The Inner Nuclear Membrane Protein Nemp1 Is a New Type of RanGTP-Binding Protein in Eukaryotes

The inner nuclear membrane (INM) protein Nemp1/TMEM194A has previously been suggested to be involved in eye development in Xenopus, and contains two evolutionarily conserved sequences in the transmembrane domains (TMs) and the C-terminal region, named region A and region B, respectively. To elucidate the molecular nature of Nemp1, we analyzed its interacting proteins through those conserved regions. First, we found that Nemp1 interacts with itself and lamin through the TMs and region A, respectively. Colocalization of Nemp1 and lamin at the INM suggests that the interaction with lamin participates in the INM localization of Nemp1. Secondly, through yeast two-hybrid screening using region B as bait, we identified the small GTPase Ran as a probable Nemp1-binding partner. GST pulldown and co-immunoprecipitation assays using region B and Ran mutants revealed that region B binds directly to the GTP-bound Ran through its effector domain. Immunostaining experiments using transfected COS-7 cells revealed that full-length Nemp1 recruits Ran near the nuclear envelope, suggesting a role for Nemp1 in the accumulation of RanGTP at the nuclear periphery. At the neurula-to-tailbud stages of Xenopus embryos, nemp1 expression overlapped with ran in several regions including the eye vesicles. Co-knockdown using antisense morpholino oligos for nemp1 and ran caused reduction of cell densities and severe eye defects more strongly than either single knockdown alone, suggesting their functional interaction. Finally we show that Arabidopsis thaliana Nemp1-orthologous proteins interact with A. thaliana Ran, suggesting their evolutionally conserved physical and functional interactions possibly in basic cellular functions including nuclear transportation. Taken together, we conclude that Nemp1 represents a new type of RanGTP-binding protein.


Introduction
The nuclear envelope (NE) is not only the boundary that separates the nuclear and cytoplasmic compartments of eukaryotic cells but it also plays regulatory roles in chromatin organization and gene expression through its nucleoplasmic surface [1]. The NE is composed of double nuclear membranes, nuclear pore complexes (NPCs), and a fibrous protein meshwork called the nuclear lamina. The regulatory role of the NE is mainly attributed to NPC proteins and the inner nuclear membrane (INM) proteins [2]. To further elucidate the role of the NE, it is important to identify and characterize INM proteins, because a large number of putative integral NE proteins remain to be analyzed [3].
INM proteins such as Emerin and MAN1 have been shown to bind to lamins and hence reside on the INM [2]. In terms of the function of these INM proteins, Emerin and MAN1 contain LEM domains that interact with the BAF (barrier to autointegration factor) to anchor chromatin to the NE. They also bind to signal transducers at the INM to modulate BMP/TGFβ and Wnt signaling. For example, MAN1 binds to R-Smad and attenuates BMP/TGF-β signaling [4], and Emerin binds to β-catenin and down-regulates Wnt signaling [5]. Thus, INM proteins play regulatory roles in signal transduction in addition to gene regulation, chromatin organization, and NE formation [6].
NPCs mediate the bidirectional transport of proteins and RNAs across the NE. Nuclear transport proteins, such as importin β/karyopherin β, exportin 1/Crm1, and the small GTPase Ran facilitate the transport of proteins through NPCs [7]. Ran exists in a GTP-bound (RanGTP) and a GDP-bound (RanGDP) state, which are enriched in the nucleus and the cytoplasm, respectively. Their differential localizations are maintained by RCC1 (the nucleotide exchange factor for Ran), which binds to the chromatin in the nucleus, and by RanGAP1 (the Ran GTPase activating protein) in the cytoplasm. Importins and exportins function as transporters for various cargos. For example, transcription factors that can interact with importin β are transported as cargos into the nucleus. In the nucleus, they are dissociated from importin β upon the binding of importin β to RanGTP through the effector domain of Ran. Such nuclear transport mechanisms are conserved between animals and plants [8]. Other than nuclear transportation, Ran is also involved in controlling mitotic checkpoints, spindle assembly, and NE re-assembly through its interactions with importins [9], suggesting that Ran is a crucial factor throughout every stage of the cell cycle.
Recently, we identified a new INM protein, Nemp1 (also known as TMEM194A), which is expressed in the anterior neural plate in Xenopus [10]. Nemp1 contains an evolutionarily conserved region A within its transmembrane domains (TMs) and region B within its C-terminal region, but does not contain any known domains. We have shown that (i) Nemp1 is localized to the INM; (ii) region B faces the nucleoplasm and binds to BAF through a BAF binding site (BBS); (iii) both overexpression and knockdown of Nemp1 in Xenopus embryos reduce the expression of early eye-specific genes, resulting in severe eye defects; and (iv) Nemp1 activity requires regions A, a Lys-Arg-rich (KR) sequence, and region B [10]. Thus, our data suggest that a proper level of Nemp1 at the INM is required for eye development. However, the molecular function of Nemp1 remains to be clarified.
In this study, we analyzed the molecular nature of Nemp1 in terms of the functional roles of region A, the KR sequence, and region B using Xenopus laevis and mouse (Mus musculus) Nemp1, designated as Xl_Nemp1 and Mm_Nemp1, respectively. To the best of our knowledge, our study is the first report Nemp1 as a new type of Ran-binding protein that exhibits a fundamental cellular function.

Ethics statement
The work in this paper was conducted using protocols approved by the Animal Care and Use Committee of the University of Tokyo (permit number: 01-13).

cDNA cloning and plasmid constructs
A full-length cDNA clone (attBpBC-mKIAA0286) of Mus musculus (Mm) Nemp1 (Mm_Nemp1) (accession no. NM_001113211) was obtained from the Kazusa DNA Research Institute. Mm_ran (NM_009391) and Xenopus laevis (Xl) ran (NM_001086713) were isolated from the mouse 11-day embryo MatchMaker cDNA library (Clontech) and Xenopus total RNA at the neurula stages, respectively. Arabidopsis thaliana (At) nemp genes (NM_102639; NM_001037091; NM_114844) and At_ran2 (NM_122009) were isolated from Arabidopsis total RNA (a gift from Dr. S. Sawa). Plasmid constructs were made with HA, Myc, and FLAGtagged vectors, which were derived from pCSf107mT [11] and pCS2+. Two-round PCR-based mutagenesis was performed for making point-mutated, deleted, or chimeric constructs. All constructs and vectors used for this study are listed in S1 Table. Yeast two-hybrid screening assay The yeast MatchMaker Two-Hybrid System (Clontech) was used to screen the mouse 11-day embryo MatchMaker cDNA library using Mm_Bt (334 to 437a.a of Mm_Nemp1) as bait. The bait plasmid pGBKT7-Mm_Nemp1_Bt and the cDNA library were sequentially transformed into the yeast strain AH109. Transformants (9 × 10 6 ) were plated and screened on 100 mm-diameter plates with medium lacking leucine, tryptophan, and adenine. Colonies were picked and checked for ß-galactosidase production by using a filter assay with 5-bromo-4-chloro-3-indolyl-ß-D-galactopyranoside. Plasmid purification was done from the positive clones, and a second round of interaction screening was performed to confirm the interactions. The inserts from the positive clones were sequenced.

Microinjection experiments using Xenopus embryos
Fertilization and manipulation of Xenopus laevis embryos and microinjection of mRNA or morpholino oligo (MO) were carried out as described previously [10]. Embryos were staged according to the criteria of Nieuwkoop and Faber [12]. Antisense MOs for Xl_nemp1 (nemp1-MOs) [10] or Xl_ran (ranMO) were obtained from Gene Tools LLC. ranMO is complementary to the sequence encompassing the translation start sites of both homoeologs of ran (Xl_ran-a: NM_001086713 and Xl_ran-b: NM_001135075) (5'-CTTGAGGTTCTCCTTGGGCTGC CAT-3'). Standard control MO (stdMO; Gene Tools LLC) was used as negative control. MOs were dissolved in water and heated at 65°C for 10 min before use. mRNAs or MOs were injected into a dorsoanimal blastomere at the 4 cell stage, in which the injected area was fated to the anterior neural plate. FITC-dextran (50 ng/embryo) was used as a tracer.

Whole-mount in situ hybridization
Whole-mount in situ hybridization (WISH) was performed according to Harland [13]. Antisense Xl_ran RNA probes were transcribed with T7 RNA polymerase from SalI-linearized pGEM-T-Xl_ran.

Purification of recombinant proteins and GST pulldown assays
GST fusion constructs for Mm_Bt (GST-Mm_Bt) and Myc-tagged Mm_RanQ69L (Myc-RanQ69L) were made using pGEX6Pmcs. Purification of GST-fusion proteins and cleavage of a GST portion from GST-Ran were carried out as described previously [14]. Loading of GTP to recombinant RanQ69L was carried out in binding buffer (20 mm Tris-HCl, pH 8.0, 50 mm NaCl, 2.5 mm MgCl 2 , 0.5% NP-40, 1% BSA and 10% (v/v) glycerol) containing 2 mM GTP by incubating at room temperature for 30 min in a final volume of 50 μl, then diluting to 250 μl in binding buffer. For GST pulldown assays, GST-Mm_Bt attached to glutathione-Sepharose beads was incubated at 4°C for 2 h with cell lysate (see below) or with GTP-loaded RanQ69L in 300 μl of binding buffer. The beads were washed 4 times with binding buffer. Pulled down proteins were analyzed by western blotting with FluoroTrans membranes (Pall corporation) and the appropriate antibodies as described [10].

In vitro alkaline phosphatase assays
When calf intestine phosphatase (CIAP) (New England Biolabs: NEB) was used, Xenopus embryos overexpressing HA-tagged Xl_Nemp1 (Xl_Nemp1-HA) were lysed in lysis buffer A. Lysates were incubated with anti-HA antibody at 4°C for 1 h, then added with protein G-agarose beads, and incubated for another 1.5 h. The beads were washed 3 times with lysis buffer A, once with NEBuffer 3 (NEB), and incubated in NEBuffer 3 containing 0.5 u/ml of CIAP for 3 h at room temperature. When λ protein phosphatase (NEB) was used, Xenopus embryos overexpressing mouse Nemp1-HA (Mm_Nemp1-HA) were lysed in lysis buffer A without EDTA. Lysates were incubated with λ protein phosphatase in NEBuffer for Protein MetalloPhosphatases (NEB) for 45 min at 30°C. Treated samples were analyzed by western blotting with anti-HA antibody.

Measuring cell densities and ratios of mitotic cells in Xenopus embryos
MOs and FITC-dextran (50 ng/embryo) as a tracer were injected into the dorsoanimal region of four-cell-stage embryos. Injected embryos were fixed at the late gastrula to early neurula stages (stages12. [5][6][7][8][9][10][11][12][13] and immunostained using rabbit anti-phospho-Histone H3 (Ser10) antibody (Millipore) as primary antibody and Alexa Fluor 555-conjugated antibody (Molecular Probes) as secondary antibody, as described previously [11]. Nuclei were stained with DAPI. Confocal microscopic analyses were performed with LSM 710 (Zeiss). Five embryos from each experimental group were used for counting the number of nuclei and pH3-positive nuclei in more than two separate regions (a total area was more than 0.1 mm 2 ) of MO-injected and FITC-positive regions of each embryo. For rescue experiments, embryos were co-injected with MOs and nemp1, ran, or globin (negative control) mRNA together with EGFP-HA mRNA (200 pg/embryo) as a tracer, fixed at stages12.5-13, stained with DAPI and anti-HA antibody. Nuclei were counted in EGFP-HA positive areas. The statistical significance (P-value) was calculated using Student's or Welch's t-test after comparison of the variances of a set of data by F-test.

RT-quantitative PCR (RT-qPCR)
Total RNA was isolated from mock or Mm_Nemp1-HA-transfected COS-7 cells (derived from African green monkey) using Trizol (Life Technologies) and digested with RQ1 DNase (Promega). Quantitative PCR analysis was performed using a StepOnePlus real-time PCR system (Applied Biosystems) with PCR primers for Nemp1 and GAPDH (glyceraldehyde 3-phosphate dehydrogenase). Nemp1-RT primers were designed for conserved sequences in nemp1 CDSs of mouse, green monkey (Chlorocebus sabaeus) (XM_008003729), and human (NM_001130963). Green monkey GAPDH (Cs_GAPDH) primers were designed for conserved sequences in gapdh CDSs of green monkey (XM_007967342) and human (NM_001289745), and used as an internal control. Real-time PCR assays were performed in triplicate using the following primers:

Region A of Nemp1 plays a role in its colocalization with lamins
In this study, we used both Xl_Nemp1 and Mm_Nemp1, in which two conserved regions, region A and region B show 67% and 80% identities, but the KR sequence is unique to Xl_Nemp1 [10] (Fig 1A).
Because region A of Xl_Nemp1 is sufficient for its nuclear envelope (NE) localization [10], we examined the interaction of Nemp1 with lamins and the NPC component Nup153 using specific mouse monoclonal antibodies. Nemp1 was transfected to COS-7 cells, and colocalization was analyzed by confocal analysis. Expression levels of exogenous Mm_Nemp1-HA were much higher than those of endogenous nemp1 in COS-7 cells as assayed by RT-qPCR (S2 Table), implying that the behavior of tagged proteins was not affected by the endogenous protein. The data showed that the punctated staining of Mm_Nemp1-HA significantly colocalized with lamins at the NE ( Fig 1B; upper panels, indicated by white arrows) but not as significantly with Nup153 at the NPC ( Fig 1B; lower panels, indicated by magenta arrows). Using deletion constructs of Xl_Nemp1, we tested which region of Nemp1 is required for its colocalization with lamins and found that Xl_ΔBt-HA, but not Xl_ΔA-HA, colocalized with lamins ( Fig 1C). These data suggest that Nemp1 colocalizes with lamins through region A.
We have previously suggested that the KR sequence is a nuclear localization signal (NLS), which is required for nuclear localization of the C terminal region (Ct; see Fig 1A) [10]. To examine NLS function, the KR sequence was fused to GST-mRFP-HA, which alone cannot be transported into the nucleus. In COS-7 cells, GST-mRFP-HA fused with KR (KRa or KRb from the homoeologs) or SV40NLS as a positive control exclusively localized to nuclei, but KRa(ΔR) or KRm (from Mm_Nemp1) did not. These data clearly indicate that the KR sequence functions as a NLS (S1 Fig A). On the other hand, although the KR sequence is not present in Mm_Nemp1, Mm_Bt (the region B plus its downstream region; see Fig 1A) was localized to the nucleus (S1 Fig B), which is different from Xl_Bt [10], suggesting the possibility that Mm_Bt itself has a cryptic NLS. Therefore, to test this possibility, Mm_Bt was fused to GST-mRFP-HA. This GST-mRFP-Mm_Bt-HA protein exhibited, though in a part of cells, nuclear localization (S1 Fig B), implying that Mm_Bt can exhibit NLS function under some conditions. These data suggest the possibility that the KR sequence in Xl_Nemp1 and the Cterminal region in Mm_Nemp1 as well as region A for association with lamins participate in the INM localization of Nemp1.

Nemp1 oligomerizes with itself and INM proteins thorough TMs
The INM proteins MAN1 and Emerin have been shown to be associated with each other in vitro [16]. Therefore, we examined the interaction of Nemp1 with MAN1 and Emerin, and with itself by co-IP assays using embryos overexpressing HA-tagged Xl_Nemp1 with either Myc-tagged Xl_Nemp1, MAN1, or Emerin (Fig 2A). We found that Xl_Nemp1 forms a complex with itself and to a lesser extent with MAN1 or Emerin (Fig 2A). Deletion analysis revealed that HA-tagged WT, ΔN, ΔA, ΔB, SP+A, and SP+TM but not ΔTM, N, or Ct (see S2 Fig for structures of constructs) were coimmunoprecipitated with Myc-tagged WT, indicating that the TMs are both required and sufficient for the oligomerization of Nemp1 (Fig 2B). The punctated staining of tagged Nemp1 at the nuclear membrane might reflect the oligomerization ability of Nemp1 (see Fig 1B). These data suggest that Nemp1 complexes could be formed through the TMs in the NE and perhaps the ER, and might directly associate with MAN1 and Emerin or indirectly through the nuclear lamina because all three can associate with lamin. were used (see [10] for more detail). After immunoprecipitation against Myc, western blotting was performed with anti-Myc or-HA antibody as indicated below each panel.

The region B of Nemp1 directly binds to Ran
Because the previous study has shown that region B faces the nucleoplasm and is required for the eye-reducing activity of Nemp1 in Xenopus embryos [10], we searched for region B-interacting proteins using the yeast two-hybrid system. We subcloned Mm_Bt as bait (Fig 3A) for screening a mouse embryonic cDNA library. As a result, we identified several candidate proteins, such as Ran and Ubc9 (not shown), which consistently interacted with Mm_Bt in yeast ( Fig 3A). To assess the interaction of Nemp1 with Ran in the Xenopus embryo and also in vitro, we performed co-IP and GST pulldown assays using HA-or Myc-tagged proteins. In parallel, we also tested whether Nemp1 binds to either RanGTP or RanGDP using the GTPand GDP-bound mutant forms RanQ69L and RanT24N, respectively. For co-IP analysis, Xenopus embryos were coinjected with mRNAs encoding for Mm_Bt-HA and Myc-Mm_Ran constructs. As shown in Fig 3B, Myc-Ran (WT) and Myc-RanQ69L (GTP-bound form mutant) but not Myc-RanT24N (GDP-bound form mutant) coimmunoprecipitated with Mm_Bt-HA, suggesting that Nemp1 specifically forms a complex with RanGTP through region B in the embryo. Similarly in GST pulldown analysis, HA-Ran and HA-Myc-RanQ69L (GTP form) but not HA-RanT24N (GDP form) from embryonic lysates were pulled down by recombinant GST-Mm_Bt that was purified from bacterial lysates (Fig 3C), indicating that Mm_Bt specifically interacts with RanGTP. To analyze direct interactions, we bacterially synthesized and purified recombinant Myc-RanQ69L by cleaving the GST moiety. Fig 3D shows that Myc-RanQ69L was pulled down by GST-Mm_Bt in comparison with GST alone, demonstrating the direct interaction between Mm_Bt and RanQ69L (GTP form).
We then examined the region of RanGTP that binds to region B. RanGTP is known to bind to both importin β and RanBP1. These interactions are disrupted in Mm_RanT42A, which has a point mutation in its effector domain [17]. In addition, the interaction between RanGTP and RanBP1 is abolished in Mm_RanΔC, which lacks the highly conserved acidic C-terminal tail of Ran (the DEDDDL sequence) [18]. Therefore, to examine whether these regions of Ran interact with Nemp1, we performed co-IP assays using these two mutant constructs. Fig 3E shows that the ΔC mutant but not the T42A mutant co-immunoprecipitates with Mm_Bt, suggesting that region B, similarly to importin β interacts with the effector domain of Ran. Therefore, we next tested whether region B competes for the interaction between importin β and RanQ69L. As expected, importin β was pulled down with GST-RanQ69L, and this interaction was reduced by the addition of recombinant Mm_Bt protein at a high concentration (Fig 3F). These data suggest that region B directly interacts with the same surface of RanGTP as importin β.
To determine a minimal Ran-binding region within the Bt region, we next performed co-IP experiments using HA-tagged deletion constructs of Mm_Bt, which were stabilized by fusing to EGFP. Interactions with Myc-Mm_Ran were detected with Bt and B but not with Ba, Bb, or Bt2 constructs (Fig 4A). Furthermore, the deletion of the BBS (ΔBBS) in Xl_Bt abolished its interaction with Myc-Xl_Ran (Fig 4B). These data suggest that a secondary or ternary structure of region B is required for Ran binding and that the BBS is required for Ran binding as well as BAF binding.
To assess the interaction between Nemp1 and Ran, we used full-length Nemp1 to perform co-IP and confocal microscopic analyses. As shown in Fig 5A, HA-tagged Nemp1 was coimmunoprecipitated with Myc-tagged Ran (WT), RanQ69L, and RanΔC, but not with RanT24N and RanT42A. This data is consistent with that using the Bt region (see Fig 3B and 3E). We next analyzed the localization of Nemp1 and Ran using co-immunostaining of tagged proteins in COS-7 cells by confocal microscopy. Myc-Ran alone was uniformly distributed in the nucleus (Fig 5B; upper panels). By contrast, when coexpressed with Nemp1-HA, Myc-Ran accumulated at the nuclear periphery and colocalized with Nemp1 at the NE (Fig 5B; lower panels).    Taken together, our data suggest that Nemp1 at the INM directly interacts with RanGTP in the nucleoplasm.

Phosphorylation of Nemp1
We found that there were shifted bands of Mm_Nemp1 in western blotting of embryo lysates (see Fig 5A and S3 Fig; indicated by asterisks in the panels for Nemp1-HA). We also found in the database from comprehensive analyses of phosphoproteins and phosphorylation sites [19][20][21][22][23], in which human Nemp1 was phosphorylated at multiple sites (Ser368, Ser378, Ser382, Ser424, Ser425). Some of these phosphorylated serines (Ser368, Ser378, Ser382) are located within region B and are evolutionarily conserved among vertebrates, and Ser-378 in the BAF binding sites is also conserved in the vertebrate paralog Nemp2 (see S4 Fig,S7A and S7B Fig). We thus hypothesized that phosphorylation at this site might modulate its interaction with RanGTP. We first examined whether Xl_Nemp1-HA is phosphorylated during early Xenopus development when cells are actively divided. Western blotting analysis of embryonic lysates containing the phosphatase inhibitor NaF revealed that shifted bands were strongly detected at the blastula stage (stage 9), and the intensity of these bands was reduced at neurula-to-tailbud stages (stages 14, 17, and 25), suggesting that the modification of Nemp1 occurs in proliferating cells ( Fig 6A). Furthermore, phosphatase treatments of immunoprecipitates or lysates abolished shifted bands of both Xl_Nemp1 ( Fig 6B) and Mm_Nemp1 (Fig 6C), indicating that modifications of Nemp1 are phosphorylation. To seek phosphorylation sites, we mutated Ser-366, Ser376, Ser380, Ser419, and Ser420 in Mm_Nemp1, which is relevant to the phosphorylated serines in human Nemp1, to Ala or Glu to produce the constructs 5SA or 5SE, respectively. Fig 6D and S3 Fig show that only the upper shifted band was abolished in 5SA and 5SE mutants, suggesting that all or some of these five serine residues function as either phosphorylation sites (probable Ser366-Pro367 and Ser380-Pro381 as Cyclin/Cdk sites) or recognition sites or both, and that other phosphorylation sites exist in Nemp1. Moreover, both the 5SA and 5SE mutations abolished the interaction with Ran, suggesting that all or some of these five serines of Mm_Nemp1 are involved in the interaction with Ran.

Nemp1 and Ran cooperate to function in early Xenopus development
Binding of Nemp1 to Ran prompted us to examine the function of their association in Xenopus embryos. Although the expression of ran is reported during the development of Xenopus tropicalis [24], we re-examined this by WISH with X. laevis embryos using a short chromogenic reaction to reduce staining intensity. Relatively strong ran expression was detected within the animal pole region at the four-cell stage, then in the anterior neural plate at the neurula stage, and within the head region including the otic vesicles, branchial arches, and the tail region at the tailbud stage (Fig 7A). These expression patterns were similar to those of nemp1 in Xenopus embryos [10], consistent with the interaction of Nemp1 with Ran.
To elucidate the cooperative role of Nemp1 and Ran in Xenopus eye development, we knocked down both Nemp1 and Ran activities by injecting antisense morpholino oligos (MOs), nemp1MOs [10] and ranMO. We designed ranMO to be complementary to the sequence encompassing the translation start sites of both X. laevis homoeologs of ran. We confirmed that ranMO specifically inhibited protein synthesis from Xl_Ran-Myc mRNA containing the MO target sequence but not from Myc-Xl_Ran mRNA without the target (S5 Fig). Injection of nemp1MOs or ranMO alone exhibited weak activity for inhibiting eye development, whereas the co-knockdown with nemp1 and ran elicited more severe eye defects than either individual knockdown alone (Fig 7B). This data suggests the functional interaction between Nemp1 and Ran.
Because nemp1 and ran are expressed in the anterior neural region and eye vesicles, in which cells highly proliferate, we next performed loss-of-and gain-of-function experiments for nemp1 and ran to examine their effects on cell densities and ratios of mitotic cells at the late ). Lysates were subjected to western blotting with anti-HA or β tubulin antibody (loading control). uc, uninjected control. B. In vitro alkaline phosphatase assay of Xl_Nemp1. Lysates were prepared at the late blastula stage (stage 9). Xl_Nemp1-HA was immunoprecipitated by anti-HA antibody and treated with (+) or without (-) calf intestinal alkaline phosphatase (CIAP). C. In vitro alkaline phosphatase assay of Mm_Nemp1. Lysates were prepared at the mid blastula stage (stages 8-8.5), and treated with (+) or without (-) λ protein phosphatase (λPP). D. Co-IP of Mm_Ran with phosphorylation site mutants of Mm_Nemp1. mRNA for Myc-tagged Mm_Ran was injected into Xenopus embryos with mRNA for HA-tagged Mm_Nemp1, its alanine mutant (5SA), or its glutamic acid mutant (5SE). Injected embryos were collected at the mid blastula stage and lysed with lysis buffer B. This data is the same as lanes 1, 2, 7, and 8 shown in S3 Fig Black arrowheads,  Nemp1 Is a RanGTP-Binding Protein gastrula to early neurula stages (stages12.5-13) (Fig 8 and S6 Fig). Injection of nemp1MOs (10 or 20 ng/embryo) but not stdMO or ranMO significantly reduced cell densities (Fig 8A and 8B) and was likely to increase nuclear size (compare panels a' and d' in Fig 8A). Co-injection of ranMO with nemp1MOs further reduced cell densities compared to single knockdowns (Fig 8B). The reduction by nemp1MOs was significantly rescued by low doses of nemp1 mRNA (Fig 8C) and the reduction by both MOs tended to be rescued by nemp1and ran mRNAs (Fig 8D). These data suggest again the functional synergism between Nemp1 and Ran. Similar to eye phenotypes [10], high doses of nemp1 mRNA as well as nemp1 MO reduced cell densities (Fig 8E), suggesting that a proper level of Nemp1 is required for normal functions. Supporting the reduction in cell density, single knockdown of nemp1 or co-knockdown of nemp1 and ran as well as overexpression of nemp1 by mRNA injection tended to decrease mitotic rates, which were determined using antiphosphohistone-H3 antibody (S6A and S6B Fig). Thus, it is likely that the reduction in cell density at the neurula stage is caused by the reduction in cell cycle progression by knockdown or overexpression of Nemp1 and Ran. Taken together, the data suggest that nemp1 and ran function cooperatively in proper cell cycle progression and eye development in Xenopus.

Interaction between Nemp and Ran is evolutionally conserved in Arabidopsis
Database search and phylogenetic analysis revealed that Nemp proteins exist not only in eumetazoans including Nematostella vectensis (sea anemone) but also in a plant, Arabidopsis  (S7A and S7C Fig). To examine whether or not the interaction of Nemp with Ran is conserved in eukaryotes, we performed co-IP assays for the three Arabidopsis Nemp proteins (At_Nemp-A, At_Nemp-B, and At_Nemp-C; S7 FigC) and At_Ran2. At_Ran2 is one of the four Ran proteins in Arabidopsis and is the most related to vertebrate orthologs (75% amino acid identity). As shown in Fig 9A, At_Ran2 coimmunoprecipitated significantly with the Cterminal region of At_Nemp-A, weakly with those of At_Nemp-B, and barely with At_Nemp-C. Notably, Fig 9B shows that Mm_Bt did not interact with At_Ran2, and conversely, At_Nemps (the Ct regions) did not interact with Mm_Ran, suggesting that Nemp and Ran coevolved to interact with each other. Based on the conserved interaction between Nemp and Ran in eukaryotes, the role of Nemp as a Ran-interacting protein might be related to basic cellular functions, such as the nuclear import system.

Discussion
A previous study has shown that the signal peptide and TMs are necessary and sufficient for Nemp1 to localize at the NE [10]. The localization of nuclear membrane proteins to the INM is postulated to be facilitated by the following two mechanisms: diffusion-retention and importin α/β-mediated transport. For example, in the former case, MAN1 moves diffusely to the INM from the ER and is retained by the binding of its N-terminal domain to lamins [16,25]. In the latter case, Heh2, a yeast homolog of vertebrate LEM2, contains a canonical NLS-like sequence within its N-terminal nucleoplasmic domain and is transported to the INM by importin α/β complexes [26]. We have shown that Nemp1 colocalizes with lamina through region A (Fig 1), and that the C-terminal region of Nemp1 exhibits nuclear localization activity (S1 Fig). Our data suggest that Nemp1 localizes at the INM via both diffusion-retention and NLS-dependent transport mechanisms.
We have found that Nemp1 specifically interacts with RanGTP via region B (Figs 3 and 4). RanGTP is known to interact with various factors that are associated with nuclear transport and spindle formation, including importins, exportins, RanBP1, and RanBP2. The following two types of RanGTP-binding motifs have been reported: importin β and RanBP1/2 motifs [27,28]. As the conserved motif of region B is different from these two types, Nemp1 might represent a new type of Ran-binding motif. Furthermore, Nemp1 is the first identified nuclear membrane protein reported to bind to RanGTP.
What is the function of Nemp1? We have shown that the coexpression of Mm_Nemp1 promotes the accumulation of Ran at the nuclear envelope (NE) in COS-7 cells (Fig 5B). Based on this observation, it is possible to speculate that the role of Nemp1 is to promote the accumulation of RanGTP at the nuclear periphery. This idea is supported by observations in C. elegans and Arabidopsis that endogenous Ran localizes at the NE during interphase [29,30]. The peripherally biased distribution of RanGTP in the nucleus might be important for efficient dissociation of cargo-importin complexes, which are imported through the NPC and could immediately encounter enriched levels of RanGTP. Nemp1 Is a RanGTP-Binding Protein example, lamin filaments are depolymerized upon phosphorylation of lamin B proteins [31,32], and the LBR (lamin B receptor) is dissociated from chromatin upon phosphorylation of LBR [33]. Similar to these NE proteins, our data suggest the possibility that phosphorylation of Nemp1 occurs in mitosis (Fig 6A,6B and 6C). In addition, serine residues at possible phosphorylation sites are involved in the interaction with Ran ( Fig 6D). These data suggest that the interaction of Nemp1 with Ran can be regulated by phosphorylation.
A previous study revealed that Nemp1 interacts with BAF through the BBS and that the BBS is required for the eye-reducing activity of overexpressed Nemp1 [10]. Recently, it was reported that in the absence of DNA, there is no interaction of BAF with BBS-containing proteins, such as CRX and MAN1 [34]. Nemp1 might also indirectly bind to BAF via DNA. In this study, we have shown that the deletion of BBS abolishes the interaction of the Bt region with Ran ( Fig 4B). Furthermore, both BAF [10] and Ran accumulate at the nuclear periphery via Nemp1 (Fig 5B). These data suggest that the eye-reducing activity of overexpressed Nemp1 is mediated through its interactions with BAF and Ran. Although Nemp1 is likely to play a role in promoting the accumulation of Ran at the nuclear periphery, the excessive accumulation of Ran as well as of BAF at the nuclear periphery by overexpression of Nemp1 likely perturbs their normal functions. Conversely, the co-knockdown of nemp1 and ran elicits reduction of cell density and eye defects more significantly than the individual knockdown for nemp1, supporting their functional interaction.
How is Nemp1 involved in eye development? While there seems to be no direct relationship between the function of Nemp1/Ran and eye development, we propose two possibilities. First, Nemp1 might regulate the nuclear transport of eye-specific transcription factors. In humans, the importin β family consists of 20 members [35], of which importin 13 is known to function Nemp1 Is a RanGTP-Binding Protein in the import of several transcriptional factors, such as Pax6 and Crx, which are important for eye development [36]. Knockdown of nemp1 reduced the expression of early eye marker genes, rax and pax6 [10]. Therefore, it is conceivable that Nemp1 controls importin13-mediated transport. The second possibility is that Nemp1 is associated with cell proliferation, which is important for eye development. This possibility is based on the fact that knockdown of nemp1 caused the reduction of cell densities (Fig 8A and 8B), and that expression patterns of nemp1 and ran (Fig 7A) are similar to those of cell cycle regulators, cyclin D1, cyclin E, and cdk4 [37]. These data suggest that increased levels of Nemp1 and Ran are necessary for maintaining an actively proliferative state, in which nuclear uptake is more likely to be active than in non-proliferating cells [38].
Our results show that the functional interaction between Nemp1 and Ran is required for proper eye development in Xenopus and that their physical interaction is conserved, even in eyeless organisms, such as C. elegans [39] and plants (Fig 9). Furthermore, nemp1 and ran are co-expressed in tissues other than the eyes in Xenopus embryos. Therefore, the interaction between Nemp1 and Ran at the nuclear periphery likely plays important general roles in regulating the nuclear transport of proteins during cellular proliferation and differentiation. We conclude from the results that the inner nuclear membrane protein Nemp1 represents a new type of RanGTP-binding protein and that this interaction might control the nuclear transport of molecules in eukaryotes. were transfected with the HA-tagged mouse Bt construct (Mm_Bt-HA) or GST-mRFP-Mm_Bt-HA, fixed, and stained with anti-HA antibody (red) and SYTOX Green for DNA. GST-mRFP-Mm_Bt-HA exhibited cytoplasmic localization, but also nuclear localization in some cases. Scale bars, 5 μm. We have previously shown that Myc-tagged Xl_Ct and KR constructs but not Xl_Bt (see Fig 1A) is localized to the nucleus, suggesting NLS function of the KR sequence [10]. Therefore, we systematically examined the nuclear localization activity of KR, using GST-mRFP-HA, which alone cannot be transported into the nucleus because of its large molecular mass (122 kDa as a dimer under the native conditions). GST-mRFP-HA was fused with short peptides related to the KR sequence, and the fusion constructs were analyzed for their ability to localize to the nucleus. As shown in S1 Fig A, GST-mRFP-HA alone was localized in the cytoplasm, whereas the SV40NLS fusion, which served as a positive control, exhibited nuclear localization. Similarly, the KR fusion proteins, KRa and KRb, which were derived from the Xenopus homoeologs of Nemp1, Nemp1a and Nemp1b, exhibited nuclear localization, whereas the KRa(ΔR) fusion did not, indicating that both KRa and KRb sequences function as NLSs (S1 Fig B) and that the first Arg residue of the RKIKXKRAK (X is R or L) motif is required for this activity. We also analyzed a short sequence from Mm_Nemp1, whose position corresponds to that of KR in Xl_Nemp1, named KRm, though KRm does not contain a canonical NLS sequence. As expected, KRm did not elicit NLS function (S1 Fig A). However, although Mm_Bt does not have a canonical NLS sequence either, HA-tagged Mm_Bt exhibited nuclear localization (S1 Fig B; upper panels). Therefore, we analyzed NLS function of Mm_Bt suing GST-mRFP-HA, and observed that GST-mRFP-Mm_Bt-HA exhibited weak nuclear localization (middle panels, compare with GST-mRFP-HA negative control), and, in a few cases, it was exclusively localized to the nucleus (lower panels), suggesting that Mm_Bt could have NLS function. Thus, we conclude that the C terminal region of Nemp1 proteins (that is, KR in Xenopus and Bt in mouse) exhibits NLS function. (TIF) S2 Fig. Diagram of deletion constructs of Xl_Nemp1. These deletion constructs were used in Fig 2B. Blue, signal peptides (SP); magenta, transmembrane domains (TMs); green, KR sequence; yellow, region B. (TIF) S3 Fig. Co-IP of Nemp1 with Ran using Xenopus embryos. This is the original data for Figs 5A and 6D. mRNA for HA-tagged Mm_Nemp1 or its mutants (5SA, 5SE) was coinjected into Xenopus embryos with mRNA for a Myc-tagged construct of Mm_Ran (WT) or its mutants (T24N, Q69L, T42A, ΔC). Injected embryos were collected at the mid blastula stage (stages 8-8.5) and lysed with lysis buffer B for Co-IP. Black arrowheads, modified forms of Nem-p1-HA. Note that WT Nemp1 has two major modified bands (lane 2) and co-expression with Ran(T24N; a GDP form) enhanced these modifications (lane 3). Also note that the upper modified band disappeared in 5SA and 5SE constructs (lanes 7, 8), suggesting that all or some of these five serine residues are involved in modification (phosphorylation) by functioning as either phosphorylation sites or recognition sites or both, and that there are other phosphorylation sites besides there five serine residues. (TIF)  Fig 8. (A) Reduction of the ratio of mitotic cells by co-knockdown of Nemp1 and Ran. Similar tendencies were obtained from the three experiments and statistically significant differences was observed in one of them. Nuclei stained with DAPI or immunostained for phospho histone H3 were counted in FITC-positive areas. (B) Reduction of the ratio of mitotic cells by overexpression of Nemp1. DAPI-stained nuclei were counted in EGFP-HA positive areas. Ã , P<0.05; ÃÃÃ , P <0.005; error bars, standard deviation. (TIF)