A self-inhibitory interaction within Nup155 and membrane binding are required for nuclear pore complex formation

Nuclear pore complexes (NPCs) are the essential gatekeepers of the nuclear envelope. They form transport gates in the double membrane of the nuclear envelope, which restrict diffusion of macromolecules and, when required, allow highly efficient directed transport of proteins, nucleic acids and RNA–protein complexes between the cytoplasm and the nucleoplasm. In vertebrates, about 30 different proteins, called nucleoporins or Nups, contribute to the structure of the eightfold symmetric NPCs, each in 8, 16 or more copies (Beck and Hurt, 2017; Ori et al., 2013) making up a total mass of ∼125 MDa. How more than 500 individual components assemble into a macro-complex, and how they integrate in the double membrane of the nuclear envelope, is a research question yet to be answered.

Roughly, the different nucleoporins can be categorized into three different classes: transmembrane nucleoporins anchor the NPC in the pore membrane, structural nucleoporins assemble as the scaffold of the pore, and largely unstructured nucleoporins containing FG (phenylalanine-glycine) repeats form the permeability barrier and transport gate of the pore. The NPC structural scaffold can be viewed as a stack of three rings (for review, see Beck and Hurt, 2017): a cytoplasmic and a nucleoplasmic ring are both largely formed by an evolutionary conserved subcomplex of the NPC, which is known as the Y-complex or, in vertebrates, the Nup107–Nup160 complex (Bui et al., 2013; von Appen et al., 2015). Sandwiched in between these two outer rings is the inner ring, which is mainly formed by multiple copies of the Nup93 complex, which consists of Nup93, Nup155, Nup53 and the two paralogues Nup205 and Nup188 (for a review, see Vollmer and Antonin, 2014). The Nup93 complex links the pore membrane with the central channel. Nup53 and Nup155 interact with the pore membrane directly (Vollmer et al., 2012; von Appen et al., 2015) and via the transmembrane nucleoporins POM121 and NDC1 (Eisenhardt et al., 2014a; Mansfeld et al., 2006; Mitchell et al., 2010). In turn, the Nup62 complex, which forms a large part of the central transport channel of the NPC, is recruited by the N-terminal region of Nup93 (Chug et al., 2015; Sachdev et al., 2012).

In vertebrates, NPCs assemble following two distinct assembly pathways, depending on the cell cycle state (for a recent review, see Weberruss and Antonin, 2016): at the end of mitosis, NPCs reassemble on the decondensing chromatin in parallel with the nuclear envelope. After nuclear envelope assembly is completed, NPCs form de novo into the intact nuclear envelope during interphase. Mitotic NPC assembly, best studied in a cell-free system using Xenopus egg extracts (Lohka, 1998), is initiated on the chromatin by MEL-28 (also known as ELYS and AHCTF1), which acts as a seeding point for NPC formation (Franz et al., 2007; Galy et al., 2006; Rasala et al., 2006). ELYS recruits the Y-complex, a major structural component of the outer rings to the assembling pores. Then, membrane connection is established: this probably happens via the transmembrane nucleoporin POM121 and NDC1 (Antonin et al., 2005; Mansfeld et al., 2006; Rasala et al., 2008), which are the next components to be recruited. The following known assembly step involves components of the Nup93 complex. In contrast to the Y-complex, which is recruited as a preformed unit, at least in Xenopus egg extracts, components of the Nup93 complex are added in a sequential manner: first, Nup53 binds to assembling NPCs via its membrane binding motif and its interaction with NDC1 (Eisenhardt et al., 2014a; Vollmer et al., 2012). Nup53 recruits Nup155 and Nup93 (Eisenhardt et al., 2014a; Hawryluk-Gara et al., 2005; Sachdev et al., 2012), the latter in a complex with one of the two paralogues Nup188 or Nup205 (Theerthagiri et al., 2010). Nup93 additionally recruits the Nup62 complex as a major component of the central channel (Chug et al., 2015; Sachdev et al., 2012). The later events in this assembly pathway are less well defined. Also less clear is which assembly pathway interphase NPC formation takes.

Here, we focus on the conserved NPC component Nup155. Vertebrate Nup155 and its two Saccharomyces cerevisiae homologues, Nup170 and Nup157, each possess an N-terminal β-propeller region (Devos et al., 2004; Lin et al., 2016; Mans et al., 2004; Seo et al., 2013). The C-terminal half of the protein is formed by a large α-helical domain (Flemming et al., 2009; Lin et al., 2016; Whittle and Schwartz, 2009). In Caenorhabditis elegans, Drosophila and mouse, NUP155 is an essential gene (Galy et al., 2003; Kamath et al., 2003; Kiger et al., 1999; Zhang et al., 2008), most likely because of its key function in NPC assembly (Franz et al., 2005). Yeast lacking both of the Nup155 homologues Nup170 and Nup157 are not viable (Aitchison et al., 1995; Makio et al., 2009). We demonstrate here that direct membrane interaction of vertebrate Nup155 is required for NPC formation. Nup155 mutants or fragments that cannot bind to membranes are not able to replace the endogenous protein in NPC assembly. We further show that the C-terminal region of Nup155 has an auto-inhibitory function: it binds to the N-terminus of Nup155, decreasing the strength of the Nup155–Nup53 interaction to a level that does not support NPC assembly. In the course of mitotic NPC assembly, the auto-inhibitory effect of the C-terminal interaction of Nup155 must be overcome by Nup93 binding to Nup53, triggering conformational changes that allow further progression of NPC assembly.

We have previously shown by cryo-EM of the human NPC that a loop within the Nup155 β-propeller region dips into the pore membrane and is required for membrane binding of the recombinant protein in liposome flotation assays (von Appen et al., 2015). In order to test whether the membrane binding capability of Nup155 has an important role in NPC assembly, stability or function, we aimed to replace the Nup155 with protein versions defective in membrane interaction. For this, we depleted Nup155 from Xenopus laevis egg extracts using specific antibodies (Fig. 1A). Because the Nup93 complex does not exist as a pre-assembled complex in Xenopus egg extracts, Nup155 depletion does not co-deplete Nup53, Nup93, Nup188 and Nup205 or other nucleoporins (Fig. 1A). Xenopus egg extracts are widely used to faithfully reconstitute nuclear envelope and NPC assembly in vitro when sperm DNA is added as a chromatin template (Eisenhardt et al., 2014b; Lohka, 1998). In mock-depleted extracts, a closed nuclear envelope was formed on the chromatin template, indicated by smooth membrane staining; however, upon Nup155 depletion, membrane vesicles bound to the chromatin surface, but failed to form a closed nuclear envelope (Fig. 1B,C) as previously reported (Franz et al., 2005). This is due to a failure of NPC assembly, as observed upon depletion of a number of nucleoporins, including Nup93, Nup53, POM121 and NDC1 (Antonin et al., 2005; Grandi et al., 1997; Hawryluk-Gara et al., 2008; Mansfeld et al., 2006; Sachdev et al., 2012; Vollmer et al., 2012). Indeed, NPCs were absent on the chromatin. This was indicated by the strong reduction in mAB414 staining, an antibody which recognizes FG-repeat-containing nucleoporins (Davis and Blobel, 1986) and is widely used as a general marker for NPC assembly. Similarly, RNAi-mediated downregulation of Nup155 reportedly decreased mAB414 staining and NPC numbers in tissue culture cells (Mitchell et al., 2010).

In vitro-assembled nuclei grow in size once a nuclear envelope and NPCs are formed. This process requires the newly formed NPCs to be competent for nuclear import and leads to a further decondensation of the chromatin template (Philpott et al., 1991). Consistent with a block in nuclear envelope and NPC assembly, the chromatin templates remain small in the Nup155-depleted extracts, in comparison to the mock control (Fig. 1B, Franz et al., 2005).

We next added back Nup155 to Nup155-depleted extracts with mRNA encoding Xenopus Nup155, resulting in expression of the protein. Addition of the mRNA encoding wild-type Nup155 restored nuclear envelope and NPC formation, showing the point specificity of the depletion phenotype. Addition of mRNA encoding Nup155 with a point mutation (L258D) as well as a 10 amino acid deletion (Δ258–267), both defective in direct membrane binding (von Appen et al., 2015), did not restore nuclear envelope or NPC formation. Consistently, as a result of the lack of nuclear import, chromatin templates did not grow in size, in contrast to the wild-type rescue. This indicates that membrane binding of Nup155 is required for in vitro NPC assembly.

Interestingly, a point mutation within human Nup155, an arginine to histidine exchange in position 391, has been linked to atrial fibrillation and early sudden cardiac death (Zhang et al., 2008). To test whether this mutation would impact Nup155 function in NPC assembly, we generated the corresponding mutation in Xenopus Nup155, R385H. Addition of the mRNA encoding this mutant restored nuclear envelope and NPC formation in Nup155-depleted extracts to similar levels as the wild-type Nup155 mRNA (Fig. 1B). This indicates that, at least in the in vitro system, the Nup155 R385H mutant is functional in NPC assembly.

Nup155 predictably contains two distinct structural regions, similar to its yeast homologues Nup157p and Nup170p (Flemming et al., 2009; Kosinski et al., 2016; Seo et al., 2013; Whittle and Schwartz, 2009): an N-terminal β-propeller, which interacts with Nup53 and POM121 (Mitchell et al., 2010), and a C-terminal α-solenoid region, which binds the export cofactor Gle1 and the nucleoporin Nup98 (Lin et al., 2016). Interestingly, addition of the N-terminal β-propeller region of Nup155 (aa 1–589) rescued the depletion phenotype of the nuclear envelope assembly (Fig. 2). A closed nuclear envelope, indicated by smooth membrane staining, formed around the chromatin template. Nup155 staining was not detectable on the chromatin template. Despite the rescue of the membrane phenotype, it is possible that the N-terminal part of Nup155 is indeed absent from the nuclei. However, we cannot exclude the possibility that the fragment cannot be recognized by the antibody in immunofluorescence. The nuclei assembled in the presence of the N-terminal fragment showed only reduced staining of the NPC marker mAB414 with respect to the mock sample, similar to the Nup155-depleted nuclei (Fig. 2A). This could indicate that the Nup62 complex, as one of the NPC components recognized by the antibody, was not recruited to the assembling NPCs when the N-terminal β-propeller region of Nup155 replaced the full-length protein. Indeed, when analyzing Nup155-depleted nuclei supplemented with the N-terminal region of Nup155 (aa 1–589) with a variety of antibodies against nucleoporins, we detected on the chromatin MEL28/ELYS as well as Nup107, a member of the Y-complex (Fig. 3A). Nup53 was present on the nuclei in the absence of Nup155, consistent with the notion that it can be recruited to NPCs even if it cannot interact with Nup155 (Eisenhardt et al., 2014a). Nup153, which is part of the nuclear basket structure of the NPC and interacts with the Y-complex (Vasu et al., 2001) was also detected at the NPCs of these nuclei. The weak mAB414 staining observed might have therefore been Nup153, which is also recognized by this antibody (Fig. 3A). Nup214, also recognized by the mAB414 antibody, is strongly reduced on the surface of Nup155-depleted nuclei when analyzed with a Nup214-specific antibody, although to a lesser extent than other NPC components. Nup93, as well as the Nup62 complex component Nup62, were absent from the assembling pores, which was not due to depletion from the extract (Fig. 3B). Additionally, Nup98 was unable to localize on chromatin in both Nup155-depleted and addback nuclei. Since many of the FG-containing nucleoporins tested were not found on the chromatin in the absence of full-length Nup155, and especially considering that the Nup62 complex and Nup98, which are required for the nuclear transport function of the NPC, these nuclei were unable to import an EGFP-tagged nuclear import substrate, in contrast to the mock control (Fig. 3C). This lack of nuclear import is not due to a co-depletion of import factors, such as importin-α and -β (Fig. 3B). Together, these data suggest that nuclei in which endogenous Nup155 had been replaced by the N-terminal β-propeller fragment form a part of the inner ring components, but lack Nup93 and Nup98 as well as Nup188, Nup205 and the Nup62 complex, which are all recruited via Nup93 and Nup98. Since depletion of Nup93 in Xenopus egg extracts co-depletes Nup205 and Nup188 (Sachdev et al., 2012), it has been hard to establish whether Nup205 or Nup188 can be recruited by additional binding partners: indeed, biochemical analysis in Chaetomium identified novel bridge-like interactions between Nup53, Nup98, Nup155 and Nup205 homologues (Lin et al., 2016). Our work confirms that the Nup53–Nup155 interaction and recruitment to the assembling NPCs is not sufficient for proper localization of Nup98, Nup205 and Nup188 at NPCs at the end of mitosis, and that Nup93 is required for this.

The nuclei assembled in Nup155-depleted extracts supplemented with Nup155(1–589) showed a closed nuclear envelope and lacked Nup93 (Figs 2 and 3). This was surprising in the light of our previous finding that depletion of Nup93 does not allow for closed nuclear envelope formation (Sachdev et al., 2012; Theerthagiri et al., 2010) since removal of Nup93 did not co-deplete Nup53 and importantly Nup155 from the extracts. We therefore wondered whether the C-terminus of Nup155 has a negative, potentially autoinhibitory, influence on nuclear envelope formation. This domain might impede some nucleoporin interactions necessary for inner pore ring assembly.

We tested this idea by checking whether the N- and C-terminal domains of Xenopus Nup155 can interact as shown for the yeast homologue Nup170 (Flemming et al., 2009). GST pulldowns using the recombinant, GST-tagged Nup155 N-terminal β-propeller region (aa 1–589) and SUMO-tagged C-terminal fragments of Nup155 (aa 589–1344 and aa 504–1388) indeed showed an interaction between both regions (Fig. 4A). Note that because of difficulties in expression of some of the nucleoporin fragments, we used SUMO-tagged versions to increase their solubility in this and the following pulldown experiments. Next, we tested whether the C-terminus of Nup155 interferes with the described interaction between the N-terminus of Nup155 and the C-terminal region of Nup53 (Amlacher et al., 2011; Eisenhardt et al., 2014a; Lin et al., 2016). When titrating increasing amounts of SUMO–Nup155(504–1388), the interaction between GST–Nup53(162–320) and SUMO–Nup155(1–589) was weakened (Fig. 4B).

The nuclei assembled in Nup155-depleted extracts supplemented with Nup155(1–589) resembled nuclei we previously observed when replacing full-length Nup93 with its C-terminal domain (Sachdev et al., 2012). These nuclei contained large parts of the structural backbone of the NPC including the Y-complex, Nup53 and Nup155 but lacked the Nup62 complex, the main component of the central transport channel, required for nuclear import. We therefore speculated that the C-terminus of Nup93 could overcome the inhibitory effect of the Nup155 C-terminus in NPC structural backbone assembly. We tested this hypothesis by performing GST pulldowns using Nup155(1–589) and Nup53(1–312), which allows for the Nup155 interaction but also includes its N-terminal region, which is important for Nup93 binding (Lin et al., 2016; Sachdev et al., 2012). As expected, addition of the C-terminus of Nup155 weakened the interaction (Fig. 5). This effect is overcome by addition of the C-terminal region of Nup93 [SUMO–Nup93(608–820)], which enhances the Nup53–Nup155 interaction, as reported previously (Sachdev et al., 2012). Notably, the Nup93- and Nup155-binding regions within Nup53 have been mapped to different regions. Together, these pulldown data indicate that the N- and C-terminal regions of Nup155 can interact in isolation, and that this interaction weakens the binding of Nup155 to Nup53.

Here, we show that direct interaction of Nup155 with membranes is required for NPC formation. Surprisingly, the N-terminal β-propeller region of Nup155 is, at least in vitro, sufficient to rescue the nuclear envelope-depletion phenotype of Nup155, indicating that assembly of the structural backbone of the NPC can take place. Furthermore, self-interaction between the N- and C-terminal regions of Nup155 has an auto-inhibitory function, preventing efficient Nup155–Nup53 interaction, which is nonetheless required for further functional NPC assembly.

Nup155 belongs to a group of nucleoporins such as Nup53, Nup133 and Nup153 (Drin et al., 2007; Vollmer et al., 2015, 2012), which can directly interact with the pore membrane without containing a transmembrane region. Direct membrane interaction of both Nup53 and Nup155 is crucial for NPC assembly (Vollmer et al., 2012 and this study), probably because the membrane-interaction motifs also stabilize the highly curved nuclear membranes. Cryo-electron tomographic reconstruction of the human nuclear pore indeed suggests that Nup155 interacts with the pore membrane (von Appen et al., 2015). Modelling of the yeast Nup53 and Nup170 structures into the human tomographic reconstruction proposes that this mode of membrane interaction is conserved for both proteins (Lin et al., 2016). Indeed, genetic and biochemical studies also suggest that Nup170 and Nup157 (Aitchison et al., 1995; Miao et al., 2006; Tcheperegine et al., 1999), as well as Nup53 (Marelli et al., 1998, 2001; Miao et al., 2006; Onischenko et al., 2009; Tcheperegine et al., 1999), are located close to the pore membrane. In the filamentous fungus Aspergillus nidulans, no genes encoding the homologues of vertebrate Nup53 or yeast Nup59 have been identified, and NDC1 is non-essential (Osmani et al., 2006). In contrast, Nup170 is encoded by an essential gene in A. nidulans, and in the semi-open mitosis in this organism, in which NPCs partially disassemble, Nup170, its interaction partner Gle1 and the Y-complex are the only non-transmembrane proteins that remain at the pore (Osmani et al., 2006). It is likely that Aspergillus Nup170 interacts directly with the pore membrane to fulfil its function in NPCs: the corresponding leucine in position 258 is conserved in A. nidulans, as well as the hydrophobic nature of residues 258–267 that we deleted in our experiments.

In vertebrates, Nup53 interacts with both Nup155 and Nup93 (Eisenhardt et al., 2014a; Hawryluk-Gara et al., 2005), thereby indirectly connecting these two nucleoporins, which do not detectably bind each other (Eisenhardt et al., 2014a; Mitchell et al., 2010; Sachdev et al., 2012). Nup93, in turn, interacts with either Nup188 or Nup205 (Theerthagiri et al., 2010) and recruits the Nup62 complex (Chug et al., 2015; Sachdev et al., 2012) to the assembling NPCs (Fig. 6). This interaction network is evolutionarily conserved: biochemical work using purified recombinant Chaetomium thermophilum nucleoporins established that, in this fungus, Nup170 interacts with the C-terminus of Nup53 (Amlacher et al., 2011; Lin et al., 2016). This dimer then binds a second dimer formed by Nic96 (homologue to metazoan Nup93) and either Nup188 or Nup192 (homologue to metazoan Nup205) in a mutually exclusive way, resulting in formation of either a Nup170–Nup53–Nic96–Nup188 or a Nup170–Nup53–Nic96–Nup192 tetramer (Lin et al., 2016; Stuwe et al., 2014).

Re-addition of the N-terminal β-propeller region of Nup155 rescued the nuclear envelope, but not the NPC assembly phenotype of the Nup155 depletion. We have previously observed a similar phenotype when adding the C-terminal fragment of Nup93 to Nup93-depleted extracts (Sachdev et al., 2012). In that study, the structural backbone of the NPC could form, including the Y-complex as outer ring, as well as Nup53 and Nup155 as inner ring components. The C-terminus of Nup93 was suggested to stabilize the Nup53–Nup155 interaction. Given that the N-terminal region of Nup93 is required for recruitment of the Nup62 complex, which forms a large part of the central channel of the NPC, these components were missing from the nuclei. Similarly to the previously observed phenotype, here we show that Y-complex components and Nup53 were present when the N-terminus of Nup155 was added to Nup155-depleted nuclei. The Nup93 complexes, and thus the Nup62 complexes, were missing. Surprisingly, in comparison to the Nup93-depletion phenotype, which does not support assembly of the NPC structural backbone, the Nup155 fragment allowed this. A possible explanation could be that the C-terminus of Nup155 prevents NPC assembly in the absence of Nup93, probably because it has an inhibitory function that can be overcome by the presence of the C-terminal region of Nup93 (Fig. 6). In line with this hypothesis, overexpression of the C-terminal domain of Nup170 is toxic in S. cerevisiae, but toxicity can be rescued by overexpression of the N-terminal domain (Flemming et al., 2009). As the separated N- and C-terminal domains of Nup170 reciprocally interact in yeast two hybrid and pulldown assays, it has been suggested that, in the presence of an excess of the C-terminal domain of Nup170, the Nup170–Nup53 dimer does not assemble correctly. Indeed, our pulldown assays indicate that the C-terminus of Nup155 interferes with Nup155 binding to Nup53 by interacting with the N-terminal Nup155 β-propeller domain. Interestingly, the presence of Nup93, or more precisely its C-terminal region, can overcome the auto-inhibitory function of Nup155. Binding of Nup93 to the N-terminus of Nup53 might induce a change in Nup53, which overrides the auto-inhibitory effect of Nup155 C-terminus and strengthens the Nup53–Nup155 interaction. The auto-inhibition of Nup155 might constitute a crucial control point in the mitotic NPC assembly path: upon binding of Nup93, Nup53 signals that the assembly is proceeding correctly, and the Nup155–Nup53–Nup93–Nup188/205 complex is able to further recruit the Nup62 complex. It remains unclear why Nup93 is not detected on the chromatin template in the nuclei assembled in depleted extracts in which the N-terminal domain of Nup155 replaced the full-length protein. It is possible that during assembly of the structural NPC backbone Nup93 undergoes some structural rearrangements which need to be supported by the C-terminus of Nup155. Alternatively, in the absence of the auto-inhibitory function of the C-terminus of Nup155, protein interactions between the inner ring components of the NPC could occur ectopically, resulting in a lack of Nup93 on chromatin templates.

A homozygous R391H Nup155 mutation in humans has been linked to atrial fibrillation and early sudden cardiac death (Zhang et al., 2008). The R391H mutation was suggested to affect NPC localization of Nup155. Indeed, in a targeting assay, in which Nup155 R391H was ectopically localized to chromatin, its interaction partners Nup53 and POM121 showed less co-recruitment compared with the wild-type Nup155 control (Schwartz et al., 2015). However, given that Nup155 is essential in C. elegans, Drosophila and mouse, most likely because of its crucial role in NPC assembly, it seems questionable whether organisms are viable without Nup155 being able to localize to assembling NPCs. Indeed, our results suggest that the Xenopus laevis protein with the corresponding R385H mutation supports NPC assembly and is localized to the assembled NPCs. Most likely, the homozygous R391H mutation in humans causes the observed pathological phenotype for reasons other than defective NPC assembly. We would rather speculate that the homozygous mutation causes a milder deviation in NPC function that affects heart cells most severely, resulting in the cardiac defects. Indeed, a similar pattern has been observed for Nup93, another essential component of the inner ring module of NPCs: mutations causing a nephrotic phenotype were found in combination where at least one allele would support NPC assembly but caused a defect in NPC transport function (Braun et al., 2016). Patients with Nup93 mutations blocking NPC assembly on both alleles were not found; in all likelihood, this condition is not compatible with life, given the essential function of Nup93.

In conclusion, our work provides evidence for the requirement of the N-terminal membrane binding domain of Nup155 for proper nuclear envelope and NPC formation. Although required and sufficient for nuclear envelope formation, this N-terminal domain is not sufficient for proper NPC assembly, as recruitment of Nup93 and members of the Nup62 complexes are not supported after its addition back to Nup155-depleted nuclei. Yet, the assembly of a large part of the structural NPC backbone, including the outer-ring-forming Y-complex, is possible. We demonstrate an auto-inhibitory function of the C-terminal domain of Nup155 in inner pore ring assembly, which is required for assembly of functional NPCs. Finally, mutations of Nup155 associated with pathological condition of atrial fibrillation did not correlate with nuclear assembly defects in vitro, therefore indicating that a different malfunction of the protein is responsible for this phenotype. Identifying this malfunction remains an interesting research avenue.

GST fusions of Xenopus laevis Nup53 fragments (Vollmer et al., 2012) and the SUMO-tagged Xenopus laevis Nup93 fragment (Sachdev et al., 2012) have been described. N-terminal (aa 1–589) and C-terminal (aa 589–1388 and 504–1388) Xenopus laevis Nup155 fragments were cloned as codon-optimized sequences for expression in E. coli into a modified pET28a vector with a yeast SUMO solubility tag or a GST tag, which is followed by a tobacco etch virus (TEV) cleavage site. A Xenopus laevis Nup53 fragment (aa 162–267) serving as a control prey was also cloned into a modified pET28a vector with a yeast SUMO solubility tag. All proteins were expressed in E. coli by autoinduction at 18°C and purified by Ni-affinity chromatography. Proteins were dialyzed to PBS and employed in pulldown experiments.

For addback experiments to egg extracts, the N-terminal (aa 1–589) Nup155 fragment was cloned into a modified pET28a vector with a yeast SUMO solubility tag followed by a TEV cleavage site. After expression in E. coli by autoinduction at 18°C and by Ni-affinity chromatography purification, the SUMO tag was cleaved off by incubating the purified protein with TEV protease. The His6-tagged SUMO moiety and the TEV protease were then removed via further incubation with Ni-affinity beads. Unbound material containing tag-cleaved protein was dialyzed against sucrose buffer with high salt (250 mM sucrose, 750 mM KCl, 10 mM HEPES-KOH, 2.5 mM MgCl₂).

For pulldown assays in PBS buffer, 0.5 µM GST baits were incubated with 0.5 µM of prey proteins in a 750 µl volume for 1 h at 4°C. Where indicated, 1 µM or 3 µM His6–SUMO-tagged Nup155 504–1388 and 3 µM His6–SUMO-tagged Nup93 608–820 were added. After 1 h, 20 µl of the samples were removed for western blot analysis as input. The remaining sample was incubated for 2 h with 30 µl of 50% slurry of GSH-Sepharose 4B (GE Healthcare). GSH beads were washed five times with PBS and eluted in 30 µl total volume by TEV protease cleavage (0.5 mg/ml) for 1 h at 25°C. The TEV protease cleaves the GST fusions between the GST moiety and Nup53 or Nup155 fragment.

The input and elutions were analyzed by western blotting using a mouse monoclonal anti-His6 antibody (Roche, 11922416001; 1:2000) which recognizes the His6–SUMO-tagged fusion proteins. For quantification, ECL signals were determined and the signal ratio of eluate/input was plotted.

Preparation of high-speed interphase extracts, sperm heads, floated labelled and unlabelled membranes required for in vitro nuclear assembly as well as immunofluorescence experiments were carried out as described (Eisenhardt et al., 2014b). Fluorescence images were acquired using a confocal microscope [FV1000; Olympus; equipped with a photomultiplier (model R7862; Hamamatsu)] using 405, 488 and 559 nm laser lines and a 60× NA 1.35 oil-immersion objective lens. Rabbit polyclonal antibodies against Nup107 (Walther et al., 2003), Nup160 (Franz et al., 2007), MEL28/ELYS (Franz et al., 2007), Nup53 (Theerthagiri et al., 2010), Nup205 (Theerthagiri et al., 2010), Nup188 (Theerthagiri et al., 2010), Nup155 (Franz et al., 2005), Nup98 (Theerthagiri et al., 2010) have been described and were employed at 1:1000 dilution for western blotting and 1:100 for immunofluorescence, Nup153 (Vollmer et al., 2015) 1:100 for immunofluorescence, mouse monoclonal antibodies mAB414 and anti-His6 antibodies were purchased from Covance (MMS-120R, used 1:2000 for immunofluorescence, and 1:10,000 for detection of Nup62, Nup214 and Nup153 in western blot) and Roche (11922416001, used 1:1000 for western blotting), respectively. The rabbit polyclonal antibodies against Nup62 and Nup214 were kind gifts from Birthe Fahrenkrog (Laboratoire du Biologie du Noyau, Institut de Biologie & de Médecine Moléculaire Université Libre de Bruxelles, Charleroi, Belgium) or Maarten Fornerod (Departments of Pediatric Oncology and Cell Biology, Faculty Molecular Medicine and Nanobiology, Erasmus University Medical Center, Rotterdam, The Netherlands), respectively, and were used at a concentration of 1:50 or 1:100, respectively, for immunofluorescence.

The anti-Nup155 antibody for depletion was generated in rabbits using full-length Xenopus laevis Nup155 (see Vollmer et al., 2015). Anti-Nup155 beads and mock beads were generated by coupling anti-Nup155 serum and unspecific rabbit IgG, respectively, to Protein-A Sepharose resin (GE Healthcare) with 10 mM dimethyl pimelimidate (Thermo Fisher Scientific). Nup155 and mock immunodepletions for addback of Nup155 fragments were performed by mixing X. laevis high-speed extracts with beads at a 1:0.8 ratio twice rotating at 8°C for 15 min each.

For rescue experiments with full-length Nup155 and Nup155 mutants, mRNA encoding Xenopus laevis Nup155 (GenBank accession NM_001087331.1) was prepared using the mMESSAGE mMachine kit (Life Technologies) and added to extracts to a final concentration of 300 ng/μl. For rescue experiments with the Nup155 N-terminal domain (aa 1–589), the fragment was added to the reactions to a final concentration of 100 nM. To compensate for volume change, the same volume of high-salt sucrose buffer was added to the mock reactions.

To test nuclear import, a NLS–2×GFP protein and its nuclear localization sequence (NLS) mutant as negative control, both cloned into the pTrcHisB vector (Yang et al., 2004), were expressed in E. coli and purified by Ni-affinity and size exclusion (Superdex 200; GE Healthcare) chromatography. Proteins were dialyzed to sucrose buffed and added to the assembly reaction in final concentration of 300 nM.

We are grateful to Birthe Fahrenkrog for providing the Nup62 antibody and Maarten Fornerod for the Nup214 antibody. We thank Katharina Schellhaus, Hideki Yokoyama and Daniel Moreno-Andrés for critical reading of the manuscript.