The Nup155-mediated organisation of inner nuclear membrane proteins is independent of Nup155 anchoring to the metazoan nuclear pore complex

The nuclear pore complexe (NPC), a structure composed of several copies of ∼30 different proteins called nucleoporins or Nups, regulates the trafficking of macromolecules between the nucleus and the cytoplasm (Tran and Wente, 2006). Beyond its transport role, the NPC has several other cellular functions, such as nuclear membrane formation (Hetzer et al., 2005). Both Nup155 and Nup53 are essential for nuclear envelope (NE) biogenesis in higher eukaryotes (Franz et al., 2005; Hawryluk-Gara et al., 2008; Ródenas et al., 2009). The Nup155–Nup53 interaction seems to have an evolutionarily conserved function as the Nup155-binding site of human Nup53 is able to restore nuclear membrane assembly in Xenopus extracts depleted of endogenous Nup53 (Hawryluk-Gara et al., 2008). Interestingly, this region overlaps with the part of Nup53 that is absent in a C. elegans mutant that exhibits severe NE defects (Ródenas et al., 2009). How Nup155 and Nup53 mediate the formation of the NE is not yet understood.

The in vivo study of Nup155 is severely hampered by its essential function in cell survival. To overcome this problem, we sought to take advantage of the hypomorphic situation in cell culture RNAi systems. We inactivated the gene by treating Drosophila S2 and KC cells with Nup155 dsRNA and stained the cells for lamin B receptor (LBR) and otefin, two inner nuclear membrane (INM) proteins for which we had antibodies available. While LBR and otefin were restricted to the NE in control (untreated) cells, they were partially mislocalised to the cytoplasm in dsRNA-treated cells (Fig. 1A,B, and data not shown). Treating the cells with Nup155i-2, a second set of Nup155 dsRNA that had no overlapping sequence with the original one (Nup155i-1), resulted in a similar defect (Fig. 1C). Neither Nup155i-1 nor Nup155i-2 altered the levels of LBR or otefin significantly (supplementary material Fig. S1). Finally, expression of V5-tagged Nup155 (Nup155-V5) in Nup155 RNAi cells was able to restore the normal distribution of LBR (Fig. 1C). The localisation of INM proteins might be affected by alterations in the nuclear lamina as binding to lamin targets many proteins to the INM. To assess the lamina status, we checked the level of lamin Dm0 protein in Triton X-100-soluble and -insoluble fractions and also carried out fluorescence recovery after photobleaching (FRAP) for EGFP–lamin Dm0 (supplementary material Fig. S2). We could not detect any significant difference between control and Nup155 RNAi cells using either of methods. Based on the overall data, we concluded that LBR and otefin mislocalisation phenotype is due to Nup155 inactivation. This is consistent with the recent finding that Nup155 is required for the targeting of INM proteins to the NE in human HeLa cells (Mitchell et al., 2010).

Do nuclear transport deficiencies account for the nuclear membrane protein mislocalisation phenotype? To address this question, we treated the cells with Nup155 dsRNA and stained them for nuclear proteins such as CREB-binding protein, proteasome activator 11S REG gamma and mRNA-binding protein NonA. None of these proteins was mislocalised to the cytoplasm in Nup155 RNAi cells (supplementary material Fig. S3). Similarly, there was no clear change in the import and export of soluble reporter cargos between control cells and Nup155-inactivated cells in previous studies (Sabri et al., 2007; Mitchell et al., 2010), suggesting that the general protein transport is not perturbed. The localisation of mAb414-interacting proteins, the NPC proteins that function in nuclear transport, was also unaffected (supplementary material Fig. S3). A recent study has shown that the lack of the NPC protein Nup188 accelerates INM protein translocation through the pore and that the Nup188 depletion from in vitro NE assembly reactions causes a three-fold change in nuclear size (Theerthagiri et al., 2010). However, there was no significant size difference between the nuclei of untreated and Nup155-dsRNA-treated cells (supplementary material Fig. S4), indicating that Nup155 is probably not a major player in INM protein transport pathways.

As a first approach to understand how the interaction of Nup155 with its neighbouring Nups contributes to the function of Nup155, we stained Nup93, Nup53 and NDC1 RNAi cells with anti-Nup155 antibody (supplementary material Fig. S5). Nup155 was completely mislocalised to large cytoplasmic foci in Nup93 dsRNA-treated cells, whereas there was no change in the Nup155 localisation in Nup53 RNAi cells. Nup155, as well as lamin clustered into a cap-like structure at one side of the NE in NDC1 RNAi cells. This is consistent with the essential function of NDC1 in NPC assembly and NE formation (Mansfeld et al., 2006), indicating that the effect of NDC1 on Nup155 localisation might be indirect.

GST pull-down assays showed that both endogenous and recombinant Nup155 proteins interacted with Nup93 (Fig. 2A; supplementary material Fig. S6). To map the Nup93-binding domain of Nup155, we used an in vitro translation system to generate several radiolabeled Nup155-deleted proteins (Fig. 2B) and incubated Nup155 products with GST or GST–Nup93 (Fig. 2C). Nup155ΔC [amino acids (aa)1–986], which lacked the C-terminal domain of Nup155, was still able to bind to Nup93. However, the lack of the N-terminal part of the protein eliminated the binding of Nup155ΔN (aa508–1365) to Nup93 to a large extent. We concluded that the N-terminal part of Nup155, which has a predicted β-propeller structure (Fig. 2D), contains the major binding site for Nup93. In agreement with this concept, Nup93 mainly interacted with Nup155N (aa1–539) in our binding assays. Interestingly, the β-propeller domain of Nup155 also binds to POM121 in vertebrates (Mitchell et al., 2010), indicating that this region of Nup155 is the major NPC-binding site.

Finally to determine the functional significance of the interaction of Nup93 with Nup155, we decided to perform RNAi rescue experiments. We overexpressed the V5-tagged version of Nup155 proteins in S2 cells (Fig. 3A; supplementary material Fig. S7) and analysed whether the overexpression of these products restored the LBR localisation of Nup155 RNAi cells (Fig. 3B,C). Nup155 full-length was localised to the nuclear rim, and in the cytoplasm near the nuclear membrane, and rescued the LBR mislocalisation phenotype of Nup155 dsRNA-treated cells (69.8% rescue). Nup155N, which exhibited a localisation similar to that of Nup155 full-length failed to restore LBR distribution (0.1% rescue). The lack of the N-terminal part of the protein displaced Nup155ΔN into the nucleus, indicating that in the absence of β-propeller region, Nup155 is no longer sequestered to the nuclear membrane tightly. Remarkably, however, Nup155ΔN was capable of even better rescue than Nup155 full-length (83.0% compared to 69.8%). From experiments described above, we concluded that the binding of Nup93 to the Nup155 β-propeller plays no role in LBR targeting to the NE while it is necessary for Nup155 anchoring to the NPC.

We divided Nup155ΔN, a part of protein with a predicted α-solenoid structure (Fig. 2D) and an overlapping sequence with the HDAC4-binding domain of human Nup155 (Kehat et al., 2011), into two parts: Nup155C (aa959–1365) was localised in small cytoplasmic foci whereas Nup155M (aa508–986) appeared in the nucleus (Fig. 3A). We extracted chromatin from the cells that expressed Nup155M. As Fig. 4A shows, Nup155M was specifically co-purified with chromatin proteins. Consistently, a histone H3 antibody co-precipitated Nup155M from chromatin extract (Fig. 4B). We concluded that the nuclear localisation of Nup155M is probably associated with its chromatin-binding activity. In order to understand the functional significance of chromatin-binding region of Nup155, we carried out RNAi rescue experiments (Fig. 3B,C). Nup155M restored the normal localisation of LBR in Nup155-inactivated cells (82.9% rescue) while Nup155ΔM could not rescue.

We also took a synchronization approach based on the UV-induced inactivation of colcemid in Drosophila cells (Schubiger and Edgar, 1994), which successfully enhanced the proportion of post-metaphase mitotic figures (Fig. 4C). We then prepared chromatin from untreated and treated cells and analysed the extracts by immunoblotting using anti-Nup155 antibody. As Fig. 4D shows, chromatin derived from colcemid-treated cells had two times more Nup155 than control chromatin suggesting that Nup155 has a stronger affinity for chromatin before interphase.

Drosophila Nup53 dsRNA-treated cells exhibited LBR mislocalisation phenotype moderately, but consistently (supplementary material Table S1), implying that both Nup53 and Nup155 influence INM protein localisation. There are several studies indicating that Nup53 and Nup155 directly bind to each other (e.g. Hawryluk-Gara et al., 2008; Onischenko et al., 2009). So it is conceivable that Nup53 and Nup155, as parts of one protein complex, function in the same pathway to direct INM proteins to the NE. Using GST pull-down, Drosophila Nup53-binding sites were mapped to Nup155N and Nup155M (Fig. 2C). Because Nup155N has no contribution in LBR localisation, one might expect that the interaction of Nup155M with Nup53 plays a major role in the targeting of LBR to the NE. More functional studies are needed, however, to examine this expectation.

The absence of NPCs does not seem to prevent nuclear membrane formation as Xenopus extracts depleted of Nup107 or MEL-28 are still able to produce NEs that are devoid of NPCs (Harel et al., 2003; Walther et al., 2003; Franz et al., 2007). This and the finding that it is possible to make NPC-free nuclear membranes by treating the in vitro NE assembly reactions with the calcium chelator BAPTA (Macaulay and Forbes, 1996), raise an important question of whether the essential function of nucleoporins in NE formation is mediated through an NPC-independent pathway. Here, we provide the first functional mapping of metazoan Nup155. We show that the Nup155 β-propeller plays no role in the targeting of LBR to nuclear membranes while it is necessary for the anchoring of Nup155 to the pore. The α-solenoid part of Nup155, which is required for the normal localisation of LBR, displays a chromatin-binding activity that is stronger at the end of mitosis. All together, our findings raise an intriguing possibility that the role of Nup155 in INM organization is mediated by the chromatin-binding function of the protein. Although, we are still unable to establish whether the INM protein mislocalisation of Nup155-depleted cells occurs during post-mitotic NE formation or it is an interphase phenotype. Further studies will be required to identify the chromatin proteins that interact with Nup155 and to specify the time point when these interactions become essential for INM protein organization.

S2 and KC167 lines were obtained from DGRC, Indiana University. RNAi and RNAi rescues were carried out as described (Sabri et al., 2007). For Nup155 rescue assays, Nup155i-2, a dsRNA product that targeted the 3′ UTR of Nup155 was used. For transfection, cells were transfected with pAC5.1/V5 plasmids (Invitrogen) that encoded C-terminally V5-tagged Nup155 proteins using the Effectene transfection reagent (Qiagen).

Antibody staining of cells was performed as described previously (Sabri et al., 2007). Images were collected by a BioRad Radiance 2000 confocal microscope using a 63× (1.4 NA) objective. Each image represents a single XY section of 0.8 µm thickness, which was taken through the centre of the cell. The dynamic of the nuclear lamina was assessed as previously described in Broers et al. using fluorescence recovery after photobleaching (FRAP) (Broers et al., 1999). The images were acquired using a LSM 700 confocal microscope, with a 63× (1.4 NA) objective and a pinhole setting of 1 AU. Cells were recorded before and at 0.2 sec intervals after bleaching for up to 10 sec and the intensity of the bleached area was monitored. Finally, cells were recorded 120 sec after bleaching. All confocal data were processed with ImageJ.

In vitro translations of Nup155 products were performed using the corresponding pET28 plasmid (Novagen), [³⁵S]methionine (Perkin-Elmer) and T7 TNT in vitro translation system (Promega). For pull-down assays, pGEX-5X plasmids (GE Healthcare) that expressed GST or N-terminally tagged GST–Nup53 and GST–Nup93 were used. Protein expression and binding assays were carried out as described previously (Roth et al., 2003). Protein extraction and western blots were performed as described previously (Sabri et al., 2007). For evaluating radioactively labelled bound pools, gels were exposed in a FLA 3000 phosphorimager (Fuji) and resulted digital images were quantified by ImageJ.

Semi-synchronized cells were prepared by modifying a protocol from Stubblefield et al. (Stubblefield et al., 1967). In brief, cells were cultured at the concentration of 1×10⁶ cells per ml 18 hr prior to the treatment. Cells then were treated with colcemid (Sigma) at a concentration of 0.03 µg/ml of culture for 6 hr. The untreated and colcemid-treated cells were exposed to UV light for 30 sec and kept in fresh medium for 60 min before being used for either chromatin preparation or staining with antibodies for phosphorylated histone H3 and α-tubulin to analyse mitotic configurations. Chromatin extraction and chromatin IP were performed as described previously (Tyagi et al., 2009).

The fractionation was carried out according to a protocol from Kasahara et al. (Kasahara et al., 1991). In brief, cells were lysed in 20 mM Tris-HCl pH 7.2, 1% Triton X-100, 1 mM EDTA, 1 mM EGTA, and protease inhibitors and then spun down for 1 hr at 100,000 g. The Triton X-100-insoluble proteins were lysed in 9.5 M urea containing 2% CHAPS and 71 mM DTT.

We used primary antibodies against LBR and otefin (Wagner et al., 2004), lamin Dm0 and GP210 (DSHB, University of Iowa), Nup155 (Gigliotti et al., 1998), CBP (Lilja et al., 2003), REG gamma (Masson et al., 2001), NonA (Buchenau et al., 1993), TPR (a gift from V. Cordes, Max Planck Institute for Biophysical Chemistry, Göttingen, Germany), RanGAP (Merrill et al., 1999), histone H3 (Abcam), phosphorylated histone H3 (Abcam), α-tubulin (Sigma), V5 (Invitrogen), GST (Santa Cruz), His (Qiagen) and GFP (Roche).

We are grateful to S. Gigliotti, M. Mannervik, P. Young, B. Ganetzky, V. Cordes, H. Saumweber, DSHB, and DGRC for providing antibodies and cDNA clones, and N. Visa and S. H. Gee for critical reading of the manuscript. We are indebted to G. Krohne for EGFP-lamin Dm0 construct. We thank the University of Gothenburg image core facility.