Border Control at the Nucleus: Biogenesis and Organization of the Nuclear Membrane and Pore Complexes

Abstract

Over the last decade, the nuclear envelope (NE) has emerged as a key component in the organization and function of the nuclear genome. As many as 100 different proteins are thought to specifically localize to this double membrane that separates the cytoplasm and the nucleoplasm of eukaryotic cells. Selective portals through the NE are formed at sites where the inner and outer nuclear membranes are fused, and the coincident assembly of ~30 proteins into nuclear pore complexes occurs. These nuclear pore complexes are essential for the control of nucleocytoplasmic exchange. Many of the NE and nuclear pore proteins are thought to play crucial roles in gene regulation and thus are increasingly linked to human diseases.

Introduction

A landmark event in the evolution of the eukaryotic nucleus was the acquisition of the nuclear envelope (NE). Formation of this distinct double lipid bilayer encased the eukaryotic genome and resulted in physical boundaries facing the cytoplasm, the endoplasmic reticulum (ER) lumen, and the nuclear chromatin. Each of these environments requires specialized NE protein and lipid compositions for execution of critical cellular functions. Of particular importance, the barrier formed by the NE restricts nuclear access and delivery of nuclear transcription products to the cytoplasm. To achieve entry and exit of macromolecules, the inner and outer nuclear membranes are fused at discrete locations to form nuclear pores. However, nuclear pore complexes (NPCs) embedded in these pores generate a selective permeability barrier to molecules. Thus, to allow nucleocytoplasmic trafficking across the border, the biogenesis of the nuclear membrane barrier is inherently linked to the biogenesis of NPCs. In this review, we will highlight the current conceptual framework for achieving this special border control through NE and NPC biogenesis mechanisms. This will include insights from the analysis of dynamics and regulation during the cell division cycle, as well as assembly and growth during interphase and in nondividing cells.

Architecture of the NE Boundary

The membrane topology of the NE is unique among eukaryotic cell membranes. It is composed of a double lipid bilayer, each forming a flat, spheroid membrane sheet. In metazoan cells, these sheets measure several hundred square micrometers in area, and they are juxtaposed by a perinuclear space with an apparently even separation of ~30 to 50 nm (Hetzer et al., 2005). The outer nuclear membrane (ONM) is continuous with the ribosome-associated endoplasmic reticulum (ER), thus allowing for direct insertion of NE membrane proteins and translocation of proteins into the perinuclear space. The inner nuclear membrane (INM) harbors a unique set of membrane proteins, perhaps as many as ~70 polypeptides, some of which interact with chromatin or, in metazoan cells, the nuclear lamina (Schirmer et al., 2003; Schirmer and Gerace, 2005). The latter is a scaffold structure of intermediate filaments with critical functions in nuclear stability and chromatin organization (Gruenbaum et al., 2003). In yeast cells, there is no evidence for an intranuclear intermediate filament network (Taddei et al., 2004). Mutations in the genes encoding INM proteins or their mislocalization are frequently linked to aberrant nuclear functions (Burke and Stewart, 2002), with growing relevance to our understanding of various human diseases. Human pathological consequences for perturbations in NE proteins range from muscular dystrophies, cardiomyopathies, and neurodegenerative disorders to the premature aging syndrome progeria (Gruenbaum et al., 2005; Worman and Bonne, 2007; Zhang et al., 2008).

In order to accommodate molecular transport across this massive membrane barrier, the INM and ONM are fused at specific sites to form aqueous pores. Pore density and distribution in intact NEs varies greatly between different cell and tissue types. In the budding yeast Saccharomyces cerevisiae, NPC density peaks in S-phase at an average of ~14.6 NPCs/mm² (Winey et al., 1997), and in cultured human HeLa cells, a density of ~11 NPCs/mm² is observed (Maul and Deaven, 1977); however, both of these cell types are notably less than the densely packed NPCs in Xenopus oocytes, where more than 50 pores are present per square micrometer (Scheer, 1973). NPCs in both the budding yeast and rat kidney cell NEs are non-randomly distributed (Maul et al., 1971; Winey et al., 1997). For budding yeast NPCs, some NE regions have relatively higher density (e.g., near spindle pole bodies during mitosis) and others show lower densities (e.g., in the NE at the bud neck of a dividing anaphase cell) (Winey et al., 1997). Based on the time course of growth arrest for yeast NPC assembly mutants, the density of NPCs in budding yeast is at least three times the capacity needed to maintain nuclear transport at a level needed for viability (Gomez-Ospina et al., 2000; Makio et al., 2009). NPC distribution can be affected by the lamina [with a Drosophila lamin Dm(0) mutant showing clustered NPCs ; Lenz-Bohme et al., 1997], by perturbations in NPC components (as in multiple budding yeast NPC mutants; summarized in Dawson et al., 2009; Doye and Hurt, 1997), or by different developmental stages that might be linked to heterochromatin state (for example, in spermatocytes; Fawcett and Chemes, 1979). Additional work is required to reveal whether regulated changes in NPC density or distribution have functional consequences in situ.

Insights into the Nuclear Pore Complex Framework

Parallel studies in multiple model systems such as fungi, amoeba, nematodes, amphibian, and human cultured cells have made direct contributions to our understanding of NPC structure, function, and biogenesis. Although there are certainly organism-specific features, the principle NPC properties are evolutionarily conserved (Stoffler et al., 2003; Tran and Wente, 2006). With the Dictyostelium nuclear pore measuring ~105 nm in diameter with ~30 nm between the INM and ONM (Beck et al., 2007), the continuous membrane of each pore is inherently highly curved. Specific pore membrane proteins (Poms) are concentrated in this membrane and presumably facilitate the anchoring of peripheral proteins termed nucleoporins (Nups) (Rout et al., 2000). A recent computational model of budding yeast NPC structure predicts that it is assembled from 456 individual pieces (Alber et al., 2007b). Taken in the context of a reported cryo-EM tomography structure at 58 Å resolution for the Dictyostelium NPC (Beck et al., 2007) and extensive biochemical data for Nup-Nup interactions (Alber et al., 2007a), the structural and compositional view of the NPC framework is emerging more and more clearly.

The ~30 distinct Nups including Poms are estimated to each be present in multiples of eight copies (based on the NPC structural symmetry) and to associate within the NPC in specific structural modules (Alber et al., 2007b; Cronshaw et al., 2002; Rout et al., 2000) (Figure 1G). Two respective outer peripheral rings are formed from the oligomerization of the heptameric yeast (y) Nup84 subcomplex (the Nup107-160 complex in metazoans [m]) (Lutzmann et al., 2002), and the central inner ring(s) is presumably formed from the association of yNup170/157, yNup188, and yNup192 (mNup155, mNup188, and mNup205). Linker Nups (yNup82 and yNic96; mNup88 and mNup93) bridge between these central and outer ring structures. Importantly, new work indicates that the central ring via yNup170 also associates with yNup53/yNup59 to connect with the transmembrane protein complex yNdc1, yPom152, and yPom34 (Flemming et al., 2009; Makio et al., 2009; Onischenko et al., 2009). A separate “membrane” ring formed from the luminal domains of these Poms opposes the luminal side of the pore (Beck et al., 2007). These rings and linkers are aligned such that eight radially displayed perpendicular spoke-like structures resolve. The cryo-EM structure also shows these assembled spokes harbor peripheral channels near the NE. Finally, the Nups of the FG family (see below), which represent at least 11 of the 30 Nups, are layered along the inner surface of the spokes and localized in distinct filaments extending from both NPC faces (Fahrenkrog and Aebi, 2003; Rout et al., 2000). The resulting ~60 MDa NPC measures ~200 nm from the cytoplasmic filaments to a nuclear basket ring, and is one of the largest protein complexes in eukaryotic cells (Stoffler et al., 2003).

Figure 1. De Novo NPC Biogenesis.

Distinct steps in the sequential self-assembly pathway are shown (supporting references are cited in the text).

(A) At the closed nuclear envelope (NE), specific Nups are assembled into subcomplexes and recruited to preassembly sites on the inner nuclear membrane (INM)or outer nuclear membrane (ONM).

(B) The RanGTPase is required on both sides of the NE to mediate nuclear localization via preexisting NPCs and to release assembly factors and/or Nups.

(C) Proteins at preassembly sites on the INM and ONM mediate close apposition of the membranes. Poms are shown here as one possible mediator of this apposition; other possible mediators are discussed in the text.

(D) Pore formation requires that the luminal leaflets of the INM and ONM resolve and fuse.

(E) Nascent pore membrane curvature could be transiently stabilized by the reticulons (RTNs).

(F) Recruitment of Nup subcomplexes results in a stable membrane coat and linkage of the Poms to the central ring Nups via yNup170/157 and yNup53/59.

(G) Assembly of the peripheral Nups forms the filamentous cytoplasmic and nuclear structures, with the permeability barrier from FG domains and a central 9 nm aqueous channel.

Recently, several groups have made significant progress on obtaining high resolution structures for essential Nups domains and cocrystals of critical interaction partners. Multiple Nups are predicted to contain β-propeller and/or α-helical domains and have structural homologies with membrane vesicle coatomer proteins such as clathrin (Devos et al., 2004, 2006). These similarities are supported by structural studies of yNup84/mNup107-160 subcomplex members and yNic96 (Berke et al., 2004; Boehmer et al., 2008; Brohawn et al., 2008; Debler et al., 2008; Hsia et al., 2007; Jeudy and Schwartz, 2007; Schrader et al., 2008) and have fueled the prior speculations that these Nups associate with and maintain the curved pore membrane. However, two alternative models are proposed for the structure that the yNup84/mNup107-160 subcomplex forms when it oligomerizes in the NPC. It could arrange as either a fence-like concentric cylinder that forms a novel membrane coat structure (Debler et al., 2008; Hsia et al., 2007) or as a coat lattice with a structure more similar to that formed by COP1/COPII on vesicles (Brohawn et al., 2008). Regardless, these studies have revealed potential evolutionary conserved principles of protein-induced or stabilized membrane curvature and are critical for our understanding of pore assembly and function. Future mutagenesis and structural studies should be able to resolve these models and validate the state of the oligomerized NPC substructures.

The overall NPC structure has a functionally defined 9 nm aqueous central channel that allows the diffusion of small molecules; however, transport of larger proteins and RNA requires a facilitated mechanism (Pante and Kann, 2002) (Figure 1G). For such soluble macromolecules, the translocation properties of the NPC are dependent on at least two components (Stewart, 2007; Weis, 2007). One element is the FG Nups that harbor domains with extensive repeats of phenylalanine-glycine (FG) or related FxFG and GLFG derivatives and are distributed throughout the central NPC channel. Second, these FG repeats provide binding sites for cargo-bound shuttling receptors. Translocation is dependent on these receptors, which have the capacity for both transient FG binding, and regulated by nucleotide-dependent switch factors for directional release/binding of cargo (Stewart, 2007). For most protein and RNA transport pathways, the RanGTPase provides this directionality control, whereas for mRNA export, the DEAD-box protein Dbp5 ATPase is required (Tran and Wente, 2006). The selective permeability properties of the NPC have been recapitulated in vitro by two biochemically distinct approaches: with the formation of a hydrogel comprised of a recombinant FG domain fragment (Frey and Gorlich, 2007), and more recently, with recombinant FG domains anchored in ~30 nm diameter nanopores of a 6 mm thick polycarbonate membrane (Jovanovic-Talisman et al., 2009). In both cases, solely the FG domains are required to generate an artificial permeability barrier. However, the FG domains themselves comprise only ~10% of the total NPC mass. What does this mean for the functional roles of the remaining ~90% of the NPC architecture? The assembled NPC framework is exquisitely designed to coincidentally provide not only a scaffold for the FG domains to position throughout the central channel, but also a platform for critical steps in transport directionality (by RanGTPase and Dbp5) and other processes (such as SUMO-modification of cargo; Terry et al., 2007) and most critically, a mechanism for forming, maintaining, and disassembling the NE pore.

De Novo Assembly of NPCs into Intact NEs

Nuclear pore biogenesis occurs in proliferating cells and also during cell differentiation or in response to changes in metabolic activities. For example, with no cell division, NPC density nearly doubles in lymphocytes upon stimulation with phytohemaagglutinin (Maul et al., 1971). NPC assembly occurs continuously during the closed mitosis of the budding yeast cell cycle (Winey et al., 1997), with the daughter cell NE showing an elevated rate of new NPC insertion relative to the mother cell NE (Shcheprova et al., 2008). Some speculation had existed regarding whether such pore biogenesis in NEs occurs de novo by INM/ONM fusion or by duplication and splitting of existing NPCs. By coupling in vitro assembly assays and high resolution imaging, it was shown that the new pores and NPCs are formed de novo in intact NEs (D'Angelo et al., 2006).

Classic electron microscopy studies have suggested that de novo NPC biogenesis is based on discrete steps at the INM and ONM (Goldberg et al., 1997). This includes potential inward dimpling of the respective nuclear membranes, fusion of the bilayers, and expansion of the nascent pore by incorporation of the peripheral membrane Nups (Figure 1). A series of very recent studies provides compelling molecular evidence for such a stepwise mechanism. First, distinct Nup subcomplexes are preassembled in the cytoplasm or on the cytoplasmic ONM face; this specifically includes Nups that comprise the cytoplasmic filaments (Makio et al., 2009). Meanwhile, on the nucleoplasmic INM face, Nups of the nuclear basket are localized and preassembled, presumably by import through existing NPCs. Interestingly, the yNup84/mNup107-160 complex is required on both sides of the NE (D'Angelo et al., 2006). The RanGTPase cycle is also required, as well as two transport receptors, yKap95 and yKap121 (D'Angelo et al., 2006; Lusk et al., 2002; Ryan et al., 2003). RanGTP interacts with the Kaps and acts as a molecular switch for binding and release of Kap-associated proteins (Stewart, 2007). The INM localization of Nups, Poms, and assembly factors might be mediated by classic Ran-Kap-dependent import through preexisting NPCs. For example, in budding yeast, yNup53 is imported by yKap121 (Lusk et al., 2002; Marelli et al., 2001). RanGTP might also be required for a cytoplasmic assembly step to release assembly factors and provide spatial cues for the delivery of Nups to the biogenesis sites on the NE (D'Angelo et al., 2006; Ryan et al., 2003).

The next critical step involves fusion of the INM and ONM, and this mechanism is the least resolved and most challenging to date. Models have long speculated that the luminal domains of Poms might facilitate pore formation. However, it is perplexing that the three known Poms in fungi (Ndc1, Pom152, and Pom34) are not required for NPC biogenesis in Aspergillus nidulans when its Nup84 subcomplex is intact (Liu et al., 2009). Potential unidentified membrane proteins might be present and compensate for the known Poms. Recent work has documented a role for the ER membrane curving proteins yRtn1, yYop1, and mRtn4a in de novo NPC assembly (Dawson et al., 2009). The yRtn1 and mRtn4a are members of the reticulon (RTN) family that are structurally related to yYop1 and metazoan DP1 (Oertle et al., 2003). These evolutionarily conserved proteins share a domain of 200 amino acids containing two hydrophobic segments, which are thought to form a wedge-like hairpin within one leaflet of a lipid bilayer (Shibata et al., 2008a). Such asymmetric insertion into the bilayer might trigger spontaneous deformation or stabilize preexisting membrane curvature (Sheetz et al., 1976). Alternatively, multiple Nups have putative membrane interaction, amphipathic α-helical domains, specifically members of the yNup84/mNup107-160 complex, yNup170, and yNup53 (Drin et al., 2007; Marelli et al., 2001; Patel and Rexach, 2008). For pore formation, the transmembrane proteins and/or the membrane-associated proteins could initiate localized membrane deformations that trigger electrostatic interactions between the lipid bilayers. Pore formation ultimately requires that the luminal leaflets of the INM and ONM be disrupted and resolved (Figure 1).

Protein-mediated forces are required not only to induce fusion but also to stabilize the highly curved pore membrane and prevent the pore from expanding catastrophically. A recent study suggests that these proteins could transiently stabilize the nascent curved pore (Dawson et al., 2009). Based on work in the budding yeast model, the yNdc1-yPom proteins interact with yNup170/yNup157 and yNup53/yNup59 proteins and mediate subsequent insertion of preassembled INM and ONM Nup subcomplexes into nascent pores (Flemming et al., 2009; Makio et al., 2009; Onischenko et al., 2009). Linkage of the Pom, RTN, yNup170/yNup157, and yNup53/yNup59 steps is further indicated by genetic suppression of the pom34Δ nup59Δ lethal mutant by RTN1 overexpression and of the rtn1Δ yop1Δ NPC clustering mutants by NDC1 or POM152 overexpression (Dawson et al., 2009). An essential event in NPC de novo biogenesis is the incorporation of the yNup84/mNup107–160 subcomplex (D'Angelo et al., 2006; Siniossoglou et al., 2000). The structural similarities between Nups and COPII proteins suggest that yNup84/mNup107–160 subcomplexes maintain the curved pore membrane by oligomerizing to form a coat (as described above). The mature NPC is ultimately formed by the sequential recruitment and insertion of the full cohort of peripheral Nups. The ordering and coordination of all these assembly steps will require a more in-depth understanding of the nearest neighbor interactions between Poms and Nups, as well as the timing of Nup recruitment during de novo assembly.

In addition to the proteins, the intrinsic properties of the NE membrane are also potentially important for NPC biogenesis. Insights into such roles have come from the analysis of yeast mutants. Inhibition of acetyl-CoA carboxylase (yAcc1) activity alters very long chain fatty acid levels, and the NE and NPC morphology and nuclear transport activity of acc1-1 mutant cells are coincidentally severely perturbed (Schneiter et al., 1996). More recently, a NPC biogenesis role for a novel ER and NE integral membrane protein yApq12 was reported (Scarcelli et al., 2007). Intriguingly, altering cellular membrane fluidity rescues the apq12Δ Nup mislocalization defect and apq12Δ acc1-1 double mutants have enhanced growth defects. This suggests that yApq12 function connects NE dynamics and NPC formation. Based on the differential rescue of apq12Δ versus nup170/nup157 mutant NPC phenotypes by altering membrane fluidity (Makio et al., 2009), there are also emerging functional distinctions for such a membrane role and the early Nup-mediated assembly steps. Concentrations of certain lipid types might promote or stabilize the pore membrane bilayer curvature and affect intrinsic membrane fluidity properties that facilitate pore formation. Understanding this coupling between the NE membranes, Poms, and Nups, will be required to pinpoint further steps in de novo NPC assembly mechanism.

Border Control Loss during Mitotic NE Breakdown and NPC Disassembly

In both S. cerevisiae and Schizosaccharomyces pombe, the NE remains intact in mitosis and the spindle is assembled within the confines of the NE. As eukaryotes evolved into multicellular organisms, a continuum of mechanisms in which the NE becomes transiently permeabilized has been developed. For instance, in the filamentous fungus A. nidulans, the NE persists but the NPCs partially disassemble, leaving scaffold structures intact in the mitotically incomplete nuclear pores. A more extreme case of NEBD can be found in Ustilago myadis, a corn smut fungus, in which the NE is ruptured by a dynein-mediated process (Straube et al., 2006), and peripheral Nups and Pom152 are dispersed during prophase (Theisen et al., 2008).

The most extreme form of mitotic NE breakdown (NEBD) is found in multicellular organisms in which nuclear integrity and compartmentalization is completely lost (Figure 2). NEBD initiates at the transition from prophase to prometaphase and is marked by retraction of the nuclear membrane from chromatin and NPC disassembly. This inherently results in nucleoplasmic and cytoplasmic contents mixing. (Lenart et al., 2003). The fate of NE components during this dramatic cellular reorganization has been analyzed in mammalian tissue culture cells (Dultz et al., 2008). NPCs disassemble rapidly in a stepwise manner: first, the dismantling of the peripheral Nups, followed by a wave of synchronous Nup dissociation. The ~30 Nups, as well as the lamins, are released into the mitotic cytoplasm either as single polypeptides or in the form of stable subcomplexes (Gerace and Blobel, 1980; Hetzer et al., 2005). Nuclear membranes are detached and completely removed from chromatin (Beaudouin et al., 2002; Salina et al., 2002) and, together with membrane proteins, dispersed into the mitotic ER (Anderson and Hetzer, 2007; Ellenberg et al., 1997; Puhka et al., 2007).

Figure 2. Cell Cycle Dynamics of the Nuclear Envelope Border.

Nuclear envelope breakdown (NEBD) and nuclear envelope (NE) formation are both observed in cultured human osteosarcoma (U2OS) cells as they progress through the cell cycle. The NE is visualized by Pom121-GFP (green), whereas the chromatin is visualized by histone H2B-cherry (red). Image courtesy of Jesse Vargas.

Whereas NEBD is often described as the catastrophic collapse of interphase cell organization, it should be kept in mind that this event reflects a highly coordinated transition from one cellular state to another. Control of NEBD by mitotic entry is likely triggered by signaling pathways that render it irreversible until anaphase (see below). Consistent with this idea, NEBD appears to be regulated at multiple levels by phosphorylation. Essentially all characterized specific NE-associated proteins, including NPC proteins, lamins, and INM proteins, are phosphorylated coincident with mitosis (Kutay and Hetzer, 2008). One of the key players, Cdk1/cyclin B, phosphorylates lamins directly and triggers lamina depolymerization (Buendia et al., 2001). In a similar manner, Nup phosphorylation seems to initiate NPC disassembly (Beausoleil et al., 2004; Blethrow et al., 2008; Favreau et al., 1996; Glavy et al., 2007; Nousiainen et al., 2006; Olsen et al., 2006). In addition to cyclin B1, several other kinases have been implicated in NEBD, including NIMA (De Souza et al., 2004), Aurora A (Hachet et al., 2007; Portier et al., 2007), and a cyclin A2/Cdk complex (Gong et al., 2007). Beyond providing a temporal signal for starting NEBD, phosphorylation of NE proteins also potentially interferes with protein-protein interactions and thereby leads to massive dissociation events within macromolecular NE structures.

Cell cycle-regulated phosphorylation of NE proteins might not be sufficient to drive NEBD. Unobstructed access to the mitotic spindle by chromosomes requires NE membrane clearing and invokes yet another cellular machine. In somatic mammalian cells, interactions of microtubules with the NE generate mechanical forces that contribute to the rupture of the lamina in a dynein-mediated process (Beaudouin et al., 2002; Muhlhausser and Kutay, 2007; Salina et al., 2002). However, nuclear disassembly can occur in the absence of microtubules (Lenart et al., 2003), indicating that additional levels of complexity likely exist.

Recent data suggest that membranes are not simply passive bystanders that get pulled away from chromatin. On the contrary, active ER membrane remodeling by the GTPase Rab 5 is critical for NEBD (Audhya et al., 2007). Furthermore, the Pom gp210 (Galy et al., 2008) as well as RTNs, a class of membrane-bending ER proteins, have been implicated in NEBD (Audhya et al., 2007). The requirement for RTNs in ER tubule formation (Voeltz et al., 2006) suggests that a tug-of-war kind of mechanism between membrane sheet versus tubule structures might be critical for NEBD. Whether the COPI coatomer complex, which has also been implicated in NEBD (Liu et al., 2003), participates in the same pathway still needs to be characterized. Finally, the RanGTPase system in conjunction with the transport receptor importin β has been shown to play a pivotal role in controlling NEBD (Muhlhausser and Kutay, 2007). Whether importin β acts as negative regulator of a NEBD-associated dissociation step remains speculative.

Mitotic Functions of Nucleoporins

One of the exciting new concepts that has emerged is the realization that a number of NE proteins with well-characterized roles in interphase also have important functions during mitosis (for review see Guttinger et al., 2009). In budding yeast, molecular rearrangements in the NPC control the subcellular distribution of molecules that direct the progression through mitosis. These M-phase-specific NPC alterations allow yNup53 to bind to the import receptor yKap121, thereby slowing its translocation through the pore and cargo release.

More dramatic examples of cell cycle-dependent functions of NPC components are found in metazoa. For instance, in interphase, the multimeric mNup107–160 complex is critical for pore function and assembly (D'Angelo et al., 2006). In contrast, during mitosis, the mNup107–160 complex and ELYS/MEL-28 can be detected in association with kinetochores, and mNup107–160 is also found with spindle poles and proximal microtubules (Galy et al., 2006; Loiodice et al., 2004; Rasala et al., 2006). This mitotic localization is functionally important, as the absence of the mNup107–160 complex perturbs bipolar spindle formation (Orjalo et al., 2006) and inhibits mNup358 (also known as RanBP2) recruitment to kinetochores. In interphase, mNup358 is part of the NPC cytoplasmic filaments and its SUMO E3 ligase activity modifies RanGAP, the GTPase activating protein for Ran (Joseph et al., 2002; Pichler et al., 2002). Interestingly, mNup358/SUMO-RanGAP complexes are also found with spindle microtubules and microtubule-bound kinetochores (Joseph et al., 2004; Salina et al., 2003), and the absence of mNup358 alters chromosome alignment and spindle assembly (Joseph et al., 2004; Salina et al., 2003). However, in a Xenopus in vitro spindle assembly assay, the mNup358/SUMO-RanGAP complex does not associate with kinetochores (Arnaoutov and Dasso, 2005). Thus, the mNup107–160 complex might have an independent role at kinetochores. Based on the interactions between mNup133 (a mNup107–160 member) and CENP-F (a kinetochore protein), this function might be linked to the dynein partners Nde1 and Nde1l, which also bind CENP-F (Vergnolle and Taylor, 2007; Zuccolo et al., 2007).

Roles for other Nups in cell cycle and mitotic progression have also been reported (Blower et al., 2005; Chakraborty et al., 2008; Jeganathan et al., 2005), and such multifunctional states might well extend to proteins associated with NE structure and function. Depletion of lamin B, a type V intermediate-filament protein and a component of the nuclear lamina, also results in mitotic spindle defects (Zheng and Tsai, 2006). Interestingly, the spindle-associated lamin B appears to be present in a membranous, matrix-like network and seems to facilitate spindle microtubule organization in a dynein-dependent manner (Ma et al., 2009; Tsai et al., 2006). Although the mechanistic details of spindle matrix function with respect to lamin B remain to be determined, these results contribute to an emerging paradigm for structural components of the interphase nucleus to have roles in mitosis.

Mitotic Membranes

As mentioned, chromosomes that are aligned in the metaphase plate are essentially membrane-free (Anderson and Hetzer, 2007; Ellenberg et al., 1997). All transmembrane NE proteins, including the three known metazoan Poms gp210, mNdc1, and mPom121, reside in the mitotic ER (Anderson and Hetzer, 2007; Ellenberg et al., 1997; Puhka et al., 2007). Thus, the precursor membrane of the interphase NE is the mitotic ER (see below). Recent three-dimensional modeling of the ER by electron tomography reveals that the mitotic ER remains an intact network of membrane tubules and is essentially free of sheets (Puhka et al., 2007). In contrast, using rapid live-cell 3D imaging, the ER was found to be largely cisternal with a small fraction remaining tubular (Lu et al., 2009). Several lines of evidence suggest that tubulation of the mitotic ER is directly linked to NE cell cycle dynamics. For instance, recent data suggests that the intrinsic propensity of the ER to oscillate between tubules and sheets is utilized during mitosis and affects the fate of the NE during C. elegans mitosis (Audhya et al., 2007).

Although not essential for viability, knockdown of the membrane-bending ER proteins RTNs and DP1/yYop1 results in reorganization of the ER network into sheets (Voeltz et al., 2006). It remains to be tested if (Puhka et al., 2007) RTN-dependent tubulation might be a critical step in NEBD. In contrast, NE reformation represents membrane sheet formation and (Anderson and Hetzer, 2008). Activities that regulate cell cycle ER tubulation must exist. As the oligomerization state of RTNs is critical for tubule formation (Shibata et al., 2008b), it is possible that a mechanism exists that would interfere with RTN oligomerization. Consistent with this hypothesis, RTN oligomers continuously form and disassemble in vivo (Shibata et al., 2008b). It is also unclear if the roles of RTNs in both NE formation and de novo NPC assembly involve distinct RTN functions. Because the membrane topology of tubules is different from a forming pore, it is likely that RTNs are organized in a different ways in these curved membranes. Indeed, yeast genetic results suggest that the yRtn1 and yYop1 role in de novo NPC formation is separate from their role in tubular ER maintenance (Dawson et al., 2009). More information about RTN function is required to address these questions.

Reassembly of the NE Boundary following Open Mitosis

Over the last years, advanced imaging methods such a high-resolution time-lapse microscopy (Anderson and Hetzer, 2007; Ellenberg et al., 1997) and EM tomography (Puhka et al., 2007) have changed the view of the major topological aspects of NE formation. It has become evident that NE formation in living cells involves the reshaping of ER membranes and not, as previously thought, vesicle fusion (Anderson and Hetzer, 2007). The idea that vesicles are the precursors for the NE largely stems from in vitro systems based on Xenopus egg extracts (Lohka and Masui, 1983). This system has been extremely useful in discovering the molecular players involved in NE formation (Hetzer et al., 2005; Vasu and Forbes, 2001). However, nuclear reconstitution in a cell-free system does not necessarily recapitulate the in vivo situation. This is particularly true if the precursor is a delicate membrane system such as the ER, which during cell fractionation is typically fragmented into vesicles. Recent studies suggest that an intact tubular ER is required for NE formation at the end of mitosis (Anderson and Hetzer, 2008). This conclusion is supported by both the requirement for RTNs and live cell imaging studies, which revealed that the ER is targeted to chromosomes via tubule-end binding and subsequently immobilized on the chromatin surface. This chromatin-bound network then flattens and seals into a closed NE. It is currently unclear if the displacement of RTNs from the ER tubules occurs by an active mechanism. Previously observed inhibition of NE formation by nonhydrolyzable GTP analogs (Boman et al., 1992) and the dependence on SNARE-mediated membrane fusion (Baur et al., 2007) could be explained by defects in ER reconstitution due to a block in ER fusion.

If NE formation does not occur by vesicle fusion, then what is the mechanism by which the nuclear membrane is formed? In vitro data suggests that targeting of membranes to chromatin might be regulated by NE-specific transmembrane proteins that bind to DNA and/or chromatin (Anderson and Hetzer, 2007; Ulbert et al., 2006). It is relatively straight forward to envision that the formation or tethering of flat membrane sheets on chromatin could be initiated by the recruitment of DNA-binding INM proteins (Ulbert et al., 2006). Furthermore, many INM proteins share a common sequence motif referred to as the LEM domain (based on its presence in Lamina-associated polypeptide [LAP2], Emerin, and MAN1) (Shumaker et al., 2001), which has been shown to bind to the chromatin-associated protein barrier-to-autointegration factor (BAF) (Segura-Totten and Wilson, 2004). The depletion of BAF leads to the mislocalization of lamin, emerin, and MAN1 and consequently to defects in postmitotic NE assembly (Gorjanacz et al., 2007). In mammalian cells, interactions between BAF and LAP2α, a soluble LEM protein, have been suggested to be critical for recruiting membrane-anchored LEM proteins such as emerin and LAP2β to chromatin in late anaphase (Dechat et al., 2004). It will be interesting to determine the rate-limiting steps of NE formation and to determine how many INM proteins are involved in enclosing the chromatin mass. Since this process occurs within a few minutes and involves massive amounts or ER membranes (Anderson and Hetzer, 2008), this mechanism is likely mediated by abundant INM and chromatin proteins. However, it is also possible that NE formation is driven by protein-DNA interactions that potentially are linked to chromatin decondensation (i.e., exposure of free DNA).

Although it is conceivable that the reshaping of the ER network into sheets is a critical step in NE formation, it is unknown how sealing of independently targeted ER patches is achieved to form a fully closed NE. However, this step is only critical if the NE is a completely closed membrane sphere. One could argue that a reforming mammalian cell nucleus does not require a full sealing of the NE as the coincidently formation of nuclear pores (i.e., holes in the envelope) is obligatory. Instead, as membranes spread on the chromatin surface, the remaining holes could become stabilized and eventually occupied by the assembly of NPCs. Alternatively, the final NE sealing step might involve chromatin-associated Nups (Walther et al., 2003a) such as mPom121 and mNdc1 (Antonin et al., 2005; Mansfeld et al., 2006; Stavru et al., 2006), which have been shown to bind to DNA (Ulbert et al., 2006) and drive NE flattening (Anderson et al., 2009), or an as yet unidentified “sealing” machinery. Future studies are required to test these models.

Postmitotic Assembly of NPCs

The assembly of NPCs into the reforming NE has been analyzed in some detail. Yet, it is currently unclear if postmitotic and interphase assembly (described above) occur by identical or distinct mechanisms (Table 1). However, there is general agreement about the critical roles of the mNup107–160 complex as well as mNup53, mNup155, and the RanGTPase in both assembly events (D'Angelo and Hetzer, 2008). At the end of mitosis, the mNup107–160 complex, mNup153, and mNup50 associate with chromatin prior to membrane recruitment (Walther et al., 2003a; Walther et al., 2003b). This might allow the formation of prepores that are subsequently engulfed by the forming NE.

Table 1.

Nuclear Envelope and Nuclear Pore Complex Biogenesis Factors

	Yeast (S. cerevisiae)	Metazoan	De Novo NPC	Postmitotic NPC	Proposed Role
NPC specific	Nup84 subcomplex	Nup107-160 subcomplex	+	+	Forms membrane coat
	-	ELYs/MEL-28	nd	+	Recruits mNup107-160 to chromatin
	Gsp1/Ran	Ran	+	+	Targeted localization/release of Nups
	Kap95	karyopherin β/importin β	+	+	Mediate Nup localization
	Nic96	Nup93	+	+	NPC biogenesis/ linking central-outer NPC rings
	Nup192	Nup205	+	+
	Apq12	-	+	nd	Nuclear membrane dynamics
	Kap121	importin-5	+	nd	Nup53 import/localization
	Ndc1	Ndc1	+	+	INM/ONM fusion
	Pom152	-	+	+	Anchors Nup subcomplexes
	Pom34	-	+	nd	Anchors Nup subcomplexes
	-	Pom121	nd	+	INM/ONM fusion, early NPC assembly steps; NE formation from ER
	Nup170	Nup155	+	+	Insertion of cytoplasmic Nups; INM/ONM fusion
	Nup157		+	+
Dual NE/NPC	Nup59	Nup35	+	+	Links yNup192/mNup205, yNup170/mNup155, and Poms;Lamin assembly
	Nup53	Nup53	+	+	Links yNup192/mNup205, yNup170/mNup155, and Poms;maintain NE integrity
	-	Pom121	nd	+	INM/ONM fusion, anchor Nups; postmitotic NE formation
	-	gp210	nd	+	Unresolved
	Rtn1, Yop1	Reticulons	+	nd	Pore membrane stabilization; NE growth/formation
NE specific	-	BAF	nr	+	NE assembly and proper lamin localization
	-	VRK-1	nr	+	Regulates BAF localization
	-	LBR	nr	+	Targeting of membranes to chromatin
	-	DNA-binding INM proteins	nr	+	Membrane spreading around chromatin

The Nup MEL28/ELYS is critical for the recruitment of the mNup107–160 complex to chromatin (Franz et al., 2007; Gillespie et al., 2007; Rasala et al., 2008). Interestingly, orthologs of MEL28/ELYS have not been identified in budding yeast (Liu et al., 2009), which might indicate a specific requirement for ELYS in postmitotic NPC assembly. In support of this hypothesis, annulate lamellae, NPCs that reside in the ER, can form without ELYS (Franz et al., 2007). Thus, recruitment of the mNup107–160 complex to chromatin represents potentially a rate-limiting step in postmitotic NPC biogenesis. Contacts between the chromatin-associated Nups and the NE membrane are most likely established by mPom121 and mNdc1 (Rasala et al., 2008). As such, the postmitotic assembly of NPCs and the NE would be carefully coordinated and in synchrony (Antonin et al., 2005).

The formation of a pore intermediate, consisting of mNup107–160, mNup53, mPom121, and mNdc1, likely spans the INM and ONM. As in de novo assembly, identification of a membrane-sensing domain in the mNup107–160 complex member mNup133 suggests that unconventional interactions might be involved in pore biogenesis and require direct protein-membrane interactions (Drin et al., 2007). Next, additional Nups, like the mNup93 and mNup62 subcomplexes, are incorporated (Dultz et al., 2008). Because members of the mNup62 complex and several other FG Nups are required for formation of the NPC permeability barrier and mediate selective transport, the transport competence of NPCs is observed concomitantly with the association of the mNup93 and mNup62 subcomplexes (Dultz et al., 2008). The depletion of Nup155 has been shown to prevent the accumulation of several nucleoporins (including Nup107 and Pom121) at the NE, suggesting that it plays an early step in NPC assembly (Franz et al., 2005). The final step of postmitotic NPC assembly involves the addition of more peripheral Nups such as mNup214, mNup153, Tpr, and mNup50, as well as the membrane protein gp210. Further information from X-ray crystallography, EM tomography, biochemistry, live-cell imaging, and modeling will be necessary to decipher the complex protein-protein networks that are established during the assembly process.

Maturation of the NE in Interphase

In addition to the doubling of pores in interphase, the NE undergoes a series of molecular maturation events between G1 and G2 such as NE expansion and NPC doubling, first reported more than 30 years ago (Maul et al., 1972). Furthermore, INM proteins that are synthesized in the ER need to be trafficked to the INM. Recent data suggests that targeted INM localization of membrane proteins involves components of the nuclear transport machinery. The INM proteins might move via NPCs through the pore membrane (King et al., 2006; Lusk et al., 2007) in a temperature- and energy-dependent process (Ohba et al., 2004). Although no consensus model has been formulated, it is possible that the lateral NPC channels identified in EM tomography provide passage channels for these proteins (Beck et al., 2007).

NE expansion might not be a simple consequence of an increased DNA content due to replication requiring a larger nucleus. Dynamic remodeling of the NE might serve additional functions. Invaginations of the NE and cell types with highly lobulated nuclei have been observed (Schermelleh et al., 2008). It is tempting to speculate that these invaginations represent a necessity to optimize NE chromatin interactions. This idea is supported by an increasing number of reports proposing active roles for NE proteins in gene regulation and chromatin organization (Akhtar and Gasser, 2007). Remodeling of the NE could be triggered by regulation of the phospholipid biosynthetic pathways. In budding yeast, nuclear membrane structure is altered by changes in the expression levels of diacylglycerol kinase yDgk1 or phosphatidic acid phosphatase yPah1 (Siniossoglou, 2009).

Further understanding of NE and NPC biogenesis is also critical given the involvement of assembly factors in particular aspects of tissue-specific development and human pathologies. A recent study identified a mutation in human Nup155, a component of the scaffold Nup93-205 complex, which underlies atrial fibrillation (AF), an inherited form of clinical arrhythmia that can lead to sudden cardiac death (Zhang et al., 2008). Another example is mouse Nup133, which was shown to play a role in embryonic development of the neural lineage (Lupu et al., 2008). Furthermore, genetic mutation of the zebrafish ELYS inhibits normal development and proliferation of the retina and the intestine (Davuluri et al., 2008; de Jong-Curtain et al., 2008). It remains to be seen if these effects reflect a role for these Nups in gene regulation. Alternatively, changes in pore number and composition might result in alterations of specific transport events critical to different tissues.

Finally, maturation of the NPC in nondividing postmitotic cells has only very recently been investigated (D'Angelo et al., 2009). Remarkably, the mNup107–160 core Nup substructures of the NPC do not turn over during the entire lifespan of differentiated cells. As Nup gene expression and protein synthesis are also diminished, an age-dependent deterioration NPC occurs, and the nuclear permeability barrier is compromised. This could have a range of impacts as the spatial segregation of signaling factors and the facilitated nuclear transport of a number of essential macromolecules is critical for cell viability. Thus, the lack of a replacement mechanism for NPC scaffold components comes at a significant physiological cost for long-lived postmitotic cells.

Perspective

Since its existence was first suspected almost 100 years ago (Kite, 1913), the nuclear membrane has been a central focus of biomedical research. Initially, it was assumed that the NE was only a passive membrane barrier separating the nucleoplasm and cytoplasm of eukaryotic cells. However, the NE and NPCs are much more than border guards. Progress over the years has gradually changed this view, and more recently, the importance of the NE and NPCs in signal transduction, gene regulation, and chromatin organization has become evident. The most recent increase in scientific interest centers around the diverse roles NE proteins play in regulating cell division, differentiation, and aging. It is of note that most of the ~100 proteins thought to specifically localize to the NE have not been fully investigated. Thus, for this new and emerging field, we predict that the potential for an accelerated pace of future discoveries in nuclear cell biology is tremendously high.