Special Reviews

Blurring the Boundary: The Nuclear Envelope Extends Its Reach

See allHide authors and affiliations

Science  30 Nov 2007:
Vol. 318, Issue 5855, pp. 1408-1412
DOI: 10.1126/science.1142034


The past decade has seen a complete rethinking of the traditional view of the nuclear envelope as simply a passive enclosure for the chromosomes. The convergence of several lines of clinical and basic research has revealed additional roles in both signaling and mitotic progression. It is becoming apparent that the nuclear envelope defines not only nuclear organization but also that of the cytoskeleton and, in this way, integrates both nuclear and cytoplasmic architecture.

The nuclear envelope (NE) represents the interface between the nucleus and cytoplasm and is essential in maintaining the unique identity of each compartment (1). To perform this role, the NE contains channels (the nuclear pores) capable of mediating selective exchange of macromolecules. In fact, the NE displays the properties of a molecular sieve. Small molecules, corresponding to globular proteins of less than ∼40 kD, may pass across the NE relatively unimpeded. Larger molecules, however, need to display specific signals to enter or exit the nucleus (2).

The NE is composed of several elements (Fig. 1), the most prominent of which are inner and outer nuclear membranes (INM and ONM). These are separated by the perinuclear space (PNS), a regular gap of 30 to 50 nm. Annular junctions between the INM and ONM form channels or pores that traverse the NE (3). These channels are occupied by nuclear pore complexes (NPCs). Each NPC, of which there are several thousand in vertebrate somatic cells, contains multiple copies of ∼30-protein subunits or nucleoporins (Nups) and weighs in at more than 60 megadaltons (MD). Although NPCs were first recognized in the 1960s (4), their suggested role as mediators of nucleocytoplasmic transport was finally definitively demonstrated 20 years later (5).

Fig. 1.

Overview of nuclear envelope organization. Some selected INM proteins, including lamina-associated polypeptides 1 and 2 (LAP1 and LAP2) and LBR, are shown; these proteins interact with HP1 and BAF and provide links to chromatin. Lamin filaments are thought to form the basic structure of the nuclear lamina. Continuities between the INM and ONM at the periphery of each NPC are shown. The ONM, which is continuous with the ER, is characterized by cytoskeleton-associated nesprin proteins that are tethered by Sun1 and Sun2 in the INM.

Despite their continuities, the nuclear membranes are biochemically distinct. The INM contains a unique array of integral membrane proteins (6, 7), whereas the ONM shares many of its functions with the peripheral endoplasmic reticulum (ER), to which it has multiple connections. Thus the ER, ONM, and INM represent discrete domains within a single membrane system, with the PNS being an extension of the ER lumen.

The final feature of the NE is the nuclear lamina (8). In vertebrate somatic cells, this is a 10- to 20-nm-thick protein meshwork associated with the nuclear face of the INM and is primarily composed of A- and B-type lamins. These are members of the intermediate filament (IF) family and, as do all IF proteins, feature a coiled coil flanked by nonhelical head and tail domains. In Xenopus oocytes, the lamina consists of a network of IF-like filaments (9). However, the generality of this organization is uncertain, because the oocyte contains only a single major lamin species, whereas somatic cells can contain up to five. This issue is more than just academic, because lamin defects are linked to several human diseases (see below).

The standard view of the NE has been that its functions are largely passive: as a selectively permeable container for the genome and as a bystander that needs to be moved out of the way during mitosis. Recent developments, however, reveal that the NE plays far more active roles throughout the cell cycle. The recognition that certain diseases are linked to defects in both lamins and INM proteins has highlighted an essential purpose for the NE in defining interphase nuclear architecture. Related studies reveal that the NE also functions as a signaling node with an active role in mechanotransduction. Moreover, changes in nuclear lamina organization lead to altered cytoplasmic mechanics. The implication is that the NE integrates both nuclear and cytoplasmic architecture by providing a dynamic link between nuclear components and the cytoskeleton.

Mitotic Events

During the “open” mitosis of higher cells, the NE undergoes major structural reorganization (10). Because the mitotic spindle is assembled in the cytoplasm, the NE must be partially or completely disassembled, a process that is mediated by phosphorylation of multiple NE components. NE disassembly, in this way, permits spindle microtubules to gain access to the mitotic chromosomes. On completion of chromatid segregation, all of the disassembled NE components are reutilized to assemble NEs in each daughter cell.

Disruption of the nuclear membranes and dispersal of their constituents throughout the peripheral ER is facilitated by astral microtubules and ONM-associated cytoplasmic dynein, a microtubule motor protein (11). There is also a poorly understood involvement of COP1, a vesicle coat protein complex associated with the Golgi apparatus (12, 13). COP1 displays an arrangement of structural motifs, featuring β-propeller and α-solenoid folds, that is also found in the Nup107-Nup160 complex of nucleoporins, a component of the NPC central framework. This raises intriguing questions about the evolutionary and mechanistic relations among these proteins (14).

An emerging theme is the display of dual-interphase versus mitotic functions of several NE components. Both A- and B-type lamins are dispersed throughout the cell during mitosis. Although A-type lamins behave as soluble proteins, B-type lamins, which are constitutively farnesylated, remain largely membrane-associated (15). Surprisingly, the B-type lamins play an active role in mitotic progression by contributing to the formation of a spindle matrix (16).

Some Nups also display such dual functions. Nup358 is a component of short filaments that form the cytoplasmic aspect of the NPC. Following NE breakdown (NEB), however, Nup358 relocates to kinetochores (17). In the absence of Nup358, kinetochores are malformed, which leads to chromatid segregation defects (18, 19). The Nup107-Nup160 complex exhibits similar mitotic behavior (20, 21). Evidently, there is a reciprocal relation between the spindle and the NE. On the one hand, dynein and astral microtubules facilitate NEB; on the other, disassembled NE components contribute to maturation of a functional spindle.

Because budding yeast uses an intranuclear spindle, it does not undergo NEB. Nonetheless, NE rearrangements still play an essential role in yeast mitotic progression. The NPC subunit Nup53 is involved in the regulation of cytokinesis through SUMOylation of septin proteins in the bud neck (22, 23). Furthermore, the kinetochore-associated checkpoint proteins MAD1 and MAD2 are immobilized at NPCs during interphase and evidently contribute to NPC functionality (24). The latter findings, as well as those in vertebrates, raise some fundamental questions about evolutionary relations between kinetochores and NPCs.

NE reassembly commences in anaphase, with stepwise recruitment of NE components. The earliest event involves association of the Nup107-Nup160 complex with the chromatin surface (10). This is followed by the appearance of INM proteins and the formation of a continuous double membrane. These membrane components originate in the peripheral ER network (25). NPC proteins also play a crucial role in membrane assembly. Interplay between Nup107-160 and the NPC membrane protein POM121 seems to function as a membrane assembly checkpoint, which ensures that a sealed NE does not form in the absence of functional NPCs (26).

The Nuclear Envelope and Disease

The past decade has seen the emergence of links between the NE and both infectious and hereditary diseases. Viruses, in particular, have developed strategies for subverting, and even bypassing, host-cell nuclear transport pathways. Both adenovirus and influenza A employ virally encoded adaptor proteins to import their genomes into the nucleus (27, 28). Parvovirus, in contrast, resorts to transient disruption of the nuclear membranes to achieve the same result (29). Nuclear export of HIV RNA uses the leucine-rich nuclear export signal receptor exportin-Crm1, by using the HIV Rev protein as an adaptor (30). However, herpes simplex virus capsids, which are assembled within the nucleoplasm, exit the nucleus by budding through the lamina and INM into the PNS (31). In this way, herpesvirus capsids bypass NPCs and conventional export pathways. Even vesicular stomatitis virus, the replication of which is exclusively cytoplasmic, preys on the nucleocytoplasmic transport system. It uses its own matrix protein (M) to block cellular mRNA export. M protein forms a ternary complex with Rae1, an mRNA export factor, and Nup98. To counter this, both Rae1 and Nup98 expression is up-regulated by interferon (32). In this way, the NE is one of several fronts on which antiviral warfare is fought.

Undoubtedly, it is the identification of about eight different diseases (laminopathies) resulting from mutations within the lamin A (LMNA) gene, along with a further seven diseases or anomalies due to defects in other NE proteins, that has revitalized interest in the NE (33). The laminopathies can be grouped into three categories: those primarily affecting striated muscle (skeletal and cardiac) and peripheral nerves, those affecting skeletal and fat development and homeostasis, and those resulting in premature aging or progeria (33). LMNA is unique in that mutation of no other gene results in so many diverse phenotypes. This is all the more puzzling because LMNA is expressed in most adult cell types, yet different mutations result in tissue-specific pathologies. It is surprising that only one disease (autosomal dominant leukodystrophy) is linked to the LMNB1 gene (a duplication) (34). This implies that mammals may be more sensitive to perturbations of the constitutively expressed B-type lamins, a notion supported by the late embryonic or perinatal lethality observed in Lmnb1-null mice (35).

An intriguing aspect of laminopathies is that most of the pathologies only become apparent after birth, despite the fact that LMNA expression commences during embryogenesis (36). This has led to speculation that A-type lamins are important in postnatal tissue homeostasis, particularly those of mesenchymal origin that undergo steady turnover during postnatal life (37). Lamins might therefore be involved in regulating the balance between stem cell turnover and differentiation. Nevertheless, we still do not fully understand how different mutations result in tissue-specific pathologies. Clues have come from the study of cells derived from mice engineered to carry the same mutations in their Lmna gene as those causing disease in humans. Mutations causing muscle laminopathies, such as autosomal-dominant Emery-Dreifuss muscular dystrophy (EDMD2) and dilated cardiomyopathy (CMD1A), are found throughout LMNA and frequently disrupt lamin assembly (38). There are several consequences to this. First, mechanical integrity of the nucleus is compromised; the nucleus becomes weaker and more susceptible to mechanical stress (39). Second, disruption of the lamina affects mechanical properties of the cytoplasm, which results in reduction in cytoplasmic elasticity and viscosity (39, 40). Third, the centrosome is lost from the vicinity of the nucleus (40). Last, a disrupted lamina results in redistribution of heterochromatin and NPCs and affects the retention of some nuclear membrane proteins. The best example of the latter is emerin, an INM protein, which may relocate to the peripheral ER in LMNA mutant cells (41). Indeed, the X-linked form of EDMD is linked to mutations in the emerin gene (EMD) (42). This suggests that loss of emerin from the nucleus contributes to some of the dystrophic and cardiac conduction-defect laminopathies. Precisely what function (or functions) emerin performs at the NE is unclear, although evidence from mice lacking emerin suggests that it may be important for the proper function of retinoblastoma protein (pRb) and MyoD in muscle differentiation (43). Other proteins such as Sun2 and certain nesprin isoforms may also be lost from the NE when A-type lamin expression is compromised (44, 45). Their mislocalization may be linked to the changes in cytoplasmic mechanics and centrosome positioning (see below).

A fifth consequence of disrupting lamina assembly is that signaling systems, such as the stress-responsive nuclear factor κB pathway, are compromised, which makes the affected cells less resistant to mechanical stresses (46). The pleiotropic effects of lamina disruption may help explain why the muscle-based laminopathies display such variation in severity and penetrance. It may also suggest why some 50% of individuals diagnosed with EDMD carry no mutations in either the LMNA or EMD genes, and therefore, nonmuscle laminopathy may arise from defects in other proteins whose function is directly or indirectly dependent on an intact lamina (47).

At present, we have few insights into the molecular bases of the nonmuscle laminopathies, including peripheral neuropathy (CMT2B1), lipodystrophy (FPLD), and mandibuloacral dysplasia (MAD). In contrast to the muscle laminopathies, each of these diseases is associated with specific amino acid changes. None appears to affect lamina assembly, although posttranslational processing of lamin A may be affected in FPLD (48, 49). Because the amino acids that are mutated are at the surface of either the rod domain (CMT2B1) or the C-terminal globular domain (FPLD and MAD), it is possible that these may perturb interactions with other nuclear proteins. If so, such proteins remain to be identified, although for FPLD the lipogenic activator, sterol regulatory element–binding protein (SREBP-1), has been suggested as a possible candidate (50).

The last and perhaps most fascinating group of laminopathies are the progeric diseases, Hutchinson-Gilford progeria syndrome (HGPS) and atypical Werner's syndrome. The most common mutation resulting in HGPS causes an inframe deletion of 50 amino acids in the lamin A protein, resulting in the removal of a critical endoproteolytic cleavage site (51, 52). Consequently, posttranslational processing of lamin A is impaired, with the truncated lamin protein (progerin) retaining a farnesyl residue at its C terminus. Expression of progerin has devastating consequences, with affected children displaying growth retardation, skeletal and skin defects, and alopecia. Death occurs in the mid-teens due to arteriosclerosis. The principal cellular effects of progerin are an accumulation and thickening of the lamina and impaired mitosis (53, 54). These alterations may be associated with defects in DNA repair mechanisms, as well as alterations to chromatin, all of which may contribute to a decline in cell proliferation observed in cultured fibroblasts from progeric patients (5557). Any decrease in proliferative capacity could result in a failure of some tissues to regenerate, which may be the underlying basis of progeria. Programs are under way to determine the efficacy of farnesyl transferase inhibitors in treating progeria. However, it is worth noting that we have not addressed the cellular and molecular consequences of other, even more rare, LMNA mutations that are associated with HGPS or atypical Werner's and whether they, too, feature defective farnesylation (58).

The Emerging Role of the NE in Cellular Signaling

The recognition of multiple human diseases linked to defects in NE proteins has led to the reevaluation of the NE as a signaling platform. There is abundant evidence for NE involvement in the regulation of DNA replication (59) and RNA polymerase II–dependent gene expression (60, 61). Lamins are known to associate with gene regulatory proteins such as pRb (6264), barrier-to-autointegration factor (BAF) (65), histones (66), SREBP-1 (48, 50), and c-Fos (67). In addition, several integral INM proteins such as emerin, lamin B receptor (LBR), MAN1, and lamina-associated polypeptide 2β (Lap2β) interact with transcriptional repressors and/or chromatin modifiers (61). Emerin alone associates with gene-regulatory proteins such as BAF (68), germ cell–less (GCL) (65), β-catenin (69), Btf (70), YT521-B (71), and Lmo7 (72). These interactions may all serve to modulate gene expression and cell differentiation. A case in point is the nuclear retention and protection from proteasomal degradation of the hypophosphorylated active form of the tumor suppressor, pRb (73). This probably occurs through interaction of pRb with Lap2α and A-type lamins (74).

The best-characterized example of an INM protein involved in cell signaling is MAN1 (75). A highly conserved region of the MAN1 nucleoplasmic C-terminal domain mediates interaction with the MH2 domain of the R-Smads, transcriptional regulators downstream of transforming growth factor–β (TGFβ) and bone morphogenic protein (BMP) (7680). MAN1 binding of R-Smads attenuates TGFβ-BMP signaling. Although the exact mechanism for this remains unclear, it likely involves interference in Smad hetero-oligomerization and/or phosphorylation, both necessary for Smad function (75). Mutations in MAN1 are linked to osteopoikilosis, Buschke-Ollendorff syndrome, and melorheostosis (76), three autosomal dominant diseases with the shared pathology of increased bone density; they could be accounted for, in part, by reduced capacity to modulate TGFβ-BMP signaling, as well as hyperactivation of Smad transcriptional activity (78).

Of the dozen or so INM proteins characterized to date, virtually all interact with regulatory molecules, chromatin components, or both. However, there is limited evidence of their contribution to cell physiology in either health or disease. The situation will likely become even more complex as ∼60 additional, largely uncharacterized, nuclear membrane proteins (81), possibly including INM calcium channels (82, 83), are included in the equation. It seems increasingly clear that the signaling and regulatory pathways that form a nexus at the NE will be deciphered using systems-based approaches.

Two recently characterized classes of INM and ONM transmembrane proteins interact across the PNS and associate with both the lamina and cytoskeleton (Fig. 2). The notion of such translumenal interactions arose from studies in Caenorhabditis elegans, where localization of a large ONM actin-binding protein, Anc-1, was found to be dependent on Unc-84, an INM protein (84). Unc-84 is a member of the SUN domain family of proteins (85), of which there are at least five in vertebrates. Two of these, Sun1 and Sun2, are INM proteins and are expressed in somatic cells. Both Sun1 and Sun2 interact with A-type lamins (44, 45, 86, 87) and function as tethers for members of the nesprin family of spectrin-repeat proteins in the ONM. The nesprins (also known as Syne proteins), Anc-1, and a Drosophila ONM protein, Klarsicht, all feature a short, highly conserved C-terminal KASH domain (Klarsicht, Anc-1, Syne homology), which resides within the PNS (88). It is the KASH domain that interacts with the Sun1/2 lumenal domain and forms a translumenal link analogous to that proposed for Unc-84 and Anc-1. The nesprins are encoded by at least three genes and encompass a bewildering array of splice variants (8992). The largest isoforms (up to 1 MD) of nesprins 1 and 2 contain N-terminal actin-binding domains, whereas nesprin 3 contains a binding site for plectin, an IF-associated cytolinker. Some smaller nesprin isoforms are also found in the INM. Together, SUN proteins and ONM nesprins comprise the LINC complex (linker of nucleoskeleton and cytoskeleton) (44, 45, 86, 87). Emerin might also be considered a peripheral component of the LINC complex because it associates with both INM nesprins (9395) and lamins (68, 96), as well as with nuclear actin (9799), nuclear myosin I, and αII-spectrin (100). Despite similarities, Sun1 and Sun2 are segregated within the INM. Sun1, in particular, is concentrated in a halo around NPCs (101). Although the significance of this localization is not understood, it does provide an explanation for the reported association of microfilaments and IFs with NPCs (102).

Fig. 2.

The LINC complex in vertebrate cells may integrate nuclear components, including chromatin domains (blue), with cell surface and extracellular structures, including focal adhesions and junctional complexes. This is mediated by elements of the cytoskeleton, including both microfilaments and intermediate filaments. The LINC complex contains ONM nesprins that are tethered, via translumenal interactions, by INM SUN proteins. Sun1, shown in the diagram, likely exists as a dimer. Its nucleoplasmic domain binds A-type lamins and histone H2B, as well as the chromatin modifier hALP (108).

A decade ago, mechanical coupling of the nucleus and extracellular matrix (ECM) was observed (103). Physical displacement of cell surface integrins caused deformation of the nucleoplasm at a distance of several microns. The LINC complex now provides a mechanism to explain these observations and contributes to a model in which the cytoskeleton and ECM are in physical association with the NE and nucleoplasmic contents. This has obvious implications in terms of mechanotransduction and could provide a molecular basis for the changes in cytoplasmic organization and mechanics that are observed in certain laminopathy models (39, 40). Conversely, extracellular cues could directly modulate nuclear positioning, chromatin remodeling, or even gene expression. Perturbation of LINC complex proteins alters positioning of nuclei in skeletal muscle (104, 105). During cell migration, is the nucleus simply a passenger or a central anchor on which the cytoskeleton can organize to extend or retract processes? Deciphering the mechanisms behind such relations and their role in cell physiology seems likely to occur during the next phase in our ongoing quest to understand the NE.


With the exception of its periodic convulsion during mitosis, the NE presents us with a seemingly bland façade, which buttresses the widely held view that the role of the NE is largely passive. However, studies from several quarters have led to a reevaluation and expansion of our understanding of NE function. The link between human diseases and NE proteins has provided unexpected insight into their roles in normal cellular physiology. Although A-type lamins have been considered little more than NE structural proteins, such a simplistic view is incapable of accounting for all of the laminopathy-associated pathologies. Studies on LMNA-associated disease mechanisms have led to the finding that alterations in nuclear lamina organization can cause cytoskeletal changes potentially involving the LINC complex of SUN proteins and nesprins. This opens up possibilities of signaling routes across the NE that may bypass NPCs. On the other hand, certain nucleoporins display alternate mitotic functions that are essential for mitotic progression. We are also beginning to recognize a complex interplay between disparate NE components. Assembly of NPCs during telophase is coupled to membrane assembly by what has all the hallmarks of a Nup-dependent checkpoint. Sun1 and, hence, a subpopulation of LINC complexes are associated with NPCs and provide them with a connection to both the nuclear lamina and the cytoskeleton. This synopsis only scratches the surface of the complex networks of interactions that are being revealed at the nuclear periphery. During the coming years, we may anticipate further elucidation of both structural and signaling networks at the nuclear margins and renewed vision of the NE as a central element in cellular physiology.

Note added in proof: Two recent papers by P. Muhlhausser and U. Kutay (106) and G. Salpingidou et al. (107) further enhance our view of NE-cytoskeletal interactions.

References and Notes

View Abstract

Navigate This Article