A Novel Nuclear Interactor of ARF and MDM2 (NIAM) That Maintains Chromosomal Stability^*

ARF, the alternative reading frame product of the INK4a/ARF locus (1), is a potent tumor suppressor and key activator of p53 (2). Normal cells typically do not express ARF, but it is dramatically induced by activated oncoproteins such as RAS and MYC (2). Once expressed, ARF protects cells against oncogenic transformation by stimulating p53-mediated expression of apoptotic and growth inhibitory genes (3-5). Together, p53 and ARF are two of the most commonly inactivated genes in human cancers (6, 7). Notably, ARF also acts through other regulators besides p53 to suppress cancer, as revealed by its ability to inhibit the growth of cells lacking functional p53 (8-12) and to limit tumorigenesis in p53-null mice (8). The fact that ARF and p53 are both inactivated in a significant percentage of human tumors also suggests distinct tumor suppressor activities for each protein (13-15).

Our understanding of ARF signaling is incomplete but improving as more ARF-associated proteins are identified (for review, see Ref. 16). It is well established that ARF induces p53 transcriptional activity by binding and inactivating MDM2 (mouse double minute-2; HDM2 in humans) (2). MDM2 is a ubiquitin-protein isopeptide ligase (E3)³ and p53 transcriptional target that acts in a negative autoregulatory feedback loop to restrict p53 activity (17). Although the majority of ARF resides in nucleoli, accumulating evidence suggests that an active fraction of ARF stimulates p53 by binding MDM2 in the nucleoplasm (18). In agreement with this idea, recent work shows that ARF is bound and sequestered in the nucleolus by nucleophosmin (also called B23), thereby inhibiting ARF-p53 signaling by restricting ARF-MDM2 association in the nucleoplasm (19-22). Additional complexity to ARF-p53 signaling is suggested by the fact that ARF associates with several other proteins that directly regulate p53 (23-27). Notably, ARF also activates a delayed, less potent, p53-independent cell cycle arrest. The regulation of this pathway is not well understood but is presumably mediated by the individual or combined actions of many different ARF-binding proteins (16). Recent studies imply important roles for such diverse factors as ARF-binding protein-1 (27), nucleophosmin (28-30), MYC (31, 32), NF-κB (33, 34), and ATM (ataxia telangiectasia mutated)/ATR (ATM- and Rad3-related) (34, 35), highlighting the pleiotropic activities of ARF.

This study explores a new ARF-associated protein, NIAM (nuclear interactor of ARF and MDM2), which we show has intrinsic anti-proliferative activity that intersects with the ARF-MDM2-p53 pathway. NIAM can relocalize ARF into the nucleoplasm, stabilizes and activates p53, and is down-regulated by MDM2-mediated ubiquitination and degradation. Interestingly, NIAM normally helps maintain chromosomal stability. This is significant because a defining feature of most cancers is genetic instability, and NIAM is down-regulated in a variety of human tumors. Together, these findings suggest that NIAM may be a novel modifier or suppressor of tumorigenesis.

MATERIALS AND METHODS

Cell Culture—Most cell types were maintained in standard Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 2-4 mm glutamine, and 100 μg/ml penicillin/streptomycin, including human 293T; PANC-1, and MIA PaCa-2 pancreatic carcinoma (both provided by Joseph Cullen, University of Iowa); U2OS osteosarcoma; U2OS-derived, Narf6-expressing, isopropyl β-d-thiogalactopyranoside (IPTG)-inducible p14^ARF (provided by Gordon Peters, London Research Institute); MCF-7 breast adenocarcinoma (provided by Lori Wallrath, University of Iowa); and mouse NIH 3T3 fibroblasts. Primary mouse embryonic fibroblasts (MEFs; wild-type and p53^-/- (provided by Tyler Jacks, Massachusetts Institute of Technology) and p53^-/-/Mdm2^-/- and p53^-/-/Mdm2^-/-/Arf^-/- (provided by Gerry Zambetti, St. Jude Children's Research Hospital)) were grown in the same standard medium supplemented with 0.1 mm nonessential amino acids and 55 μm 2-mercaptoethanol, whereas MIA PaCa-2 cell medium was supplemented with 2.5% horse serum. Human MDA-MB-231 breast cancer (provided by Lori Wallrath), 22Rv1 prostate cancer (provided by Michael Henry, University of Iowa), and BxPC-3 pancreatic cancer (provided by Joseph Cullen) cells were grown in RPMI 1640 medium supplemented with 10% fetal bovine serum, 2 mm glutamine, and 100 μg/ml penicillin/streptomycin, with additional 2.5 g/liter d-glucose, 10 mm HEPES, and 1 mm sodium pyruvate for BxPC-3 cells. Cells were treated with 50 μm MG132 (Calbiochem) for 3 h to inhibit the proteasome.

DNA Constructs—Mouse and human wild-type NIAM cDNAs were cloned by reverse transcription (RT)-PCR from normal liver and pancreas cDNA pools (Clontech), respectively. Primers containing ClaI and EcoRI restriction enzyme sites were used to amplify the murine (forward, 5′-CCATCGATAGCGTGTTGAGCGGCCTGGCC-3′; and reverse, 5′-GGAATTCAATCTGAAGACTGAATTG-3′) and human (forward, 5′-ACATCGATACCATGAGCCTGCTGGACG-3′; and reverse, 5′-GGAATTCAATCTGAAGACTGAATTGGGC-3′) NIAM cDNAs (Integrated DNA Technologies, Coralville, IA). PCR products were directly ligated into the pCRIITOPO vector (Invitrogen); sequenced; and subcloned into various expression vectors, including pXM-HA (36), pSRα-MSCV-IRES-GFP (provided by Martine Roussel, St. Jude Children's Research Hospital), pGEX-4T-2 (Amersham Biosciences), and pcDNA3.1/CT-GFP and pcDNA3 (Invitrogen). Expression plasmids encoding wild-type and mutant forms of ARF and MDM2 have been described previously (9, 37, 38).

Protein Expression—Bacterial glutathione S-transferase (GST)-tagged fusion proteins were generated as described previously (39). For mammalian cell expression, cells were transfected using FuGENE 6 (Roche Applied Science) or a modified calcium phosphate precipitation procedure (40). Production of ecotropic retroviruses in 293T cells and infections of mouse fibroblasts were performed as described (1, 41).

RT-PCR Analyses—NIAM mRNA expression was measured in human tissue cDNA panels (Clontech) by semiquantitative PCR for 34 cycles with the same primers used to clone fulllength NIAM (see above). As a control, the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase was amplified for 22 cycles using the forward primer 5′-TGAAGGTCGGAGTCAACGGATTTGGT-3′ and the reverse primer 5′-CATGTGGGCCATGAGGTCCACCAC-3′. For quantitative assays, real-time RT-PCR was performed with total cellular RNA that was isolated using an RNeasy mini kit (Qiagen Inc., Valencia, CA) and quantified using a BioPhotometer (Eppendorf, Westbury, NY). Reverse transcription of NIAM mRNA was performed with 1 μg of total RNA using a high capacity cDNA archive kit (Applied Biosystems, Foster City, CA) as specified by the manufacturer. Subsequent PCRs consisted of 40 ng of cDNA added to 12.5 μl of SYBR Green PCR Master Mix (Applied Biosystems), 1.25 μl (0.6 μm) of gene-specific NIAM primers (forward, 5′-CAGGATGAAAAAGCTCCCGAA-3′; and reverse, 5′-GCAGCATTTTCAAACACCGTG-3′), and PCR-grade water to a total reaction volume of 25 μl. PCR was performed as follows: 50 °C for 2 min, 95 °C for 10 min, and 40 cycles at 95 °C for 15 s, with annealing and extension at 60 °C for 1 min on an ABI 7000 real-time sequence detection system (Applied Biosystems). -Fold differences in NIAM mRNA expression were calculated using NIAM expression calibrated to 18 S ribosomal RNA expression and computed using ABI relative quantitation software (Applied Biosystems).

In Vitro Binding Assays—Coupled in vitro transcription and translation of plasmids containing human or mouse NIAM were performed using the TnT kit (Promega Corp., Madison, WI). ³⁵S-Labeled proteins were incubated for 2 h at 4 °C with equivalent amounts of GST or GST-human ARF proteins that were previously bound to glutathione-Sepharose. Protein complexes were washed four times with Nonidet P-40 buffer (50 mm Tris (pH 8.0), 120 mm NaCl, 1 mm EDTA, 0.5% Nonidet P-40, and 0.1 mm Na₃VO₄) supplemented with 1 mm NaF, 10 μm leupeptin, and 30 μm phenylmethylsulfonyl fluoride and resolved by SDS-PAGE. Gels were fixed with 30% methanol and 10% acetic acid, stained with Coomassie Blue, and dried. Autoradiography and phosphorimaging (GE Healthcare) were used to detect protein interactions.

Immunoprecipitation and Western Blot Analyses—Frozen cell pellets were lysed on ice in Nonidet P-40 buffer and briefly vortexed, and lysates were incubated on ice for 15 min prior to sonication (1 × 5-s pulse) and clarification by centrifugation at 14,000 rpm for 10 min at 4 °C. Immunoprecipitations were performed using protein A- or G-Sepharose plus antibodies against mouse p19^ARF (1), human p14^ARF (Sigma, mouse monoclonal, clone DCS-240, 2 μg/immunoprecipitation), the hemagglutinin (HA) epitope (Roche Applied Science, rat monoclonal, clone 3F10, conjugated to agarose, 10 μl/immunoprecipitation), NIAM (rabbit polyclonal against a C-terminal peptide of NIAM (NH₂-CLKSPSQASPIQSSD-COOH), referred to as NIAM-C, 10 μg/immunoprecipitation), and p16^INK4a (Santa Cruz Biotechnology, Inc. (Santa Cruz, CA), rabbit polyclonal, catalog no. sc-1207, 2.5 μg/immunoprecipitation). Immune complexes were washed four times with Nonidet P-40 buffer, separated by SDS-PAGE, transferred to polyvinylidene difluoride membranes, and analyzed by immunoblotting. Equivalent amounts of total cellular protein (50-100 μg/lane) were loaded on gels as a control for relative protein expression levels.

Proteins were detected on membranes by ECL (Amersham Biosciences) with antibodies against mouse p19^ARF (Oncogene Research Products (San Diego, CA), rabbit polyclonal, Ab-1, 0.7 μg/ml), human p14^ARF (2 μg/ml), MDM2 (monoclonal, clone 2A10, 1:50 dilution; and Oncogene Research Products, mouse monoclonal, Ab-1, 1 μg/ml), p53 (Oncogene Research Products, sheep polyclonal, Ab-7, 1:2,500 dilution), anti-p21 (Pharmingen, mouse monoclonal, clone 6B6, 1.5 μg/ml), green fluorescent protein (GFP; Abcam (Cambridge, MA), rabbit polyclonal, ab290, 1:5000 dilution), γ-tubulin (Sigma, mouse monoclonal, clone GTU-88, 1:10,000 dilution), glyceraldehyde-3-phosphate dehydrogenase (Abcam, mouse monoclonal, ab8245, 1:10,000 dilution), HA (conjugated to horseradish peroxidase, 1:1000 dilution), and NIAM-C (1.5 μg/ml).

Colony Formation Assays—Narf6 cells were transfected with the pXM-HA vector or pXM-HA-NIAM (human wild-type) along with one-tenth the amount of GFP (pEGFP-C3, Clontech). Forty-eight hours post-transfection, GFP-positive cells were sorted using a FACSDiva instrument (BD Biosciences) and plated in triplicate at 1 × 10³ cells/well in a 6-well tissue culture dish with or without 1 mm IPTG to induce ARF expression. Dishes were cultured for 14 days before colonies were washed twice with phosphate-buffered saline fixed with methanol, and stained with Giemsa.

Subcellular Localization—Narf6 cells were transfected with the pXM-HA vector or pXM-HA-NIAM (mouse or human wild-type) on glass coverslips in a 6-well dish. One day posttransfection, cells were treated with or without 1 mm IPTG to induce ARF expression, and 16-24 h later, cells were fixed with 4% paraformaldehyde and permeabilized with 0.1% Triton X-100. Human ARF was stained using anti-human p14^ARF antibody (1 μg/ml), biotinylated anti-mouse IgG (Amersham Biosciences, 1:200 dilution), and Texas Red-conjugated streptavidin (Amersham Biosciences, 1:200 dilution). Exogenous HA-NIAM was detected using HA-fluorescein isothiocyanate-conjugated IgG (Roche Applied Science, 1:100 dilution). Nuclei were visualized by staining with 4′, 6-diamidino-2-phenylindole (Sigma) at 1 μg/ml for 1 min, and samples were analyzed by confocal microscopy (Zeiss LSM 510).

Cell Cycle Analyses—Cell cycle progression into S phase was measured by bromodeoxyuridine (BrdUrd) incorporation into replicating DNA (9). Briefly, 10 μm BrdUrd was added for 22-24 h to the medium of cells 1 day post-transfection (3 days post-infection of p53^-/-/Mdm2^-/-/Arf^-/--NIAM short hairpin RNA (shRNA; referred to as shNIAM) or p53^-/-/Mdm2^-/-/Arf^-/--control shRNA (referred to as shCON) MEFs). Cells (3.5 × 10⁴/well in 8-well poly-l-lysinecoated chamber slides; BD Biosciences) were fixed in methanol/acetone (1:1) at -20 °C, treated with 1.5 n HCl, and sequentially stained with anti-BrdUrd antibody (Abcam, sheep polyclonal, 8 μg/ml), biotinylated anti-sheep IgG (Amersham Biosciences, 1:200 dilution), Texas Red-conjugated streptavidin (1:200 dilution). Exogenous HA-NIAM and nuclei were stained as described above, and the BrdUrd positivity of NIAM expressors was quantified by counting under a Zeiss fluorescence microscope. The statistical significance of the data was analyzed by a unpaired, two-tailed, equal variance Student's t test. To determine the cell cycle distributions of GFP-positive and GFP-negative cells after transfections with GFP, GFP-NIAM, and GFP-ARF constructs, cells were sorted on a FACS-Diva instrument, stained with propidium iodide, and assayed for DNA content using a FACScan (BD Biosciences) as described (24). Flow cytometric profiles were analyzed using ModFit LT Version 2.0 software.

p53 Reporter Assays—A p53 reporter construct containing p53-responsive promoter elements fused to the firefly luciferase gene (Stratagene, La Jolla, CA) was cotransfected into U2OS cells with varying amounts of pXM-HA-NIAM with or without pcDNA3-p14^ARF (human; 3 μg) plus varying amounts of empty pXM-HA for a constant amount of DNA (12 μg) per transfection. A pRL-SV40 construct containing Renilla luciferase (Promega, 80 ng) was included in all transfections to normalize for transfection efficiency. Luciferase activity was measured in duplicate or triplicate from at least three independent experiments 48 h post-transfection using the Dual-Luciferase reporter assay system (Promega Corp.) and a Sirius V3.1 luminometer (Berthold Detection Systems, Pforzheim, Germany).

Ubiquitination Assays—In vivo ubiquitination of HA-NIAM was measured in U2OS cells cotransfected with His-tagged ubiquitin (provided by Mary Horne, University of Iowa) and pXM-HA or pXM-HA-NIAM plus empty vector, pcDNA3-HDM2 (wild-type), or an HDM2 mutant (Ala^466-473; provided by Karen Vousden, Beatson Institute for Cancer Research, Glasgow, Scotland, UK). Two days post-transfection, cells were treated with or without proteasome inhibitors (clasto-lactacystin β-lactone at 5 μm for 4 h or MG132 at 50 μm for 3 h). Cells were lysed in 6 m guanidinium chloride, 0.1 m Na₂HPO₄/NaH₂PO₄, and 0.01 m Tris-HCl (pH 8.0) plus 5 mm imidazole and 10 mm β-mercaptoethanol. Lysates were sonicated, and ubiquitinated proteins were bound for 2 h at room temperature on 50 μl of nickel-nitrilotriacetic acid-agarose (prewashed with the lysis buffer) as described (42, 43). Samples were washed extensively as described (43), and His-ubiquitin-conjugated proteins were eluted with 200 mm imidazole in 5% SDS, 0.15 m Tris-HCl (pH 6.7), 30% glycerol, and 0.72 m β-mercaptoethanol. Eluates were resolved by SDS-PAGE, and proteins were detected by immunoblotting with anti-NIAM antibody (both anti-HA and anti-NIAM-C antibodies were used separately to obtain confirmatory results).

RNA Interference—NIAM (shNIAM, 5′-ACTGGAAGTTCTGAAGAAA-3′) and control (shCON, 5′-ACTGGAAGTGCTGAAGAAA-3′) target shRNAs were designed using OligoEngine targeting software (www.oligoengine.com) according to criteria established by Brummelkamp et al. (44). Complementary oligonucleotides (Integrated DNA Technologies) were annealed (3 μg each in 100 mm potassium acetate, 30 mm HEPES-KOH (pH 7.4), and 2 mm magnesium acetate) and cloned into pSUPER.retro.neo.GFP (OligoEngine). Stable polyclonal populations of NIH 3T3 cells and primary MEFs expressing the NIAM knockdown (shNIAM) or control point mutant (shCON) shRNA targeting constructs were generated by transfection or retroviral infection, respectively, followed by cell sorting for GFP-positive cells on a FACSDiva instrument. Stable NIH 3T3 populations were further selected in 0.8 mg/ml Geneticin (G418) for 2 weeks and maintained in 0.4 mg/ml G418 thereafter.

Metaphase Analyses—Metaphase spreads were prepared by standard protocols as described previously (45). Briefly, exponentially growing MEFs (p53^-/-/Mdm2^-/- and p53^-/-/Mdm2^-/-/Arf^-/-) lacking or expressing NIAM were treated at passage 14 with Colcemid (Invitrogen) for 4 h. Cells were then stained with both propidium iodide (Sigma) and 4′, 6-diamidino-2-phenylindole and mounted with VECTASHIELD (Vector Laboratories, Burlingame, CA). Blinded counts were performed on at least 50 separate metaphase spreads by fluorescence microscopy (Nikon, Melville, NY) with MetaVue software (Molecular Devices Corp., Sunnyvale, CA).

RESULTS

Identification of a Novel ARF- and MDM2-interacting Protein, NIAM—To identify novel factors within the ARF signaling pathways, we performed a yeast two-hybrid screen and discovered several new ARF-associated proteins (46). The most frequently identified ARF partner in the screen was NIAM, which we named based on findings presented herein that it is a nuclear interactor of ARF and MDM2. Full-length mouse and human NIAM cDNAs, which have nearly 90% nucleotide homology, were cloned by RT-PCR after identifying identical cDNA sequences within the data bases.⁴ Semiquantitative RT-PCR analyses showed that the NIAM transcript is widely expressed at low levels in most human tissues, with the highest expression observed in the pancreas, lung, and liver (Fig. 1A). When translated in vitro, the cDNAs encoded polypeptides of ∼50 kDa (human NIAM) and 55-60 kDa (mouse NIAM) that bound directly with purified GST-tagged ARF (Fig. 1B). Interestingly, the mouse and human NIAM proteins (84% amino acid identity) are predicted to contain nuclear/nucleolar localization sequences; a lysine-rich region; and two poorly understood domains rich in phenylalanine, tyrosine, and arginine (FYRN and FYRC) thought to mediate protein-protein associations (Fig. 1C) (47).

To examine the association between endogenous NIAM and ARF proteins, we generated a polyclonal antibody against the C terminus of NIAM and screened a large number of cell lines and non-transformed primary cells for NIAM expression. We consistently observed very low to undetectable levels of endogenous NIAM protein expression, particularly in human cancer cell lines (data not shown). Notably, NIAM expression was much higher in MEFs lacking p53 compared with wild-type controls (Fig. 2A). Endogenous NIAM expression was also elevated in immortalized NIH 3T3 fibroblasts compared with primary wild-type MEFs, although these levels were minor in comparison with exogenous HA-NIAM levels in transfected NIH 3T3 cells. We took advantage of the fact that p53^-/- MEFs also express high amounts of ARF (Fig. 2A) and examined endogenous NIAM-ARF complexes (Fig. 2B). In this experiment, cells were infected with retroviruses encoding either GFP vector alone or untagged NIAM plus GFP, and reciprocal immunoprecipitation and immunoblot analyses were performed using antibodies to ARF and NIAM. Complexes between endogenous NIAM and ARF were readily observed (lane 5), and overexpression of NIAM enhanced coprecipitation of the two molecules (lanes 6 and 8). Used as a negative control, the alternative product of the INK4a/ARF locus, p16^INK4a, failed to associate with NIAM (lanes 3-6).

NIAM Is a Ubiquitinated MDM2-binding Protein Maintained at Low Levels by p53 and MDM2—The observation that NIAM levels are increased in p53-null MEFs suggested that NIAM might be down-regulated by the proteasome via MDM2, an E3 ligase and p53 transcriptional target, the expression of which is generally decreased in cells lacking p53 (48). Consistent with this idea, endogenous NIAM expression was markedly stabilized by exposure of wild-type MEFs to the proteasome inhibitor MG132 (Fig. 3A). Interestingly, NIAM levels were also increased by MG132 treatment of p53/Mdm2-null MEFs, suggesting that other E3 ligases besides (or in addition to) MDM2 may regulate NIAM.

To directly test whether MDM2 negatively regulates NIAM expression, U2OS cells were transfected with exogenous HA-NIAM and either wild-type or E3-deficient (Ala^466-473) MDM2 (Fig. 3B). Western blotting of cell lysates (upper panels) showed that wild-type MDM2 caused a dramatic reduction in NIAM expression (lane 4), whereas NIAM levels were unchanged by mutant MDM2 (lane 3). Notably, MDM2-mediated down-regulation of NIAM was partially reversed by MG132 and by ARF, which is known to block MDM2 E3 activity (49). Coexpression of His-tagged ubiquitin in these cells allowed us to test whether MDM2 promotes NIAM ubiquitination (lower panel). An NIAM immunoblot of ubiquitinated proteins collected on nickel-agarose under denaturing conditions showed that NIAM is a ubiquitinated protein (lane 2) and that wild-type MDM2 (lane 4), but not ligase-deficient MDM2 (lane 3), markedly enhances its ubiquitination. These results correlated with coprecipitation of MDM2 and NIAM complexes from cells (Fig. 3C) and direct MDM2-NIAM association in vitro (data not shown). Overall, these results show that NIAM ubiquitination and degradation are promoted by MDM2 and inhibited by ARF.

NIAM Collaborates with ARF to Inhibit Cell Proliferation—The fact that NIAM is expressed at low levels and downregulated by the growth-promoting factor MDM2 suggested that it may be a growth inhibitor. To test this idea, empty vector or HA-tagged wild-type NIAM was introduced with GFP (10:1 NIAM/GFP ratio) into U2OS cells expressing IPTG-inducible ARF (Fig. 4). Successfully transfected, GFP-positive cells were collected by cell sorting, and the long-term proliferative capability of these cells in the presence or absence of ARF was measured in colony formation assays (Fig. 4A). HA-NIAM significantly inhibited U2OS cell proliferation in the absence of ARF compared with the vector control (p = 0.01), although it was clearly a less potent inhibitor than ARF. Notably, combined expression of ARF and NIAM yielded a greater arrest than either protein alone (p = 0.0005), which appeared to be an additive rather than synergistic effect. Confocal microscopy analyses of HA-NIAM and ARF subcellular localization in these cells revealed that NIAM is a nuclear protein (Fig. 4B, panels iv and vii) that redistributes a significant fraction of ARF into the nucleoplasm in 30-50% of cells (panels v and viii and bar graph). Normally, the majority of ARF resides in nucleoli, as seen in control cells lacking NIAM overexpression (panel ii). Thus, increased NIAM expression and NIAM-ARF association coincide with reduced rates of cell proliferation and mobilization of ARF into the nucleoplasm.

NIAM Stabilizes and Activates p53 but Does Not Require It to Inhibit Cell Growth—Recent evidence suggests that nucleoplasmic ARF is more effective in binding MDM2, activating p53, and inhibiting growth compared with nucleolar ARF (18-20). Therefore, we tested whether NIAM collaborates with ARF by promoting its ability to activate p53. We first measured p53 activity in reporter assays and found that combined expression of HA-NIAM and ARF together did not activate p53 above levels induced by ARF alone (Fig. 5A). Unexpectedly, expression of HA-NIAM alone (in these ARF-null cells) stimulated p53 transactivation activity, and it did so in a dose-dependent manner (Fig. 5, A and B). To more fully assess the effects of NIAM on p53, we examined the expression of endogenous p53 and its targets, MDM2 and p21, by immunoblotting U2OS cells expressing exogenous HA-NIAM (Fig. 5C). In agreement with the reporter assays, NIAM stabilized p53 and induced expression of MDM2 and p21, yet its coexpression with ARF caused no apparent enhancement of ARF-mediated p53 stabilization or activation of its target genes.

To determine the specific effect of NIAM on the cell cycle, we expressed GFP-tagged NIAM in U2OS cells, sorted for GFP-positive cells, and measured DNA content by flow cytometry (Fig. 6A). Untransfected GFP-negative cells and controls expressing GFP alone continued to proliferate, whereas cells expressing GFP-NIAM or GFP-ARF (positive control) accumulated in the G₁ phase of the cell cycle. No apoptosis was observed in either NIAM or ARF overexpressors. As expected from earlier results, the G₁ phase arrest induced by GFP-NIAM coincided with p53 stabilization and up-regulation of its target gene, Mdm2 (Fig. 6B). To test whether NIAM requires p53 to inhibit growth, we compared the ability of NIAM to block BrdUrd incorporation in several different cell types that either lack or express functional p53. With the exception of p53^-/- MEFs, all cells also lack ARF. Fig. 6C shows that NIAM effectively blocked the proliferation of cells expressing wild-type p53 (NIH 3T3, U2OS, 22Rv1 prostate cancer, and MCF-7 breast cancer) as well as those expressing mutant p53 (MDA-MB-231 breast cancer and PANC-1 pancreatic adenocarcinoma) or null for p53 (p53^-/- MEFs). These results show that NIAM can activate p53 and collaborate with ARF to restrict proliferation, but it does not require either protein to inhibit DNA synthesis.

NIAM and ARF Act in Separate Anti-proliferative Pathways—One possibility implied by the above data is that NIAM may function downstream of both ARF and p53 but can feedback to regulate their localization and activity. To determine whether NIAM is a downstream component of ARF signaling, we stably knocked down NIAM in NIH 3T3 cells and assessed the ability of ARF to inhibit growth. These cells were chosen because they express detectable levels of the endogenous protein (see Fig. 2). The Western blots in Fig. 7A show effective NIAM knockdown in polyclonal populations expressing an shRNA directed against NIAM (shNIAM) but not in control cells expressing a point mutant shRNA construct (shCON). Loss of NIAM did not alter the basal rate of cell proliferation (negative data not shown) or the ability of ARF to cause cell cycle arrest (Fig. 7B). Indeed, ARF caused a robust G₁ and G₂ phase cell cycle arrest (Fig. 7B) and efficiently induced p21 and MDM2 expression regardless of NIAM levels (Fig. 7A). ARF also effectively arrested p53^-/-/Mdm2^-/-/Arf^-/- MEFs lacking NIAM (Fig. 7C). These results show that NIAM is nonessential for ARF-mediated growth inhibition, whether it be p53-dependent or p53-independent. Combined with earlier findings, these data suggest that ARF and NIAM function in separate anti-proliferative pathways that intersect mechanistically and exert additive growth-suppressive effects when concomitantly activated.

NIAM Helps Maintain Genomic Stability—Chromosomal instability (CIN) is characterized by abnormal number (aneuploidy) and structures of chromosomes and is a significant feature of advanced stages of cancer (50, 51). CIN is associated with the missegregation of chromosomes during mitosis (52, 53), which can result from loss of p53 (54). During the course of our experiments, we noticed that triple knock-out (TKO) p53^-/-/Mdm2^-/-/Arf^-/- MEFs stably expressing the NIAM knockdown construct (shNIAM) displayed elevated tetraploidy that accumulated over time in culture compared with identically passaged control (shCON) populations (Fig. 8, A and B). Specifically, 85% of passage 14 TKO shNIAM cells exhibited tetraploidy compared with 63% of shCON cells (Fig. 8B). It should be noted that p53-null MEFs, such as TKO cells, are known to have high basal levels of tetraploidy such as that seen in early passage 4 cells. Because tetraploidy (an exact duplication of all chromosomes) often precedes development of a highly aneuploid state because of subsequent gains or losses of individual chromosomes (55), we tested whether NIAM loss in higher passage populations increased the number of cells with >4N DNA content (Fig. 8C). For these experiments, similarly passaged double knock-out (DKO) p53^-/-/Mdm2^-/- MEFs stably expressing shNIAM or shCON constructs (Fig. 8A) were also included to determine whether the expression of endogenous ARF alters the effects of NIAM loss. NIAM silencing caused increased aneuploidy in both DKO and TKO shNIAM cells; however, the percentage of polyploid cells was remarkably higher in ARF-null TKO populations compared with ARF-positive DKO cells (Fig. 8C).

The increased aneuploidy observed in shNIAM populations correlated with a similar rise in the number of abnormal chromosomal structures (Fig. 8D). The major chromosomal aberrations that were observed in these cells included detached centromeres, small chromosomal fragments, and chromosomal or chromatid breaks (Table 1). And once again, the frequency of chromosome alterations caused by NIAM knockdown was reduced in DKO compared with TKO cells (Fig. 8D and Table 1). Although rare, chromosome exchanges were also seen in some shNIAM cells, whereas none were observed in shCON cells, as illustrated in an example metaphase spread from TKO shNIAM cells (Fig. 8E). Altogether, we found that reduced NIAM expression enhanced both numerical and structural CIN and that this effect was exacerbated in an ARF-null setting. These results strongly support a role for both NIAM and ARF in the preservation of genetic stability.

NIAM mRNA Expression Is Decreased in Human Tumors—The prime importance of ARF-p53 signaling to tumor suppression is well established (2). Thus, the functional link of NIAM with these regulators, its own distinct growth inhibitory activity, and its ability to reduce CIN imply that it may also be a tumor suppressor. Consistent with that notion, data base searches (including the cancer microarray data base called ONCOMINE at www.oncomine.org) (56) revealed that NIAM mRNA expression is significantly reduced in many human cancers (Table 2), including carcinomas of the lung (57, 58), liver (59), breast (58, 60), pancreas (61), and kidney (62) as well as chronic lymphocytic leukemia and diffuse large B-cell lymphoma (63). For these analyses, the original GenBank™ sequence homologous to human NIAM (Gene ID 84897)⁴ was used to search the data bases. Interestingly, most of the cancers with reduced NIAM mRNA represent advanced epithelially derived tumors that are frequently not detected in patients until late stages of tumor development. NIAM mRNA is also more highly down-regulated in nodepositive, high-grade breast cancer (60).

To begin validating the microarray findings, we examined NIAM mRNA expression by quantitative real-time RT-PCR in a small panel of pancreatic adenocarcinoma cell lines. NIAM mRNA levels were in fact down-regulated in the tumorderived cells compared with the normal pancreatic epithelial cell line H6c7 (Fig. 9). These data, the microarray results, and our inability to detect NIAM protein in most human cancer cell lines examined suggest that loss of NIAM may be an essential step during the development and/or progression of human tumors.

DISCUSSION

This study represents the initial characterization of NIAM, a novel growth inhibitory protein we discovered in a search for new ARF-binding partners (46). Our data suggest that NIAM and ARF act in separate but convergent anti-proliferative pathways that are required for maintaining chromosomal stability. On the basis of these findings and the fact that NIAM mRNA expression is down-regulated in many different types of human cancers, we speculate that NIAM may be a new modifier or suppressor of tumorigenesis.

The mechanistic links between NIAM and the ARF-MDM2-p53 pathway are numerous (Fig. 10). The most obvious connection is the direct physical association between NIAM and ARF or between NIAM and MDM2, the functional consequences of which are quite distinct. MDM2 binds and negatively regulates NIAM by promoting its ubiquitination and proteasomal degradation, greatly restricting the expression of endogenous NIAM in cells. Conversely, ARF can block the effects of MDM2 and at least partially stabilize NIAM, perhaps by inhibiting the E3 activity of MDM2 (49) or by competing with MDM2 for NIAM binding. Ultimately, a negative feedback loop involving ARF-p53 activation and up-regulation of MDM2 is predicted to return NIAM to its low basal level of expression.

Maintenance of NIAM at low levels may be necessary for cell viability because NIAM can stabilize and activate p53. How NIAM stimulates p53 has not yet been defined. NIAM binding to MDM2 could limit MDM2-p53 association and thereby facilitate p53 stabilization, or NIAM could directly inhibit the ability of MDM2 or other ubiquitin ligases (COP1, PIRH2, and ARF-binding protein-1/MULE) to target p53 for degradation (27, 64, 65). The evidence supporting either possibility is currently circumstantial, but it may be relevant that NIAM is targeted by (and thus likely associates with) other E3 ligases besides MDM2. Notably, ARF is not required for any of these events because NIAM activates p53 and is degraded by MDM2 in ARF-null U2OS cells. This situation is reminiscent of cyclin G₁, another cell cycle regulator that binds ARF and MDM2, transiently activates p53 in an ARF-independent manner, and is down-regulated by MDM2 (24). Unlike cyclin G₁, however, NIAM does not appear to be a p53-responsive gene.⁵

NIAM also influences ARF function. When introduced into ARF-expressing cells, NIAM binds ARF, redistributes a significant fraction of it into the nucleoplasm, and enhances ARF-mediated growth suppression in long-term assays. The ability of NIAM to mobilize ARF into the nucleoplasm, where it effectively activates p53 (18, 19), suggests that it might collaborate with ARF by transiently amplifying or sustaining ARF-p53 signaling. We favor the latter possibility because altered NIAM expression (either knockdown or overexpression) had no apparent impact on transient ARF-mediated p53 activation or cell cycle arrest, whereas it did enhance growth suppression by ARF in colony formation assays. In both settings, the balance between ARF and MDM2 levels is expected to dictate the magnitude and timing of the effects of NIAM on p53, so comprehensive time course studies may be required to determine how NIAM impacts the longevity of ARF-p53 signaling. Given that persistent rather than transient stimulation of p53 is vital for p53 tumor suppressor activity following DNA damage and that this long-term response is entirely dependent on ARF (66), NIAM may provide substantial protection against cancer if it does in fact sustain ARF-p53 signaling.

Notably, NIAM does not require either ARF or p53 to inhibit DNA synthesis. When considered with the NIAM knockdown data, such results strongly suggest that NIAM and ARF act in separate anti-proliferative pathways that converge because of their association and regulation of the common target proteins p53 and MDM2. The question is what is the physiological significance of NIAM to ARF and p53 signaling and vice versa? Unfortunately, teasing apart the individual contributions of the numerous regulators that impact ARF signaling, such as NIAM, is complicated because ARF has so many weapons in its arsenal to suppress proliferation (16). Consequently, loss of one effector or regulator typically has little effect. This is especially true in cells expressing wild-type p53, where ARF activates a maelstrom of p53-responsive anti-proliferative genes. For instance, although p21 is normally a dominant factor in mediating ARF-induced arrest, it is dispensable because of the compensatory actions of other p53-activated genes that exert a similar albeit distinct cell cycle arrest (41, 67). Consistent with this result, homozygous deletion of p21 has no effect on tumor latency or life span in ARF-deficient mice, but its loss does alter the spectrum of tumors that develop (68). Such data suggest that p21 acts more as a tumor modifier than as a classical tumor suppressor, similar to another p53 target gene, bax (69). The same may be true for NIAM because it influences but is not essential for ARF signaling, and future genetic studies will address this possibility.

The fact that NIAM mRNA is highly expressed in the pancreas, liver, and lung predicts important biological roles in those cell types. Indeed, NIAM mRNA expression is dramatically reduced in hepatocellular carcinomas and lung adenocarcinomas (58, 59), and real-time RT-PCR data confirmed its down-regulation in several pancreatic cancer cell lines. NIAM may also have an essential function in blood cells. Microarray studies show that its transcript is most highly expressed in normal peripheral blood cells compared with other tissues (58), whereas it is markedly reduced in diffuse large B-cell lymphomas and chronic lymphocytic leukemias compared with benign lymphoid tissue (Table 2). Given its low basal level of expression and ability to activate p53, particular developmental signals and/or cellular stresses may normally regulate NIAM expression in these tissues. In support of this idea, NIAM was previously identified as one of many transforming growth factor-β1-responsive genes (70), and we found that it is induced in transforming growth factor-β1-arrested WEHI231 B-cells.⁶ Notably, transforming growth factor-β1 is a key regulator of development and tumorigenesis (71, 72).

Nearly all cancers exhibit aneuploidy, a hallmark of CIN that is largely caused by defective mitotic spindle checkpoints (55, 73). We found that loss of NIAM in primary MEFs markedly enhanced both numerical (aneuploidy) and structural CIN, supporting an important role for NIAM in maintaining genetic stability. Although neither ARF nor p53 was required for this activity, the loss of ARF further exacerbated the genetic instability initiated by loss of NIAM. Thus, both NIAM and ARF participate in stabilizing the genome, providing insight into the potential physiological significance of their association. Other work has indicated a role for ARF in maintaining genomic stability. Khan et al. (74) originally showed that microtubule disruption causes increased polyploidy in Arf^-/- MEFs, suggesting that ARF is at least partially required for proper mitotic checkpoint control. Indeed, ARF mediates a DNA damage-induced G₂/M phase checkpoint through activation of ATM (35), which is intriguing because ATM is critical for suppressing CIN and because its loss defines a genomic instability syndrome in humans (75). Most recently, Wang et al. (45) showed that the acquisition of chromosomal aberrations in ARF-null MEFs could be reduced by decreased expression of MDM2. This implies a role for MDM2 in promoting CIN, which may arise from its ability to inhibit DNA double-strand break repair (76). Because MDM2 down-regulates NIAM and because NIAM loss accelerates CIN, this provides yet another provocative connection between these regulators and cancer.

This study has described a novel growth inhibitor that helps maintain chromosomal stability and intersects with ARF-MDM2-p53 signaling at multiple levels. Future studies will be pivotal in validating the down-regulation of NIAM in human tumors and establishing whether NIAM is in fact a new suppressor of tumorigenesis.