals is fully demonstrated by classic examples from metazoan NF-κB and p53 signaling. Nuclear import of the transcription factor NF-κB heterodimer p65-p50 is regulated by intermolecular masking of its NLS (22). Crystal structure studies reveal that the NF-κB inhibitor IκBα directly occludes the NLS of p65-p50 (23, 24). This results in the cytoplasmic localization of the heterodimer; however, in response to proinflammatory stimuli, IκBα is subsequently degraded and NF-κB is released for nuclear import (Fig. 3A) (22). The tumor suppressor p53 shuttles between the nucleus and the cytoplasm in a cell cycle–dependent manner (25). In response to cellular stress, p53 is retained in the nucleus via an “NES masking” mechanism whereby p53 homotetramerizes, burying the NESs and blocking interaction with the export Kapβ Crm1 (Fig. 3A). The nucleocytoplasmic shuttling of p53 is further abrogated by stress-induced poly(ADP-ribosyl)ation that prevents p53 interaction with Crm1 (26). Defects in p53 nuclear retention are associated with a number of neoplasms (27), illustrating the importance of proper cellular localization for single cargoes.

Fig. 3.

Regulation of nucleocytoplasmic trafficking is dynamic and diverse. (A) Intermolecular and intramolecular interactions (NF-κB or p53, respectively) regulate trafficking of single cargoes. (B) Transport of individual cargoes is regulated by posttranslational modifications (ubiquitination of UbcM2) and posttranscriptional modifications (mRNA splicing). (C) Expression patterns of transport receptors alters cargo translocation. The expression of Kapα subtypes dictates which cargoes (color-coded hexagons) are imported. (D) Influenza virus sequesters the mRNA export receptor mNxf1-Nxt1, inhibiting cellular mRNA export. (E) NPC composition broadly affects transport. Influenza virus down-regulates mNup98, thus removing a key binding site for cellular mRNA export factors. (F) The NPC permeability barrier blocks free diffusion of large molecules (>40 kD or ∼8 nm); smaller molecules move through independent of a transport receptor.

The affinity of a protein cargo for its cognate Kapβ receptor or Kapα adaptor influences its nucleocytoplasmic transport efficiency (28) and represents a subtle effector of transport regulation. However, cargo-receptor binding that is too strong impedes cargo delivery and release (29). More broadly, cargo-receptor complex formation can be directly affected by posttranslational changes that involve specific modifications to NLSs or NESs. The S. cerevisiae transcription factor Pho4 undergoes phosphorylation-dependent interactions with Kapβs to regulate its localization in response to phosphate availability (30). For cyclin B1, phosphorylation of its NES at the onset of mitosis leads to nuclear retention by blocking interaction with Crm1 (31). In addition to phosphorylation, posttranslational modifications such as methylation and ubiquitination can regulate the localization of a cargo. Methylation of the NLS in RNA helicase A is crucial for its proper import (32). The phosphatase and tumor suppressor PTEN is efficiently imported after monoubiquitination of specific lysine residues (33). Interestingly, the ubiquitin-conjugating enzyme UbcM2 is imported only when charged with ubiquitin, implicating proper enzyme activation as a transport “trigger” (Fig. 3B) (34). Thus, the use of posttranslational modifications to regulate protein transport allows cyclical and timely responses to cellular states.

The paradigm of discrete signal elements that regulate nuclear transport is also readily evident for various RNA cargoes. tRNA functions in the cytoplasm, and its export via direct interaction with the Kapβ exportin-t is tightly linked to its maturation state (35). Nuclear export of a subset of mRNAs is directly regulated by cis-acting RNA signal elements. Type D retroviral transcripts use a constitutive transport element (CTE), which directly binds the mNxf1 transport receptor (10). Curiously, a CTE also exists within a splice variant of the mRNA encoding mNxf1 (36). The mRNAs for some gene products linked to cell cycle control harbor a unique, conserved structural element in their 3′ untranslated region (3′UTR) that is recognized by the translation initiation factor eIF4E (37). Binding of eIF4E to this element targets these mRNAs to an mNxf1-independent export pathway. In addition to promoting export, nucleic acid sequence elements regulate RNA nuclear retention and effectively modulate gene expression. The mRNA of the tumor-associated cytokine MSF contains a 3′ UTR signal element that mediates nuclear retention until posttranscriptional processing in response to transforming growth factor–β1 signaling triggers export (38). Similarly, a 3′UTR element in an alternate transcript of the amino acid transporter mCAT2 results in nuclear retention until stress conditions allow posttranscriptional processing and export (39). RNA signal elements can also direct nuclear import; a hexameric sequence within the microRNA miR-29b is both necessary and sufficient for its import (40). Sequence and structural elements therefore directly affect transport of specific RNAs.

Essential to the maturation of mRNAs and ribosomal subunits is a series of regulated, large-scale processing steps that control interactions with appropriate transport receptor(s). This ensures that only mature, fully processed cargo is exported. For mRNA, specific proteins are cotranscriptionally recruited to form a messenger ribonucleoprotein particle (mRNP) with serial changes to the mRNP during splicing, capping, and polyadenylation (19). Regulated mRNP maturation is required for export, with deficiencies resulting in nuclear mRNP retention (Fig. 3B). A critical step in the mRNP export process is the recruitment of the transport receptor yMex67-Mtr2 (mNxf1-Nxt1) and targeting to the NPC (19). Controlled, directional release of the mRNP into the cytoplasm, and alterations to the mRNP composition during export, are potentially coordinated by spatial activation of the RNA-dependent ATPase yDbp5 through NPC-associated yGle1 and inositol hexakisphosphate (41, 42). The requirement for a soluble inositide in mRNA export might also allow modulation of transport by phospholipase C signaling (43). In a parallel manner, ribosomal subunits undergo sequential, regulated processing in the nucleolus and nucleoplasm to form rRNA-protein complexes that are selectively exported to the cytoplasm (19). Hence, cargo signals, modifications, and composition can play essential roles in controlling the transport of individual proteins, RNAs, and RNA-protein complexes.

Control via Modulation of Transport Receptors

Although changing the transport dynamics of a single cargo is accomplished in several ways, the regulation of a suite of cargoes can perhaps be more effectively accomplished by altering distinct transport receptors. S. cerevisiae has at least 17 different transport receptors, including the 14 Kapβ family members, a single Kapα, yNtf2, and the yMex67-Mtr2 heterodimer (10, 17). Metazoa have ∼20 known Kapβ family members, multiple Kapα isoforms, mNtf2, and up to four known mNxf family members (17, 44). Each of these receptors recognizes a specific subset of NLSs and/or NESs and thereby governs the transport of select cargoes.

Several studies have documented that transport might be regulated by changes in receptor or adaptor expression levels (Fig. 2, transport receptor). Differences in transport receptor expression patterns during development have been observed in Caenorhabditis elegans and Drosophila melanogaster (22). Tissue-specific expression of the three Drosophila Kapαs during gametogenesis and early embryogenesis directly regulates the cargoes that are imported (Fig. 3C) (18). Indeed, aberrant Kapα expression patterns result in defects in fertility and embryogenesis. Human Kapαs also show tissue-specific expression patterns (4547). Coordinated switching between Kapα subtypes has been implicated in regulating the neural differentiation of mouse embryonic stem cells (48). In this case, the temporal expression of Kapαs is critical for stage-specific nuclear import of transcription factors associated with developmental progression. Not only do Kapαs allow regulation of which cargoes are imported (Fig. 3C), but use of a Kapα adaptor with Kapβ broadens the dynamic range of import above that which a Kapβ can accomplish alone (49).

Deregulation of transport receptor shuttling pathways occurs in several disease states. Over-expression of Kapβ and/or Kapα family members is detected in colon, breast, and lung cancers (50, 51), and deregulation of Kapα2 expression in melanoma cells and breast cancers correlates with poor survival and prognosis (51, 52). However, it is unclear what cargoes are redistributed in cells overexpressing Kapα2 and how this affects tumor progression. In viral pathogenesis, sequestration of Kapα1 or Kapα2 by Ebola virus or SARS-CoV, respectively, impedes the cellular antiviral response (53, 54). Thus, spatiotemporal control of transport receptor levels directly affects development and is linked to disease mechanisms.

The interplay between Kapβs for NPC binding sites is also emerging as a mechanism for regulating transport. As the abundance of a Kapβ and its cognate cargo(es) varies, so too does the efficiency of nuclear import (5557). Increasing the Kapβ level increases transport rates (55, 57); however, the process is saturable for a given Kapβ (58). It is also possible that different individual transport receptors compete for a limited number of NPC binding sites. Import rates measured in intact cells are lower than rates in permeabilized cells by a factor of ∼10, presumably because most competing Kapβsare removed from the in vitro system (55). Transport receptors also show preferences for different FG domains both in vivo and in vitro, as summarized in (59, 60). Genetic manipulation of the NPC FG-repeat composition in S. cerevisiae further reveals that perturbations of single transport receptors are associated with specific FG deletions (59, 60). This indicates that the NPC has multiple independent FG pathways, each used by different transport receptors. Moreover, with up- or down-regulation of transport receptor levels, nucleocytoplasmic translocation could also be affected by competition of transport receptors for FG pathways.

Large-Scale Regulatory Mechanisms at the NPC

The FG-Nups establish avenues for efficient trafficking of specific transport receptors, as highlighted above. In addition, the structural and barrier properties of NPCs broadly influence the transport of multiple receptors and cargoes simultaneously (Fig. 2, NPC). The integrity of this NPC barrier is potentially regulated by changes in Nup composition. Both the filamentous fungus Aspergillis nidulans and the starfish Asterina miniata undergo a semi-open mitosis (i.e., incomplete nuclear envelope breakdown) and disassemble only a fraction of their Nups during mitosis (12). The specific Nups are presumably targeted for disassembly by phosphorylation and result in mitotic NPCs that are more permeable. Furthermore, organisms that undergo a closed mitosis, such as S. cerevisiae, initiate changes to the NPC structure in a cell cycle–dependent, Nup phosphorylation–mediated mechanism to alter specific transport events (61, 62). The NPC substructural relocalization of yNup53 alters yKap121 binding sites, with direct consequences on nuclear import of yKap121 cargoes (61). Because yKap121 also contributes to the proper subcellular localization of the small ubiquitin-like modifier (SUMO) protease yUlp1, this could be the basis for modified septin deSUMOylation patterns in nup53 and kap121 mutants (63). Such temporal variation in NPC composition also appears in several developmental scenarios, wherein expression patterns of mNup50 and the pore membrane protein gp210 are restricted to specific tissues in mice (64, 65). Similarly, expression of the Nup mbo (homologous to mNup88 and yNup82) is tissue-specific in Drosophila (66), and mbo expression has an inhibitory effect on Crm1-mediated export (67). Even within a single S. cerevisiae nucleus, the differential localization of NPC-associated proteins suggests that subsets of NPCs have specialized transport functions (68).

Nuclear entry is critical for replication of most DNA and some RNA viruses, including poliovirus, rhinovirus, influenza, and vesicular stomatitis virus (69, 70), with each virus targeting a subset of Nups for degradation or selective inhibition. Disruption of NPCs skews the efficiency of individual transport receptors to favor trafficking of viral components. For example, influenza virus selectively inhibits the mRNA export factors mNxf1-Nxt1 (Fig. 3D), mRae1/Gle2, and mE1B-AP5, and also down-regulates mNup98 (Fig. 3E) (70). Cellular mRNA export requires these export factors and a specific binding site on mNup98, and thus is effectively disrupted. However, as viral RNA export is not dependent on these host mRNA export factors or binding sites, it is unaffected, as summarized in (70). This viral mechanism is in direct competition with interferon-γ–mediated up-regulation of mNup98 (71). Ultimately, the balance between mNup98 induction and degradation affects the cellular outcome of the viral infection (70).

In addition to changes in Nup association affecting the translocation of specific receptors and the permeability barrier, the physical shape of NPCs might also be linked to transport regulation. Electron microscopy studies have found that NPCs vary in diameter and structure (7275). However, it is unclear whether these different structures represent assembly intermediates, subsets of NPCs with specific functions, changes in NPC structure for specialized cargo, or consequences of specimen preparation. Despite a central aqueous channel diameter of ∼10 nm (10), the NPC efficiently transports large cargoes such as ∼1 to 3 MD (∼25 nm diameter) ribosomal subunits and proteasomal precursors (19, 76). Artificial substrates up to 39 nm in diameter, such as transport receptor–coated gold particles, can also be translocated (9). Interestingly, the x-ray crystal structures of mNup58 and mNup45 indicate that these two Nups might slide along their common α-helical interaction surface (77). The interaction surface has the potential to shift by ∼11 Å. Moreover, with both mNup58 and mNup45 located in the central NPC channel, circumferential Nup sliding is estimated to potentially increase the NPC diameter by up to ∼30 Å, which might function to either regulate or accommodate large cargo translocation. The biophysical impact of this sliding on the interactions between other Nups is unclear, as are the mechanisms that could instigate these changes. Overall, there are multiple mechanisms by which changes to the NPC composition could broadly affect transport capacity and pore structure.

Next Steps in Understanding Transport Regulation

We have outlined the importance of regulated nucleocytoplasmic trafficking and have discussed how control is accomplished at multiple levels (Fig. 2). However, there are unanswered questions at each level. At the most fundamental stage, continued investigation is needed to further delineate the basic transport machinery. All the receptors used by various cellular cargoes and the precise signal sequences recognized are still not fully identified. Adding to the complexity, a single receptor can bind more than one signal motif without a readily identifiable primary consensus sequence (78). We also do not completely know how complex cargo molecules such as mRNPs are selected for targeting to the NPC through coupled maturation and quality control steps (19). Elucidating these cargo signals and how they are recognized by specific transport receptors will require multidisciplinary bioinformatic, structural, and cell biological approaches. As these cargoes and receptors are prime targets for regulation, this information could provide insight into a broad range of biological processes.

Given the hierarchical role of the NPC as the sole portal for nucleocytoplasmic exchange, understanding the NPC translocation mechanism is a priority. Specifically, studies are needed to resolve the biophysical nature of the permeability barrier and explain how translocation proceeds through this barrier (Fig. 3F). This will also require models that accommodate all the documented active and diffusive transport capacities of the NPC, as well as considerations of how interactions within and among cargoes, transport receptors, and Nups affect the stability of the NPC barrier. With this knowledge in hand, the field could make rational predictions and conduct tests for how transport would be affected by changes in Nup composition during viral infection, cell cycle transitions, signaling cascades, or cell differentiation.

There are clearly emerging examples of biological conditions under which cells broadly alter nuclear transport. This includes the interferon-γ–mediated up-regulation of mNup98, the cell cycle control of yNup53, and the change in cargo trafficking during neural stem cell differentiation (48, 61, 71). However, many of the mechanisms by which this occurs are not well defined. For example, it is not fully understood how heat shock–induced mRNA is preferentially exported under stress conditions in S. cerevisiae [as summarized in (79)]. Future studies of cell fate determination and development will reveal the prevalence of spatiotemporal regulation of transport receptor, adaptor, and Nup levels and the cargoes whose trafficking is affected. A further challenge is to integrate knowledge of this multitiered, hierarchical regulatory scheme involving cargoes, transport receptors, and the NPC into a global systems biology perspective of the coordination of cellular responses to internal and external stimuli.

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article