Sculpting nuclear envelope identity from the endoplasmic reticulum during the cell cycle

ABSTRACT

The nuclear envelope (NE) regulates nuclear functions, including transcription, nucleocytoplasmic transport, and protein quality control. While the outer membrane of the NE is directly continuous with the endoplasmic reticulum (ER), the NE has an overall distinct protein composition from the ER, which is crucial for its functions. During open mitosis in higher eukaryotes, the NE disassembles during mitotic entry and then reforms as a functional territory at the end of mitosis to reestablish nucleocytoplasmic compartmentalization. In this review, we examine the known mechanisms by which the functional NE reconstitutes from the mitotic ER in the continuous ER–NE endomembrane system during open mitosis. Furthermore, based on recent findings indicating that the NE possesses unique lipid metabolism and quality control mechanisms distinct from those of the ER, we explore the maintenance of NE identity and homeostasis during interphase. We also highlight the potential significance of membrane junctions between the ER and NE.

KEYWORDS:

• Cell cycle
• endoplasmic reticulum
• ER–NE junction
• mitosis
• nuclear assembly
• nuclear envelope
• nuclear pore complex

Introduction

The nucleus, first identified by Robert Brown in 1833 [1], serves as the principal defining organelle within a eukaryotic cell. This organelle harbors genetic material and is surrounded by the nuclear envelope (NE). The NE consists of two lipid bilayers called the inner and outer nuclear membranes (INM and ONM, respectively) and separates the nucleus from the cytoplasm. The NE is perforated by nuclear pore complexes (NPCs), which are large protein channels composed of multiple copies of nucleoporins that mediate macromolecular transport across the NE. The NE acts as a physical barrier that protects the genome within the nucleus and plays a crucial role in maintaining the integrity and functionality of the nucleus.

The outer membrane of the NE is continuous with the endoplasmic reticulum (ER) [2,3], where most lipids and transmembrane proteins are synthesized. Despite its continuity with the ER, the NE contains a distinct set of transmembrane proteins that define its identity. Proteins, including lamin B receptor (LBR), emerin, and SUN domain proteins, localize specifically to the INM through their interactions with chromatin and other nuclear scaffold proteins [4]. Proteins called nesprins localize to the ONM through their interactions in the perinuclear space with SUN proteins at the INM. Nesprins and SUN proteins form the linker of nucleoskeleton and cytoskeleton (LINC) complex [4], which stabilizes the NE against cytoplasmic forces and contributes to nuclear positioning and mechanotransduction [5]. The INM proteins are involved in chromosome organization and gene expression, and play important roles in cellular differentiation and tissue development [6]. Dysfunction of NE proteins leads to human diseases including laminopathies, muscular dystrophies, and premature aging disorders [6–8].

During open mitosis in higher eukaryotes, the NE undergoes disassembly in a process known as nuclear envelope breakdown (NEBD), which leads to the dispersal of NE and NPC proteins to the mitotic ER and cytosol [4]. Toward the end of mitosis, when the chromosomes segregate into daughter cells, the NE and NPCs reassemble around the separated sets of chromosomes. Efficient reformation of a functional NE is crucial for restarting essential nuclear functions such as transcription and replication, and maintaining genome integrity. Here, we explore the current understanding of NE disassembly, the fate of disassembled NE components, and the re-establishment of the functional NE during open mitosis to examine how NE identity is lost and regained. Additionally, we discuss how NE identity is maintained within the continuous ER–NE endomembrane system throughout interphase. Recent findings that the NE has a distinct lipid composition and metabolism from the ER [9] and that the junctions connecting the NE and the ER have highly-constricted morphology [10] open up the intriguing possibility that a novel mechanism might keep the NE in a distinct environment.

NE breakdown and its absorption into the mitotic ER: is NE identity completely lost?

In eukaryotic cells, there are different types of mitosis depending on the degree of NE disassembly. In cells that undergo closed mitosis, such as Saccharomyces cerevisiae and Schizosaccharomyces pombe, the mitotic spindle assembles inside the nucleus and the NE remains intact. In cells with semi-open mitosis, such as Drosophila melanogaster and Caenorhabditis elegans, the NE does not completely disassemble but fenestrates so that spindle microtubules emanating from centrosomes can reach the chromosomes. During both closed and semi-open mitosis, NE components, such as transmembrane nucleoporins and lamins, are partially disassembled in the NE [11–13]. In contrast, in cells that undergo open mitosis, such as mammalian cells, the NE fully disassembles (in a process called nuclear envelope breakdown, NEBD) during mitotic entry, detaches from the chromatin, and disperses into the mitotic ER network [4]. NEBD allows cytoplasmic spindle microtubules to access chromosomes, which is crucial for chromosome alignment on the metaphase spindle and subsequent segregation during mitosis [14,15]. In this section, we describe the molecular regulation of NE disassembly during open mitosis and discuss whether the NE identity is completely lost.

NEBD is induced by the mitotic phosphorylation of NPCs, lamins, and INM proteins, which triggers their disassembly and release from the nuclear periphery. Mitotic kinases such as cyclin-dependent kinase 1 (CDK1) and polo-like kinase 1 (PLK1) phosphorylate the NPC components NUP98 and NUP53 during prophase, which promotes kinase recruitment to the NPCs to phosphorylate other nucleoporins [16–19]. The kinase recruitment and multivalent phosphorylation of nucleoporins disrupt the binding interfaces between neighboring nucleoporins, facilitating NPC disassembly into subcomplexes [14]. The NPC disassembly occurs very rapidly; within 5 min most soluble nucleoporins are dispersed in the mitotic cytosol, and transmembrane nucleoporins are absorbed in the mitotic ER [20]. A sub-fraction of soluble nucleoporins, such as members of the NUP107–160 complex and NUP358, relocate to kinetochores during mitosis to function in spindle assembly and chromosome segregation [21]. For nuclear lamins, their mitotic phosphorylation causes depolymerization, leading to the dispersal of A-type lamins across the cytoplasm, while B-type lamins remain associated with mitotic ER membranes [22,23]. For INM proteins, such as lamin B receptor (LBR), lamina-associated polypeptide 2 (LAP2), emerin, MAN1 domain (LEM domain) proteins, and SUN1, their phosphorylation disrupts their interactions with chromatin, the nuclear lamina, and other INM proteins [24,25]. This disruption allows the INM proteins to separate from the nuclear periphery. Phosphorylation of the barrier-to-autointegration factor (BAF), which in its unphosphorylated form connects LEM domain proteins to chromatin in interphase, also promotes the disruption of the links between LEM domain proteins and chromatin during mitosis [26]. A list of mitotic kinases/phosphatases and their substrates involved in NE breakdown/assembly is summarized in another review [24].

In addition to the mitotic phosphorylation of NE components, the cytoskeleton also plays a key role in NEBD. During the late G2 phase, the centromere protein CENP-F and the nucleoporin NUP358 are phosphorylated by CDK1, which allows the minus end-directed motor dynein to associate with NE [27]. The dynein-dependent forces pull parts of the NE membranes along microtubules toward the centrosomes, facilitating NE separation from chromatin during prometaphase [28–31]. The actin cytoskeleton also contributes to NEBD. In starfish oocytes, the F-actin shell has been shown to assemble in the lamina and promote NE separation from the lamina via ARP2/3-driven actin protrusions [32]. In human U2OS and RPE cell lines, an actin network is formed around the NE during prophase/prometaphase via interactions with the outer nuclear membrane proteins, nesprins [33]. While the contraction of this actin network by myosin-II facilitates chromosome congression to the spindle equator in a microtubule-independent manner [33], the extent to which the actomyosin network contributes to NE disassembly in mammalian cells remains unknown.

Whether the NE completely disassembles or remains partially intact during open mitosis has been unclear. Previous studies using conventional live-cell and immunofluorescence microscopy have shown that most of the NE proteins are evenly distributed throughout the mitotic ER network and cytoplasm [20,28,30,34–36], with the exception of the INM proteins SAMP1 and SUN2 that are partially enriched at the spindle poles [37,38] and members of the NUP107–160 complex and NUP358 that partially relocate to kinetochores [21]. However, recent studies using advanced light and electron microscopy and biochemical approaches have suggested that Y-complexes of the NPC remain partially intact and are associated with the mitotic ER after NEBD (Figure 1, metaphase) [39,40]. Thus, some NPC subcomplexes may not disassemble into individual nucleoporins after phosphorylation by mitotic kinases. Instead, they appear to remain partially assembled and be ready for incorporation into newly forming NPCs during mitotic exit (Figure 1, metaphase). The LINC complexes may also remain partially assembled in the mitotic ER during metaphase. A study using immunoprecipitation showed that while SUN1 loses its connection with the nuclear lamina during mitosis and is dispersed into the mitotic ER, its self-oligomerisation and interaction with nesprin-2 are maintained [25]. This suggests that while the INM proteins are dispersed in the mitotic ER due to mitotic phosphorylation, the LINC complexes, at least a fraction of them, remain assembled within the ER lumen throughout mitosis (Figure 1, metaphase).

Figure 1. Dynamic remodeling of the endoplasmic reticulum (ER) and nuclear envelope (NE) from metaphase to interphase.

In metaphase, the NE is absorbed into the ER (green), which is excluded from the chromosome mass (dark grey). Y-complexes of the nuclear pore complex (NPC) and the linker of nucleoskeleton and cytoskeleton (LINC) complexes remain partially intact and associated with the mitotic ER. In anaphase, the ER membranes begin to interact with the surface of segregating chromosomes. In telophase, the membranes enclose the chromosomes and the NPCs assemble in the reforming NE. The central spindle (light blue) is present in anaphase/telophase and excludes most ER membranes, except for some tubular ER. In interphase, the NPCs are fully assembled, and LINC complexes are established throughout the NE. The left half of the cells show the ER morphology of cell types with a higher sheet abundance (such as HeLa cells), while the right half show cell types with a higher tubular ER morphology (such as U2OS cells). The ER–NE junctions begin to become constricted in telophase and form a constricted morphology in interphase. The ER–microtubule association is present throughout all mitotic stages (highlighted in anaphase). ONM: outer nuclear membrane; INM: inner nuclear membrane.

In summary, NEBD is driven by (i) the disassembly of NPCs and the release of lamins and INM proteins from the chromatin induced by mitotic phosphorylation, (ii) the pulling force of microtubules via NE-associated dynein, and potentially (iii) actomyosin network contraction. After NEBD, soluble NE components are dispersed within the cytosol, whereas most membrane-bound NE components are absorbed by the mitotic ER via diffusion. During open mitosis, NE identity is transiently lost, but the NPCs and LINC complexes remain partly assembled.

The mitotic ER: its exclusion from chromosomes, role in spindle function, and morphology

After NEBD, NE components absorbed into the mitotic ER are kept away from the chromosome/spindle area until mid/late anaphase, when they reassemble around the chromosomes (Figure 1, metaphase/anaphase). This ER exclusion from the chromosome/spindle area is important because if the chromosomes are misaligned outside the ER exclusion zone at metaphase, they are surrounded by ER membranes, which prevents the efficient attachment of spindle microtubules to chromosomes and causes a higher tendency for chromosome missegregation and micronucleus formation [41]. Spindle microtubules and ER-shaping proteins are key players in excluding the ER from chromosomes and in organizing the mitotic ER architecture.

During mitosis, microtubules reorganize into mitotic spindles, from which the microtubule plus ends extend toward the periphery of the cell and cell center. While the ER is tethered to microtubules in interphase, ER–microtubule tethering is mostly abolished during mitosis, as the ER-localized microtubule-binding proteins STIM1 and CLIMP-63 are phosphorylated [42–44]. Overexpression of a phosphorylation-deficient mutant of STIM1 results in ER association with spindle microtubules in mitotic cells [42], indicating that mitotic phosphorylation of STIM1 contributes to ER exclusion from the spindle microtubules. In addition, the ER membrane proteins REEP3 and REEP4 are suggested to actively clear the ER from the central spindle by transporting ER membranes along the microtubules toward the spindle poles [45]. Knockdown of REEP3/4 inhibits ER exclusion from the chromatin/spindle area and causes defects in chromosome segregation, cytokinesis, and NE morphology [45], highlighting the importance of ER clearance from the chromatin/spindles for the subsequent progression of mitosis. How REEP3/4 activity is regulated during the cell cycle and whether it is regulated via mitotic phosphorylation remains elusive. The amount of ER membranes seems to also affect the efficient clearance of the ER from the mitotic spindle area. When ER membrane production is increased by the knockout of the lipid metabolism regulator CTDNEP1, chromosome movement speed slows down and misaligned chromosomes cannot be corrected effectively, leading to micronuclei formation in daughter cells [46]. Thus, modulation of ER abundance as well as active ER exclusion from the spindle area ensures faithful chromosome segregation and NE reformation during mitotic exit.

While most of the ER is excluded from spindle microtubules during cell division, a small fraction is observed at the spindle poles and a subpopulation of microtubules during mitosis (Figure 1, anaphase) [47–49]. When ER membranes in the vicinity of the spindle poles are exclusively and acutely disrupted in mitotic cells in Drosophila embryos, the spindle becomes smaller, spindle poles detach from the microtubules, and sister chromatid separation velocity slows down [47]. In C. elegans embryos, ER membranes associated with centrosomes have been shown to regulate microtubule dynamics and spindle function, and potentially assist in NE breakdown [50]. These data suggest a general function of the ER membranes that are associated with spindle poles in terms of spindle size and function in metazoans. More studies, especially in mammalian cells, are required to address the specific functions of the ER associated with spindle poles and spindle microtubules during cell division.

The proteins that regulate the ER–spindle pole association are not well understood, as most ER–microtubule-binding proteins are phosphorylated during mitosis to inhibit their binding to microtubules. A recent study identified a kinase called TAOK2 (thousand and one amino acid kinase 2) as a new ER–microtubule tether that regulates the association of the ER with microtubules during interphase and mitosis [51]. TAOK2 is active in mitosis, and the expression of its kinase-dead mutant increases ER tethering to spindle poles in HEK293T cells. The increased ER–spindle pole association causes severe mitotic defects, such as monopolar, multipolar, and aberrant spindles. In TAOK2-knockout cells, ER tethering to the spindle poles is drastically reduced, and bipolar spindle formation is impaired. Therefore, unlike the known ER-microtubule tether proteins such as STIM1 and CLIMP-63, TAOK2 remains associated with the ER during mitosis and ensures proper spindle function by controlling the amount of ER at the spindle poles through its kinase activity. Whether TAOK2 is the only ER-microtubule tether that is active in mitosis remains to be explored. Interestingly, the INM proteins SAMP1 (also called TMEM201) and SUN2 have been reported to contribute to mitotic spindle formation. A small fraction of SAMP1 and SUN2 are enriched at the spindle poles during mitosis [37,38]. SAMP1 interacts with γ-tubulin, and its depletion decreases levels of β- and γ-tubulin on the mitotic spindle, which leads to an increase in chromosome missegregation and prolonged metaphase stages [38]. Knockdown of SUN2 reduces the number of astral microtubules, causes elongation of spindles, and prolongs mitosis duration [37]. These new studies have revealed that, although mitotic ER is largely excluded from the chromosome/spindle area, a small fraction of ER and the associated proteins contribute to the establishment and function of the mitotic spindle. It remains to be determined whether and how TAOK2, SAMP1, and SUN2 work cooperatively and if additional factors are involved.

How the continuous membrane network of the ER is shaped and maintained in interphase is well understood. The ER tubules are maintained by membrane curvature-stabilizing proteins including reticulons and REEPs [52], the ER sheets are stabilized by ER-shaping proteins including CLIMP-63 and SIGMAR1 [53,54], and fusion between ER tubules is mediated by the dynamin-related guanosine triphosphatase (GTPase) atlastin [55,56]. The transition from interphase to mitosis leads to significant changes in ER morphology. How exactly the ER morphology changes from interphase to metaphase has been the subject of debate [35,57–59]. Nonetheless, the consistent results in these studies are that (i) the ER remains continuous during mitosis and (ii) the mitotic ER contains holes (or fenestrations). Membrane fenestrations in the mitotic ER are suggested to be critical for the efficient assembly of NPCs during NE reformation in late anaphase [39,60,61]. The morphological changes in the ER during the transition from interphase to mitosis are partly due to the drastic reorganization of microtubules into spindles, as well as the dissociation of the ER from microtubules via mitotic phosphorylation. Furthermore, the mitotic phosphorylation of ER-shaping proteins such as reticulon-4 (RTN4) and lunapark (LNPK) may regulate ER reorganization during mitosis [62,63]. RTN4 promotes the formation of ER tubules, and its knockdown leads to a significant increase in the proportion of sheet-like ER during interphase [52,64]. In-silico predictions have identified kinase recognition motifs in RTN4 (e.g. substrate motifs of CDK1 and NIMA-related kinase 6) [62]. However, it remains unclear how RTN4 phosphorylation regulates its ER-shaping competence and how it contributes to mitotic ER remodeling. LNPK stabilizes three-way junctions at the ER via self-oligomerisation [65–67]. The phosphorylation of LNPK weakens its oligomerisation competence, and the expression of its phosphomimetic mutant abolishes its localization at three-way ER–ER junctions in interphase cells [63,68]. Thus, mitotic phosphorylation of LNPK may destabilize three-way ER–ER junctions to facilitate ER reorganization during mitosis. However, the specific kinase responsible for LNPK phosphorylation during mitosis awaits discovery. Moreover, whether other key ER morphology regulators, such as atlastins, are regulated by phosphorylation during mitosis remains unclear. Further studies are required to understand how ER-shaping proteins are inactivated and activated by mitotic phosphorylation and dephosphorylation to enable dynamic remodeling of the ER during the transition between interphase and mitosis.

NE reformation from mitotic ER: when is the NE identity re-established?

At the onset of anaphase, chromosomes start to segregate from the metaphase plate toward the spindle poles via pulling forces generated by kinetochore microtubule depolymerization [69,70] (Figure 1, anaphase). At this stage, chromosomes are clustered by the spindle-mediated poleward pulling [71], as well as the chromosome-periphery organizing protein Ki-67 which undergoes structural reorganization during mitotic exit that increases the adhesion of the chromosome surface [72]. A chromokinesin KIF22 might also contribute to the anaphase chromosome clustering in early embryogenesis [73]. Subsequently, in late anaphase, ER-derived membranes come in contact with the surface of the segregating chromosomes to enclose them and eventually reform the functional NE.

The initial ER–chromosome contact occurs in two distinct ways, depending on the mitotic ER morphology [61] (Figure 1, anaphase). In cells with abundant ER sheets, such as HeLa cells, the ER initially touches the periphery of the chromosome mass (‘non-core’ region), while the central region of the chromosome mass (‘core’ region) remains free from ER membranes due to the presence of dense kinetochore microtubules and the central spindle on the chromosome surface [60,74,75]. In cells with a reticulated ER, such as U2OS cells, ER membranes infiltrate the mitotic spindle guided by actin filaments and make contact with both the core and non-core regions of chromosomes [61,76]. In both HeLa and U2OS cells, initial ER–chromosome contact occurs at mid-anaphase (3–4 min after anaphase onset) [60,77,78]. At this time point, no INM proteins or nucleoporins are enriched in the chromosome-associated ER [20,29,34,75,76,78–81], and the chromosome-associated ER is morphologically indistinguishable from the rest of the ER [10,60,82]. At late anaphase/early telophase (4–8 min after anaphase onset), the ER–chromosome contacts increase rapidly (Figure 1, anaphase/telophase), and the chromosome-associated ER membranes start to accumulate NE proteins including lamin B, INM proteins (such as LEM domain proteins and LBR), and some nucleoporins (inner and outer ring nucleoporins, NUP153, and a transmembrane nucleoporin POM121) [20,29,34,74–76,79–81]. At this time, BAF is recruited to the chromosome surface to constrain membrane attachment to the surface and prevent nuclear fragmentation [83]. Lamin A/C enriches in the core region, whereas lamin B starts accumulating in the non-core region [84]. Interestingly, reticulons-3 and 4, the ER proteins that preferentially localize to ER tubules, are displaced from the chromatin-associated ER to the surrounding ER tubules [75,76,85,86], likely due to their inability to localize on the flat sheet of the chromosome-associated membranes [52]. Therefore, the progressive recruitment of NE proteins and the simultaneous exclusion of some ER-shaping proteins specialize the chromosome-associated membranes into precursor NE membranes. In early telophase, the chromosome-associated ER becomes morphologically distinguishable from the rest of the ER, as a large number of NPCs are formed via the postmitotic assembly pathway (discussed in the following section) [58,60]. However, the distance between ONM and INM is still highly variable (20–60 nm) and not as confined as in interphase (around 30 nm) [60], most likely because LINC complex components, such as SUN1, are not recruited at this stage and start to be recruited to the nuclear periphery at a later stage (mid/late telophase) [34,76]. Altogether, ER-derived membranes attached to segregated chromosomes become distinct from the rest of the ER in terms of composition and morphology during anaphase and telophase, which is critical for re-establishing NE identity after open mitosis.

What mediates the attachment of ER membranes to the chromosome surface during mitotic exit at the molecular level? After anaphase onset, the phosphorylation events that induce NEBD at prophase are reversed by the inactivation of mitotic kinases such as CDK1 and the reactivation of phosphatases such as protein phosphatase 1 (PP1) and 2A (PP2A) [15,24]. This makes lamins, INM proteins, and nucleoporins competent for rebinding to chromatin and to each other. For example, PP1, with its regulatory subunit called Repo-Man, participates in lamina reassembly by dephosphorylating lamin A/C which is distributed to the cytosol in mitosis [87]. PP1 also dephosphorylates lamin B which remains associated with membrane during mitosis in cooperation with ER-localized A-kinase anchoring protein 149 (AKAP149) that targets PP1 to the NE [88]. In addition to the mitotic dephosphorylation, the small GTPase Ran and nuclear transport receptors also regulate membrane–chromosome association. Nuclear transport receptors importin-β1 and β2 bind to lamins, INM proteins, and nucleoporins to prevent their promiscuous aggregation and interactions with other proteins in the cytoplasm [89]. When importins interact with the GTP-bound form of Ran (RanGTP), which is enriched on the chromosome surface, it triggers the release of NE proteins from importins and allows them to bind to chromatin [90–93]. Taken together, the restored binding capability of multiple NE proteins to the chromosome, mediated by mitotic dephosphorylation and the RanGTP-induced release of importins, drives the attachment and spread of ER membranes on the chromosome surface.

ER-shaping proteins may also contribute to the efficiency of membrane attachment to chromosomes. Overexpression of the ER tubule-shaping proteins reticulons-3 and 4 and deleted in polyposis locus protein 1 (DP1) delays membrane sealing on the chromatin, whereas the depletion of reticulons accelerates it [76]. However, it remains unclear (i) whether and how ER-shaping activities are modulated in a timely manner by mitotic kinases/phosphatases and (ii) the degree to which other ER-shaping proteins, such as atlastins and CLIMP-63, affect NE assembly.

Re-establishment of a functional nuclear envelope at the end of mitosis

To re-establish nucleo-cytoplasmic compartmentalization after mitosis, the nucleus must regain selective transport competence across the NE. In the final step of NE assembly, the NPCs assemble in the holes of reforming NE, and the rest of the holes in the NE are sealed to enclose the chromosome (Figure 1, telophase).

NE sealing during mitotic exit involves the fusion of NE membranes and disassembly of spindle microtubules attached to the chromosome surface. The endosomal sorting complex required for transport-III (ESCRT-III) machinery plays a key role. In the ESCRT-dependent pathway, the INM protein LEM2 accumulates in the NE holes enclosing the spindle microtubules that remain connected to the chromosomes. LEM2 recruits the ESCRT partner protein CHMP7, ESCRT-III machinery, and microtubule-severing ATPase spastin [11,94–98]. These proteins function cooperatively in microtubule disassembly and membrane fusion to seal the NE. It has been proposed that LEM2 condenses into a liquid-like phase and forms O-rings with CHMP7 around the spindle microtubules that facilitate nuclear sealing [99]. Mitotic phosphorylation of CHMP7 by CDK1 ensures timely and local LEM2 accumulation in reforming NE during mitotic exit [100]. Studies using fission yeast have suggested ESCRT-independent mechanisms for NE sealing. In Schizosaccharomyces japonicus, a fission yeast that undergoes partial mitotic NEBD, the increased activity of a fatty acid desaturase promoted NE sealing in cells lacking Cmp7 (CHMP7 in humans) [101]. In S. pombe, a fission yeast that undergoes closed mitosis, increased expression of a fatty acid elongase rescues NE permeability defects in the absence of Lem2 [102]. During the meiosis of the C. elegans oocyte and mitosis of the C. elegans embryo, the production of glycerolipids at the NE has been shown to contribute to NE sealing in cooperation with the ESCRT machinery [103]. Given that lipid composition can facilitate membrane fusion by altering the biophysical properties of the lipid bilayer [101,104], the modulation of lipid composition would also be important for NE sealing during open mitosis. Collectively, ESCRT-mediated membrane fusion and modulation of NE lipid composition would drive NE sealing at the end of mitosis.

In cells undergoing open mitosis, the NPCs assemble concomitantly with the assembly and sealing of the NE. NPC components assemble in a synchronous and stepwise manner [14,92,105]. Studies using time-resolved electron microscopy have revealed that NPC assembly proceeds by radial dilation of small membrane openings that extend to larger NPC-sized channels [60]. At 10 min post-anaphase, major NPC subcomplexes, such as outer and inner rings and the central channel, are established, while the cytoplasmic NUP214, the cytoplasmic filament component NUP358, and the nuclear basket component TPR have not yet been incorporated [20,60,78,80]. NUP214 and TPR are incorporated into NPCs 10–20 min post-anaphase, and NUP358 is recruited 20–30 min after anaphase [20,60,78,80]. In HeLa cells, 2000 to 3000 NPCs form within 10 min of anaphase onset [60]. Importantly, the NE can mediate active nuclear transport of nuclear localization signal (NLS)-bearing proteins 10 min post anaphase [20,78,106]. Therefore, the rapid assembly of major NPC subcomplexes and the NE sealing allow the nucleus to re-establish selective nucleocytoplasmic transport.

In addition to selective transport, another mechanism facilitates protein compartmentalization between the cytoplasm and the nucleoplasm during mitotic exit. It has been shown that, before the NE fully assembles around the chromosomes, large macromolecules such as Dextran larger than 40 kDa, 20S proteasomes, and mature ribosomes are excluded from anaphase chromosomes [107–109]. This macromolecular exclusion is due to chromosome condensation/clustering that is mediated, at least in part, by condensin II and Ki-67 [107,108]. Thus, chromosome clustering followed by rapid NE/NPC assembly enables the efficient re-establishment of nucleocytoplasmic compartmentalization. In this way, cells that undergo open mitosis, in which the cytoplasm and nucleoplasm transiently mix after NEBD, are able to separate the cytoplasmic and nuclear processes effectively at the end of mitosis.

Nuclear envelope expansion during interphase

In eukaryotic cells, the nucleus grows as the cells progress through the cell cycle [110]. Studies using quantitative live cell imaging have shown that the nuclear surface area increases by 2–5 fold before entering the next round of mitosis in somatic cultured mammalian cells [111–113], as well as in embryos of Drosophila [114] and see urchin [115]. To enable substantial nuclear growth in interphase, large amounts of NE proteins and lipids must be supplied to the nucleus. In early embryos, maternally deposited NE components in the cytosol are utilized for nuclear growth [114]. For example, it has been shown that a rapid nuclear growth in early Drosophila embryos is ensured by a direct incorporation of stacked sheets of ER membranes filled with pre-assembled NPCs (structures termed annulate lamellae) into the NE [114]. In somatic cells, NE proteins and lipids must be newly synthesized to sustain nuclear growth during interphase. Nuclear membrane proteins are produced by ribosomes bound to the ER and ONM, and NE lipids are synthesized by lipid-metabolizing enzymes localized to the ER [116]. Considering that the surface area of the ONM is approximately 2–10% of that of the entire ER in typical somatic mammalian cells [117–120], the majority of NE lipids and proteins are expected to be synthesized in the ER and subsequently transported to the NE [121]. After being newly synthesized in the ER, NE proteins and lipids diffuse within the ER network [31,121,122]. Because the ER membrane is continuous with the ONM, the NE components can translocate to the NE by diffusion through the membranous junctions between the ER and NE (termed ER–NE junctions, Figure 1) without the need for energy-requiring processes, such as vesicle-mediated transport [31,122]. Once NE proteins reach the NE, they are retained by interacting with nuclear binding partners, including lamins and chromatin [31,121,122]. In general, nuclear growth is sustained by a constant supply of NE proteins and lipids from the ER to the NE via a diffusion and retention-based mechanism in somatic cells.

Recent studies have shown that the INM is an active site for lipid metabolism [9]. A study using budding yeast, Drosophila, and mammalian cells has shown that PCYT1A, a rate-limiting enzyme of phosphatidylcholine (PC) that localizes in the nucleus, restores PC levels in response to lipid packing stress presumably by synthesizing PC at the INM [123]. Other studies using mammalian cells have shown that a phosphatidic acid phosphatase lipin and its major phosphatase CTDNEP1 regulate the synthesis of the glycerolipids diacylglycerol and triacylglycerol at the INM [124,125]. This lipid biosynthesis regulated by CTDNEP1 is coordinated with the ESCRT-III-mediated NE remodeling to ensure closure of the NE during post-meiotic NE reformation in C. elegans embryos [103]. Studies using the budding yeast S. cerevisiae and fission yeast S. pombe have shown that local lipid synthesis at the INM is required for NE expansion during closed mitosis, and the activity of enzymes involved in lipid biosynthesis is timely regulated by mitotic phosphorylation and SUMOylation [126,127]. However, it remains uncertain whether local lipid synthesis at the INM also contributes to NE expansion in higher eukaryotes with open and semi-open mitosis.

Potential roles of ER–NE junctions in maintaining NE identity

NE proteins and lipids synthesized in the ER are transported to the NE through ER–NE junctions. While the membrane and lumen of the ER and NE are continuous (Figure 1, interphase), some evidence suggests that there may be selectivity in the ER-to-NE transport of certain proteins and lipids, at least in mammalian cells. For example, a study using immuno-electron microscopy reported that a secretory protein called Apomucin, which forms oligomers with molecular weights ranging from 500 to 1000 kDa [128], localized predominantly in the ER lumen rather than in the NE lumen in the cells of porcine submandibular glands [129]. Another study using cryofixation and freeze-fracture replica labeling EM showed that the major acidic phospholipid phosphatidylserine is highly enriched in the ER membrane but not in the ONM in mouse embryonic fibroblasts [130]. Other studies using cryofixation and electron microscopy have demonstrated that ER–NE junctions have a highly constricted hourglass morphology with an inner diameter of ~10 nm in several mouse and human cells, as well as in plant cells [10,131] (Figure 1, interphase). The constricted morphology of ER–NE junctions forms rapidly after open mitosis [10] (Figure 1, telophase). Given that ER–NE junctions have a constricted neck of ~10 nm and that the average diameter of single human proteins is ~5 nm, multimeric protein complexes such as Apomucin and large protein aggregates might not be able to diffuse through ER–NE junctions. The highly curved membranes of ER–NE junctions could also prevent the free diffusion of cylindrical-shaped lipids, such as phosphatidylserine, as these lipids favor the formation of planar monolayers [132]. Interestingly, in budding yeast, an asymmetric distribution of phosphatidylserine between the ER and ONM has not been observed [130], which is consistent with the fact that ER–NE junctions in budding yeast are not reported to be constricted [10,133] (Figure 2). Instead, diacylglycerol (DAG), which has an inverted conical shape, has been proposed to be enriched in the INM of budding yeast [134]. It remains elusive whether DAG is also enriched in the ONM and how it is retained in the INM. Altogether, these studies raise the intriguing possibility that constricted ER–NE junctions in higher eukaryotes could restrain the passage of certain lipids, large macromolecular complexes, or even misfolded proteins that are prone to forming large aggregates in the ER lumen [135]. Furthermore, studies using fission yeast have suggested that Lem2 and Lnp1 (LEM2 and LNPK in human, respectively) regulate lipid flow between the ER and NE to restrict the excess NE expansion and prevent nuclear membrane defects [136,137]. More experiments are needed to examine whether the constricted hourglass morphology of ER–NE junctions emerges only in higher eukaryotes, and to verify that the specialized features of ER–NE junctions play key roles in maintaining NE identity.

Figure 2. Comparison of ER–NE organization in lower eukaryotes such as yeast and in metazoans such as mammalian cells.

Features of the ER and NE in budding yeast and mammalian cells. While the ER (light green) is continuous with the NE (dark green) in all eukaryotes, the junctions that connect the two subcellular compartments (ER–NE junctions) differ in their structure and size between yeast and mammalian cells. The mammalian cells have an additional lamin meshwork (light blue) beneath the INM which is absent in budding yeast. Further, the yeast nucleus forms a stable connection with the vacuole (orange) at the region adjoining the nucleolus (grey). In contrast, no such permanent association is known to exist between the nucleus and lysosome (orange) in mammalian cells. The figure also highlights various modes of lysosome/vacuole mediated degradation of nuclear and ER components. In yeast, the ONM, INM, and NPCs are degraded by autophagic engulfment, whereas in mammalian cells only ONM proteins and lamins are known to be degraded by autophagy. The key differences between yeast and mammalian cells are summarized in Table 1.

Table 1. The key different features between budding yeast and mammalian cells.

Feature	Budding yeast	Mammalian cells
Nuclear diameter	2–4 µm	5–15 µm
ER–NE junction [10,133]	Not constricted (20–40 nm in inner diameter, morphologically similar to ER–ER junctions)	Constricted (~10 nm in inner diameter, morphologically distinct from ER–ER junctions)
Nucleus–vacuole (lysosome) junction	Permanently present (mediated by Nvj1-Vac8 [157])	Absent, but a transient association could form temporarily
Nucleolus	Single territory anchored to the nuclear periphery	Multiple nucleoplasmic territories
Centrosome	Embedded in the nuclear membrane, and called spindle pole body (SPB)	Non-membrane bound, and called primary microtubule organizing center (MTOC)
Lamina meshwork	Absent	Present (composed of laminA/C, and laminB)
ER-phagy receptor	ScAtg40p [164]	ATL3 [165], CCPG1 [166], FAM134B [167], RTN3L [168], SEC62 [169], and TEX264 [170,171]
Nucleophagy receptor	ScAtg39p [153]	Possible candidates: SEC62 [160], TEX264 [172]

Nuclear envelope protein homeostasis

Once a functional nucleus is re-established after mitosis, its structure and function must be maintained for the rest of the cell cycle. Deleterious material accumulated in the nucleus must be eliminated in order to prevent genotoxicity and proteotoxicity [138]. Failure to remove the aberrant accumulation of proteins in the nucleus causes abnormal nuclear morphology and loss of nuclear integrity [39,139–143]. For example, the accumulation of a neurodegenerative disease-causing mutant of Tau, a microtubule-associated protein, at the INM induces inward invaginations of the nuclear lamina and ONM/INM, which in turn leads to compromised NPC function, loss of nuclear integrity, and the accumulation of polyadenylated mRNAs adjacent to the invaginations in Drosophila [144,145]. Such inward nuclear folds have also been observed in cells of patients diagnosed with Alzheimer’s disease [145]. Another example of a neuromuscular degenerative disorder is early-onset isolated (DYT1) dystonia, which is caused by mutations in the ER/NE protein Torsin1 that trigger nuclear membrane herniations [146]. The ultrastructural defects of NE and spatial misorganization of NPCs in the absence of functional torsin are reported in Drosophila [147], C. elegans [148], mouse [149], and human [150]. The mutant torsin concentrates at the NE, causing nuclear deformities and dysfunction [151]. The selective elimination of certain other NE proteins, such as ANC-1 (orthologue of nesprin-2), has also been shown to promote longevity and resistance to stress in C. elegans [141]. Therefore, the removal of damaged and malfunctional proteins that accumulate in the NE as a result of disease and aging is vital to ensure nuclear function and homeostasis.

Nuclear protein degradation is mediated by autophagy and the ubiquitin-proteasome system, similar to most cytoplasmic proteins. Autophagy-dependent pathways allow cells to eliminate large parts of the nucleus, including NE components, through autophagosomes and lysosomes (vacuoles in plants and fungi) [152]. Macronucleophagy is the selective degradation of nuclear components, which is dependent on the core autophagy machinery. The molecular mechanisms underlying macronucleophagy in yeast have been elucidated. The NE-localized protein Atg39 acts as a receptor for selective autophagy, which mediates the degradation of nuclear components upon nutrient deprivation [153]. The cargoes degraded by Atg39 include nucleoplasmic (Tal1), nucleolar (Nop1), INM (Src1/Heh1), and ONM proteins (HMG1), as well as scaffold nucleoporins (Nup133 and Nup192) [153–155]. The amphipathic helices present in the perinuclear C-terminus of Atg39 allow evagination of the nuclear membrane [153]. The nuclear membrane evaginations turn into double-membrane vesicles, which are sequestered into autophagosomes via the interaction of Atg39 with other autophagy machinery components, such as Atg8 and Atg11 [153,156] (Figure 2). In addition to Atg39-dependent degradation, studies using yeast have shown that nuclear components can be directly delivered to the vacuole for degradation via an Atg39-independent pathway referred to as micronucleophagy. This degradation occurs at a site called the nucleus-vacuole junction which is formed by direct physical contact between the nucleus and vacuole via the membrane proteins Nvj1 and Vac8 [157] (Figure 2). Electron microscopy studies have provided ultrastructural evidence that parts of the nucleus (including the ONM and INM) are engulfed via autophagosomes and vacuoles in yeast [154,156,158].

In mammalian cells, selective autophagy of several nuclear components has been reported to occur in response to oncogene activation, DNA damage, cellular senescence, and recovery from ER stress [159–162]. The nuclear lamina is degraded by macroautophagy in human primary cells upon oncogenic insults [159]. The lipidated form of LC3 (an autophagosomal orthologue of yeast ATG8) directly interacts with lamin-B1 and lamin-associated chromatin domains to mediate their degradation [159]. Such selective elimination of nuclear components has also been observed in mouse myoblasts and fibroblasts. Upon ER stress, the INM protein emerin translocates to the ER and is then delivered to lysosomes via vesicle-mediated transport in C2C12 myoblast cells [163]. During recovery from ER stress, only the ONM protein nesprin-3, and not the INM proteins, SUN1 and SUN2, is delivered to the lysosomes in mouse embryonic fibroblasts [160]. The lysosomal delivery of nesprin-3 is dependent on LC3-lipidation facilitated by ATG5 [160]. However, ultrastructural evidence for the direct engulfment of the NE via the autophagy machinery is lacking in mammalian cells, and it remains unclear whether and how the ONM would be deformed and engulfed via the autophagy machinery, as reported in yeast [138] (Figure 2). Moreover, a homolog or functional ortholog of Atg39 that mediates autophagy of nuclear proteins remains to be identified [152]. The key differences between yeast and mammals are summarized in Table 1.

In addition to autophagy-mediated degradation, deleterious NE components are cleared by the ubiquitin-proteasome system. In yeast, misfolded INM proteins and missorted ER proteins are primed for elimination by the ubiquitin ligase complex Asi1/2/3 protein complex via INM-associated degradation (INMAD) [173–175]. The ASI complex localizes to the INM and functions not only in protein degradation at the INM, but also in maintaining the expression level and localization of its substrates, such as the transmembrane nucleoporin Pom33 [176,177]. The ubiquitinated membrane proteins are extracted from the INM by the AAA ATPase Cdc48/p97 and subsequently delivered to the proteasome for degradation [173,174,178]. In mammalian cells, ASI-equivalent protein complexes for INMAD have not been reported, but a couple of E3 ubiquitin ligases are involved in the degradation of INM proteins [179]. For example, the E3 ligase Rnf5, which localizes in the ER and NE at steady state, relocalizes from the ER to the NE in response to increased levels of misfolded proteins in the INM to mediate their degradation [180]. Another example is the soluble ubiquitin ligase SCFβ-TRCP, which ubiquitinates the phosphorylated form of SUN2 in the INM and mediates its proteasomal degradation by p97 [181,182]. Although such ASI-independent INMAD pathways have been reported for only a few INM proteins in mammalian cells, it is likely that ubiquitin ligases control the turnover of other INM proteins to coordinate protein homeostasis in the NE.

Conclusions and perspectives

The ER and NE attain a distinct functional identity after mitosis and that is maintained during the cell cycle. It poses many open questions regarding how these two interconnected membrane-bound organelles adopt and retain their identity. In this review, we presented the current understanding of NE disassembly and its absorption into the ER during mitosis. Phosphorylation-dependent release of nuclear and NE proteins and mechanical forces imposed by cytoskeletal elements facilitate NEBD during mitosis. The disintegrated NE components are absorbed into the ER and remain excluded from chromosomes to ensure efficient chromosome segregation. Importantly, NE identity is transiently lost during open mitosis, although NPC subcomplexes and LINC complexes remain partly assembled. We also discussed the known mechanisms of how NE identity is re-established during mitotic exit. Mitotic dephosphorylation and RanGTP-induced release allows lamins, INM proteins, and nucleoporins to bind to chromatin, which specializes the membrane associated with chromatin into a precursor NE. At the same time, ER-shaping proteins such as reticulons are excluded from the precursor NE. While studies have reported the roles of ER-shaping proteins in spindle assembly and NE reformation, how the mitosis specific roles of the ER are modulated in a timely manner remains poorly understood. Novel roles of proteins such as TAOK2 that remains associated with the mitotic ER and regulates spindle function are beginning to emerge [51]. Future studies on mitotic specific post-translational modifications of ER-shaping and ER-spindle tether proteins will provide mechanistic insights into the roles of the ER in mitosis.

In interphase, the enrichment of ONM/INM proteins defines NE identity. Accumulating evidence suggests that, while the ER and NE are a continuous endomembrane system, unique lipid composition and metabolism at the NE also affect NE functions. While the current evidence is limited to only a few glycerolipids [123–125,130,134], a quantitative comparison of the overall lipid composition between the ER and NE would provide a general understanding of how the distinct lipid species contribute to the specialized functions and quality control of the ER and NE. Based on the recent findings that ER–NE junctions have a highly-constricted morphology in mammalian cells [10], further studies elucidating the role of ER–NE junctions in the exchange of lipids and proteins between the ER and NE will shed new light on the mechanism underlying NE identity and its maintenance. Advances in light and electron microscopy [183,184], as well as organelle-specific proteomics [185] and lipidomics [186], will facilitate the visualization and quantification of the dynamic ER/NE remodeling and composition, which leads to a better understanding of how ER/NE identity is established and maintained in health and disease.

Acknowledgments

We thank Tamara Völkerer for comments on the manuscript. P.D. is a recipient of VIP2 fellowship from the European Union’s Horizon 2020 research and innovation program under the Marie Sklodowska-Curie (grant agreement no. 847548). H.B.T. received a DOC Fellowship of the Austrian Academy of Sciences (no. 25951) and a Max Perutz PhD Fellowship (University of Vienna and the Medical University of Vienna). S.O. is supported by laboratory startup funding from the Medical University of Vienna, by the Vienna Science and Technology Fund (WWTF; project LS19-001), and the Austrian Science Fund (FWF) grant (P 36743).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

No data was used for the research described in the article.

Additional information

Funding

The work was supported by the HORIZON EUROPE Marie Sklodowska-Curie Actions [847548]; WWTF [LS19-001]; Austrian Research Fund (FWF) [P 36743]; Österreichischen Akademie der Wissenschaften [25951].