A role for gp210 in mitotic nuclear-envelope breakdown

The nucleus, the defining organelle of eukaryotic cells, is bordered by a nuclear envelope (NE), which consists of two membranes – the outer and inner nuclear membranes. Embedded in these are nuclear pore complexes (NPCs), which allow transport between the cytoplasm and nucleoplasm during interphase (Burke and Ellenberg, 2002). In metazoa, the NE is underlain by the lamina, which stabilizes the structure and is dynamic during the cell cycle. The NE breaks down during mitosis and reassembles around the separated chromatin to form the daughter nuclei.

Currently, there is evidence that several processes contribute to NE breakdown (NEBD). In somatic mammalian cells, mechanical forces generated by interactions of the mitotic spindle with the NE have been shown to tear open the nuclear lamina in a dynein-dependent process (Beaudouin et al., 2002; Salina et al., 2002). This process facilitates NEBD but is not essential as NEBD can also occur in the absence of microtubules (Beaudouin et al., 2002; Cotter et al., 2007; Georgatos et al., 1997; Salina et al., 2002).

In Drosophila embryos and starfish oocytes, NEBD is initiated by disassembly of NPCs. Electron microscopy (EM) of nuclei in rapidly dividing syncytial Drosophila embryos showed NPCs lacking cytoplasmic fibrils in prophase when the nuclear membranes are still intact (Kiseleva et al., 2001). In starfish oocytes, the diffusion barrier of the NE is disrupted in a zone that spreads from the animal pole and results in NEBD (Terasaki et al., 2001). In this system, NEBD proceeds in two clearly defined phases of nuclear permeabilization (Lenart et al., 2003). During the first phase, peripheral nucleoporins are sequentially released from the nuclear rim, indicative of stepwise NPC disassembly. Nucleoporin release coincides with an increasing permeability for macromolecules of diameter up to ∼40 nm, while the NE remains still intact at the ultrastructural level. In the second phase, a wave of permeabilization spreads across the nuclear envelope. During this wave, macromolecules as large as 100 nm in diameter can enter the nucleus. Although the lamina and nuclear membranes still appear to be intact at this time by light microscopy, the NE was shown by electron microscopy to be fenestrated by gaps of ∼200 nm (Lenart et al., 2003).

Currently, it is not known precisely what triggers NPC disassembly. The mammalian NPC comprises 30 different proteins named nucleoporins (Cronshaw et al., 2002). Although the NPC is anchored in the membranes of the NE, only three of the known vertebrate nucleoporins are transmembrane proteins: NDC1, POM121 and gp210 (Antonin et al., 2005; Cronshaw et al., 2002; Mansfeld et al., 2006; Stavru et al., 2006a; Stavru et al., 2006b). The first two play important roles in postmitotic NPC assembly (Antonin et al., 2005; Mansfeld et al., 2006; Stavru et al., 2006a), whereas the function of gp210 remains largely obscure. In Xenopus laevis egg extracts, gp210 is dispensable for postmitotic NPC and NE assembly (Antonin et al., 2005). The downregulation of gp210 by RNAi, however, affects the viability of HeLa cells and development of the Caenorhabditis elegans embryo (Cohen et al., 2003). Unlike POM121, the association of gp210 with the NPC is short lived and highly dynamic (Rabut et al., 2004), and gp210 is recruited late during the process of NE-NPC assembly (Bodoor et al., 1999). These properties are not consistent with the proposed functions of gp210 in membrane fusion and pore dilation during NPC formation (Drummond and Wilson, 2002) or in NPC anchoring in the NE membrane (Cohen et al., 2003). gp210 is phosphorylated on a conserved phosphorylation site in the short C-terminal tail facing the cytoplasm during mitosis, possibly by cyclin-B–cdc2 (Favreau et al., 1996).

Here, we show that gp210 plays an important role in NEBD both in vivo in C. elegans and in vitro in Xenopus egg extracts. In C. elegans, downregulation of gp210 by RNAi or deletion of the gene encoding gp210 leads to the formation of twinned nuclei in the early embryo as a consequence of defective NEBD.

Our previous systematic identification of C. elegans nucleoporins by BLAST searches and their functional analysis by RNAi-mediated depletion and differential interference contrast (DIC) time-lapse microscopy did not address the function of the C. elegans transmembrane nucleoporins (Galy et al., 2003). NDC1 has been shown to be required for NPC assembly and membrane association in vertebrates and C. elegans (Mansfeld et al., 2006; Stavru et al., 2006a). RNAi-mediated depletion of gp210, another transmembrane nucleoporin, affects viability of both HeLa cells and C. elegans embryos and leads to abnormal NPC distribution and nuclear envelope morphology (Cohen et al., 2003). These data suggested a role for gp210 in anchoring and stabilizing NPCs in the NE as well as in defining NE membrane spacing. However, no effect on NPC assembly could be detected upon depletion of gp210 in vitro (Antonin et al., 2005). Surprisingly, the phenotypes associated with gp210 depletion in a genome-wide RNAi screen were a defect in chromatin segregation and the presence of karyomeres (Sonnichsen et al., 2005). In order to resolve this apparent discrepancy, we analyzed the effect of gp210 depletion during the first zygotic division of the C. elegans embryo using time-lapse microscopy. Here, DIC microscopy recordings consistently revealed the appearance of two nuclei in both P1 and AB cells (Fig. 1A). Each of the two nuclei in each cell was spherical but smaller than the normal single nucleus in cells of this stage in the control-RNAi embryos. Twinned nuclei were always seen at the two-cell stage and frequently also in the next few cell generations (data not shown).

To demonstrate that this phenotype was not due to an RNAi-mediated off-target effect, we have carried out analogous experiments in a gp210 mutant. Sequencing of gp210(tm2320)-derived cDNA revealed that the mutant gp210 pre-mRNA is spliced correctly after the deletion event but harbors a premature stop codon due to a frame-shift mutation. Unlike the pre-immune serum, gp210 antiserum detected a protein of the expected size in the N2 extract that was absent from gp210(tm2320) extracts, showing the specificity of the serum and the absence of full-length protein in gp210(tm2320) nematodes (Fig. 1B). Embryos expressing GFP-LEM-2 (XA3507) were treated with either gp210 dsRNA [gp210(RNAi)] or control dsRNA [control(RNAi)] and immunostained using the antiserum against gp210 and the mAb414 monoclonal antibody specific for nucleoporins containing FG repeats. gp210(tm2320) embryos expressing GFP-LEM-2 were also stained and compared with the control XA3507 embryos. In XA3507 and control-RNAi embryos, gp210 was localized in the endoplasmic reticulum (ER) and enriched in the NE, where it colocalized with the mAb414 signal (Fig. 1C). This staining was absent from both gp210(tm2320) GFP::lem-2 embryos and gp210(RNAi) XA3507 embryos (Fig. 1C), indicating that the signal was specific for gp210. Currently, we cannot exclude that the truncated gp210 mRNA in gp210(tm2320) nematodes is translated into a truncated protein, but the lack of any signal in the immunostaining suggests that the protein is indeed absent. Furthermore, a truncated form of gp210 is unlikely to be functional as it would lack the C-terminal transmembrane region.

Analogous to our RNAi observations, the absence of functional gp210 in gp210(tm2320) embryos also led to the formation of twinned nuclei visible by DIC time-lapse microscopy (Fig. 1A) and surrounded by a mAb414 signal (Fig. 1C). The analysis of gp210(tm2320) and gp210(RNAi) embryos by transmission electron microscopy (TEM) confirmed the presence of twinned nuclei surrounded by a closed NE containing NPCs (Fig. 2C and Fig. 1D, respectively). We also observed in both gp210(tm2320) and gp210(RNAi) embryos the presence of stacked membranes adjacent to the NE in the cytoplasm (data not shown and Fig. 1D). Together, these results confirm that the phenotype of possessing twinned nuclei was induced by the absence of full-length gp210 protein.

We next analyzed how the twinned nuclei formed in gp210(RNAi) embryos. To this end, we performed time-lapse confocal microscopy on gp210(RNAi) embryos expressing YFP-lamin. YFP-lamin staining at the NE disappeared quickly during prometaphase of control-RNAi-treated embryos and was completely absent around the spindle in metaphase, whereas YFP-lamin remained enriched at the NE around the spindle area during all steps of mitosis in gp210(RNAi) embryos (Fig. 2A and supplementary material Movie 1, time 00:00 and 01:00), suggesting that mitotic depolymerization of lamin was affected. The lamin was still not depolymerized when the NE was already reassembling in the daughter cells (Fig. 2A and supplementary material Movie 1, time 02:40 and 18:20, note that frequently only one of each twinned pair of nuclei is in focus). The presence of lamin remnants throughout mitosis might physically prevent complete mixing of pronuclear chromosomes during the first mitosis. This indeed appeared to be the case, as seen when we observed chromosome separation in gp210(RNAi) embryos expressing GFP-labeled histone H2B (Fig. 2B and supplementary material Movie 2). The maternal and paternal chromosomes remained physically separate on the metaphase plate (Fig. 2B and supplementary material Movie 2, time 00:00) and during anaphase segregation (Fig. 2B and supplementary material Movie 2, time 00:20). This in turn led to the formation of two nuclei in each daughter cell (Fig. 2B and supplementary material Movie 2, time 05:00). EM analysis of fixed gp210(tm2320) embryos in mitosis confirmed the presence of a NE separating the condensed chromatin masses in a figure resembling a split metaphase (Fig. 2C). As suggested by the immunofluorescence analysis of the gp210(RNAi) and mutant embryos, formation of the NE was not affected, as judged by the distribution of GFP-LEM2 protein around the chromatin (Fig. 1C). Furthermore, assembly of NPCs was not abolished, as judged by NPC staining at the nuclear periphery of interphase nuclei (Fig. 1C) as well as the presence of NPCs by TEM (Fig. 1D and Fig. 2C). However, the mAb414 NPC staining appeared slightly reduced and discontinuous in the absence of gp210 (Fig. 1C, Fig. 2D). An immunofluorescence analysis of different steps in mitosis of several NPC components, conducted on embryos expressing YFP-lamin, revealed that some mAb414-positive structures appeared between the chromatin masses during telophase and overlapped with the lamin-positive structures that remained from the previous division. Unlike NPCs assembled around chromatin, these NPC-like structures, which stained with mAb414 antibody and associated with the remaining lamin, were negative for NUP153 (Fig. 2D). This suggests that, unlike NPCs assembled around chromatin, these are not complete NPCs. Interestingly, the integral inner nuclear membrane protein emerin fused to GFP was also enriched on structures between the separating chromatin masses, as well as close to the centrosomes, in gp210(RNAi) embryos (Fig. 3A and supplementary material Movie 3). TEM analysis revealed the presence of abnormal paired stacks of NE-like structures in the cytoplasm of gp210(RNAi) and gp210(tm2320) embryos (Fig. 3B and data not shown). These structures were similar to those sometimes seen transiently during NEBD of wild-type embryos but were very abundant during exit from mitosis and even during interphase in gp210(RNAi) embryos (Fig. 3B and data not shown).

We next asked whether the remaining lamin and associated NE proteins could act as a physical barrier that prevents chromosome mixing during metaphase and consecutive steps of mitosis and thus leads to the production of two separate chromosome masses and twinned nuclei. To analyze this, we tested whether the gp210-dsRNA-induced phenotype of twinned nuclei could be suppressed by co-depleting lamin. The experiment was carried out in embryos expressing GFP fused to LEM-2, an inner nuclear membrane protein. As expected, gp210-dsRNA-mediated depletion led to the formation of twinned nuclei at the two-cell and four-cell stage, but we did not observe twinned nuclei in embryos co-depleted for lamin and gp210 (Fig. 3C).

In summary, these data show that depletion of gp210 prevents lamin depolymerization and that, in turn, the remnant lamin structures block pronuclear chromosome mixing, ultimately leading to the formation of twinned nuclei and cytoplasmic NE-like structures.

The delay in lamin depolymerization and NEBD induced upon gp210 depletion or truncation suggested that gp210 was required for NEBD. Vertebrate gp210 fused to GFP is a very dynamic nucleoporin that is able to diffuse within the NE membrane and ER and interacts transiently with the NPC (Rabut et al., 2004). Depletion of RAB-5 and YOP-1/RET-1 as well as gp210 was recently shown to induce the formation of twinned nuclei in C. elegans embryos (Audhya et al., 2007). RAB-5 and YOP-1/RET-1 depletion also affected dynamics of the peripheral ER, whereas gp210 did not. Furthermore RAB-5 and YOP-1/RET1 were localized in the peripheral ER (Audhya et al., 2007; Voeltz et al., 2006) and, the authors suggested, act on NEBD by preventing inner nuclear membrane protein diffusion into peripheral ER, whereas no clear function could be assigned to gp210 (Audhya et al., 2007). We analyzed the localization of C. elegans gp210 at different time points in cell division. We described previously that MEL-28 and NUP107, two stably associated nucleoporins essential for NPC assembly, are released from the nuclear periphery during prophase and relocalize to kinetochores (Franz et al., 2005; Galy et al., 2006). We also observed previously that GFP-NUP155 and mAb414 antigens are dispersed throughout the cytoplasm at metaphase (Franz et al., 2005). These data suggested that NPCs are largely disassembled when cells reach metaphase. Unexpectedly, immunostaining of wild-type embryos revealed that gp210 was enriched at NEBD and in the remnants of the NE that surround the mitotic spindle late during metaphase in the absence of nucleoporins recognized by mAb414 (Fig. 4) or antibodies against NUP153 (data not shown).

In summary, gp210 remains associated with the NE after several steps of NPC disassembly have occurred. Gp210 might therefore act in NEBD at any time point during the process.

Our data show that NEBD is compromised in early C. elegans embryos in the absence of gp210. As NEBD is known to be facilitated by pulling forces emanating from the centrosomes as well as by signaling events (see Introduction) (Beaudouin et al., 2002), it is difficult to determine the exact role of gp210 in vivo in order to define the primary defect induced upon gp210 inactivation. We therefore analyzed in vitro assembled nuclei in Xenopus laevis egg extracts. These nuclei can be disassembled when either extract derived from eggs arrested with cytostatic factor (CSF) or a nondegradable mutant of cyclin B (cyclin B Δ90) is added. In both cases, the nuclear envelope of the interphase nucleus is lost and the chromatin condenses (Fig. 5A,B, left panels).

To interfere with the function of gp210, we added Fab fragments derived from antibodies raised against the C-terminal (nucleoplasmic) part of gp210 to the extracts when initiating the assembly reaction [see Antonin et al. for a detailed description of the antibody (Antonin et al., 2005)]. In these reactions, nuclei assembled as expected, as indicated by the membrane stain and the decondensed chromatin (Fig. 5A,B, third panels) (Antonin et al., 2005). However, these nuclei did not undergo NEBD when mitotic extract or cyclin B Δ90 was added (fourth panels). By contrast, nuclei assembled and disassembled when a Fab fragment derived from antibodies against the N-terminus (lumenal part) of gp210 was added (Fig. 5A,B, first and second panels). From the gp210 membrane topology, it is known that the epitope of these antibodies is located in the intermembrane space of the NE (and is thus inaccessible). They therefore served as a control for this and all of the following experiments. Identical results were obtained using intact IgG proteins (data not shown). Although the inhibitory Fab fragments prevented NEBD, they did not prevent the extracts from progressing into mitosis when cyclin B Δ90 was added, as judged by histone H1 kinase assays (see supplementary material Fig. S1).

To examine the specificity of the inhibitory effect, we tested whether Fab fragments directed against several other nucleoporins would have a similar effect. For this, we assembled nuclei, added Fab fragments for 10 minutes at the end of the assembly reaction and initiated NEBD. In contrast to Fab fragments against gp210, Fab fragments against either the transmembrane nucleoporin POM121 or against NUP205 or NUP153 did not affect NEBD (Fig. 5C).

NEBD is thought to be initiated by partial NPC disassembly (Lenart et al., 2003; Terasaki et al., 2001). We examined whether the nuclei blocked for NEBD had lost nucleoporins. Assembly and disassembly reactions were performed as in Fig. 5A and the samples were stained with several antibodies against NUP proteins. Interphase nuclei assembled in the presence of the inhibitory Fab fragments showed a robust staining for all NUP proteins tested, indicating that the gp210 Fab fragments do not interfere with NPC assembly. Furthermore, the nuclei blocked with Fab against gp210 stained for NUP153, mAb414 (Fig. 4D) and all other NUP proteins tested (POM121, gp210, NUP214, NUP205, NUP107, TPR; see supplementary material Fig. S2), showing that, like the NE, the NPCs were not disassembled despite the addition of the mitotic extracts.

Next, we tested whether, although the NE stayed intact in the presence of the inhibitory Fab fragment, the chromatin had nevertheless proceeded into a mitotic state: however, chromatin in the nuclei blocked in NEBD was not positive for phospho-histone H3 (Fig. 5E) or phospho-histone H1 (see supplementary material Fig. S3).

In summary, these data show that Fab fragments directed against the cytoplasmic C-terminus of gp210 block NEBD in vitro. The block seems to occur very early in NEBD as the NUP proteins examined were retained and the nuclei remained seemingly in an interphase state. This was confirmed by the fact that no entry of dextran-70 could be observed in inhibited nuclei, unlike control nuclei, when progression into mitosis was initiated (supplementary material Fig. S4).

The experiments described so far indicated that the nuclei blocked before NEBD remained in an interphase state. However, we wished to test whether the inhibitory Fab might block an interphase function of the nucleus that would lead to a block in NEBD. First, we tested whether DNA replication was affected by the inhibitory Fab fragments. As shown in Fig. 6A, foci containing biotinylated dUTP are labeled in nuclei assembled in the presence of both control and inhibitory Fab fragments, indicating that DNA in both reactions is replicated and thus that this nuclear function is intact.

NPCs fulfil an important function in nuclear import. Therefore, we checked whether the inhibitory Fab fragment would block nuclear import. A fluorescently labeled import substrate carrying several nuclear localization signals was efficiently imported into nuclei assembled in the presence of both control or inhibitory Fab fragments, whereas a control substrate was not (Fig. 6B). As nuclei grew even in the presence of the inhibitory Fab (Fig. 5 and supplementary material Fig. S4) and the nuclei replicate, the existence of a general import defect could be excluded.

We cannot, however, rule out that the import of a specific factor required for NEBD was perturbed. One obvious candidate that might act as such a factor is cyclin-B–cdc2, which is imported into the nucleus in early prophase (Hagting et al., 1999). To check for import of cyclin B Δ90, we added biotinylated, non-degradable cyclin B Δ90 to the reaction and fixed the samples 10 minutes before control nuclei disassemble. Cyclin B Δ90 was detected both in control and inhibited nuclei, indicating that the protein is imported into the nucleus even in the presence of gp210 Fab fragments (Fig. 6C).

As general nuclear import was not affected and the inhibitory Fab fragments did not block the interphase function of the nucleus in general, we determined whether the antibody had a direct effect on gp210. We therefore asked whether gp210 itself would still correctly localize in the presence of the inhibitory Fab. Immunofluorescence experiments showed that, like all other NUP proteins tested, gp210 is localized to the nuclear rim in the presence of the inhibitory Fab (supplementary material Fig. S2). However, as these experiments could not distinguish NPC versus NE localization and as it is known that gp210 is quite dynamic in interphase (i.e. it must diffuse in and out of NPCs) (Rabut et al., 2004), we turned to electron microscopy. For this, nuclei were assembled in the presence of control and inhibitory Fab fragments. The reaction was stopped by fixation and the samples processed for immuno-electron microscopy. gp210 was found to localize both to the NPC and the outer nuclear membrane, and the distribution did not change in the presence of the inhibitory Fab (supplementary material Fig. S5).

The gp210 C-terminal domain is phosphorylated during mitosis. As the Fab fragment is directed against this region, we asked whether this modification is affected by the presence of the inhibitory Fab. To test this, we incubated membranes with control or inhibitory Fab fragments and added mitotic extracts in the presence of ³²P-labeled ATP. The reaction was stopped and gp210 was immunoprecipitated using the lumenal antibody. The control Fab fragments had no effect, whereas the Fab fragments against gp210 strongly reduced mitotic phosphorylation of gp210 (Fig. 7).

As cyclin-B–cdc2 is thought to be responsible for mitotic phosphorylation of gp210, and as Fab fragments that were able to reduce this phosphorylation blocked NEBD, we determined in C. elegans embryos whether depletion of cyclin B would mimic the loss of gp210 function. Interestingly, the RNAi treatments that, like gp210, are reported to lead to defects in chromatin segregation and the presence of karyomeres as well as the production of twinned nuclei at the two-cell stage (www.phenobank.org) (Sonnichsen et al., 2005) include RNAi against several genes encoding cyclin B proteins. We therefore used RNAi-mediated depletion of cyclin B (encoded by the cyb-1 gene) in GFP-β–tubulin-expressing embryos. We observed the formation of twinned nuclei that were surrounded by an intact and closed NE, as assayed by the exclusion of soluble fluorescent GFP-β–tubulin from the nuclear space (Fig. 8A). The distribution of GFP-LEM2 protein around the twinned nuclei also appeared normal (supplementary material Fig. S6). Furthermore, depletion of cyclin B also prevented YFP-lamin depolymerization without blocking progression through mitosis (Fig. 8B). These results obtained upon cyclin B depletion phenocopied the depletion of gp210 and showed that cyclin B is important for NEBD in C. elegans, possibly through one of its potential substrates, gp210. We speculate that phosphorylation of the gp210 C-terminal domain by cyclin-B–cdc2 in vivo is required for NEBD. In the absence of functional, phosphorylated gp210, lamin depolymerization does not proceed normally and this might directly result in delayed or blocked NEBD.

Here, we have shown that NEBD is affected both in C. elegans and in Xenopus when the transmembrane nucleoporin gp210 is depleted, mutated or functionally blocked with Fab fragments. In early C. elegans embryos, we observed twinned postmitotic nuclei upon RNAi depletion or mutation of gp210. This phenotype was not reported in previous gp210 RNAi-mediated depletions of nematodes (Cohen et al., 2003) or HeLa cells (Cohen et al., 2003; Mansfeld et al., 2006; Stavru et al., 2006b). The phenotype was specific, however, and not an RNAi off-target effect as an identical phenotype was observed in the gp210(tm2320) mutant and depletion of a variety of other nucleoporins did not produce this phenotype (Galy et al., 2003). The efficiency of the RNAi-mediated depletion appeared good as judged by immunofluorescence analysis using a specific Ce-gp210 polyclonal antibody but was possibly incomplete as we observed only 7% embryonic lethality compared with the 15% reported by Cohen and colleagues (Cohen et al., 2003) and 40% for the gp210(tm2320) mutant. Despite this potentially incomplete depletion, most (>80%) of the RNAi-treated embryos as well as gp210(tm2320) embryos displayed twinned nuclei at the two-cell stage. Surprisingly, the viable escapers did not show any obvious post-embryonic phenotype. Analysis of the dead embryos (complicated by the fact that they could only be identified after survivors hatched) showed that death occurred when the embryos had at most ∼80 cells (data not shown). Surviving gp210(tm2320) mutant nematodes had no detectable full-length gp210. This suggests that gp210 is not absolutely essential and confirm previous observations in mouse that neither cell viability nor the stable association of NUP107 and POM121 with the NPC during interphase are affected in cells that do not express gp210 (Eriksson et al., 2004; Olsson et al., 2004). Similarly, human cultured cells depleted of gp210 by RNAi were viable (Stavru et al., 2006b). This conclusion is probably also valid for the majority of C. elegans embryos (Cohen et al., 2003) (this paper), although it is not possible to rule out that viable escapers reflect the presence of a residual amount of gp210 protein in both RNAi-treated or gp210(tm2320) embryos.

The nuclear-twinning phenotype was reproduced from mother to daughter cells as long as the twinned nuclei and centrosomes were positioned in a configuration resembling the first division, with the two nuclei contacting each other and the centrosomes positioned on both sides of the contact region. This reproducible configuration, and the need for synchronous NEBD before chromosome congression on the metaphase plate, made it possible to recognize this unexpected function for gp210 in NEBD. During C. elegans early development, cell-division cycles are short, and preventing lamin depolymerization by gp210 depletion had strong consequences for mitosis. In most cases, the nuclear-twinning phenotype was no longer observed after several cell divisions. This correlated with the centrosomes being localized more distal to the contact region between the twinned nuclei. Once this configuration was observed, the twinned nuclei appeared to be resolved by the mitotic spindle and twinned daughter nuclei were no longer produced. It is possible either that gp210 is only required during the earliest zygotic divisions or that the lengthening of the cell cycle in later divisions allowed sufficient time for NEBD to proceed to completion, even if it occurred more slowly in the absence of gp210. It is nevertheless surprising that even a small number of such defective mitoses do not lead to more frequent chromosome loss and subsequent embryonic lethality at a higher rate than 7%. A nuclear-twinning phenotype was not observed upon siRNA-mediated depletion of gp210 from vertebrate cells (Cohen et al., 2003; Mansfeld et al., 2006; Stavru et al., 2006b), but a block or delay in depolymerization of the lamina in a single nucleus during mitosis would not lead to the formation of twinned nuclei. It will be interesting to test whether siRNA depletion of vertebrate gp210 affects lamin depolymerization and slows down entry into mitosis.

A delay in NEBD and the formation of twinned nuclei in C. elegans embryos has been associated recently with RNAi-mediated depletions of both gp210 and proteins involved in controlling the dynamics of the peripheral ER (Audhya et al., 2007). While it was suggested that the ER dynamics were required for NE membrane redistribution during mitosis, the contribution of gp210 in NEBD was not clear. Our data obtained in gp210(tm2320) mutants confirmed that gp210 is required for several aspects of NE disassembly, including depolymerization of lamins, and revealed that lamins are indeed responsible for the production of twinned daughter nuclei. We show that the remaining lamins, associated NE and NPCs act as a physical barrier preventing chromosome mixing and that lamin is required for the production of twinned daughter nuclei. Lamin depolymerization is a late step of NEBD that occurs after NPC disassembly and NE fenestration. Although a lack of lamin depolymerization was easily observable in vivo, it is likely that gp210 is required also for earlier events in NE disassembly. Furthermore, TEM revealed the abnormal presence of remaining NE and a large number of stacked NE-like structures during anaphase and telophase. We noticed that similar, but more fragmented, stacked structures were produced transiently in wild-type embryos during prometaphase, and these have also been described during germinal vesicle NEBD in mouse oocytes (Calarco et al., 1972). In gp210-RNAi-depleted or mutated embryos, these structures were enriched in the region that separated the chromatin masses and close to the centrosomes, similar to the remnants of the lamina. Time-lapse confocal microscopy and immunostaining revealed the presence of other NE components (NPCs, LEM-2 and emerin) associated with these structures. Several of the NPC components tested were present in these structures during late anaphase and telophase but absent during early anaphase, suggesting that NPCs are largely disassembled during anaphase but can reassemble in the remaining NE at later stages of mitosis. The assembly of ectopic NPCs suggests the presence of incomplete NPC remnants serving as seeding points for NPC reassembly. The distribution of lamin, emerin, LEM2 and NPCs was normal at the pronuclear stage. Furthermore, it was clear from our dynamic analysis of the sequence of events that the aberrant NPCs and NE marker distribution were due to a problem in disassembly of the pronuclear or nuclear envelope rather than in reassembling a functionally and structurally normal NE and NPCs, confirming that gp210 is not required for NPC assembly (Eriksson et al., 2004; Mansfeld et al., 2006; Stavru et al., 2006b). Our experiments do not support the hypothesis that loss of, or blocking, gp210 function interferes with nuclear import as accumulation of an import substrate and cyclin B was normal, and the DNA was replicated and the nuclei grew, which are both indicators of normal nuclear functioning, including protein import. However, as multiple import pathways exist, we cannot formally rule out there being defects in the import of specific substrates.

The in vivo localization of gp210 during the cell cycle was consistent with its role in NEBD. gp210 was enriched in the region of the NE just before and during NEBD and remained accumulated around the spindle until telophase. This suggests the existence of a subdomain of the ER enriched in some transmembrane NE proteins that might be required for efficient NEBD. Interestingly, the redistribution of the inner nuclear membrane protein LBR into the ER as well as the release of peripheral soluble components of the NPC are early steps of NEBD (Ellenberg et al., 1997), suggesting that the NPC diffusion pore size might potentially be modified for both soluble and transmembrane proteins and that gp210 could play an active role in this process. The in vivo C. elegans results were consistent with our in vitro data showing that all known steps of NEBD were inhibited in the presence of gp210 Fab fragments. While blocking gp210 in vitro inhibited NPC disassembly and NEBD efficiently, in vivo NEBD was delayed but not completely blocked, possibly due to parallel mechanisms of NEBD such as tearing of the NE by microtubules. We therefore conclude that gp210 is required for NPC disassembly, an early step of NEBD.

Our experiments suggest that the inhibitory effect of the Fab fragments in vitro correlates with a block of gp210 phosphorylation. It has been proposed that cyclin-B–cdc2 is responsible for phosphorylation of gp210 (Favreau et al., 1996), and cyclin-B–cdc2 is an important trigger for progression of the cell cycle into mitosis (Ohi and Gould, 1999). RNAi-mediated depletion of cyclin B from C. elegans phenocopied the gp210 depletion effect on NEBD. We suggest that gp210 phosphorylation is therefore indeed an early step required for efficient progression of NEBD. It would potentially have been interesting to test the effect of phosphorylation-defective mutants of gp210 on NEBD, but this was not feasible technically. Mutants of gp210 that replace the phosphorylated serine with alanine or glutamate resulted in a change in the dynamics of gp210 association with NPCs (Onischenko et al., 2007) and were therefore not suitable for such an analysis. Interestingly, in Aspergillus nidulans, in which NPCs are only partially disassembled at the onset of mitosis, the release of NUP98 coincides with its phosphorylation by NIMA, which is a downstream target of cdc2 (De Souza et al., 2003). Similarly, in Drosophila, CDC2 activity is required for NPC disassembly (Onischenko et al., 2005).

In summary, we have shown that gp210 is an important factor in NPC disassembly and NEBD in vivo in C. elegans and in vitro in Xenopus egg extracts. We suggest that phosphorylation of gp210 either directly by, or downstream of, cyclin B is a key early event in NEBD that is required to efficiently destabilize the NPCs and the nuclear lamina.

Antibodies against gp210, POM121 (Antonin et al., 2005), NUP93, NUP205 (Franz et al., 2007), NUP107 (Walther et al., 2003), NUP214 (Walther et al., 2002), NUP153 (Walther et al., 2001) have been described previously; antibodies against TPR were from Volker Cordes (Max-Planck-Institut für Biophysikalische Chemie, Germany) and antibodies against phospho-histone 1 and phospho-histone 3 (Upstate, Charlottesville, USA), lamin (Immuquest, Cleveland, UK) and mAb414 (Babco, Cambridge, UK) were commercial.

Antibodies against C. elegans gp210/NPP12 were raised in rabbits using a hexa-histidine tagged fragment from residues 371-798 of the C elegans protein. Antibodies against NUP153/NPP-7 have been described previously (Galy et al., 2003).

The C. elegans Bristol strain N2 was used for the EM analysis of the NE upon RNAi treatment. AZ212 GFP::hisH2B (Praitis et al., 2001), XA3502 GFP::lmn-1 (Galy et al., 2003) and XA3507 GFP::lem-2 (Praitis et al., 2001) XA3541 GFP::lem2/GFP::tbb-2 (Franz et al., 2005) have been described previously. The npp12/gp210(tm2320) mutant was provided by S. Mitani from the Japanese National Bioresource Project and back-crossed four times with N2. The resulting strain was crossed with XA3507 to generate XA3562 gp210(tm2320) GFP::lem-2. Single-gene dsRNA interference was triggered by feeding (Askjaer et al., 2002) L4 nematodes with bacteria expressing dsRNA corresponding to nucleotides 2701 to 3601 of the gp210/npp-12/T23H2.1 ORF (Galy et al., 2003) for 48 hours at 20°C. For double RNAi depletion, in vitro transcribed dsRNA using as a template a PCR product obtained using T7 primers on the gp210/npp-12/T23H2.1 feeding construct in injection buffer (20 mM KPO₄, 3 mM potassium citrate, 2% PEG 6000, pH 7.5) was injected in one gonad of young adults and incubated on RNAi plates with bacteria expressing dsRNA corresponding to nucleotides 1 to 765 of lamin/lmn-1/DY3.2 cDNA for 30-32 hours at 20°C.

Live-embryo imaging was performed on a Leica confocal microscope TCS SP2, as described previously (Galy et al., 2006). For immunofluorescence, the embryos were fixed and prepared for staining (Askjaer et al., 2002) with antibodies diluted in PBS with 0.1% Tween-20. Cy5-conjugated donkey anti-rat secondary antibody (Jackson Immunoresearch Laboratories, West Grove, PA; 1:500); and goat anti-mouse Alexa Fluor 488 and 546 (Molecular Probes, Eugene, OR; 1:1000). For DNA staining, Hoechst 33258 (Sigma, Deisenhofen, Germany) was used at 1 μg/ml. Confocal images were obtained with a Leica TCS SP2. Wide-field fluorescent images were obtained on a Leica DMRXA coupled to a Hamamatsu ORCAII-ER CCD camera controlled by Openlab software (Improvision, Coventry, UK).

Extracts were prepared from N2 wild-type and gp210(tm2320) embryos, as described previously (Cheeseman and Desai, 2005). Equal amounts were loaded onto SDS-PAGE gels and detected by western blotting using either antiserum against gp210 or NUP107, and peroxidase-linked donkey anti-rabbit antibodies (GE Healthcare).

XA3562 gp210/npp-12(tm2320)GFP::lem-2 embryos were observed with a TCS SP2 Leica confocal microscope, then transferred to Leica membrane carriers filled with 20% BSA and cryoimmobilized at given time points with a Leica EMPact 2 high-pressure freezer (Leica, Vienna, Austria). Embryos were freeze substituted using a Leica AFS freeze substitution machine in 1% osmium tetroxide (Merck, Darmstadt, Germany), 0.1% uranyl acetate (Fluka), and 5% water in acetone and processed using the same program as described below.

Control and gp210 RNAi-treated nematodes were transferred to planchettes of 100 μm depth fill with 20% BSA in M9 buffer, cryoimmobilized immediately with a HPM 010 high-pressure freezer (Bal-tec, Balzers, Liechtenstein) and transferred under liquid nitrogen for freeze substitution in 1% osmium tetroxide (Merck, Darmstadt, Germany) plus 0.8% uranyl acetate (SPI, West Chester, PA) in acetone containing 2% water in an AFS (Leica, Vienna, Austria). Substitution was performed for 48 hours at –90°C, samples were warmed up (5°C/hour) to –30°C (3 hours) and then warmed up to 0°C before removal of the substitution medium and embedding in EPON. TEM on in vitro nuclei was performed as described previously (Franz et al., 2005).

When NEBD was initiated using mitotic extracts, nuclear assembly was performed as described previously (Walther et al., 2001). After 2 hours, an equal volume (20 μl) of CSF extracts (Desai et al., 1999) cleared by centrifugation for 10 minutes at 200,000 g and 0.4 μl energy mix (50 mM ATP, 500 mM creatine phosphate and 10 mg/ml creatine kinase) was added.

When NEBD was initiated using cyclin B Δ90, nuclear assembly was performed as described previously, with the modification that crude extracts were centrifuged for 10 minutes at 200,000 g to obtain cytosol (Antonin et al., 2005), and disassembly initiated by addition of cyclin B Δ90 (Glotzer et al., 1991).

Membrane staining and immunofluorescence were performed as described previously (Antonin et al., 2005) but using Alexa-488-labeled protein A (Molecular Probes Eugene, OR) to exclude detection of the Fab fragments used for the inhibition. Import reactions (Walther et al., 2001), the replication assay (Walther et al., 2002) and the histone H1 kinase assay were all performed as described previously (Felix et al., 1993).

10 μl of a membrane fraction derived form Xenopus egg extracts were incubated with 5 μg Fab fragments or control buffer on ice for 10 minutes. Then 20 μl of interphase or CSF-arrested extract pre-cleared by centrifugation for 10 minutes at 200,000 g and 1 μl of [γ-³²P]ATP were added. The samples were incubated at 20°C for 20 minutes. The reaction was stopped by solubilization with 500 μl of 50 mM sodium phosphate pH 7.4, 500 mM NaCl, 1% TritonX-100 in the presence of 12.5 mM NaF, 0.5 μM microcystin, 40 mM glycerophosphate, 1 mM orthovanadate, 10 mM EDTA, 0.5 mM PMSF and protease inhibitor cocktail (Roche, Mannheim, Germany) for 10 minutes on ice. Samples were cleared by centrifugation for 10 minutes at 17,000 g and gp210 immunoprecipitated with the `lumenal' antiserum. Immunoprecipitated material was analyzed by SDS-PAGE followed by autoradiography and western blotting.

We are indebted to Peter Askjaer for generating the lamin RNAi feeding construct and Thomas Müller-Reichert for his generous help with EM techniques. We thank Shohei Mitani from the Japanese National Bioresource Project for providing the gp210(tm2320) C. elegans strain used in this work and Antony Hyman for sharing his initial observation of the gp210 RNAi phenotype. V.G. was supported by EMBL and funded by the Fondation pour la Recherche Médicale. W.A. was supported by the Ernst Schering Research Foundation.