Retinoblastoma-binding Protein 4-regulated Classical Nuclear Transport Is Involved in Cellular Senescence*

Nucleocytoplasmic trafficking is a fundamental cellular process in eukaryotic cells. Here, we demonstrated that retinoblastoma-binding protein 4 (RBBP4) functions as a novel regulatory factor to increase the efficiency of importin α/β-mediated nuclear import. RBBP4 accelerates the release of importin β1 from importin α via competitive binding to the importin β-binding domain of importin α in the presence of RanGTP. Therefore, it facilitates importin α/β-mediated nuclear import. We showed that the importin α/β pathway is down-regulated in replicative senescent cells, concomitant with a decrease in RBBP4 level. Knockdown of RBBP4 caused both suppression of nuclear transport and induction of cellular senescence. This is the first report to identify a factor that competes with importin β1 to bind to importin α, and it demonstrates that the loss of this factor can trigger cellular senescence.

Nucleocytoplasmic trafficking, which involves transport of cargo through the nuclear pore complexes embedded in the nuclear membrane, is a fundamental cellular process in eukaryotic cells (1). Importin protein family members are known to play essential roles in mediating nuclear import. Importin ␣ functions as an adaptor molecule to connect a classical nuclear localization signal (cNLS) 3 -bearing cargo protein with importin ␤1, a carrier molecule, in the cytoplasm. The trimeric transport complex of cNLS-cargo⅐importin ␣⅐␤1 is then translocated into the nucleus through the nuclear pore complex (2)(3)(4). In the nucleus, the binding of the GTPbound form of Ran (RanGTP) to importin ␤1 triggers the dissociation of the complex (5). Then, importin ␣ is ferried to the cytoplasm in an export complex containing cellular apoptosis susceptibility gene product (CAS, also known as CSE1L) and RanGTP. The importin ␤1⅐RanGTP complex also returns to the cytoplasm, and the importins are reused for the next round of transport.
Importin ␣ is composed of an importin ␤ binding (IBB) domain in the flexible N-terminal region, a core domain composed of 10 tandem armadillo repeats, which contain the major NLS-binding site at ARM2-4 and the minor NLS-binding site at ARM6 -8 (3,6,7), and a C-terminal domain, which is the binding region for CAS (8,9). The IBB domain is known to regulate cNLS binding via an autoinhibitory mechanism (6); however, not much is known about the proteins that interact with the IBB domain, except for importin ␤1.
Cellular senescence was first described as the irreversible growth arrest of normal primary cells after prolonged proliferation in cell culture (10). Recently, it has been shown that retinoblastoma-binding protein 4 (RBBP4, also referred to as RbAp48) is associated with cellular senescence (11,12). The level of RBBP4 protein is decreased in aged primary human fibroblast cells, and the knockdown of RBBP4 and RBBP7, a molecule closely related to RBBP4, causes aging-associated chromatin defects (11). Furthermore, reduction of RBBP4 has been observed in the aging human hippocampus and is correlated with memory loss. Conversely, the regulation of RBBP4 ameliorated age-related hippocampus-based memory loss (12). These results suggest that RBBP4 is an aging-associated molecule.
RBBP4 is a WD-repeat protein that functions in the nucleus. It binds directly to the histone H3⅐H4 complex (13) and serves as the histone binding domain of several different complexes, such as the HDAC complex (14), the NuRD complex (15), the CAF-1 complex (16), and the PRC2 (17), all of which regulate chromatin structure and/or function. As mentioned above, although a reduction in RBBP4 level can induce chromatin defects associated with aging (11), it remains unclear how RBBP4 contributes to cellular senescence.
In this study, we show that RBBP4, which was identified as an importin ␣-binding protein using mass spectrometry in our comprehensive proteomics study (18), accelerates the release of importin ␤1 from importin ␣ through its interaction with the IBB domain of importin ␣. Consequently, RBBP4 maintains the efficiency of importin ␣/␤-mediated nuclear transport. Furthermore, knockdown of RBBP4 impairs nuclear import effi-ciency and induces cellular senescence, implying that the loss of RBBP4 is linked to cellular senescence via the down-regulation of nuclear transport efficiency. These findings provide novel evidence for a close connection between the regulation of nuclear transport and cellular senescence.

Experimental Procedures
Importin ␣ Gene Nomenclature-The importin ␣ gene nomenclature is complicated due to inconsistencies in the usage of the terms "importin" and "karyopherin." Furthermore, differences in the number of importin ␣ genes between human and mouse have led to homologs with different names. Therefore, we uniformly used the human nomenclature for the three importin ␣ proteins tested in this study as follows: importin ␣1 (KPNA2), importin ␣3 (KPNA4), and importin ␣5 (KPNA1).
Cell Culture and Transfection-Human diploid fibroblasts (TIG-1 cells) were obtained from the Human Science Research Resource Bank (23) or Japanese Collection of Research Bioresources Cell Bank. TIG-1 cells, HeLa cells, and HEK-293F cells were cultured at 37°C in 10% CO 2 , in Dulbecco's modified Eagle's medium (DMEM) (Sigma) supplemented with 10% fetal bovine serum (FBS). TIG-1 cells were classified as young when the population doubling levels (PDL) were less than 30 and senescent when PDL was Ͼ65. The day before transfection, HeLa cells, HEK-293F cells, and TIG-1 cells were planted on 12-well plates (for immunofluorescence, Thermo Scientific, Waltham, MA) with coverslips (MATSUNAMI, Osaka, Japan) or in 6-cm dishes (for collecting cell extracts, Thermo Scientific) to be 70% confluent at the time of transfection. Plasmids were transfected into HeLa cells and HEK-293F cells using Lipofectamine 2000 or Lipofectamine 3000 (Life Technologies, Inc.) according to the manufacturer's instructions. The siRNA sequences for RBBP4 were 5Ј-CUUGUAUCAUC-GCAACAAA-3Ј and 5Ј-AAAUCAAGAUCAACCAUGA-3Ј (Ambion, Austin, TX). They were transfected into HeLa cells and TIG-1 cells using Lipofectamine RNAiMAX (Life Technologies, Inc.), according to the manufacturer's instructions.
Recombinant Proteins-The wild type RBBP4 coding sequence and the mouse-importin ␣1 deletion mutants (encoding either 61-529 amino acids (aa) (⌬IBB domain) or 1-500 aa (⌬C terminus)) were constructed using a KOD-plus-Mutagenesis Kit (Toyobo Co.). The mutants were cloned into a pGEX6P-2 vector, which is an expression vector for the production of glutathione S-transferase (GST)-fused protein.
RBBP4 was also inserted into the pGEX6P-2/hGFP plasmid (24) to produce the GFP-RBBP4 fusion protein. All mutants of the IBB domain were also constructed using a KOD-plus-mutagenesis kit. Almost all of the residues are identical in the human and mouse IBB domain of importin ␣1. The three exceptions are residues 9, 15, and 45. We mutated Glu-45 in mouse IBB domain to lysine as a substitute for Asp-45 in the human IBB domain. Protein expression and purification were performed as described previously (2,24). GST was cleaved using PreScission protease (GE Healthcare) in cleavage buffer 1 (50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA(3Na), 1 mM dithiothreitol (DTT), containing 1 g/ml each of aprotinin, leupeptin, and pepstatin) for pGEX6P vector, and thrombin (Sigma) in cleavage buffer 2 (50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 2.5 mM CaCl 2 ) for pGEX2T vector. All of the following recombinant proteins were purified in the same manner: all importin ␣ constructs, importin ␤1, GST-IBB-GFP (GST-GFP fused the IBB domain of mouse-importin ␣1, gifted from Y. Ogawa), GST-SV40T-NLS-GFP, GST-GFP, GFP and GST. The GDP-bound Ran and the GTP-bound Ran Q69L mutant were purified as described previously (25).
GST Pulldown Assay-Plasmids encoding 3ϫHA-tagged wild type or mutant RBBP4 were transfected into HEK-293F cells, and whole cell lysates were prepared 48 h after transfection. Lysates were pre-cleared with glutathione-Sepharose 4B beads (GSH beads, GE Healthcare) at 4°C for 1 h, and each pre-cleared lysate was incubated with the GST fusion proteins (50 pmol) and GSH beads at 4°C for 1 h. The beads were washed extensively with 0.1% Triton X-100 in PBS containing 1 mM DTT and 1 g/ml each of aprotinin, leupeptin, and pepstatin; and bound proteins were eluted with sample buffer for SDS-PAGE. Bacterially produced recombinant proteins (50 pmol each) were also incubated with each GST fusion protein (50 pmol) immobilized on GSH beads at 4°C for 1 h. In the experiment shown in the left panel in Fig. 7B, GST-IBB-GFP and importin ␤1 were mixed first and preincubated for 30 min, and RBBP4 was added later. In contrast, in the right panel in Fig. 7B, the three proteins (GST-IBB-GFP, importin ␤1, and RBBP4) were mixed and incubated simultaneously. Washing and elution of the bound proteins were performed as described above.
Senescence-associated ␤-Galactosidase (SA ␤-Gal) Staining-The senescence detection kit (BioVision, Mountain View, CA) was used to detect senescent cells according to the manufacturer's instructions.
Immunofluorescence and Microscopy-Cells were fixed with 3.7% formaldehyde in PBS for 15 min. After two washes in PBS, cells were treated with 0.1% Triton X-100 in PBS for 5 min. After washing with PBS twice, the cells were blocked in PBS containing 5% skim milk for 30 min and incubated in appropriately diluted (in PBS containing 5% skim milk) primary anti-bodies for 2 h at room temperature or overnight at 4°C. The following primary antibodies were used: RBBP4 (RbAp48; ab488; mouse mAb; 2.5 g/ml, Abcam, Cambridge, UK), and Ki-67 (ab833-500; rabbit polyclonal antibody; 25 g/ml, Abcam). After subsequent TBS-T washes, cells were incubated with the Alexa-Fluor-488-or Alexa-Fluor-568/594-conjugated secondary antibodies (Molecular Probes, Eugene, OR). Nuclei were counterstained with DAPI (1:10,000 in PBS, Dojindo Laboratories, Kumamoto, Japan) for 10 min. The samples were examined using a Zeiss Axiophot fluorescent microscope (Carl Zeiss, Göttingen, Germany) and a Nikon Eclipse 80i microscope (Nikon, Tokyo, Japan).
In Vitro Transport Assay and Time-lapse Analysis-HeLa cells were planted on 8-well multitest slides (MP Biomedicals, Inc., Santa Ana, CA) for 24 h prior to each assay. After washing with ice-cold transport buffer (TB: 20 mM HEPES/ NaOH (pH 7.3), 110 mM potassium acetate, 2 mM magnesium acetate, 5 mM sodium acetate, 0.5 mM EGTA/NaOH (pH 7.3), and 2 mM DTT, containing 1 g/ml each of aprotinin, leupeptin, and pepstatin), HeLa cells were incubated in ice-cold TB containing 20 g/ml digitonin for 5 min. Cells were then washed and incubated in fresh ice-cold TB for 10 min. The transport mixture (4 M cargo/substrate with 0.5 M importin ␣, 0.5 M importin ␤1, 4 M GDP-bound Ran, 500 M GTP, and an ATP regeneration system) was mixed in TB containing 2% BSA. The ATP regeneration system (0.33 M ATP, 1.5 M phosphocreatine, and 6 units/ml creatine phosphokinase) was adjusted with TB. To generate cytosol extracts for the transport reactions, 4 M cargo/substrate was mixed with 4 l of cytosol from cell extracts (10 mg/ml). The transport reactions were conducted in TB containing 2% BSA. The cytosol extracts were prepared from HeLa cells as described previously (27). After applying the transport mixture, semi-intact cells were incubated at 30°C for 30 min unless noted otherwise. Cells were then washed with TB and fixed with 3.7% formaldehyde in PBS. For the time-lapse analysis, HeLa cells were cultured on a glass bottom dish (PLL-coat, MATSUNAMI) and transfected with siRNAs 48 h before the analysis. Before the cells were permeabilized with digitonin, Hoechst 33342 (1 g/ml, Molecular Probes) was added to the dishes, followed by incubation for 15 min at room temperature. The cells were permeabilized with digitonin as described above. The dish was set on an inverted confocal microscope (Leica TCS SP8, Leica Microsystems GmbH, Germany, Wetzlar), and the import reaction was performed at 30°C. Time-lapse images were initiated at the time of the addition of the transport mixtures containing importin ␣1 and ␤1 with GST-SV40TNLS-GFP as a cargo. Signal intensities of 10 different nuclei were quantified in three independent experiments and statistically analyzed using Bonferroni's post hoc test after a two-way analysis of variance. The fluorescent scans were started in time-lapse mode, and the focal plane was adjusted to the nuclei.
Microinjection-Young and senescent TIG-1 cells were plated on coverslips (MATSUNAMI) in a 35-mm dish (Corning, Acton, MA), and GST-SV40T-NLS-GFP (2 mg/ml) and Alexa-Fluor-594-conjugated antibody (2 mg/ml, Molecular Probes, as an injection marker) were microinjected into the cytoplasm with a final concentration of 1 mg/ml within 1 min. After incubating for the indicated time at room temperature, the cells were fixed with 3.7% formaldehyde. The nuclear (Fn) and cytoplasmic (Fc) fluorescences were quantified for at least 10 different cells to enable the nuclear to cytoplasmic fluorescence ratio (Fn/c) to be determined. Statistical analysis was performed using the Student's t test. After the fluorescent examination, SA ␤-gal staining was performed and examined by an OLYMPUS CKX41 inverted microscope (OLYMPUS, Tokyo, Japan).

Results
Importin ␣1 Binds to RBBP4 but Does Not Mediate Its Nuclear Transport-By proteomic analysis, we previously identified RBBP4 as a potential importin ␣1-binding protein (18). In addition, another study also shows interaction between karyopherin ␣2 (importin ␣1) and RBBP4 (29). To validate the binding between importin ␣ and RBBP4, we performed a GST pulldown assay against three major importin ␣ subtypes, ␣1, ␣3, and ␣5. Whole cell lysates prepared from HEK-293F cells expressing HA-tagged RBBP4 (HA-RBBP4) or bacterially expressed recombinant RBBP4 protein were incubated with GST-importin ␣. As shown in Fig. 1, A and B, all three subtypes of importin ␣ bound directly to RBBP4. Additionally, the immunoprecipitation assay results clarified the interaction of the endogenous proteins in HeLa cells (Fig. 1C).
Subsequently, we used an in vitro transport assay with digitonin-permeabilized cells to examine whether importin ␣ is involved in the nuclear transport of RBBP4. We observed that cytosolic fractions of HeLa cells (27) mediated the nuclear FIGURE 1. Importin ␣ binds to RBBP4 but does not mediate its nuclear transport. A and B, GST pulldown assays using three GST-fused importin ␣ subtypes (␣1, ␣3, and ␣5) with HEK-293F cell lysates expressing HA-tagged RBBP4 (A) or with recombinant RBBP4 (B). Bound proteins were analyzed by Western blotting using the antibodies indicated. C, HeLa cell lysates were immunoprecipitated (IP) with RBBP4 or importin ␣1 antibody, followed by Western blotting (WB) using importin ␣1 or RBBP4 antibody, respectively. D and E, in vitro transport assays. Nuclear import of either GFP-RBBP4 or GST-SV40T-NLS-GFP was examined in digitonin-permeabilized HeLa cells, in HeLa cell cytosols with or without Q69LRanGTP (D), or in a reconstitution system that included importin ␣1, importin ␤1, RanGDP, and an ATP-regenerating system (E). After 30 min at 30°C, GFP-tagged proteins were visualized by fluorescence microscopy.
import of GFP-RBBP4, indicating that RBBP4 is transported into the nucleus by some cytosolic factor(s). The addition of a GTPase-deficient mutant of Ran, Q69LRanGTP, inhibited the migration of RBBP4 to the nucleus, indicating that RBBP4 is imported into the nucleus in a Ran-dependent manner (Fig.  1D). Therefore, we used recombinant proteins to test whether RBBP4 is transported into the nucleus in an importin ␣⅐␤-dependent manner. As shown in Fig. 1E, GFP-RBBP4 did not enter the nucleus in the presence of importin ␣1⅐␤1, whereas SV40 large T antigen NLS fused with GST and GFP (GST-SV40T-NLS-GFP) was efficiently transported into the nucleus. This is consistent with the fact that the cNLS mapper (30) and PSORTII (31) cNLS prediction algorithms did not detect any potential cNLS in the RBBP4 protein, indicating that there are no obvious cNLSs in the coding sequence of RBBP4. These results revealed that importin ␣1 interacts with RBBP4 but does not mediate its nuclear import.
RBBP4 Has a Novel Binding Domain for Importin ␣1-Because our results indicated that RBBP4 is unlikely to have a cNLS, we next investigated how RBBP4 binds to importin ␣1. We constructed 14 deletion mutants of HA-RBBP4 ( Fig. 2A) and conducted a pulldown assay using GST-importin ␣1. As  DECEMBER Fig. 2, B and C, most of these deletion mutants failed to interact with importin ␣1. In addition, even a deletion of 20 aa from either the N or C terminus of RBBP4 caused a significant decrease in the ability to bind importin ␣1. Thus, we could not identify any obvious importin ␣1-binding domains in RBBP4 by a conventional region deletion procedure.

RBBP4 Regulates Classical Nuclear Transport
However, we noticed an unusual cluster of basic amino acids in RBBP4 (301-310, aa WDLRNLKLKL), which is also conserved in mouse, Xenopus, and Drosophila. We constructed a deletion mutant lacking this region (⌬305-310, Fig. 2D), and we found that this region was critical for the interaction of RBBP4 with importin ␣1 (Fig. 2E). We investigated the role of the basic amino acids in the deleted sequence (NLKLKL) by substituting them with alanine or glutamic acid (K307A/K309A or K307E/K309E, Fig. 2D). However, both of these mutants were able to bind to importin ␣1. We next focused on the leucine residues. It has been reported that viral protein R of human immunodeficiency virus type 1 binds to importin ␣ and that the substitution of the leucine residues at positions 20, 22, 23, and 26 of viral protein R disrupts this interaction (32). Therefore, we created RBBP4 mutants L306A/L308A/L310A and L306E/ L308E/L310E. As shown in Fig. 2F, both of these leucine mutants show a marked decrease in binding to importin ␣1. These data indicate that these three leucine residues of RBBP4 (Leu-306/-308/-310) are crucial for its interaction with importin ␣1. Taken together, these results suggest that RBBP4 uses a novel mode of binding to importin ␣1, distinct from that of typical cNLS-containing cargoes.
IBB Domain of Importin ␣ Binds to RBBP4 -Karyophilic proteins containing cNLS(s) are recognized by importin ␣ through its major and/or minor NLS-binding sites in its central region.
To test whether RBBP4 also binds to these NLS-binding sites, we first used the ED mutant of importin ␣1 (33), which contains mutations in both the major and minor NLS-binding sites (Fig.  3A). Although SV40T-NLS-GFP failed to bind to the ED mutant, RBBP4 bound to the mutant as efficiently as it did to wild type importin ␣1 (Fig. 3B). To further clarify this interaction, we used the bimax peptides (bimax1 and bimax2), which inhibit cNLS-containing cargo binding via high affinity interactions with both NLS-binding sites (22). Although the binding of GST-SV40T-NLS-GFP to importin ␣1 was inhibited by the bimax peptides, there was no effect on the binding between RBBP4 and importin ␣1 (Fig. 3C). Collectively, these data indicate that importin ␣1 does not bind to RBBP4 through the major or the minor NLS-binding site.
To identify the RBBP4-binding site in importin ␣1, we next focused on the IBB domain and the C-terminal domain. We used the following two deletion mutants: the ⌬IBB mutant lacking the N-terminal IBB domain (1-60 aa) and the ⌬C mutant lacking the C-terminal region (501-529 aa) (Fig. 4A). In a pulldown assay, the ⌬IBB mutant showed much weaker binding to HA-RBBP4 expressed in cell lysates and to recombinant RBBP4, compared with the wild type or the ⌬C mutant of importin ␣1 (Fig. 4, B and C). Next, a recombinant protein containing the IBB domain of importin ␣1 (1-65 aa) fused to GST and GFP (GST-IBB-GFP) was used for the binding assay. As shown in Fig. 4D, the IBB domain of importin ␣1 binds to RBBP4. These results indicate that the NLKLKL sequence in RBBP4 forms a novel binding site for the IBB domain of importin ␣1 and that the leucine residues are crucial for binding.
Because we found that RBBP4, like importin ␤1, binds to the IBB domain of importin ␣1, we next examined whether RBBP4 can form a ternary complex with cNLS-cargo protein and importin ␣1. To test this idea, SV40T-NLS-GFP was used as a cNLS substrate and was incubated with GST-importin ␣1 in the presence of either RBBP4 or importin ␤1. As shown in Fig. 4E, we found that the ternary complex, SV40T-NLS-GFP⅐importin ␣1⅐RBBP4 was formed. Furthermore, we found that the amount of SV40T-NLS-GFP bound to importin ␣1 was significantly increased in the presence of RBBP4, just as it is in the presence of importin ␤1. This implies that the autoinhibition by the IBB domain of importin ␣1 to its own NLS-binding site is released by RBBP4 binding, just as it is by importin ␤1 binding (3,6,34).
Biochemical and Structural Analysis of the RBBP4-IBB Domain Interaction-Next, we examined whether RBBP4 and importin ␤1 compete with each other to bind to the IBB domain. Initially, we clarified RBBP4 did not directly interact with importin ␤1 (Fig. 5A). Then GST-IBB-GFP immobilized on a glutathione-Sepharose 4B slurry (GSH beads) was incubated with increasing amounts of importin ␤1 or GFP (as a negative control) in the presence of a constant amount of RBBP4 (Fig. 5B). As a result, RBBP4 binding was inhibited by FIGURE 3. Importin ␣1 binds to RBBP4 through a site other than the major or minor NLS-binding sites. A, schematic representation of importin ␣1 wild type (WT) and ED mutant. B, binding assay between GST-importin ␣1 (Imp␣1) and RBBP4. Either WT or the ED mutant (D192K, E396R) of GST-importin ␣1 was incubated with recombinant RBBP4 or SV40T-NLS-GFP. The SV40T-NLS-GFP was used as a positive control for binding with the WT protein and detected by an anti-GFP antibody. C, examination of binding inhibition by cNLS peptides, bimax1 and bimax2. GST-RBBP4, GST-SV40T-NLS-GFP, or GST-GFP immobilized on GSH beads was incubated with importin ␣1 in the presence of mCherry-bimax1 or bimax2 (100 pmol each). The unconjugated mCherry protein was used as a negative control.
importin ␤1 but not by GFP. These results showed that importin ␤1 specifically competes with RBBP4 for binding to the IBB domain, implying that the two proteins bind to the same region of the IBB domain.
We next modeled the binding mode of the RBBP4⅐IBB domain complex using an in silico method combined with a point mutation analysis. The structure of the importin ␤1⅐IBB domain complex (PDB code 1qgk) (35) shows that 8 of the 10 conserved residues in the IBB domain are involved in importin ␤1 binding (Fig. 5C). We hypothesized that these residues may also be involved in the binding of RBBP4. To identify other potential key interacting residues, we also performed a helical wheel analysis (Fig. 5D) and selected the following set of residues for site-directed mutagenesis: 1) a conserved hydrophobic residue, Leu-38, and neighboring non-conserved hydrophobic residues, Met-27 and Val-34, which we expect to interact with Leu-306/-308/-310; 2) conserved (Arg-31 and Lys-49) and nonconserved (Lys-42) basic residues and a non-conserved acidic residue, Asp-45 (Glu-45 in mouse importin ␣1), which are all close to Leu-38 on the helix; and 3) conserved residues in a short N-terminal segment, Arg-13 and Lys-18. The results of pulldown assays with the mutant proteins showed that muta-tion of acidic residues tends to decrease RBBP4 binding to the IBB domain, whereas mutation to basic residues tends to increase the binding (Fig. 6A). The results suggest that all of the mutated residues are involved in RBBP4⅐IBB domain binding or are located close to the binding site and that both hydrophobic and negatively charged regions in RBBP4 bind to the IBB domain (Fig. 6B).
To examine whether the N or C terminus of RBBP4 is involved in the IBB domain binding (Fig. 2C), we analyzed the structures of the IBB domain in the importin ␤1⅐IBB domain complex (PDB code 1qgk) (35) and RBBP4 in the RBBP4⅐MTA1 complex (PDB code 4pby) (36). We confirmed that the IBB domain can span between Leu-306/-308/-310 and either terminus (Fig. 2, C and F) and that the regions from Leu-306/-308/-310 to both termini are hydrophobic and negatively charged (Fig. 6B), indicating that either terminus may be involved in IBB domain binding. Based on these observations, we propose two structural models of the RBBP4⅐IBB domain complex (Fig. 6C).
RBBP4 Accelerates the Release of Importin ␤1 from the Transport Complex in the Nucleus via a RanGTP-dependent Mechanism-Because we found that RBBP4 competitively inhibits the binding of importin ␤1 to importin ␣1, we next examined whether RBBP4 can influence the nuclear transport of a cNLS-containing cargo protein (GST-SV40T-NLS-GFP) by inhibiting the formation of the importin ␣1⅐␤1 complex, using an in vitro transport assay. As shown in Fig. 7A, the nuclear migration of the cargo protein was inhibited by the addition of RBBP4 in a dose-dependent manner. These results suggest that RBBP4 may affect the regulation of importin ␣1⅐␤1-dependent nuclear transport. However, because RBBP4 is mainly localized in the nucleus in living cells, it is unlikely that RBBP4 competes with importin ␤1 for binding to importin ␣ in the cytoplasm.
Therefore, we speculated that the interaction of RBBP4 with the IBB domain of importin ␣1 would occur inside the nucleus, resulting in the dissociation of importin ␤1 from the cNLS-cargo⅐importin ␣1⅐␤1 trimeric complex. To test this, GST-IBB-GFP immobilized on GSH beads was first incubated with importin ␤1, and then increasing amounts of RBBP4 were added. Interestingly, we found that RBBP4 could not dissociate importin ␤1 from the IBB domain⅐importin ␤1 pre-formed complex (Fig. 7B, left panel), whereas RBBP4 competes with importin ␤1 when the proteins were mixed simultaneously (Fig.  7B, right panel).
Another important nuclear factor that promotes dissociation of the IBB domain and importin ␤1 is RanGTP. Therefore, we examined the cooperative effects of RanGTP on these protein interactions. As shown in Fig. 7C, the release of importin ␤1 from the IBB domain⅐importin ␤1 complex was markedly enhanced by the addition of both RBBP4 and Q69LRanGTP as compared with the single addition of Q69LRanGTP. This result indicated that RBBP4 could stimulate the dissociation of importin ␤1 from the trimeric nuclear complex of cNLS-cargo⅐importin ␣1⅐␤1, in cooperation with RanGTP, implying that RBBP4 functions in the nucleus to facilitate the release of importin ␣ from importin ␤1.
Knockdown of RBBP4 Impairs Nuclear Transport Efficiency-To understand the physiological function of RBBP4 in the classical nuclear transport machinery, the RBBP4 gene was knocked down in HeLa cells using siRNAs, and the effect of this intervention on the nuclear import of GST-SV40T-NLS-GFP was measured using time-lapse confocal microscopy. Two different siRNAs for RBBP4 (si-RBBP4-1 and si-RBBP4-2) were used to transfect HeLa cells, and after 48 h, the cells were permeabilized with digitonin for an in vitro transport assay. The nuclear uptake efficiency of the cargo protein in both the si-RBBP4-1-and si-RBBP4-2-treated cells was significantly lower than in the untreated cells (Fig. 7, D-F). These results strongly suggest that RBBP4 functions in the nucleus to release importin ␤1 from the trimeric transport complex in the presence of RanGTP, thereby affecting the nuclear transport efficiency of cNLS-containing cargo proteins.
Knockdown of RBBP4 Induces Cellular Senescence by Impairing the Nuclear Transport-RBBP4 was originally identified as an aging-associated molecule (11,12). This is consistent with our finding that the protein level or mRNA level of RBBP4 decreased in a passage-dependent manner in TIG-1 cells (Fig.  8A). Interestingly, it has been reported that cellular senescence is associated with down-regulation of active nuclear transport (37,38) and decreased protein levels of several nuclear transport-related factors (23,39); however, little is known about how the classical nuclear transport machinery is involved in cellular senescence.
To investigate a potential role for RBBP4 in senescence, we first used cytoplasmic microinjection to compare the nuclear import efficiency of GST-SV40T-NLS-GFP in young cells (PDL under 30) with that of senescent cells (PDL over 65). Senescent cells were identified by the expression of SA ␤-gal, a type of senescence marker. The results showed that the transport efficiency of cNLS-containing cargo was significantly decreased in senescent cells relative to young cells, even at the early time point (Fig. 8, B and C), suggesting that a decrease in the RBBP4 level may induce cellular senescence through the suppression of importin ␣⅐␤-dependent nuclear transport.
We next investigated whether the knockdown of RBBP4 induces cellular senescence in normal fibroblast TIG-1 cells. As shown in Fig. 8D, both si-RBBP4-1 and si-RBBP4-2 treatments led to a significant increase in the ratio of SA ␤-gal-positive cells. In addition, the Ki-67 signal, which is known as a cellular proliferation marker, was down-regulated (Fig. 8E), and the senescent marker proteins p21, p53, and p16 were up-regulated in the si-RBBP4-treated cells (Fig. 8F). Additionally, because RBBP4 is known to be a member of some chromatin-modifying complexes, we examined the effect of RBBP4 knockdown on histone modifications known to change with cellular senescence. However, as shown in Fig. 8G, the alterations of histone modifications triggered by the depletion of RBBP4 were not consistent with those observed during cellular senescence of TIG-1 cells or previous reports using different cell lines (40 -45). From these findings, we concluded that RBBP4 is a key factor for the maintenance of the classical nuclear transport efficiency, and its decrease triggers cellular senescence in TIG-1 cells. Taken together, we propose that the down-regulation of classical nuclear transport, in which RBBP4 plays an important role, is closely related to cellular senescence.

Discussion
In this study, we demonstrate that RBBP4 is a novel IBB domain-binding protein that regulates the classical nuclear transport machinery and keeps cells young. Inside the nucleus, RBBP4 works in conjunction with nuclear RanGTP to accelerate the dissociation of importin ␤1 from importin ␣1 by binding FIGURE 5. Biochemical and structural analysis of RBBP4-IBB domain interaction. A, GST pulldown assay using GST, GST-importin ␤1, and GST-importin ␣1 with RBBP4. The assays were analyzed by Western blotting. B, competition assays for GST-IBB-GFP with RBBP4 (50 pmol) in the presence of increasing amounts of importin ␤1 or GFP (50, 100, and 200 pmol, respectively). The assays were analyzed by Western blotting. GST-IBB and applied importin ␤1 or GFP are shown by Coomassie staining (CBB, Coomassie Brilliant Blue). C, multiple sequence alignment of all subtypes of human and mouse importin ␣ (IBB domains) was constructed using ClustalW (28). The underlined residues in hIMP␣1 are involved in importin ␤1 binding based on the structure of the importin ␤1⅐IBB domain complex (PDB code 1qgk), and residues boxed in thin lines are conserved among all the subtypes. D, helical wheel projection of the helix in the IBB domain was generated using a script on the web site. Conserved residues and importin ␤1-binding residues are circled in red and cyan, respectively. Importin ␤1 (PDB code 1qgk) binds to three distinct regions in the IBB domain as follows: the N-terminal short segment and two regions on the helix.
to the IBB domain of importin ␣1. This acceleration in dissociation facilitates importin ␣⅐␤-mediated nuclear protein import. This is consistent with our finding that the knockdown of RBBP4 decreased the nuclear transport efficiency of cNLS-containing cargo.
In contrast to the large number of proteins known to bind to the armadillo repeats or the C terminus of importin ␣, only a few proteins have been shown to bind to the IBB domain. Previously, we reported that Snail, a zinc finger-containing transcriptional regulator, interacts with a broad region of importin ␣ that includes the IBB domain (46) but is transported into the nucleus by importin ␤1 alone (47). In addition, we demonstrated that Snail bound to importin ␣ fails to form a complex with importin ␤1 (46), which means that importin ␣ negatively regulates the importin ␤-mediated Snail import. However, in contrast to RBBP4, there is no evidence indicating that Snail is involved in the general regulation of importin ␣⅐␤ pathway.
Although RBBP4 is the only known regulator of the importin ␣⅐␤ pathway that binds to the IBB domain of importin ␣, several other molecules that bind to the other regions of importin ␣ have also been implicated in the regulation of the pathway. Npap60 promotes the release of cNLS-cargo from importin ␣ (48), although its short isoform, Npap60S, stabilizes the complex between importin ␣ and cNLS-containing cargo (49). Recently, it was demonstrated that the interaction of the C-terminal region of importin ␣ with Nup153 promotes translocation of the cNLS⅐importin ␣⅐␤1 trimeric complex into the nucleus, suggesting that importin ␣ serves as not only an adaptor molecule but also as a transport mediator in the complex (21). Taken together, these results indicate that importin ␣⅐␤mediated nuclear transport might be regulated by these importin ␣-interacting molecules. Further studies will be required to learn how these importin ␣-binding molecules are related to physiological processes such as cellular senescence.
Cellular senescence is a state of permanent cell cycle arrest. Senescent cells are characterized by a flat and enlarged morphology, increased SA ␤-gal activity, increased expression of cyclin-dependent kinase inhibitors, and the alteration of chromatin structure (50 -53). It has been shown that chromatin modifiers, including RBBP4, are related to cellular senescence (54,55). In fact, we found that histone H3 modifications (H3K9me3 and H3K27me3) increased in si-RBBP4-treated TIG-1 cells (Fig. 8G), indicating that RBBP4 can affect the chromatin states. However, these changes were not consistent with the previous reports, in which these modifications decrease in senescent cells (40 -42). Therefore, we suppose that these changes in histone H3 modifications observed in the si-RBBP4treated TIG-1 cells might not have a direct effect on cellular senescence. Furthermore, we found that histone H4K20me3, which is known to increase in senescent cells (43), did not change in si-RBBP4-treated TIG-1 cells, although it increased in passage-dependent senescent TIG-1 cells (Fig. 8G). In addition, we found that histone methyltransferase Ezh2 increased in the si-RBBP4-treated senescent TIG-1 cells (Fig. 8G), as observed in proliferative cells (44,45). Furthermore, it has been shown that the sole knockdown of RBBP4, without simultaneous knockdown of RBBP7, has little effect on the chromatin structure in other types of cells (11). Taken together, we sup- FIGURE 6. Structural models of the complex between RBBP4 and the IBB domain. A, GST pulldown assays for GST-IBB-GFP with RBBP4. Either WT or mutants of GST-IBB-GFP were incubated with RBBP4 and then detected by Western blotting. GST-IBB-GFP proteins are shown by Coomassie staining (CBB, Coomassie Brilliant Blue). B, RBBP4-binding residues in the IBB domain. Left, electrostatic potential and hydrophobicity on the molecular surface of the RBBP4 structure (PDB code 4pby), which was obtained from the eF-site database (60). The figure is colored to identify negatively charged (red), positively charged (blue), or neutral (white) non-hydrophobic residues and negatively charged (orange), positively charged (green), or neutral (yellow) hydrophobic residues. Upper right, IBB domain structure (PDB code 1qgk), with the putative RBBP4binding residues shown as ball and stick models. Lower right, electrostatic potential and hydrophobicity on the molecular surface of the IBB domain. We predict that the hydrophobic and positively charged surface in the IBB domain binds to the hydrophobic and negatively charged surface spanning from Leu-306/-308/-310 to the C terminus of RBBP4. C, two structural models of the RBBP4⅐IBB domain complex. We used the interactive molecular viewer, jV (60), and superimposed manually the IBB domain structure in the importin ␤1⅐IBB domain complex (PDB code 1qgk) onto the RBBP4 structure in the RBBP4⅐MTA1 complex (PDB code 4pby), so that Leu-306/-308/-310 of RBBP4 would interact with Met-27/Val-34/ Leu-38 of the IBB domain, and either the N or C terminus of RBBP4 would also interact with the IBB domain. Finally, two possible models were obtained showing the C terminus of RBBP4 mediating IBB domain binding. Note that in these models, we allowed some collisions between the PP loop (Ser-347-Glu-364) of RBBP4 and the IBB domain, and loose interactions between the N-terminal short segment and RBBP4, because of possible conformational changes upon RBBP4-IBB domain binding in the PP loop and the relative position of the N-terminal short segment to the helix in the IBB domain.
pose that the knockdown of RBBP4 does not induce cellular senescence primarily through the chromatin modifications. However, cellular senescence is triggered by many causes and regulated through many pathways; therefore, we speculated that both processes, decreasing nuclear transport efficiency and altering chromatin modifications, could potentially contribute to senescence in RBBP4-depleted cells.
Several reports have shown that attenuation of the nuclear transport of macromolecules is closely linked to cellular senescence and aging disorders such as Hutchinson-Gilford progeria syndrome. For example, nuclear localization of mitogen-activated protein kinase (ERK1/2) or galectin-3 is impaired in senescent cells (37,56). Moreover, Hutchinson-Gilford progeria syndrome cells or lamin A mutant-expressing cells show FIGURE 7. Classical nuclear import of cNLS is impaired in HeLa cells in which RBBP4 is knocked down. A, in vitro transport assays using digitoninpermeabilized HeLa cells. GST-SV40T-NLS-GFP was incubated with or without importin ␣1 and ␤1 in the presence of increasing amounts of RBBP4 as indicated. After 30 min at 30°C, the GFP-tagged proteins were visualized by fluorescence microscopy. B, GST pulldown assays to examine the effect of RBBP4 on the IBB domain⅐importin ␤1 complex. GST-IBB-GFP was incubated first with importin ␤1 (200 pmol) for 30 min to form a complex (pre-incubation). Then increasing amounts of RBBP4 (25, 50, and 100 pmol) were added (B, left panel). GST-IBB-GFP, importin ␤1, and the same increasing amounts of RBBP4 were added simultaneously (B, right panel). Importin ␤1 and RBBP4 were detected by Western blotting. GST-IBB is shown by Coomassie staining. CBB, Coomassie Brilliant Blue. C, GST pulldown assays to examine the cooperative effect of Q69LRanGTP and RBBP4 against the IBB domain⅐importin ␤1 complex. GST-IBB-GFP and importin ␤1 were pre-incubated, and then Q69LRanGTP (400 pmol) was added either with or without RBBP4 (400 pmol). Importin ␤1, RBBP4, and GST-IBB were detected by Western blotting. The graph indicates the signal intensities of importin ␤1 relatively to GST-IBB (values represent averages Ϯ S.D.). D, in vitro transport assays to measure the nuclear import efficiency of GST-SV40T-NLS-GFP. HeLa cells were transfected with two siRNAs against RBBP4: si-RBBP4-1 or si-RBBP4-2. At 48 h after transfection, the cells were permeabilized with digitonin, and an in vitro transport assay was performed using importin ␣1 and ␤1 recombinant proteins. The fluorescence intensity of the cNLS substrate in the nucleus was analyzed in time-lapse imaging and is displayed for each indicated time. E, immunofluorescence staining of RBBP4 in the nucleus of HeLa cells fixed after performing transport assays. At 20 min after the incubation described in D, the cells were stained with anti-RBBP4 antibody. Green fluorescence corresponds to GST-SV40T-NLS-GFP, and red fluorescence corresponds to RBBP4. F, averaged nuclear fluorescence signal was collected for 20 min and is presented as nuclear import curves. Relative signal intensities of 10 different nuclei per experiment were measured in three independent experiments (error bars represent standard error of the mean) and statistically analyzed using Bonferroni's post hoc test after a two-way analysis of variance (*, p Ͻ 0.05; **, p Ͻ 0.001; NS, not significant). DECEMBER 4, 2015 • VOLUME 290 • NUMBER 49 reduced efficiency in importin ␤-dependent nuclear transport (57,58). The decrease of nuclear transport-related factors such as importin ␣, CAS, and Ran is also observed in senescent cells (23,38). The knockdown of Ran, CAS, or Tpr induces cellular senescence (23,59). In this study, we showed that RBBP4 acts as a novel key regulator of the classical nuclear transport pathway to keep cells young. We found that the knockdown of RBBP4 not only reduces the efficiency of classical importin ␣⅐␤-medi-ated nuclear protein transport in HeLa cells that demonstrated a no senescent state (data not shown) but it also induces cellular senescence in normal fibroblasts, even though there is no change in the protein levels of the nuclear transport factors (data not shown). We also confirmed that nuclear transport efficiency is decreased in senescent cells. Based on these findings, we propose that nuclear protein transport efficiency must be maintained to keep cells young, and thus a gradual decrease FIGURE 8. Silencing RBBP4 induces cellular senescence. A, endogenous RBBP4 in different PDLs of TIG-1 cells. p21 was used to show that cells were in senescent states, and actin was used as an internal control in Western blotting (WB). The graphs indicate the protein level or mRNA level of RBBP4 normalized to actin (Western blotting) or HPRT (quantitative RT-PCR), respectively (values represent averages Ϯ S.D.). B and C, comparisons of nuclear import ability in different stages of TIG-1 cells. Senescent cells (PDL of Ͼ65) are compared with young cells (PDL of Ͻ30). GST-SV40T-NLS-GFP and Alexa-Fluor-594-conjugated antibody as an injection marker were microinjected into the cytoplasm of both stages of TIG-1 cells. The cells were incubated at room temperature and fixed at 20 min. After the fluorescent examination, cells were stained by SA ␤-gal (B). The nuclear and cytoplasmic fluorescence intensities were determined for at least 10 different cells to enable the calculation of the nuclear to cytoplasmic ratio (Fn/c). Statistical analysis was performed using the Student's t test (C, *, p Ͻ 0.001). D-F, TIG-1 cells were transfected with two independent siRNAs (si-RBBP4-1 and si-RBBP4-2) or with control (si-control) for 48 h. Scale bars, 20 m. D, cellular senescence was detected by SA ␤-gal staining. The graph indicates the percentage of SA ␤-gal-positive cells. Values represent averages Ϯ S.D. from three experiments; n Ͼ 200. Statistical analysis was performed using the Student's t test (*, p Ͻ 0.001). E, detection of endogenous Ki-67 in the cells using a specific antibody. F, Western blotting with indicated antibodies using whole cell lysates prepared from RBBP4 knockdown TIG-1 cells. G, Western blotting for histone modifications in whole cell lysates from si-RBBP4 treated TIG-1 cells or from three different PDLs of TIG-1 cells.

RBBP4 Regulates Classical Nuclear Transport
in nuclear transport efficiency due to a variety of causes, such as the decrease of nuclear transport-related factors and the impairment of positive regulators like RBBP4, induces cellular hypofunction, leading to cellular senescence.
Author Contributions-A. T. carried out all experiments with the exception of the structural models, which Y. T. and K. M. made, and the mass analysis, which C. O. provided. A. T., Y. M., T. M., M. O., and Y. Y. designed the study and wrote the paper with comments from all other authors.