Cooperative Role of the RNA-Binding Proteins Hzf and HuR in p53 Activation

Abstract

The RNA-binding protein Hzf (hematopoietic zinc finger) plays important roles in mRNA translation in cerebellar Purkinje cells and adipocytes. We along with others have reported that the expression of the Hzf gene is transcriptionally regulated by the p53 tumor suppressor protein. We show here that Hzf regulates p53 expression in cooperation with HuR. Hzf and HuR independently interact with the 3′ untranslated region (UTR) of p53 mRNA, which facilitates the cytoplasmic localization of p53 mRNA in the presence of the ARF tumor suppressor protein. In the absence of Hzf and HuR, p53 induction by p19^ARF is significantly attenuated, and the cells consequently acquire resistance to p19^ARF. Thus, these findings demonstrate that in addition to Mdm2 inhibition, p19^ARF increases the concentration of p53 through posttranscriptional control of p53 mRNA and suggest critical roles for the RNA-binding proteins Hzf and HuR in p53 induction.

INTRODUCTION

The p53 tumor suppressor gene is frequently mutated in human cancers (11). p53 is activated in response to various cellular stresses and triggers a program that induces cell cycle arrest, apoptosis, or cellular senescence, thereby eliminating cells that have suffered irreparable damage (28, 35). Tumor-derived p53 mutations compromise the transcriptional activity of p53, allowing cells that have sustained oncogenic damage to survive. Thus, the inability to activate p53 in response to DNA damage or oncogene signals contributes to the generation of genetically unstable cells, which can lead to malignancy.

p53 is a labile protein, and its level is low in unperturbed cells. The induction of cell cycle arrest or apoptosis in response to stress requires an increase in the intracellular level of functional p53 protein. The E3 ubiquitin ligase Mdm2 (HDM2 in humans) interacts with p53 to facilitates its rapid proteasomal degradation (3, 12, 15, 25). Aberrant growth signaling caused by oncogenes induces the expression of the tumor suppressor protein p19^ARF (p14^ARF in humans), which binds to Mdm2 and antagonizes its E3 ubiquitin ligase activity, thereby stabilizing p53 protein (30). In contrast, DNA damage resulting from single- or double-strand breaks activates ATM/ATR family protein kinases (18). These kinases induce p53 or Mdm2 phosphorylation directly or indirectly through other kinases (Chk1/Chk2), which are phosphorylated and activated by ATM/ATR (4). Phosphorylation of p53 and Mdm2 interferes with their interaction and results in decreased proteasomal degradation of p53 protein. Several other proteins, including the ribosomal proteins L11 and L23 and the nucleolar protein NPM/B23 have been shown to contribute to p53 activation by inhibiting Mdm2 function (17, 27, 29, 43).

In addition to posttranslational regulation of p53 by Mdm2, it has recently become evident that translational regulation also contributes to p53 induction. The accumulation of p53 protein after DNA damage is inhibited by protein synthesis inhibitors, and the de novo synthesis of p53 protein is dramatically increased in response to stress (19, 20). The 5′ untranslated region (UTR) of p53 is predicted to form a stem-loop structure that is able to inhibit p53 translation, and p53 directly binds to this region to repress its own translation (33). Recently, Takagi et al., using a yeast three-hybrid screening assay and an RNA pulldown assay (39), reported that several proteins associate with the p53 5′ UTR. As a result, RPL26 was found to enhance p53 mRNA translation in both the absence and presence of DNA damage while nucleolin was found to repress its translation. Besides its 5′ UTR, p53 translation is also regulated through its 3′ UTR. HuR (Hu antigen R) binds to the AU-rich element (ARE) of the p53 3′ UTR and increases p53 translation in cells exposed to UV radiation (32).

Hzf was originally identified as a gene whose expression is specifically induced in hematopoietic precursor cells derived from differentiating embryonic stem cells (13). Consistent with its expression in hematopoietic precursors, mice lacking the Hzf gene display partially defective megakaryocyte maturation (24). We along with others have reported that Hzf is a transcriptional target of p53 (7, 37). Hzf expression is induced by p19^ARF or DNA damage and is thought to contribute to p53-dependent cell cycle arrest. Hzf contains three typical C₂H₂-type zinc finger domains that serve as nucleic acid-binding motifs, and Iijima et al. reported that Hzf interacts with the 3′ UTR of IP3R (inositol 1,4,5-triphosphate receptor) mRNA in cerebellar Purkinje cells, where it regulates its localization and translation (16). We have shown that Hzf is induced during adipocyte differentiation and that cells that suffer from impaired Hzf expression are incapable of undergoing efficient adipogenesis (21). Hzf associates with the 3′ UTR of C/EBPα mRNA, which facilitates its translation, and consistent with the role of C/EBPα in glucose metabolism, Hzf-deficient mice show impaired adipocyte functions (e.g., reduced adipokine production) and exhibit glucose intolerance and reduced insulin sensitivity.

During further exploration of the molecular function of Hzf, we found that HuR interacts with Hzf. HuR plays essential roles in regulating the stability, localization, and translation of its target mRNA (1, 6). Hzf and HuR independently associate with the 3′ UTR of p53 mRNA, and the absence of both the Hzf and HuR proteins results in reduced p53 protein synthesis due to impaired export of its mRNA. Thus, Hzf and HuR contribute to the posttranscriptional p53 regulation that occurs in response to signals that induce p53 activation such as p19^ARF.

MATERIALS AND METHODS

Cells and culture conditions.

NIH 3T3, Arf-inducible NIH 3T3-derived cells, MT-Arf cells (26), and 293T cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS) and 100 U/ml penicillin-streptomycin. Mouse embryonic fibroblasts (MEF) were cultured in medium supplemented with 0.1 mM nonessential amino acids, 55 μM 2-mercaptoethanol, and 10 μg/ml gentamicin instead of penicillin and streptomycin. To analyze mRNA or protein stability, the cells were treated with 5 μg/ml of actinomycin D or 100 μg/ml of cycloheximide, respectively. For the clonogenic survival assays, the cells were seeded into 10-cm-diameter culture dishes. After 10 days, the cultured cells were stained with crystal violet, and the number of colonies per dish was counted.

Retroviral expression plasmid, retroviral vector production, and infection.

Flag-Hzf-His expression retrovirus plasmids were obtained by inserting the His tag cassette into the murine stem cell virus (MSCV) Flag-Hzf protein (37). Mouse HuR cDNA was obtained by PCR using the following primers: sense, 5′-AAGAATTCTCTAATGGTTATGAAGACCAC-3′; antisense, 5′-AACTCGAGTTATTTGTGGGACTTGTTGG-3′. The PCR product was digested with EcoRI and XhoI and cloned into an MSCV vector containing a hemagglutinin (HA) epitope. For the HuR knockdown, two sets of annealed oligonucleotides (sh-HuR1, GATCCCCGTCTGTTCAGCAGCATTGGTTCAAGAGACCAATGCTGCTGAACAGACTTTTTGGAAA and AGCTTTTCCAAAAAGTCTGTTCAGCAGCATTGGTCTCTTGAACCAATGCTGCTGAACAGACGGG; sh-HuR2, GATCCCCTGTGAAAGTGATTCGTGATTTCAAGAGAATCACGAATCACTTTCACATTTTTGGAAA and AGCTTTTCCAAAAATGTGAAAGTGATTCGTGATTCTCTTGAAATCACGAATCACTTTCACAGGG) were cloned into the pSUPERretro vector plasmid (Oligoengine, Seattle, WA) in accordance with the manufacturer's instructions. sh-HuR2 was utilized in all experiments unless otherwise indicated.

For retrovirus production, 293T cells were transfected with retroviral expression plasmids (MSCV p19^ARF [26], Flag-Hzf [37], HA-HuR, pSUPERretro sh-Hzf [38], or sh-HuR) together with helper plasmids, as described previously (44). The culture supernatants were harvested 24 to 60 h after transfection, pooled, and stored on ice. Exponentially growing cells in 10-cm-diameter culture dishes were infected with 3 ml of fresh virus-containing supernatant in complete medium containing 8 μg/ml Polybrene. Infection was confirmed either by a flow cytometric assay for green fluorescent protein (GFP) expression or by selection for drug resistance.

Purification of Hzf protein complex.

NIH 3T3 cells were infected with Flag-Hzf-His retroviruses. Five days later, the cells were suspended in immunoprecipitation (IP) buffer (50 mM HEPES pH 7.5, 150 mM NaCl, 1 mM EDTA, 2.5 mM EGTA, 1 mM dithiothreitol [DTT], 0.2% Tween 20, 10% glycerol, and protease inhibitors) and sonicated for 10 s. Then, the lysates were cleared by centrifugation, and their protein concentrations were measured. Lysates containing 3 mg of protein were precleared with protein G beads for 60 min, before being immunoprecipitated overnight using Flag-agarose beads (M2-agarose; Sigma, St. Louis, MO). The beads were then recovered and washed three times with TBS-T (Tris-buffered saline, 0.1% Tween 20), and the immunoprecipitated proteins were eluted with 100 μg/ml Flag peptide (Sigma) for 30 min. Eluates were incubated with Ni-nitrilotriacetic acid (NTA) beads for 90 min, and the recovered beads were washed five times with TBS-T. Then, the proteins were eluted by incubating the beads twice with TBS containing 0.3 M imidazole for 30 min, and the purified proteins were concentrated by methanol-chloroform precipitation, resolved by SDS-PAGE (4 to 20%), and stained with silver. The specific protein bands were excised, reduced with 10 mM DTT at room temperature for 2 h, and alkylated with 40 mM iodoacetamide in a dark room for 30 min. Each sample was digested with 4 μg/ml trypsin (Trypsin Gold; Promega, Madison, WI) in 40 mM NH₄HCO₃–10% acetonitrile (ACN) at 37°C overnight. The extracted peptides were then separated via nano-liquid chromatography (LC) (Paradigm MS4; Michrom BioResources, Auburn, CA) using a reverse C₁₈ column (Magic C₁₈; Michrom BioResources), and the LC eluent was analyzed by an LCQ Advantage MAX mass spectrometer ([MS]Thermo Fisher Scientific, Waltham, MA) equipped with an ion spray source. All tandem MS (MS/MS) spectra were examined using the TurboQUEST algorithm within the Bio Works, version 3.2, software.

Immunoprecipitation, immunoblotting, and immunofluorescence.

Cell lysates prepared using IP buffer were incubated at 4°C overnight with anti-Flag (M2), anti-HA (3F10; Roche, Indianapolis, IN), anti-HuR (3A2; Santa Cruz Biotechnology, Santa Cruz, CA), or anti-Hzf (34) antibodies. Immune complexes were recovered with protein G-Sepharose (GE Healthcare, Buckinghamshire, United Kingdom) and washed three times with IP buffer. For sequential immunoprecipitation, the precleared lysates (5 mg) were immunoprecipitated with anti-Flag beads (M2 agarose; Sigma) for 3 h at 4°C. Immune complexes were recovered by centrifugation and washed four times with IP buffer and 100 μg/ml Flag peptide as described above and subjected to a second immunoprecipitation with anti-HA (3F10) for 3 h at 4°C. The immunoprecipitated complexes and total lysates were separated by SDS-PAGE and transferred to polyvinylidene difluoride (PVDF) membranes (Millipore, Billerica, MA). Proteins were then detected with antibodies to Flag tag (M2), HA tag (3F10), p19^ARF, HuR, lamin A/C, α-tubulin (5-C3, 3A2, H-110, and B-7; all from Santa Cruz Biotechnology), or p53 (IMX25; Novocastra, Newcastle Upon Avon, United Kingdom). The intensity of the bands was quantified using NIH ImageJ. For RNA analysis, recovered immune complexes were suspended in Trizol (Invitrogen, Carlsbad, CA).

Immunofluorescence was examined as described previously (5). Proteins were detected using anti-Flag (M2) or anti-HA (3F10) antibodies and visualized with Cy3-labeled secondary antibodies (Jackson ImmunoResearch, West Grove, PA), and cells were mounted using Vectashield and 4′,6′-diamidino-2-phenylindole. (DAPI; Vector Labs, Burlingame, CA).

Metabolic labeling.

For metabolic labeling, the cells were preincubated with Met-Cys-free DMEM containing 10% dialyzed FCS for 30 min and then labeled in the presence of 200 μCi/ml [³⁵S]Met-Cys (Perkin Elmer, Boston, MA). Lysates were prepared with IP buffer, precleared with protein A beads, and immunoprecipitated using anti-p53 (FL393; Santa Cruz Biotechnology). The recovered proteins were separated by SDS-PAGE and detected by autoradiography.

RNA-binding assay.

cDNA corresponding to the 5′ UTR, open reading frame (ORF), and 3′ UTR of mouse p53 were obtained by PCR using the following primers: for the 5′ UTR, 5′-TTTCCCCTCCCACGTGCTCAC-3′ (sense) and 5′-CCAGTCTTCGGAGAAGCGTG-3′ (antisense); for ORF 5′-ATGACTGCCATGGAGGAGTC-3′ (sense) and 5′-TCAGTCTGAGTCAGGCCCCAC-3′ (antisense); for the 3′ UTR, 5′-CTGCCTCTGCATCCCGTCCC-3′ (sense) and 5′-AAGTGATAACAAAATTTTATTG-3′ (antisense). The PCR products were cloned into the pGEM-T easy vector (Promega), and RNA probes were produced by in vitro transcription using a Riboprobe System (Promega) in the presence of radioactive nucleotides. Fifty nanograms of bacterially (Escherichia coli strain BL21) produced Hzf or HuR proteins (21) was preincubated for 15 min at 30°C in a buffer containing 5 mM HEPES (pH 7.5), 25 mM KCl, 2 mM MgCl₂, 3.8% glycerol, 0.2 mM DTT, 1 mM ATP, and 2.5 μg of yeast tRNA; the sample was mixed with purified probe (10⁵ cpm) and incubated for another 30 min. Samples were then subjected to UV cross-linking (0.4 J/cm²), treated with ribonucleases (2.5 μg of RNase A and 5 U of RNase ONE [Promega]) for 30 min at 37°C, separated by SDS-PAGE, and detected by autoradiography.

R-EMSA (RNA-electrophoretic mobility shift assay) was performed as previously described (21). cDNA corresponding to Hzf-binding region (see Fig. 8D) were conjugated to T7 sequence on their 5′ by PCR using following primers: 15-60, 5′-TGTAATACGACTCACTATAGGGCGACGTCCCCATCACCAGCCTCC-3′ (sense) and 5′-GAAGTCATAAGACAGCAAGG-3′ (antisense); 199-244, 5′-TGTAATACGACTCACTATAGGGCGACGAAATTCTATCCAGCCAGT-3′ (sense) and 5′-TTCATTGTAGGTGCCAGGGT-3′ (antisense). PCR products were purified, and radiolabeled RNA probes were produced using a Riboprobe System (Promega). A total of 10⁴ cpm of purified radiolabeled probes was incubated with bacterially produced His-tagged Hzf proteins, together with Hzf antibody where indicated in Fig. 8E, for 30 min at room temperature in a buffer containing 10 mM HEPES-KOH (pH 7.5), 90 mM KOAc, 1.5 mM Mg(OAc)₂, 2.5 mM DTT, 10 units RNase inhibitor, 10 μg of yeast tRNA, and 15% glycerol. Samples were separated on a 4% polyacrylamide gel containing 5% glycerol in 0.5× TBE (Tris-borate-EDTA) buffer and subjected to autoradiography.

Fig. 8.

Hzf binding is required for the expression of the p53 3′ UTR reporter in HuR-knockdown cells. (A) Schematic representation of luciferase reporter mRNA. (B) The luciferase reporters illustrated in panel A were cotransfected into NIH 3T3 cells expressing sh-HuR together with the p19^ARF plasmid. The luciferase activity in each sample was measured and plotted. (C) HuR knockdown NIH 3T3 cells were transfected with the indicated luciferase reporter plasmids together with p19^ARF expression plasmid and pRL-SV40. Cytoplasmic and nuclear mRNAs were analyzed as above. (D) Potential Hzf-binding regions in the 3′ UTR of p53 mRNA are illustrated. (E) R-EMSA was performed using radiolabeled RNA probes representing the p53 3′ UTR (residues 15 to 60 or 199 to 24). The probes were mixed with increasing quantities (75, 150, and 300 fmol) of bacterially produced wild-type of d2 Hzf proteins. Hzf antibody was used for supershift, and ARF antibody was used as a control. Complexes were separated by native PAGE and detected by autoradiography. The asterisk and arrowhead indicate Hzf-RNA complexes and supershifted complex, respectively.

Preparation of cytoplasmic, nuclear, and polysome fractions.

To prepare the cytoplasmic and nuclear fractions, the cells were suspended in a hypotonic lysis buffer (20 mM Tris, pH 7.5, 10 mM NaCl, 3 mM MgCl, 1% NP-40, and protease inhibitors) and incubated on ice for 10 min. Nuclei were pelleted by centrifugation at 2,200 × g for 5 min and washed three times with hypotonic buffer. Polysome fractions were prepared by sucrose gradient sedimentation, as previously described (21).

RNA analyses.

RNAs were prepared from cells or immune complexes using Trizol, reverse transcribed using a PrimeScript RT reagent kit (Takara, Otsu, Japan), and subjected to PCR using the following primers: for c-fos, 5′-AATCCGAAGGGAACGGAATAAGA-3′ (sense) and 5′-TGCAACGCAGACTTCTCATCT-3′ (antisense); c-myc, 5′-TCTATTTGGGGACAGTGTTC-3′ (sense) and 5′-GGTCATAGTTCCTGTTGGTG-3′) (antisense); p21^Cip1, 5′-CGAGAACGGTGGAACTTTGAC-3′ (sense) and 5′-CAGGGCTCAGGTAGACCTTG-3′ (antisense); mdm2, 5′-AGGGGAAAGATAAAGTGGAA-3′ (sense) and 5′-CCCCTGGCAGATCACACATG-3′ (antisense) (40); p27^Kip1, 5′-GGGCAGATACGAGTGGCAG-3′ (sense) and 5′-CCTGAGACCCAATTAAAGGCAC-3′ (antisense); p53, 5′-TGGAGAGTATTTCACCCTCAAGA-3′ (sense) and 5′-CTCCTCTGTAGCATGGGCATC-3′ (antisense); cyclin A2, 5′-ACATTCACACGTACCTTAGGGA-3′ (sense) and 5′-CATAGCAGCCGTGCCTACA-3′ (antisense); cyclin B1, 5′-CTCAGGGTCACTAGGAACACG-3′ (sense) and 5′-AGCTCTTCGCTGACTTTATTACC-3′ (antisense); cyclin D1, 5′-GCGTACCCTGACACCAATCTC-3′ (sense) and 5′-CTCCTCTTCGCACTTCTGCTC-3′ (antisense); GAPDH, 5′-AATGGTGAAGGTCGGTGTG-3′ (sense) and 5′-GAAGATGGTGATGGGCTTCC-3′ (antisense); β-actin, 5′-CTAAGGCCAACCGTGAAAAG-3′ (sense) and 5′-ACCAGAGGCATACAGGGACA-3′ (antisense); 18S rRNA 5′-AGTCCCTGCCCTTTGTACACA-3′ (sense) and 5′-GATCCGAGGGCCTCACTAAAC-3′ (antisense); firefly luciferase, 5′-ATGTACACGTTCGTCACATC-3′ (sense) and 5′-ACCTTTAGGCAGACCAGTAG-3′ (antisense); Renilla luciferase, 5′-TCCAGAACAAAGGAAACGGA-3′ (sense) and 5′-ATAATACACCGCGCTACTGG-3′ (antisense). Real-time PCR analysis was carried out on a Chromo4 real-time PCR system (Bio-Rad, Hercules, CA). For Northern blotting, RNAs extracted from the polysome fractions were separated on a 1% formaldehyde-agarose gel and transferred to Hybond N⁺ membranes (GE Healthcare). p53 and β-actin cDNAs were radiolabeled using a random priming kit (Roche), hybridized, and detected by autoradiography.

Luciferase reporter assay.

p53 5′ UTR and 3′ UTR (full-length and with deletions of residues 1 to 150, 1 to 300, and 15 to 60 [Δ1–150, Δ1–300, and Δ15–60, respectively]) fragments were obtained from a mouse cDNA library by PCR using the following primers: for the 5′ UTR, 5′-AAAAGCTTCGAAGCTGCGCGGGCGCGAG-3′ (sense) and 5′-AACCATGGGGGAGTTAGAGTTCTCCCGG-3′ (antisense); for the 3′ UTR, 5′-AATCTAGAGGCGCGCGGCTGCGGGACCG-3′ (sense) and 5′-AAGCTAGCGTCCAGCCCTGCTCCCAGCC-3′ (antisense); 3′ UTR (Δ1–150), 5′-AAGCTAGCCCCAAGTTGGGGAATAGGTTG-3′ (sense) and 3′ UTR antisense; for 3′ UTR (Δ1–300), 5′-AAGCTAGCATCCTCCAGGGCCTACTTTC-3′ (sense) and 3′ UTR antisense; 3′ UTR (Δ15–60), 5′-AAGCTAGCCTGCCTCTGCATCCAGGGCTGAGACACAATCCTC-3′ and 3′ UTR antisense. The 3′ UTR (Δ199–244) was obtained by PCR with template full-length p53 3′ UTR cDNA. Two PCRs were performed as follows: 3′ UTR sense mixed with Δ199–244 antisense (5′-AGGGTGAGATCTGGGCCAGCAGAGACCTGA-3′) and Δ199–244 sense (5′-GCTGGCCCAGATCTCACCCTACCCCACACC-3′) primer mixed with the 3′ UTR antisense primer. Purified PCR products were mixed and subjected to a second PCR using 3′ UTR sense and 3′ UTR antisense primers. The PCR products were digested with HindIII-NcoI (5′ UTR) or NheI-XbaI (3′ UTR) and cloned into the pGL3 control vector (Promega). Then, 3 × 10⁴ cells were plated in a 12-well culture dish and transfected using Fugene 6 (Roche Diagnostics) with a 25 ng of p19^ARF expression plasmid or an empty vector, 25 ng of luciferase reporter plasmid, and 25 ng of pRL-SV40 (where SV40 is simian virus 40) (Promega). Seventy-two hours later, cell lysates were prepared, and their luciferase activities were measured using a Dual Luciferase Reporter Assay System (Promega).

RESULTS

Hzf interacts with HuR in an RNA-dependent manner.

In an effort to identify proteins that associate with Hzf, we infected NIH 3T3 cells with a retrovirus expressing Hzf with Flag and His tags fused to its N and C termini, respectively (Flag-Hzf-His). Five days after infection, the cells were lysed, immunopurified using Flag tag antibody, eluted with Flag peptide, and subjected to His tag purification using Ni-NTA agarose. As a control for nonspecific binding, the same amount of lysate from uninfected cells was subjected to the same purification procedure. The recovered proteins were concentrated by methanol-chloroform precipitation and electrophoretically resolved in a denaturing 4 to 20% gradient polyacrylamide gel (Fig. 1 A). The gel-separated proteins were then silver stained to pinpoint bands and specifically recovered with Flag-Hzf-His. These bands were sequenced by mass spectrometry, and peptides from several RNA-binding proteins in addition to Hzf were identified. Of these, HuR was most strongly associated with Hzf. Both Hzf and HuR predominantly localized to the cell nuclei (data not shown), and we confirmed the interaction of Flag-Hzf and HA-HuR through their reciprocal coimmunoprecipitation in NIH 3T3 cells (Fig. 1B). The experiments that followed focused on the interaction between Hzf and HuR.

Fig. 1.

Hzf interacts with HuR in an RNA-dependent manner. (A) Purification of Hzf complex. NIH 3T3 cells were infected with Flag-Hzf-His expression retroviruses. Five days later, complex purification was performed from cells expressing Flag-Hzf-His or from uninfected cells (mock). The purified complexes were electrophoretically resolved on denaturing gels and visualized with silver. Specific bands recovered with Flag-Hzf-His were sequenced by mass spectrometry. Lane MW, molecular weight markers. hnRNP, heterogeneous nuclear ribonucleoprotein; G3BP1, GTPase-activating protein-binding protein 1; Pur beta, purine-rich element-binding protein beta. (B) Lysates prepared from NIH 3T3 cells infected with HA-HuR and/or Flag-Hzf retroviruses were immunoprecipitated using HA or Flag antibodies. Immune complexes were separated on denaturing gels and blotted with the same antibodies. (C) NIH 3T3 cells infected with the control or p19^ARF expression retroviruses. Three days later, lysates were prepared and precipitated with HuR (lanes 3 and 6) or Hzf (lanes 9 and 12) or without antibody (lanes 2, 5, 8, and 11). Proteins were separated on gels and blotted with HuR or Hzf antibodies. (D) NIH 3T3 cells infected with control or Flag-Hzf (wild type or an RNA-binding-defective d2 mutant) together with HA-HuR retroviruses. The lysates were treated (lanes 2, 4, and 6) or untreated (lanes 1, 3, and 5) with RNase, immunoprecipitated with Flag antibody, and blotted with Flag or HA antibodies. RNAs recovered from the same lysates were separated on agarose gel and stained with ethidium bromide (EtBr) to confirm the occurrence of RNase activity. IB, immunoblot; IP, immunoprecipitation; α, anti.

To examine the interaction of the endogenous proteins, HuR and Hzf were immunoprecipitated from NIH 3T3 cells. Both proteins were observed to coprecipitate, and the amount of Hzf-HuR complex was increased when Hzf expression was induced by p19^ARF (Fig. 1C) (37). Given that both Hzf and HuR are RNA-binding proteins, it was possible that the interaction between these proteins was mediated through RNA molecule(s). To test this possibility, wild-type (wt) or RNA-binding-defective (d2) Flag-Hzf was expressed in the NIH 3T3 cells together with HA-HuR, and lysates from these cells were treated (or not) with RNase before the immunoprecipitation step. Pretreatment with RNase completely abolished the interaction of Hzf with HuR (Fig. 1D, lanes 3 and 4). Moreover, we were unable to detect an interaction between HuR and the Hzf d2 mutant, even in the absence of RNase (lanes 5). Together, these results indicate that the interaction between Hzf and HuR is indirect and that common RNA targets of Hzf and HuR mediate the interaction of these proteins.

Hzf and HuR target p53 mRNA.

The above results prompted us to identify RNA molecules that associate with both Hzf and HuR. So far, only IP3R and C/EBPα mRNAs have been shown to associate with Hzf (16, 37), while HuR is known to interact with a variety of mRNAs including essential cell cycle regulators such as c-fos, c-myc, p53, p21, cyclin A2, cyclin B1, and cyclin D1 (1, 6). Hence, we investigated whether any of these known HuR targets are able to associate with Hzf. For this purpose, Flag-Hzf and HA-HuR were retrovirally expressed in NIH 3T3 cells, and the lysates of these cells were sequentially immunoprecipitated using Flag (first IP) and HA (second IP) antibodies to obtain a complex that included both the Hzf and HuR proteins (Fig. 2A). RNAs were extracted from the purified complexes, reverse transcribed, and subjected to PCR analysis using specific primers for c-fos, c-myc, p53, p21, cyclin A2, cyclin B1, and cyclin D1. Although c-fos, c-myc, p21, cyclin A2, and cyclin D1 mRNAs were hardly detectable, p53 and cyclin B1 mRNAs were specifically enriched in the Hzf-HuR complex (Fig. 2B). These results suggest that p53 and cyclin B1 mRNAs are common targets of Hzf and HuR proteins. The cyclin B1 mRNA level varies depending on cell cycle stage due to its transcriptional regulation by E2F, while the level of p53 mRNA hardly changes. We therefore analyzed the association between p53 mRNA and Hzf-HuR in this study.

Fig. 2.

Hzf and HuR associate with p53 both in vivo and in vitro. (A) Lysates were prepared from NIH 3T3 cells that had been infected with the indicated retroviruses and immunoprecipitated with Flag-antibody. Immune complexes were produced using Flag peptide and reimmunoprecipitated with HA antibody. Proteins were separated on denaturing gels and blotted with the same antibodies. (B) RNAs were extracted from the immune complexes, reverse transcribed, and subjected to PCR analysis using specific primers for the indicated genes. (C) An in vitro RNA-binding assay was performed using purified bacterially produced HuR or Hzf (wild type and d2 mutant) protein. The 5′ UTR, ORF, and 3′ UTR of the p53 mRNA were synthesized in vitro in the presence of ³²P-labeled nucleotides. These RNA probes were incubated with HuR (left; only the 3′ UTR probe was used) or Hzf (right) proteins followed by UV cross-linking. Samples were treated with RNase and separated on denaturing gels. RNA-protein complexes were detected by autoradiography. Hzf proteins were detected by immunoblotting with Hzf antibody. Mw, molecular weights in thousands.

HuR interacts with the ARE in the 3′ UTR of p53 mRNA (32, 41, 45). Using an in vitro RNA-binding assay involving a bacterially produced HuR protein, we confirmed that HuR is able to associate with the 3′ UTR of p53 mRNA (Fig. 2C, left). We then tested whether Hzf also directly associates with p53 mRNA. Purified wild-type or d2 mutant Hzf proteins were incubated with the 5′ UTR, ORF (open reading frame), or 3′ UTR sequence of p53 mRNA. Similarly to HuR, Hzf also specifically associated with the 3′ UTR of p53 mRNA, and a previously identified RNA-binding domain was found to be required for the binding (Fig. 2C, right).

Either Hzf or HuR is required for p19^ARF-induced p53 expression.

Next, we investigated the role of the Hzf-HuR interaction in p53 expression. NIH 3T3 derivative ARF-inducible MT-Arf cells (26) were infected with retroviruses encoding short hairpin RNA (shRNA) directed at Hzf (sh-Hzf) and/or sh-HuR to knock down the endogenous proteins. The infected cells were then selected and treated with zinc sulfate to induce exogenous p19^ARF expression. In the parental NIH 3T3 cells, zinc treatment and knockdown of Hzf and/or HuR had no effect on the basal p53 levels (data not shown) (26). In the MT-Arf cells, p19^ARF induction increased endogenous p53 protein expression as expected, and loss of Hzf or HuR alone had no detectable impact on p53 expression (Fig. 3A). However, p53 induction was attenuated in the cells with double knockdown of Hzf and HuR (Hzf/HuR-DKD). Consequently, p21^Cip1 expression was impaired in the Hzf/HuR-DKD MT-Arf cells (Fig. 3B) although the effect on Mdm2 expression was less pronounced (Fig. 3C). In contrast to the findings of the knockdown experiments, the overexpression of Hzf and/or HuR did not further increase p19^ARF-induced p53 protein expression, suggesting that the endogenous levels of Hzf and HuR are sufficient to maintain their effects on p53 expression (Fig. 3D).

Fig. 3.

Loss of Hzf and HuR expression results in impaired p53 induction by p19^ARF. (A) MT-Arf cells infected with retroviruses encoding scrambled shRNA (SCR), sh-Hzf, sh-HuR, or sh-Hzf in combination with sh-HuR were treated with zinc sulfate to induce exogenous p19^ARF protein expression. Lysates were prepared after the indicated periods of zinc treatment, and the expression of the indicated proteins was analyzed by immunoblotting. α-Tubulin was used as a loading control. (B and C) Total RNAs were extracted from MT-Arf cells that had been treated to induce p19^ARF expression for the indicated periods. The amount of p21^Cip1 (B) and mdm2 (C) mRNAs present in the cells was analyzed by real-time PCR and normalized to the amount of β-actin mRNA in each sample. (D) Effects of Hzf and/or HuR overexpression on the p19^ARF-p53 pathway. NIH 3T3 (upper) and MT-Arf cells (lower) were infected with Flag-Hzf and/or HA-HuR expression retroviruses. The cells were then treated with zinc for the indicated periods, and the expression of the indicated proteins was analyzed by immunoblotting. (E) Effects of HuR knockdown in Hzf^−/− MEF. Embryonic fibroblasts prepared from wild-type (Hzf^+/+) or Hzf^−/− mice (passage 3) were infected with retroviruses encoding sh-HuR together with the control or p19^ARF retroviruses. The sh-HuR constructs (sh-HuR1 and sh-HuR2) target different regions within the HuR open reading frame. Four days later, lysates were prepared, and the expression levels of the indicated proteins were analyzed by immunoblotting.

To confirm that the reduced p53 expression observed in the Hzf/HuR-DKD MT-Arf cells was specifically caused by the loss of Hzf and HuR, rather than being an off-target effect of the shRNA, wild type (Hzf^+/+) and Hzf knockout (Hzf^−/−) mouse embryonic fibroblasts (MEF) were infected with two independent sh-HuR retroviruses that target different regions of the HuR mRNA molecule, together with a p19^ARF-expressing retrovirus. Similarly to the above results, the induction of p53 by p19^ARF was attenuated in the absence of both Hzf and HuR expression, and both sh-HuR constructs elicited the same effect on p53 expression in the Hzf^−/− MEF (Fig. 3E). These results indicate that the impaired p53 expression observed in the Hzf/HuR-DKD MT-Arf cells was specifically caused by the loss of Hzf and HuR and was not attributed to off-target effects of the shRNA.

Hzf and HuR regulate the translation of p53 mRNA.

Both Hzf and HuR regulate the translation of their target mRNAs, and HuR has been shown to enhance the translation of p53 mRNA in cells exposed to UV irradiation (32). On the other hand, it has recently been reported that Hzf physically associates with p53 protein (7). Given that p53 expression is regulated at multiple steps, we wished to determine at which level stress-induced p53 expression is impaired in Hzf/HuR-DKD cells. We first checked whether the loss of Hzf and HuR affects p53 protein stability. MT-Arf cells that had been treated to induce p19^ARF expression were cultured in medium containing cycloheximide, and the p53 half-life was analyzed. In the presence of p19^ARF, the half-life of the p53 protein was approximately 80 min and was not affected by Hzf or HuR status, suggesting that these proteins do not participate in the regulation of p53 protein stability (Fig. 4A and B). Additionally, loss of Hzf/HuR or p19^ARF expression did not affect p53 mRNA levels (Fig. 4C). Nonetheless, pulse-labeling experiments using ³⁵S-labeled amino acids revealed that the rate of p53 protein synthesis was significantly decreased in the absence of Hzf and HuR (Fig. 4D). Considering that the loss of both Hzf and HuR did not affect either p53 protein stability or the p53 mRNA level (Fig. 4A to C), it is conceivable that the reduced de novo p53 protein synthesis observed in the Hzf/HuR-DKD cells was due to decreased translation of p53 mRNA.

Fig. 4.

Loss of Hzf and HuR decreases p53 protein synthesis without affecting protein stability or the p53 mRNA level. (A) MT-Arf cells expressing the indicated shRNA were treated with zinc for 24 h. The cells were then further treated with cycloheximide to block protein synthesis for the indicated periods, and the expression of p53 and α-tubulin was analyzed by immunoblotting. (B) The intensity of the p53 protein signal in panel A was quantified and plotted. (C) Total RNAs were prepared from MT-Arf cells expressing the indicated shRNA with or without p19^ARF induction. p53 mRNA levels were analyzed by real-time PCR. The level of p53 mRNA in each sample was normalized to that of β-actin mRNA. (D) shRNA-expressing MT-Arf cells that had been treated to induce p19^ARF expression were pulse-labeled with [³⁵S]Cys-Met for the indicated periods. The immunoprecipitated p53 (lanes 1 to 12) proteins or total cell extracts (lanes 13 to 20) were separated on SDS-PAGE and detected by autoradiography. CHX, cycloheximide.

Hzf and HuR regulate the nuclear export of p53 mRNA.

The above results suggest that Hzf and HuR regulate p53 expression posttranscriptionally. Nevertheless, neither Hzf/HuR-DKD nor p19^ARF expression affected the stability of p53 mRNA (Fig. 5A). We then examined whether these proteins directly affect the association of p53 mRNA with ribosomes. Cytoplasmic lysates were prepared from induced MT-Arf cells and fractionated into polysome/nonpolysome fractions by sucrose gradient sedimentation. RNAs were recovered from each fraction and analyzed for p53 mRNA by Northern blotting. p53 mRNA was predominantly observed in the nonpolysome fraction (Fig. 5B and C), and loss of HuR resulted in a modest decrease of p53 mRNA in polysomal fractions while simultaneous knockdown of Hzf and HuR had no further effect. These results suggest that the translation of p53 mRNA was inefficient, regardless of whether the cells expressed sh-Hzf/sh-HuR, while HuR may independently affect the translation of p53 mRNA (32). Next, we investigated the nuclear export of p53 mRNA in these cells. RNAs were prepared from the cytoplasmic and nuclear fractions, and the amounts of p53 and β-actin mRNAs were analyzed. In the control, sh-Hzf, and sh-HuR cells, we unexpectedly observed that p19^ARF enhanced the nuclear export of p53 mRNA while β-actin mRNA localization was unaffected (Fig. 6). However, p19^ARF failed to promote the nuclear export of p53 mRNA in the Hzf/HuR-DKD cells. These results suggest that in addition to p53 stabilization, p19^ARF increases p53 expression by facilitating the nuclear export of p53 mRNA and that Hzf and HuR mediate this effect.

Fig. 5.

Hzf and HuR do not affect the p53 mRNA stability or its translation. (A) MT-Arf cells expressing the indicated shRNA were treated or untreated with zinc for 24 h. The cells were then further treated with the RNA polymerase inhibitor actinomycin D for the indicated periods. Total RNAs were prepared, and the p53 mRNA level was analyzed relative to that of β-actin mRNA by real-time PCR. (B) Polysomal distribution of the p53 mRNA. Cytoplasmic lysates prepared from MT-Arf cells that had been treated to induce for p19^ARF expression were fractionated using sucrose density gradient centrifugation. Northern blot analysis was performed using p53 or β-actin as a probe. 28S and 18S rRNAs were visualized by methylene blue staining, and relative value of the optical density at 260 nm (OD₂₆₀) was plotted. (C) Intensity of the signal in each fraction was quantified using NIH ImageJ, and the relative values to total signal strength in each sample were plotted. Act D, actinomycin D.

Fig. 6.

Hzf and HuR regulate the cytoplasmic accumulation of the p53 mRNA. Cytoplasmic (C) and nuclear (N) lysates were prepared from the MT-Arf cells. The levels of p53 and β-actin relative to those of 18S rRNAs were analyzed by real-time PCR, and the ratio of cytoplasmic to nuclear mRNA was plotted. Immunoblotting using lamin A/C (nuclear marker) and α-tubulin (cytoplasmic marker) antibodies was performed to confirm the fractionation.

Hzf and HuR promote protein synthesis by interacting with the p53 mRNA 3′ UTR.

As both Hzf and HuR directly associate with the 3′ UTR of p53 mRNA (Fig. 2), we next tested whether the interaction between them promotes protein synthesis in response to p19^ARF. For this purpose, we expressed a luciferase mRNA conjugated with the 5′ or 3′ UTR of the p53 mRNA together with p19^ARF in NIH 3T3 cells. In the absence of the p53 UTR (5′ or 3′), p19^ARF had no effect on the expression of luciferase (Fig. 7A). In the presence of the p53 5′ UTR, p19^ARF promoted luciferase expression, and the loss of Hzf and/or HuR did not affect this response, suggesting that p19^ARF also has the ability to induce the expression of p53 independently of Hzf and HuR through the p53 5′ UTR (Fig. 7B). p19^ARF also accelerated luciferase expression through the p53 3′ UTR (Fig. 7C). However, the response was completely abolished in the Hzf/HuR-DKD cells. We also checked the cytoplasmic/nuclear ratio of reporter mRNA in these cells. While p19^ARF had little effect on the cytoplasmic/nuclear ratio of the reporter mRNA produced from pGL3 control or pGL3 p53 5′ UTR, significant cytoplasmic accumulation of the reporter mRNA conjugated with the p53 3′ UTR was observed in the presence of p19^ARF (Fig. 7D). This effect was abolished in Hzf/HuR-DKD cells, suggesting that p19^ARF induces the reporter expression through Hzf/HuR-dependent nuclear export of the mRNA.

Fig. 7.

Hzf and HuR regulate p53 expression through the 3′ UTR of p53 mRNA. (A to C) NIH 3T3 cells expressing the indicated shRNA were transfected with plasmids expressing firefly luciferase mRNA in conjunction with the 5′ or 3′ UTR of p53 mRNA under the regulation of an SV40 promoter/enhancer together with p19^ARF plasmids, and an SV40-driven Renilla luciferase plasmid was used as an internal standard. The luciferase activity in each sample was measured, and the relative luciferase activity (firefly/Renilla) in each sample was calculated. The relative luciferase activity in the absence of ARF was normalized to 1.0 in each group. (D) p19^ARF induces cytoplasmic localization of luciferase reporter mRNA through the p53 3′ UTR. NIH 3T3 cells expressing the indicated shRNA were transfected with plasmids expressing firefly luciferase mRNA in conjunction with the 5′ or 3′ UTR of p53 mRNA under the regulation of an SV40 promoter/enhancer together with p19^ARF plasmids, and an SV40-driven Renilla luciferase plasmid (pRL-SV40) was used as an internal standard. Three days later, cytoplasmic and nuclear mRNAs were prepared. The levels of firefly luciferase mRNA relative to those of Renilla luciferase mRNA were analyzed by real-time PCR, and the ratio of cytoplasmic to nuclear mRNA was plotted.

To narrow down the Hzf-responsive region within the p53 3′ UTR, we generated luciferase reporters conjugated to partially deleted p53 3′ UTR molecules (Fig. 8A). These reporters were expressed in HuR-knockdown NIH 3T3 cells together with p19^ARF. Although p19^ARF was able to enhance the luciferase expression conjugated to the full-length p53 3′ UTR in the HuR-knockdown cells, this effect was completely abolished by deleting the first 150 bases of the 3′ UTR (Fig. 8B). Similarly, deletion of this region was sufficient to abolish the nuclear export of the reporter mRNA (Fig. 8C), implying that Hzf interacts with an element residing in this region. Because Hzf directly binds to a specific sequence in the 3′ UTR of IP3R and C/EBPα mRNA (16, 21), we searched for potential Hzf-binding sequence in the 3′ UTR of p53 mRNA and found two regions that have modest homology to the known Hzf-binding sequence (residues 15 to 60 and 199 to 244) (Fig. 8D). To test whether Hzf binds to these sequences, we performed an R-EMSA using radiolabeled RNA corresponding to these regions. Hzf directly associates with the region of residues 15 to 60 and also with the region of residues 199 to 244 with lower affinity, while the d2 mutant failed to bind to both probes (Fig. 8E). We further detected supershifted signal in the presence of Hzf antibody, suggesting that Hzf has an ability to directly associate with, at least, residues 15 to 60 of the p53 3′ UTR. Consistent with these results, deletion of residues 15 60 from the p53 3′ UTR was sufficient to abolish the effect of p19^ARF in HuR knockdown cells, while deletion of the residues 199 to 244 had little effect (Fig. 8B and C). These results strongly suggest that Hzf and HuR mediate the effect of p19^ARF on p53 mRNA through interactions with the 3′ UTR of p53 mRNA.

Hzf and HuR double knockdown cells are resistant to p19^ARF.

As Hzf and HuR were found to be involved in p19^ARF-mediated p53 expression, we investigated whether the loss of these proteins confers resistance to p19^ARF. To this end, NIH 3T3 cells with or without shRNA were infected with p19^ARF retroviruses and analyzed for their colony-forming efficiency and proliferation. As expected, p19^ARF expression resulted in a significant reduction of the colony-forming abilities of these cells (Fig. 9A and B). Hzf knockdown had no effect on p19^ARF activity while HuR knockdown resulted in a slight increase in cell growth in the presence of p19^ARF. However, in the Hzf/HuR-DKD cells, we observed significant increases in both colony number and cell proliferation in the presence of p19^ARF, indicating that the activity of p19^ARF partly depends on the presence of Hzf or HuR. Likewise, the inhibition of HuR expression using two independent shRNA constructs conferred resistance to p19^ARF in Hzf^−/− MEF (Fig. 9C and D). Taken together, these results suggest that Hzf and HuR are involved in a cellular stress response that leads to p53 activation in which they regulate the nuclear export of p53 mRNA.

Fig. 9.

Hzf and HuR are required for cell growth inhibition by p19^ARF-p53. (A and B) NIH 3T3 cells were infected with the indicated shRNA retrovirus together with control or p19^ARF retroviruses, and a colony formation assay was performed. Then, 5,000 (control) and 50,000 (p19^ARF) cells were plated. Asterisk indicates a P value of <0.05. (C) The same cells had their proliferation assessed. (D) Embryonic fibroblasts prepared from wild-type (Hzf^+/+) or Hzf^−/− mice (passage 3) were infected with retroviruses encoding sh-HuR together with the control or p19^ARF retroviruses. The sh-HuR constructs (sh-HuR1 and sh-HuR2) target different regions within the HuR open reading frame. The cells were cultured for 10 days without being passaged. (E) Working model of the posttranscriptional control of p53 expression by Hzf and HuR. p19^ARF activates p53, thereby inducing the transcription of its target genes including Hzf. p19^ARF also triggers nuclear export of the p53 mRNA, and Hzf and HuR cooperatively act on this process.

DISCUSSION

As a critical component of the tumor suppressor pathway, the activity and expression of p53 are regulated through multiple mechanisms. While the posttranslational regulation of p53 including its phosphorylation and ubiquitylation has been extensively studied, little is known about the significance of and molecular mechanisms behind the posttranscriptional regulation of p53. We show here that Hzf and HuR cooperate to regulate p53 expression and that p53 accumulation is attenuated in the absence of Hzf and HuR (Fig. 9E).

Hzf is a target of the p53 tumor suppressor protein and participates in the p53-dependent cell cycle checkpoint response elicited by DNA damage or the ARF tumor suppressor protein (37). Hzf encodes an RNA-binding protein that is required for the correct translation and/or localization of specific mRNA in adipocytes and cerebellar Purkinje cells (16, 21). The function of Hzf in these cells appears to be independent of p53, and it remains unknown how the RNA-binding activity of Hzf contributes to the p53 pathway. We found that Hzf associates with several RNA-binding proteins in living cells by tandem affinity purification of the Hzf protein complex. Among these, the interaction between Hzf and HuR was confirmed by their reciprocal immunoprecipitation. We concluded that the interaction between Hzf and HuR is indirect and mediated through RNA molecules as RNase treatment completely abolished the interaction of these proteins. In addition, a mutant form of Hzf protein that displays defective RNA binding (d2) was not able to interact with HuR, even in the absence of RNase treatment. Thus, it is likely that common RNA ligands mediate the interaction between Hzf and HuR proteins.

HuR is a ubiquitously expressed member of the family of ELAV proteins, which are involved in diverse biological processes (10, 31). HuR has been shown to associate with a number of mRNAs, and the biological consequence of HuR association varies depending on the mRNA it is binding to (1, 14). HuR directly associates with the ARE in the 3′ UTR of p53 mRNA and contributes to the induction of p53 expression in response to UV irradiation or the von Hippel-Lindau (VHL) tumor suppressor protein by enhancing the translation of p53 mRNA (8, 32). Our data demonstrate that HuR also regulates the nuclear export of p53 mRNA in cooperation with Hzf. p53 mRNA is included in the Hzf-HuR complex, and Hzf directly associates with the 3′ UTR of the p53 mRNA independently of HuR. Thus, p53 mRNA is likely to be one of the mRNA molecules that mediate the interaction between the Hzf and HuR proteins. We unexpectedly discovered that p19^ARF accelerates the nuclear export of p53 mRNA. This indicates that p19^ARF not only stabilizes p53 by antagonizing Mdm2 E3 ligase but also strengthens p53 expression by facilitating the cytoplasmic localization of p53 mRNA. The mechanism by which p19^ARF accelerates the nuclear export of p53 mRNA is unclear; however, our data suggest that the presence of either Hzf or HuR is a prerequisite. Nonetheless, expression of Hzf was not sufficient, and a signal caused by p19^ARF is required to induce the p53 mRNA translocation. It has been recently reported that cyclin-dependent kinase 1 (CDK1) phosphorylates several serine residues on HuR protein (22, 23). The phosphorylation leads to the nuclear accumulation of HuR protein. Since activation of the ARF-p53 pathway inhibits CDK1 activity, it is possible that, in addition to Hzf induction, CDK1 inhibition contributes to the nuclear accumulation of p53 mRNA.

Hzf has been shown to directly interact with p53 protein on chromatin and modifies its target gene preference in order to induce cell cycle arrest (7) although no p53 protein was detected in the Hzf complex we purified (Fig. 1A). Nonetheless, our data strongly suggest that Hzf, by cooperating with HuR, also regulates p53 expression through the posttranscriptional control of p53 mRNA. Thus, it is possible that Hzf participates in the regulation of p53 activity at both the posttranslational and posttranscriptional levels.

Given that the ARF-p53 pathway counteracts aberrant growth signals, one might expect Hzf/HuR-DKD cells to be prone to oncogenic transformation. However, we did not observe any increase in the focus formation efficiency of Hzf/HuR-DKD MEF in the presence of oncogenic Ras (data not shown). Additionally, these cells failed to form tumors when transplanted into athymic nude mice. There are at least two possible explanations for the transformation failure of these cells despite their reduced p53 activity. First, besides p53 mRNA, HuR controls the expression of numerous genes involved in cell cycle progression. Therefore, even in the absence of the p53 barrier, oncogenic Ras may not induce sufficient signaling to elicit aberrant cell division. Second, loss of Hzf and HuR does not result in complete absence of p53 expression (Fig. 3A). Therefore, residual amounts of p53 protein might be sufficient to block the effects of oncogenic signaling. This is more likely since the expression of p53 targets was not impaired in the Hzf/HuR-DKD cells (Fig. 3B and C, p21 versus Mdm2). Thus, posttranscriptional regulation by Hzf and HuR may reinforce p53 expression and promotes its antitumor effects.

While Hzf and HuR contribute to the posttranscriptional modification of p53, a question still remains as to the circumstances in which this mechanism plays an important role. In this regard, it is interesting that HuR is downregulated during cellular senescence (42). p53 plays an essential role in cellular senescence, and MEF cells, in which p53 is inactivated or deleted, do not undergo replicative senescence. Hence, in senescent cells in which HuR expression is absent or low, Hzf may functionally compensate for HuR to maintain p53 activity. Likewise, certain types of stress are known to induce ubiquitin-mediated proteolysis of HuR (2), and Hzf may also function to offset the reduction in HuR activity.

There is little doubt that the posttranscriptional regulation of p53 contributes to the stress-induced p53 response (39). Furthermore, posttranscriptional regulation of the p53 tumor suppressor protein may be evolutionally conserved since nematode p53 (CEP-1) expression is translationally regulated by the RNA-binding protein GLD-1 in response to DNA damage (36), and Mdm2-mediated p53 ubiquitylation does not occur in this organism.

It is reasonable to speculate that Hzf and HuR have targets other than p53 mRNA. Our findings suggest that cyclin B1 mRNA also associates with these proteins. We currently have no data regarding the effects of these proteins on cyclin B1 expression. Nonetheless, considering that Hzf is a transcriptional target of p53, it is feasible that Hzf and HuR cooperatively act to regulate the expression of cell cycle genes under stress conditions. Additionally, both Hzf and HuR are involved in adipocyte differentiation (9, 21). Therefore, it is also possible that a similar posttranscriptional regulatory mechanism contributes to the functions of adipocytes as well.