B23 regulates GADD45a nuclear translocation and contributes to GADD45a-induced cell cycle G2-M arrest.

Gadd45a is an important player in cell cycle G2-M arrest in response to genotoxic stress. However, the underlying mechanism(s) by which Gadd45a exerts its role in the control of cell cycle progression remains to be further defined. Gadd45a interacts with Cdc2, dissociates the Cdc2-cyclin B1 complex, alters cyclin B1 nuclear localization, and thus inhibits the activity of Cdc2/cyclin B1 kinase. These observations indicate that Gadd45a nuclear translocation is closely associated with its role in cell cycle G2-M arrest. Gadd45a has been characterized as a nuclear protein, but it does not contain a classical nuclear localization signal, suggesting that Gadd45a nuclear translocation might be mediated through different nuclear import machinery. Here we show that Gadd45a associates directly with B23 (nucleophosmin), and the B23-interacting domain is mapped at the central region (61-100 amino acids) of the Gadd45a protein using a series of Myc tag-Gadd45a deletion mutants. Deletion of this central region disrupts Gadd45a association with B23 and abolishes Gadd45a nuclear translocation. Suppression of endogenous B23 through a short interfering RNA approach disrupts Gadd45a nuclear translocation and results in impaired Gadd45a-induced cell cycle G2-M arrest. These findings demonstrate a novel association of B23 and Gadd45a and implicate B23 as an important regulator in Gadd45a nuclear import.

Gadd45a is an important player in cell cycle G 2 -M arrest in response to genotoxic stress. However, the underlying mechanism(s) by which Gadd45a exerts its role in the control of cell cycle progression remains to be further defined. Gadd45a interacts with Cdc2, dissociates the Cdc2-cyclin B1 complex, alters cyclin B1 nuclear localization, and thus inhibits the activity of Cdc2/ cyclin B1 kinase. These observations indicate that Gadd45a nuclear translocation is closely associated with its role in cell cycle G 2 -M arrest. Gadd45a has been characterized as a nuclear protein, but it does not contain a classical nuclear localization signal, suggesting that Gadd45a nuclear translocation might be mediated through different nuclear import machinery. Here we show that Gadd45a associates directly with B23 (nucleophosmin), and the B23-interacting domain is mapped at the central region (61-100 amino acids) of the Gadd45a protein using a series of Myc tag-Gadd45a deletion mutants. Deletion of this central region disrupts Gadd45a association with B23 and abolishes Gadd45a nuclear translocation. Suppression of endogenous B23 through a short interfering RNA approach disrupts Gadd45a nuclear translocation and results in impaired Gadd45ainduced cell cycle G 2 -M arrest. These findings demonstrate a novel association of B23 and Gadd45a and implicate B23 as an important regulator in Gadd45a nuclear import.
The cell cycle checkpoint is one of the major genomic surveillance systems in mammalian cells. Inactivation of such a system results in genomic instability and malignant transformation of cells (1,2). The tumor suppressor gene p53 is implicated in the control of both cell cycle G 1 -S and G 2 -M arrests in response to genotoxic stress (3,4). In addition to the well characterized p53-p21 waf/cip1 pathway in regulating DNA damage-activated cell cycle checkpoints (5)(6)(7), the p53-Gadd45a pathway has been shown to primarily play a role in the control of G 2 -M arrest following certain DNA-damaging agents (8 -11).
The Gadd45a gene is induced by a wide spectrum of DNA-damaging agents or growth arrest signals such as ionizing radiation, UV radiation, methyl methanesulfonate (MMS), 1 hydroxyurea (12), growth factor withdraw, and serum starvation (13,14). The ionizing radiation induction of Gadd45a is transcriptionally regulated by p53 via a p53-binding site in the third intron of the Gadd45a gene (3) and strictly depends on normal cellular p53 function (15). Gadd45a has also been shown to be a downstream gene of BRCA1 (16 -18), a breast cancer-associated gene that plays roles in the control of cell cycle progression, apoptosis, DNA repair, and gene regulation. The regulation of Gadd45a by BRCA1 does not require normal p53 function (17). Gadd45a physically interacts with several important cellular proteins, including proliferating cell nuclear antigen, p21, Cdc2 (9), core histones, and MTK1/MEKK4 (9, 19 -24). The presence of Gadd45a in these protein complexes suggests that Gadd45a may play important roles in cell cycle control, DNA repair, and the regulation of signaling pathways. The role of Gadd45a in maintaining genomic stability has been demonstrated by the findings that mouse embryonic fibroblasts (MEFs) derived from Gadd45a-null mice exhibit aneuploidy, chromosomal aberrations, gene amplification, and centrosome amplification. Additionally, Gadd45a knock-out mice display increased ionizing radiation-or UV radiation-induced carcinogenesis (25)(26)(27).
Multiple lines of evidence indicate that Gadd45a is an important regulator of the cell cycle G 2 -M checkpoint in response to certain genotoxic stressors. Gadd45a interacts with Cdc2, dissociates Cdc2-cyclin B1 complexes, and suppresses Cdc2/ cyclin B1 kinase activity. Overexpression of Gadd45a results in the alteration of cyclin B1 subcellular localization and reduces nuclear distribution of cyclin B1 protein (8 -11). Gadd45a is a nuclear protein (20,21), and its nuclear translocation might be critical for its role in the control of the cell cycle G 2 -M checkpoint. However, Gadd45a protein has no classical nuclear localization signal sequence (NLS) and may utilize different mechanism(s) for its nuclear import. The machinery controlling nuclear translocation of Gadd45a protein is currently unclear.
It has been well accepted that all nuclear proteins are synthesized in the cytoplasm and need to be imported through the nuclear pore complexes into the nucleus. Active nuclear import is energy-dependent and is mediated by import receptors. Import into the nucleus can be conferred by several distinct import signals. The classical NLS consists of one or more clusters of basic amino acids and is present in a large number of pro-teins. Another major import signal is mediated by the M9 domain. NLS-and M9-containing proteins do not compete with each other for import and are recognized by distinct receptors. The active import depends on four soluble factors, namely importins ␣ and ␤, which together constitute the NLS receptor, the GTPase Ran/TC4, and NTF2 (nuclear transport factor 2) (28 -30). However, there is an NLS-independent and importinindependent nuclear import for some nuclear proteins. ␤-Catenin, a major component of the Wnt signaling pathway, has no nuclear localization sequence (NLS) and is imported into the nucleus by binding directly to the nuclear pore machinery (31,32). In addition, B23 (also called nucleophosmin, NO38, or numatrin) has been reported to function as a chaperone in the process of nuclear transport.
B23/nucleophosmin is a multifunctional nucleolar phosphoprotein and is required for assembly of ribosomes. B23 plays critical roles in the control of centrosome duplication, cell proliferation, and transcriptional regulation (33)(34)(35). Most interestingly, this protein has been shown to shuttle between the cytoplasm and nucleus and acts as a carrier protein for certain proteins, including nucleolar protein p120, nucleolin, and several viral proteins such as human T-cell lymphotrophic virus Rex, human immunodeficiency virus Rev, and human immunodeficiency virus Tat (36 -40). In the current study, the machinery that mediates Gadd45a nuclear translocation has been investigated. We show that Gadd45a physically associates with B23, and the interaction between Gadd45a and B23 is essential for the process of Gadd45a nuclear import. Disruption of Gadd45a association with B23 protein greatly abrogates Gadd45a-induced cell cycle G 2 -M arrest, indicating that Gadd45a nuclear localization is mediated through a B23 chaperone and is required for its role in the control of cell cycle progression.

EXPERIMENTAL PROCEDURES
Plasmid Clones-The following plasmid clones were used. All Myctagged Gadd45a deletion expression vectors were constructed as described previously (9,10). These Myc-tagged clones harbor different regions of Gadd45a protein. GST-Cdc2, GST-p53, GST-p21, and GST-EF1 were constructed by inserting their open reading frames into the EcoRI-XhoI sites of pGEX-5X-1 plasmid. GST-Gadd45a was made by cloning Gadd45a cDNA containing the open reading frame into the XhoI site of the pGEX-5X-1 vector. pEGFP-Gadd45a was constructed by inserting the open reading frame of Gadd45a cDNA into the pEGFP vector. Similarly, pEGFP-Gadd45a-(61-100) was constructed by inserting Gadd45a cDNA containing 61-100 amino acids into the pEGFP plasmid. The pEGFP-⌬(60 -100)-Gadd45a was made to express the truncated GFP-Gadd45a fusion protein with deletion of 61-100 amino acids. GST-B23 was provided by Dr. Kenji at the University of Cincinnati.
Cell Cultures and Treatment-The human colon carcinoma line HCT116 was grown in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS). Wild-type and Gadd45a knock-out MEFs were kindly provided by Albert Fornace of the National Institutes of Health and maintained in DMEM with 10% FBS. HCT116 Gadd45ainducible cells were established previously. Cells were grown in DMEM containing tetracycline at a concentration of 2 g/ml. Following withdrawal of tetracycline, Gadd45a protein exhibits high expression.
For cell transfection with Myc tag-Gadd45a expression vectors, 5 ϫ 10 5 HCT116 cells were seeded onto 10-mm plates before transfection. In each plate, 5 g of DNA and 15 l of Lipofectamine (Invitrogen) were added in 300 l of Opti-MEM (Invitrogen) in separate tubes. Solutions were mixed gently, allowed to sit 15 min at room temperature, diluted with 2.4 ml of Opti-MEM, and added to the plates for 6 h at 37°C. Equal volumes of media with 10% FBS were added, and plates were incubated overnight. Fresh media were added the following day, and cells were harvested 48 h later.
For UV radiation treatment, cells plated in 100-mm dishes were rinsed with PBS and irradiated with UVC to a dose of 10 J/m Ϫ2 . Following UV radiation treatment, fresh medium was added to plates, and cells were cultured in the incubator until harvest. For MMS treatment, cells were exposed to MMS (Aldrich) at 100 g/ml for 4 h, and media were then removed and replaced with fresh media. Cells were then collected at the indicated time points.
siRNA Transfection-The B23 siRNA sequence was designed as UGA UGA AAA UGA GCA CCA G. The nonspecific siRNA sequence was designed as GAC CAC GAG UAA AAG UAG U. For cell transfection with siRNA, HCT116 cells were placed onto 6-well plates 16 h prior to transfection and grown at 2 ϫ 10 5 . HCT116 cells were seeded onto 6-well plates. In one tube, 10 l of 20 M siRNA was added to a tube containing 50 l of Opti-MEM. In a separate tube, 1 l of Lipofectamine was mixed with 50 l of Opti-MEM. The two tubes were next mixed, allowed to sit at room temperature for 30 min, and then added to each well in the plate, which contained 1 ml of medium. 4 h later, 4 ml of fresh medium was added, and transfected cells were incubated for 24 -72 h until they were ready to assay for gene knock-down analysis.
RT-PCR-HCT116 cells were grown in RPMI 1640 supplemented with 10% FBS. Total RNA was isolated using RNeasy mini kit (Qiagen) according to the manufacturer's protocol. RT-PCR was carried out using an RNA PCR core kit (PerkinElmer Life Sciences). 0.5 g of total RNA in 1 l of RNase-free water was used in 20 l of RT mix containing the following: 4 l of 25 mM MgCl 2 , 2 l of 10ϫ PCR buffer, 2 l of diethyl pyrocarbonate water, 8 l of dNTP mix (2.5 mM each of dATP, dCTP, dGTP, and dTTP), 1 l of RNase inhibitor (20 units/l), 1 l of random hexamers (50 M), and 1 l of murine leukemia virus-reverse transcriptase (50 units/l). The mixture was subjected to cDNA synthesis using GeneAmp PCR System 9600 (PerkinElmer Life Sciences). 10 l of cDNA product was added to 40 l of PCR mix containing the following: 2 l of 25 mM MgCl 2 , 4 l of 10ϫ PCR buffer, 3 l of dNTP mix (2.5 mM each of dATP, dCTP, dGTP, and dTTP), 28.5 l of sterile distilled water, 0.5 l of TaqDNA polymerase (5 units/l), and 2 l of 1:1 primer mix (30 M each of upstream and downstream primers). The mixture was subjected to DNA amplification using GeneAmp PCR System 9600 (PerkinElmer Life Sciences). Finally, 30 l of PCR product was loaded on a 1% agarose gel for analysis. The primers used to amplify B23 and ␤-actin were designed as follows: 1) B23, 5Ј-CGCGGATCCCGATGGAAGATTCGA, and 5Ј-GCCG-CTCGAGTTAAAGAGACTTCC; 2) ␤-actin, 5Ј-GCGGGAAATCGTGCGT-GACATT, and 5Ј-ATGATGCTTCAACACCCAGGC.
Cellular Protein Preparation and Immunoblotting Analysis-In the preparation of nuclear protein, the exponentially growing HCT116 cells were collected, rinsed with PBS, and resuspended in 200 l of cold buffer A (10 mM Hepes, pH 7.9; 10 mM KCl; 0.1 mM EDTA; 0.1 mM EGTA; 1 mM dithiothreitol; 0.5 mM phenylmethylsulfonyl fluoride). Following vortexing, the samples were incubated on ice for 10 min and followed by addition of Nonidet P-40 to a final 0.5% concentration. After centrifugation, insoluble pellets were resuspended in 100 l of ice-cold buffer C (20 mM Hepes, pH 7.9; 400 mM KCl; 1 mM EDTA; 1 mM EGTA; 1 mM dithiothreitol; 1 mM phenylmethylsulfonyl fluoride). The samples were placed on ice and subjected to vortexing for 15 s every 10 min, for a total of 40 min. Finally, the samples were centrifuged at 14,000 ϫ g for 10 min, and the supernatant (nuclear extract) was collected for further analysis.
For preparation of whole cell protein, cells were rinsed with PBS and lysed in PBS containing100 g/ml phenylmethylsulfonyl fluoride, 2 g/ml aprotinin, 2 g/ml leupeptin and 1% Nonidet P-40 (lysis buffer). Lysates were collected by scraping and cleared by centrifugation at 4°C.
For immunoblotting analysis, 100 g of proteins were loaded onto SDS-polyacrylamide gels for electrophoresis and then transferred to Protran membranes. Membranes were blocked in 5% milk, washed with PBST (PBS with 0.1% Tween), and incubated with anti-ATF3 or actin antibodies (Santa Cruz Biotechnology, Santa Cruz, CA). Following washing and incubation with horseradish peroxidase-conjugated antirabbit or anti-mouse antibody at 1:4000 in 5% milk, membranes were washed and detected by ECL (Amersham Biosciences) and exposed to x-ray film.
For immunoprecipitation, cellular lysates were incubated with 10 l of antibody and 20 l of protein A/G-agarose beads (Santa Cruz Biotechnology, Santa Cruz, CA) at 4°C for 4 h. Immunocomplexes were washed four times with lysis buffer, loaded onto 12% SDS-polyacrylamide gels, and analyzed as described above.
Antibodies to Gadd45a, c-Myc, Cdc2, actin, ATF3, and actin were commercially provided by Santa Cruz Biotechnology (Santa Cruz Biotechnology, Santa Cruz, CA). Antibody to cyclin B1 was provided by Pharmingen.
GST fusion protein expression was induced in Escherichia coli with 0.1 mM isopropyl 1-thio-␤-D-galactopyranoside. Bacterial pellets were washed with PBS and resuspended in cold STE buffer (10 mM Tris-HCl, pH 8.0, 1 mM EDTA 150 mM NaCl). After incubation with freshly prepared lysozyme solution and Sarkosyl (0.7%), bacterial mixtures were subjected to sonication for 1 min. Following centrifugation at 16,000 rpm for 20 min, supernatants were collected and mixed with Triton X-100 to a final concentration of 2%. Glutathione-agarose beads were then added and incubated at 4°C overnight. After washing with PBS, glutathione-agarose bead-conjugated GST fusion proteins were ready for use.
Flow Cytometric Analysis-The wild-type and Gadd45aϪ/Ϫ MEFs were synchronized at the G 1 /S transition by treatment with 1 g/ml aphidicolin for 24 h. On release from the aphidicolin block, cells were treated with UV radiation and incubated for an additional 20 h in the presence of BrdUrd. Cells were collected, and the BrdUrd-positive cells were subjected to FACScan analysis. In the case of the HCT116 Gadd45a-inducible line, cells were plated into 100-mm dishes at a density of 5 ϫ 10 5 and grown in DMEM containing 2 g/ml tetracycline. 16 h later, medium was removed, and plates were washed four times followed by addition of fresh medium. After incubation for 36 h, cells were collected, washed with PBS, and fixed with 70% ethanol for 2 h at 4°C. Cells were then incubated with RNase (10 g/ml) for 30 min and stained with propidium iodine (Sigma; 50 g/ml). Cell cycle analysis was performed by using a BD Biosciences fluorescence-activated cell analyzer. At least 10,000 fluorescein isothiocyanate-positive cells were analyzed using Cellquest and Modfit programs.
Cdc2 Kinase Analysis-500 g of cellular lysate isolated from MEFs treated with UV radiation at 10 J/m 2 and MMS at 50 g/ml or Gadd45ainducible cell lysate was incubated with 10 l of cyclin B1 antibody (Pharmingen) and 20 l of protein A/G-agarose beads (Santa Cruz Biotechnology, Santa Cruz, CA) at 4°C for 6 h. Immunocomplexes were washed four times with lysis buffer and followed by kinase buffer. Histone H1 kinase assays were then performed in the presence of 10 g of histone H1 (Upstate Biotechnology, Inc., Lake Placid, NY), 15 mM MgCl 2 , 7 mM ␤-glycerol phosphate, 1.5 mM EDTA, 0.25 mM sodium orthovanadate, 0.25 mM dithiothreitol, and 10 Ci of [␥-32 P]ATP in a 30-l volume. After 15 min at 30°C, the reactions were mixed with an equal amount of standard 2ϫ SDS protein denaturing loading buffer and then size-separated on a 12% SDS-polyacrylamide gel.
Measurement of Mitotic Index-HCT116 Gadd45a-inducible cells were seeded at a density of 5 ϫ 10 5 and grown in DMEM containing 2 g/ml tetracycline and 0.4 g/ml nocodazole. Following withdrawal of tetracycline, cells were transfected with B23 siRNA and harvested at the indicated time points, fixed in methanol:acetic acid (3:1), spread on glass microscope slides, air-dried, and stained with 5% Giemsa. Nuclei exhibiting condensed, evenly staining chromosomes were scored as mitotic. At least 1000 cells were counted in each determination.

Disruption of Endogenous Gadd45a Results in Impaired Cell
Cycle G 2 -M Arrest following Certain DNA-damaging Agents-Several lines of evidence indicate that Gadd45a is one of the important components involved in the control of the cell cycle G 2 -M checkpoint. We have demonstrated previously that disruption of endogenous Gadd45a via an antisense approach results in a perturbed G 2 -M delay following DNA damage (8).
To confirm further the role of Gadd45a in G 2 -M arrest after genotoxic stress, cell cycle analysis was performed using MEF in which both Gadd45a alleles had been disrupted by homologous recombination to demonstrate that disruption of endogenous Gadd45a was sufficient to abrogate the G 2 -M checkpoint following treatment with UV radiation. A modified double labeling (PI and BrdUrd) protocol, which requires no cell cycle inhibitors, was employed in this experiment (8). Cell cycle progression was evaluated at very early passages to minimize effect because of the passage of cells in culture. As shown in Fig. 1A, wild-type MEF cells exhibited a clear G 2 -M arrest after treatment with UV radiation. In contrast, the MEFs derived from Gadd45a knock-out mice exhibited a less stringent G 2 -M arrest after UV radiation.
As described earlier, the Cdc2-cyclin B1 complex is a key regulator of the transition from G 2 to mitosis (41,42). Generally, Cdc2/cyclin B1 activity is inhibited following DNA damage, and this inhibition acts to block the G 2 -M transition (43). In Fig. 1B, the wild-type MEFs displayed a strong reduction in Cdc2 kinase activity following UV radiation or MMS, whereas the reduction was significantly attenuated in the Gadd45aϪ/Ϫ MEFs. This result correlates with the attenuated G 2 -M arrest after UV radiation described above.
In addition, HCT116 Gadd45a-inducible cells, where Gadd45a-inducible expression is controlled by a tetracycline system, were employed. Following the withdrawal of tetracycline, HCT116 exhibited a highly induced expression of the Gadd45a protein (Fig. 1C, 1st panel). The cell cycle distribution analyses were conducted in the HCT116 Gadd45a-inducible cell line. Following removal of tetracycline, Gadd45a-inducible cells were collected at 36 h and subjected to flow cytometric analysis. As shown in Fig. 1D, inducible expression of Gadd45a protein resulted in a substantial accumulation of the G 2 -M fraction, and ϳ38% of the cells were halted at the G 2 -M phase of the cell cycle following removal of tetracycline. In contrast, about 14% population of cells presented at the G 2 -M phase in the presence of tetracycline. This Gadd45a-induced G 2 -M arrest was coupled with altered nuclear localization of cyclin B1 by Gadd45a induction (Fig. 1C, 2nd panel). Collectively, Gadd45a is an important component required for cell cycle G 2 -M arrest following certain DNA-damaging agents.
Gadd45a Localizes to Both Nuclear and Cytosol Compartments-Several reports have suggested that Gadd45a is predominantly a nuclear protein (20,21). Overexpression of Gadd45a has been shown to result in dissociation of the Cdc2cyclin B1 complex and reduction of the nuclear distribution of cyclin B1 (11). Therefore, nuclear localization of Gadd45a protein might be associated with its biological function in the control of cell cycle G 2 -M arrest. In Fig. 2A, HCT116 cells were treated with UV radiation and assayed for expression of Gadd45a. In addition to the observation that Gadd45a was clearly induced by UV radiation, this protein was shown to localize to both the nuclear and cytosol compartments. To exclude protein leakage during preparations of nuclear or cytosol proteins, the detection of actin (cytosol protein) and p53 (nuclear protein) was included. The localization of these two (actin and p53) proteins confirms that no cross-contamination occurred during the fractionation. In addition, we transfected Gadd45a- (1-165), a full-length Myc tag-Gadd45a expression vector, into HCT116 cells, and we examined the Myc tag-Gadd45a fusion protein using anti-Myc antibody. In Fig. 2B, Myc tag-Gadd45a was seen to distribute in both nuclear and cytoplasmic fractions. Taken together with previous observations by others, Gadd45a is a nuclear protein but remains in both the nuclear and cytosol compartments.
The Central Region of the Gadd45a Protein Is Required for Nuclear Translocation-To rule out the possibility that Gadd45a protein is diffused into the nucleus due to its small molecular size (21 kDa), the Myc-tagged Gadd45a deletion mutants, Gadd45a-(1-71), Gadd45a-(91-165), or Gadd45a-(48 -165) were co-transfected with Gadd45a-(1-165, full length) into HCT116 cells, and an immunoblotting assay with anti-Myc antibodies was performed. As illustrated in Fig. 3A, Gadd45a-(1-71) and Gadd45a-(91-165) were only seen to localize at the cytoplasmic compartment (Fig. 3A, labeled as C). In contrast, Gadd45a-(48 -165) clearly exhibited nuclear localization (Fig. 3A, labeled as N). To examine further the Gadd45a domains involved in its nuclear translocation, six Gadd45a deletion mutants, each of which spans 40 residues with a 10-residue overlap between each contiguous peptide, were co-introduced with Gadd45a-(1-165) into HCT116 cells and subjected to immunoblotting analysis. As shown in Fig. 3B, only Gadd45a-(61-100) displayed nuclear localization, but the rest of the five deletion mutants showed exclusive cytoplasmic localization. Next, we constructed a new Gadd45a deletion mutant, ⌬(61-100)-Gadd45a, in which the region between 61 and 100 amino acids was truncated, and we examined its subcellular localization. Consistent with the results described in Fig. 3, A and B, when the amino acids from 61 to 100 were deleted, the Gadd45a protein remained in the cytoplasm (Fig. 3C), further indicating that the central region of the Gadd45a protein from amino acids 60 to 100 is required for Gadd45a nuclear translocation.
Gadd45a Interacts with B23 (Nucleophosmin), Which Transports Certain Proteins into the Nucleus-Despite its nuclear localization, Gadd45a does not have a classical NLS sequence, suggesting that Gadd45a nuclear translocation might be mediated through different nuclear import machinery. It is speculated that Gadd45a protein nuclear translocation is possibly conducted via certain carrier proteins that can act as "nuclear delivery vehicles," to which the Gadd45a protein is able to physically bind. We used the two-hybrid system to identify Gadd45a-associated proteins. The Gadd45a protein was divided into four different baits, and in total, 32 cellular proteins were characterized, of which 12 proteins were confirmed to interact with Gadd45a using in vitro biochemical methods. By using a GST pull-down assay, several cellular proteins were defined to physically associate with Gadd45a. Those proteins include Cdc2 kinase, EF-1␣, B23 (nucleophosmin), and p21 WAF/CIP (Fig. 4A). In Fig. 4B, the Gadd45a-(1-165) fusion protein expression vector or unfused Myc tag vector were transfected into HCT116 cells, and cell lysates were prepared for immunoprecipitation with antibody to actin, cyclin B1, c-Myc, Cdc2, and B23. In addition to the positive binding with Cdc2 (positive control), Myc tag-Gadd45a fusion protein was also detected after immunoprecipitation with B23 antibody but was not seen in the immunocomplexes with cyclin B1 and actin antibodies (negative controls).
To further confirm the interaction between endogenous Gadd45a and B23 proteins, we carried out immunoprecipitation assays using the antibody-directed "pull-down" approach. To enrich cellular Gadd45a protein, nuclear extracts were iso- FIG. 1. A, disruption of endogenous Gadd45a substantially abrogates DNA damage-induced cell cycle G 2 -M arrest. MEFs derived from normal and Gadd45aϪ/Ϫ mice were grown in DMEM with 10% fetal bovine serum and synchronized with aphidicolin (2 g/ml) for 18 h. Upon release from the G 1 /S boundary, cells were treated with 5 J/m 2 of UV radiation and incubated in the presence of BrdUrd for an additional 20 h. The BrdUrd-positive cells were sorted for FACScan and ModFit analysis. B, inhibition of Cdc2/cyclin B1 kinase activity in MEFs following UV radiation and MMS. Normal MEFs and Gadd45aϪ/Ϫ MEFs were irradiated with 5 J/m 2 of UV radiation or 50 g of MMS. Cellular protein was prepared 8 h later. One milligram of protein was immunoprecipitated with antibody to cyclin B1, and histone H1 kinase assays were carried out as described under "Experimental Procedures." A representative experiment is shown in C. C, altered nuclear localization of cyclin B1 and inhibition of Cdc2 activity after induction of Gadd45a. HCT116 Gadd45a-inducible cells were placed in 100-mm dishes at a density of 4 ϫ 10 5 and grown in DMEM containing tetracycline at a concentration of 2 g/ml. After withdrawal of tetracycline, cells were collected at the indicated time points for preparation of nuclear protein. 50 g of nuclear protein was used for immunoblotting analysis with antibodies to Gadd45a, cyclin B1, and Cdc2. In addition, 200 g of nuclear protein was immunoprecipitated with anti-cyclin B1 antibody, and histone H1 kinase assays were performed as described under "Experimental Procedures." Labeled histone H1 was detected by autoradiography following size separation on a SDS-PAGE gel. D, cell cycle G 2 -M arrest following inducible expression of Gadd45a protein in HCT116 cells. HCT116 Gadd45a-inducible cells were established as described under "Experimental Procedures" and grown in RPMI 1640 medium with 10% fetal bovine serum in the presence of tetracycline at a concentration of 2 g/ml. 36 h after withdrawal of tetracycline, cells were collected and subjected to flow cytometric analysis as described under "Experimental Procedures." lated from HCT116 Gadd45a-inducible cells (see Fig. 1C), which were collected 24 h after tetracycline withdrawal and were confirmed to highly express the Gadd45a protein. Cellular lysates were incubated with anti-Gadd45a, anti-B23, anti-Cdc2, anti-actin, or anti-GFP antibodies and immunoprecipitated with protein A/G-agarose beads. The immunocomplexes were then analyzed by Western blotting assay using antibodies to Gadd45a and B23. As shown in Fig. 4C, Gadd45a protein was present in the immunocomplexes precipitated by the antibodies against Cdc2 and B23. Similarly, B23 protein was detected in the immunocomplexes with both anti-Gadd45a and anti-Cdc2 antibodies. In contrast, no Gadd45a or B23 proteins were present in the anti-actin-or anti-GFP-immunoprecipitated complexes. Therefore, the results presented in Fig. 4, A-C, indicate an association of Gadd45a with B23.
To map the B23-interacting motif of Gadd45a, GST or GST-B23 was incubated with cell lysates prepared from the HCT116 cells transfected with different Gadd45a deletion mutants as follows: Gadd45a-(1-165), ⌬(60 -100)-Gadd45a, Gadd45a-(1-100), and Gadd45a-(91-165). Following immunoprecipitation, GST complexes were analyzed with anti-Myc antibody. As shown in Fig. 4D, full-length Gadd45a and Gadd45a-(1-100) were pulled down by GST-B23. In contrast, ⌬(60 -100)-Gadd45a and Gadd45a-(91-165) were unable to interact with B23, suggesting that the central region is required for the interaction of Gadd45a and B23 protein. By using the same approach, we examined the interaction of GST-B23 with a series of Myc tag-Gadd45a deletion mutants, and we found that the central region was also required for Gadd45a nuclear import (results not shown). When pEGFP-Gadd45a or pEGFP-Gadd45a-(61-100) expression vectors were transfected into HCT116 cells, the green fluorescent protein (GFP) was primarily seen in nuclei. In contrast, the GFP mainly stayed in the cytoplasm in cells transfected with pEGFP-⌬(60 -100)-Gadd45a or pEGFP empty vector (results not shown). Therefore, disruption of the central region abolishes Gadd45a nuclear translocation. FIG. 2. Gadd45a protein localizes to both nuclear and cytosol compartments. A, HCT116 cells were treated with UV radiation at 10 J/m 2 and collected at the indicated time points, rinsed with PBS, and resuspended in 200 l of cold buffer A (see "Experimental Procedures"). Following vortexing, the samples were incubated on ice for 10 min followed by addition of Nonidet P-40 to a final 0.5% concentration. After centrifugation, the supernatant was collected (cytosol protein). The insoluble pellets were resuspended in 100 l of ice-cold buffer C (see "Experimental Procedures"). The samples were placed on ice and subjected to vortexing for 15 s every 10 min, for a total of 40 min. Finally, the samples were centrifuged at 14,000 ϫ g for 10 min, and the supernatant (nuclear extract) was collected for further analysis. For immunoblotting analysis, 100 g of protein was loaded onto SDS-polyacrylamide gels for detection of the Gadd45a protein. As the loading controls, actin (cytosol protein) and p53 (nuclear protein) were included in the experiments. B, 2 ϫ 10 5 HCT116 cells were placed onto 100-mm dishes and transfected with 5 g of either a Myc tag vector or Myc tag-Gadd45a-(1-165) expression vector. 48 h later, cells were harvested for preparation of both nuclear and cytosol proteins as described in A. The subcellular localization of Gadd45a protein was detected by using anti-Myc antibody. Actin was included for a loading control.

Suppression of B23 Protein Levels Using siRNA Approach
Substantially Affects Gadd45a Nuclear Translocation-As discussed earlier, B23 is able to deliver certain important cellular or viral proteins from the cytoplasm into the nucleus (36 -40). To investigate if the B23 protein contributes to Gadd45a nuclear translocation, the siRNA approach was used to inhibit endogenous B23 expression and was followed by examination of Gadd45a nuclear localization. B23 siRNA and nonspecific siRNA were added into HCT116 cells in culture at a concentration of 40 pmol. 48 h later, cells were collected for preparation of poly(A) ϩ , and an RT-PCR assay was performed to examine the levels of B23 mRNA. As shown in Fig. 5A, endogenous B23 mRNA was substantially suppressed by B23 siRNA but not by nonspecific siRNA. Endogenous B23 protein was also examined following B23 siRNA treatment. Consistent with the results in the RT-PCR, addition of B23 siRNA was shown to greatly knock down cellular B23 expression (Fig. 5B), whereas actin protein expression remained unaltered. In contrast, nonspecific siRNA did not affect cellular B23 protein levels.
Next, Gadd45a nuclear localization following suppression of B23 endogenous protein was evaluated. In Fig. 5C, HCT116 cells were transfected with B23 siRNA or nonspecific siRNA for 48 h, treated with UV radiation, and harvested at 0, 4, and 8 h. Both cytosol and nuclear proteins were extracted and assayed for subcellular distribution of endogenous Gadd45a protein.
The results in Fig. 5C show that nuclear localization of Gadd45a protein was greatly reduced following the suppression of B23 protein expression by B23 siRNA. The addition of B23 siRNA suppressed more than 70% of the nuclear Gadd45a protein, compared with that seen in the cells treated with nonspecific siRNA. To rule out any nonspecific effect of B23 siRNA, the nuclear localization of ATF3 protein, a stress-inducible transcription factor, was also examined and did not show any alterations after B23 siRNA treatment. Additionally, Myc tag-Gadd45a was co-introduced with B23 siRNA into the HCT116 line, and cells were collected for examination of Myc tag-Gadd45a fusion protein. As shown in Fig. 5D, Myc tag-Gadd45a protein in either control cells or cells treated with nonspecific siRNA exhibited substantial nuclear accumulation. However, the cells treated with B23 siRNA displayed a weak accumulation of endogenous Gadd45a. Taken together, these results indicate that disruption of endogenous B23 protein abrogates Gadd45a nuclear localization.
Gadd45a Nuclear Localization Is Required for Gadd45ainduced Cell Cycle G 2 -M Arrest-Previous findings demonstrate that Gadd45a interacts with Cdc2 and dissociates Cdc2cyclin B1 complexes, inhibiting Cdc2 kinase activity. Recently, induction of Gadd45a has been shown to result in alterations of cyclin B1 subcellular distributions, as reflected by reduced nuclear accumulation of cyclin B1, suggesting that the place for Gadd45a targeting on the Cdc2-cyclin B1 complex should be in the nucleus. If so, Gadd45a nuclear translocation is critical for its function in cell cycle G 2 -M arrest. Therefore, experiments were carried out to determine whether Gadd45a nuclear translocation is required for Gadd45a-mediated G 2 -M arrest. To do so, HCT116 Gadd45a-inducible cell lines via the Tet-Off system were employed, and cell cycle distributions of Gadd45a-inducible cells in the presence of B23 siRNA were analyzed. In Fig.  6A, after transfection of 21-nucleotide siRNA duplexes that target B12 mRNA transcripts into HCT116 Gadd45a-inducible cells, nuclear proteins were prepared and assayed for Gadd45a- FIG. 4. Physical interaction of Gadd45a with B23. A, GST or GST-cyclin B1, GST-p53, GST-EF-1␣, GST-Cdc2, GST-B23, and GST-p21 proteins were prepared (see "Experimental Procedures") and incubated with cell lysates isolated from HCT116 cells transfected with the full-length Myc tag-Gadd45a expression vector. The GST protein pulldown complexes were washed three times with lysis buffer, analyzed by SDS-PAGE, and immunoblotted with anti-Myc antibody. B, Myc tag-Gadd45a was expressed in HCT116 cells. Whole cell protein extracts were prepared and immunoprecipitated with anti-actin, anti-cyclin B1, anti-Cdc2, anti-Myc, and anti-B23 antibodies. Following SDS-PAGE, immunoblotting assays were carried out with anti-Myc antibody. C, nuclear protein from HCT116 cells was prepared and immunoprecipitated with anti-actin, anti-Gadd45a, anti-B23, anti-Cdc2, and anti-GFP antibodies. The immunocomplexes were analyzed by SDS-PAGE and immunoblotted with antibodies against B23 and Gadd45a, respectively. The visualized bands are shown. Their estimated masses were 38 kDa for B23 and 21 kDa for Gadd45a. D, Gadd45a-(1-165) and Gadd45a deletion mutants, ⌬(60 -100)-Gadd45a, Gadd45a-(1-100), and Gadd45a-(91-165), were transfected into HCT116 cells. 48 h later, cell lysates were prepared and incubated with GST or GST-B23 proteins. Following GST protein pull-down and SDS-PAGE, an immunoblotting assay was carried out with anti-Myc antibody.
inducible expression. Similar to the observations presented in Fig. 5, nuclear accumulation of Gadd45a protein in HCT116 Gadd45a-inducible cells was significantly reduced by B23 siRNA compared with that seen in cells treated with nonspecific siRNA. Next, the B23 siRNA-treated cells were subjected to flow cytometric analysis, and the results are shown in Fig.   6B. Clearly, disruption of Gadd45a nuclear localization via employment of B23 siRNA greatly affected Gadd45a-induced cell cycle G 2 -M arrest. The cell populations of G2-M phase were reduced from 31 to 16% following B23 siRNA treatment.
The mitotic index was also measured in HCT116 Gadd45ainducible cells in the presence of B23 siRNA. To facilitate the Procedures." 48 h later, cells were rinsed with PBS and irradiated with UVC to a dose of 10 J/m Ϫ2 . Following UV radiation treatment, fresh medium was added to the plates, and cells were cultured in an incubator until harvesting at the indicated time points. Both cytosol and nuclear proteins were prepared, and immunoblotting assays were conducted to detect the levels of Gadd45a. As the controls, actin (cytosol protein) and ATF3 (nuclear protein) were examined in the same experiments. D, a Myc tag-Gadd45a expression vector was co-transfected into HCT116 cells with B23 siRNA and nonspecific siRNA. Cells were collected 48 h later for preparation of cytosol and nuclear proteins, and Western analysis was carried out using antibodies to c-Myc and actin.
FIG. 6. Suppression of endogenous B23 expression abrogates Gadd45a-induced cell cycle G 2 -M arrest. A, effect of B23 siRNA on Gadd45a-inducible expression. HCT116 Gadd45a-inducible cells were established as described previously. Cells were placed in 100-mm dishes at a density of 4 ϫ 10 5 and grown in DMEM containing tetracycline at a concentration of 2 g/ml. After withdrawal of tetracycline, cells were transfected with either B23 siRNA or nonspecific RNA and harvested at the indicated time points. Whole cell lysates were isolated and followed by immunoblotting analysis with antibody to B23 and Gadd45a. B, knockdown of B23 protein results in abrogated Gadd45a-induced cell cycle G 2 -M arrest. HCT116 Gadd45a-inducible cells were growing in DMEM with 10% fetal bovine serum in the presence of tetracycline at a concentration of 2 g/ml. After withdrawal of tetracycline, cells were transfected with B23 siRNA, collected 36 h later, and subjected to flow cytometric analysis as described under "Experimental Procedures." C, inhibition of B23 affects mitotic entry after inducible expression of Gadd45a protein. HCT116 Gadd45a-inducible cells were grown in medium with tetracycline (2 g/ml) and treated with nocodazole for 24 or 36 h. Upon the withdrawal of tetracycline, cells were treated with B23 siRNA and followed by determination of mitotic indices at the indicated time points. measurement of the mitotic index, nocodazole, a microtubule disrupter, was included in the experiments. In Fig. 6C, high mitotic indices were observed in HCT116 cells treated with nocodazole for 24 or 36 h. After withdrawal of tetracycline, mitotic indices substantially decreased, indicating that inducible expression of Gadd45a protein arrests cells in the G 2 -M transition. Addition of B23 siRNA at a concentration of 40 pmol was shown to greatly attenuate the Gadd45a-induced G 2 -M arrest. In contrast, nonspecific siRNA did not present a significant effect on Gadd45a-induced cell cycle G 2 -M arrest. Taken together with the results obtained from flow cytometric analysis, disruption of cellular B23 expression affected the G 2 -M accumulation by Gadd45a, suggesting that nuclear localization of Gadd45a protein be required for Gadd45a-induced cell cycle arrest.

DISCUSSION
The studies presented in this report further demonstrate that the stress-inducible gene Gadd45a is a nuclear protein despite the fact that Gadd45a does not contain a classical NLS. Most interestingly, Gadd45a nuclear localization was shown to be enhanced by DNA damage and is mediated through the interaction of Gadd45a with B23 proteins (nucleophosmin), which act as carrier proteins for nuclear import of certain proteins, indicating that Gadd45a nuclear translocation utilizes a distinct pathway from classical nuclear import machinery. By using Myc-tagged Gadd45a deletion fusion proteins, the B23-binding domain was mapped at the central region of the Gadd45a protein. This B23-binding motif was required for Gadd45a nuclear translocation because disruption of this domain abolished Gadd45a nuclear localization. In support of these observations, suppression of endogenous B23 proteins via employment of B23 siRNA substantially affected Gadd45a nuclear localization and greatly abrogated Gadd45a-induced cell cycle G 2 -M arrest.
Gadd45a is one of the p53-regulated genes (3,15) and has been implicated in maintaining genomic fidelity (25). Deletion of the endogenous Gadd45a gene results in severe genomic instability, including aneuploidy, chromosomal aberrations, gene amplification, centrosome amplification, and increased carcinogenesis induced by DNA damage such as ionizing radiation and UV radiation (25)(26)(27). Previous reports from our group and others have demonstrated that Gadd45a is one of the important components involved in the control of cell cycle G 2 -M arrest. Cells with disrupted Gadd45a reveal an impaired G 2 -M arrest after treatment with certain DNA-damaging agents such as UV radiation or MMS (8). Induction of Gadd45a by either microinjection or via the Tet-Off system causes cells to arrest at G 2 /M phases (8,11). In further support of these findings, MEFs derived from Gadd45a knock-outs exhibited a defect in UV-activated G 2 -M arrest. These Gadd45a-deficient MEFs also presented a weak inhibition of Cdc2/cyclin B1 kinase activity after UV radiation treatment (Fig. 1, A and B). Taken together with the observations that inducible expression of Gadd45a resulted in accumulation of G 2 /M population (Fig.  1, C and D), Gadd45a acts as a critical player in DNA damageactivated cell cycle G 2 -M arrest.
With regard to the molecular mechanism(s) by which Gadd45a functions in the control of the G 2 -M checkpoint, we have previously reported that Gadd45a interacts with Cdc2, dissociates the Cdc2-cyclin B1 complex, and inhibits Cdc2/cyclin B1 activity (9). Induction of Gadd45a leads to a reduction of nuclear cyclin B1 protein, whose nuclear localization is necessary for the completion of the G 2 -M transition (11). The Gadd45a-altered cyclin B1 nuclear localization correlates with suppression of Cdc2-cyclin B1 activity (10). These findings strongly suggest that Gadd45a nuclear translocation is critical for its role in cell cycle G 2 -M arrest. Because of the lack of a nuclear localization signal, the mechanism that mediates Gadd45a nuclear translocation remains unknown. In the current study, Gadd45a was demonstrated as a nuclear protein.
After exposure to DNA-damaging agents, HCT116 cells displayed Gadd45a protein accumulation in both cytosol and nuclear compartments ( Fig. 2A). Transfection of exogenous Myc tag-Gadd45a expression vectors into HCT116 cells also exhibited nuclear localization of Gadd45a (Fig. 2B). Additionally, Gadd45a nuclear localization was further confirmed by the evidence that green fluorescent Gadd45a fusion protein was observed primarily in the nucleus after introduction of a pGFP-Gadd45a vector into HCT116 cells (results not shown).
As discussed earlier, nuclear translocations of a large number of proteins are mediated through their NLS or M9 domains. However, there are NLS-independent and importin-independent nuclear import mechanisms for certain nuclear proteins such as ␤-catenin, a key member in the Wnt signaling pathway (31,32). Additionally, B23 protein has been found to play an important role in carrying some proteins into the nucleus (36 -40). Most interestingly, Gadd45a nuclear localization was also found to be mediated via its physical interaction with B23 protein (Fig. 5). These findings demonstrate that B23 is also involved in the nuclear translocation of certain DNA damageinducible proteins, suggesting a role for B23 protein in cellular response to genotoxic stress. Most importantly, B23-mediated Gadd45a nuclear localization was clearly shown to be required for Gadd45a-induced cell cycle G 2 -M arrest. In the presence of B23 siRNA, which strongly suppresses endogenous B23 expression, HCT116 cells exhibited a disrupted cell cycle G 2 -M arrest after inducible expression of Gadd45a protein, as reflected by reduced G 2 /M accumulations and increased mitotic index (Fig. 6).
In addition to its association with B23, Gadd45a has been found to interact with several important cellular proteins such as Cdc2 kinase, proliferating cell nuclear antigen, core histone protein, p21 WAF1/CIP1 , and MTK/MEKK4 (9, 19 -24), which play important roles in the control of cell cycle progression, DNA repair, and the regulation of signaling transduction pathways. The B23-binding domain was mapped at the central region of the Gadd45a protein. Most interestingly, this region is also critical for the interaction of Gadd45a with Cdc2 and for inhibition of Cdc2/cyclin B1 kinase activity (10). Deletion of the central region substantially abolished the capability of Gadd45a for inducing cell cycle arrest and suppressing cell growth. In addition, the p21 WAF1/CIP1 -interacting motif is localized at the central part of the Gadd45a protein as well. Collectively, the central region of Gadd45a appears to mediate the cross-talks between Gadd45a and other cell cycle regulators and coordinates the function of Gadd45a in the cellular response to genotoxic stress. In conclusion, we have made a novel observation that B23 protein, as a carrier protein for nuclear import of certain proteins, is involved in Gadd45a nuclear translocation and that B23-mediated Gadd45a nuclear translocation is required for Gadd45a-activated cell cycle G 2 -M arrest after DNA damage.