Inhibition of p63 Transcriptional Activity by p14^ARF: Functional and Physical Link between Human ARF Tumor Suppressor and a Member of the p53 Family

Abstract

The ARF/MDM2/p53 pathway is a principal defense mechanism to protect the organism from uncontrolled effects of deregulated oncogenes. Oncogenes activate ARF, which interacts with and inhibits the ubiquitin ligase MDM2, resulting in p53 stabilization and activation. Once stabilized and activated, p53 can either induce or repress a wide array of different gene targets, which in turn can regulate cell cycle, DNA repair, and a number of apoptosis-related genes. Here we show that, unlike p53, p63, a member of the p53 family, directly interacts with p14^ARF. Through this interaction ARF inhibits p63-mediated transactivation and transrepression. In p63-transfected cells, ARF, which normally localizes into nucleoli, accumulates in the nucleoplasm. Based on these observations, we suggest that stimuli inducing p14^ARF expression can, at the same time, activate p53 and impair p63 transcriptional activity, altering the pattern of p53 target gene expression. Here we show, for the first time, a physical and functional link between the p14^ARF tumor suppressor protein and p63, a member of the p53 family.

The ARF tumor suppressor protein (known as p14^ARF in humans and p19^ARF in mice) is a product of the alternative reading frame of the human INK4a locus on chromosome 9p21. The ARF tumor suppressor induces potent growth arrest or cell death in response to hyperproliferative oncogenic stimuli. ARF can activate the p53 tumor surveillance pathway by interacting with and inhibiting the p53-antagonist, MDM2 (29). ARF limits the E3 ubiquitin ligase activity of MDM2 (14), thus preventing the polyubiquitination, nuclear export, and subsequent cytoplasmic degradation of p53 (47). Once stabilized and activated, p53 can either activate or repress a wide array of different gene targets, which in turn can regulate cell cycle, DNA repair, and a number of apoptosis-related genes (28).

In addition to p53, mammalian cells contain two homologous genes, p63 and p73. These genes give rise to the expression of proteins that are highly similar to p53 in structure and function. In particular, p63 and p73 proteins can induce p53-responsive genes and elicit programmed cell death (16, 17, 25, 46).

Unlike p53, both p63 and p73 exist in multiple isoforms. In the case of p63, at least six different isotypes with widely differing transactivation potentials have been described (46). The transactivating (TA) isoforms, which resemble p53, are generated by the use of an upstream promoter; the ΔN isoforms, produced from an intronic promoter, contain the same DNA-binding and oligomerization (OD) domains as the TA isoforms but lack the transactivation domain. The ΔN isoforms contain a region of 26 amino acids at the very N-terminal end of the protein (TA2) in which a further activation function was recently mapped (8). Both the TA and ΔN isoforms have three possible carboxyl termini, termed α, β, and γ. In the C-terminal extension of the α isoforms there is a sterile alpha motif (SAM) that is found in proteins that regulate mammalian development and is thought to be involved in protein-protein interactions (36).

Despite the structural similarities, a number of functional differences were found between p53, p63, and p73 proteins that could depend on the biochemical properties of the proteins but could also derive from differences in the expression pattern.

p73 and p63 are more important during development and differentiation. In particular, p63 appears to be primarily implicated in epithelial and limbs development (37). However, it is interesting that UVB-induced DNA damage decreases levels of the dominant-negative ΔN-p63α isoform; simultaneously, the levels of the TA-p63 isoforms increase (20). Downregulation of ΔNp63α, as well as TAp63 upregulation, may be a prerequisite for UV-induced apoptosis in skin. This notion is supported by the recent observation that the TAp63γ isoform is required for p53-dependent apoptosis induced by DNA damage, implying a role for p63 in preventing stress-induced DNA damage and tumorigenesis (8, 10). Furthermore, it has recently been reported that a balance between TA and ΔN p63 isoforms is required to allow cells to respond to signals required for maturation of embryonic epidermis (18), suggesting the existence of a complex mechanism by which relative amount of the individual p63 isoforms can be regulated.

The p53 protein is a labile protein whose levels are primarily controlled by the MDM2/ARF pathway (12, 19). However, although regulation of p53 by the p14^ARF-MDM2 circuitry is well understood, very little is known about whether and how this molecular pathway regulates the functions of the p53 homologues. Remarkably, p73 can also associate with MDM2 (2), as would be expected from its strong homology with p53 and its activity is modulated by MDM2. Moreover, we have demonstrated that MDM2 induces TAp63 protein stabilization and transcriptional activation (3). Here we show, for the first time, a physical and functional association between p14^ARF and p63. Our study suggests a mechanism by which p14^ARF might differentially regulate the expression of p53-target genes through the complex network of p53-like proteins.

MATERIALS AND METHODS

Plasmids.

All of the p63 cDNAs in the pcDNA3 vector were kindly provided by H. van Bokhoven (38). A pcDNA3.1/HisTAp63γ plasmid was used in electrophoretic mobility shift assays (EMSAs). To obtain the construct TAp63Δ(297-499), pcDNA3-TAp63α was cut with EcoRI and religated in order to eliminate the last C-terminal 609 bp of TAp63α, whereas ΔNp63γΔ(1-26) and ΔNp63γΔ(1-86) were obtained by PCR with either Δ1-26F or Δ1-86F primers and the reverse primer and pcDNA3-ΔNp63γ as a template (Δ1-26F, CCGCTCGAGGACCAGCAGATTCAG; Δ1-86F, CCGCTCGAGTTCCAGCAGTCAAGC; reverse primer, GTGAATTCAGTGCCAACCTGTGGT).

The mutant plasmid pcDNA-p14^ARF(Δ1-38) was obtained as follows: a NarI-XbaI fragment was excised from pcDNA3-ARF and, after filling of the NarI site, it was cloned in EcoRV-XbaI sites of pcDNA3. The plasmid pcDNA-p14^ARF(Δ66-132) was obtained as follows. A HindIII-XbaI fragment was excised from N-p19 plasmid (4) and, after filling of the HindIII site, it was cloned in EcoRV-XbaI sites of pcDNA3.

The HindIII 1.08-kb fragment containing the apoptosis protease-activating factor 1 (Apaf1) promoter sequence was retrieved from the pGL3b-Apaf-prom(−871/+208) plasmid provided by K. Helin (23) and ligated in the HindIII site of the pCAT0 plasmid to give the Apaf1CAT plasmid.

The other plasmids have been already described (3, 11).

Cell culture and transfection.

H1299, Saos-2, COS-7, and HaCaT cells were cultured in Dulbecco modified Eagle medium supplemented with 10% fetal bovine serum. NIH 3T3 cells were cultured in Dulbecco modified Eagle medium supplemented with 10% calf serum. H1299 and NIH 3T3 cells were transfected by using Lipofectamine and Lipofectamine supplemented with Plus (Gibco), respectively. Saos-2 and COS-7 cells were transfected with Lipofectamine 2000 (Gibco). HaCaT cells were transfected with Superfect reagent (Qiagen) according to the manufacturer's instructions. The total amount of transfected DNA was kept constant by using the “empty” expression vector when necessary.

Western blotting.

At 48 h after transfection cells were lysed in 10 mM Tris-HCl (pH 7.5), 1 mM EDTA, 150 mM NaCl, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 0.5% sodium deoxycholate, and protease inhibitors. Cell lysates were incubated on ice for 30 min, and the extracts were centrifuged at 13,000 rpm for 10 min to remove cell debris. Protein concentrations were determined by the Bio-Rad protein assay. After the addition of 4× loading buffer (2% sodium dodecyl sulfate [SDS], 30% glycerol, 300 mM β-mercaptoethanol, 100 mM Tris-HCl [pH 6.8]), the samples were incubated at 95°C for 5 min and resolved by SDS-polyacrylamide gel electrophoresis. Proteins were transferred to a polyvinylidene difluoride membrane (Millipore) and probed with the following antibodies: anti-p63 (H137 or 4A4; Santa Cruz), anti-p14^ARF (C-18; Santa Cruz), anti-MDM2 (smp14; Santa Cruz), anti-p21 (C-19; Santa Cruz), anti-myc (sc40; Santa Cruz), and anti-actin (I-19; Santa Cruz). Proteins were visualized by an enhanced chemiluminescence method (Amersham).

Coimmunoprecipitations.

NIH 3T3, H1299, or HaCaT cells (5 × 10⁵/60-mm plate) were transfected with the indicated vectors. Transfected cells were harvested 48 h posttransfection, and the cell lysates were prepared as described above. Lysates containing 500 μg of proteins were precleared with 30 μl of protein A-agarose (50% slurry; Santa Cruz) and then incubated overnight at 4°C with fresh protein A-beads (30 μl) and 2 μg of anti-p63 (H137 or 4A4; Santa Cruz) or anti-p14^ARF (C-18; Santa Cruz) antibodies. The beads were washed vigorously twice with lysis buffer and once with radioimmunoprecipitation assay buffer and loaded directly onto an SDS-12% polyacrylamide gel. The immunoprecipitated proteins were detected by Western blotting. For the coimmunoprecipitations of in vitro-translated proteins, TAp63γ, p14^ARF, p14^ARF(Δ1-38) and p14^ARF(Δ66-132) proteins were in vitro translated in the presence of [³⁵S]methionine by using TnT reticulocytes from Promega with 1 μg of pcDNA3-TAp63γ and 1 μg of pcDNA3 or with 1 μg of pcDNA3-TAp63γ and 1 μg of either pcDNA3-p14^ARF, pcDNA3-p14^ARF(Δ1-38), or pcDNA3-p14^ARF(Δ66-132). Then, 40-μl portions of the individual reactions were used for immunoprecipitation with anti-His antibodies (6xHis; Clontech).

EMSA.

EMSA experiments were performed as already described (26). TAp63γ and p14^ARF proteins were in vitro translated by using TnT reticulocytes from Promega with 0.5 μg of pcDNA3.1/His-TAp63γ, 1.5 μg of pEGFP C3, 1.5 μg of pcDNA3-p14^ARF, 0.5 μg of pEGFP C3, 0.5 μg of pcDNA3.1/His-TAp63γ, and/or 1.5 μg of pcDNA3-p14^ARF. Next, 10 μl of the individual reactions was used either for the binding reaction or for Western blot analysis. The probe is a radiolabeled oligonucleotide duplex containing a p53-binding site present in the p21 promoter (p21.1 described in reference 44). A 100-fold molar excess of the same cold oligonucleotide or an oligonucleotide containing a consensus binding site for E2F1 was used for competition experiments. For the supershift antibodies against the His tag fused at the 5′ of pcDNA3.1/His-TAp63γ (6xHis), anti-p63 antibodies (H-137; Santa Cruz) or unrelated polyclonal anti-p21 antibodies (C-19; Santa Cruz) were used, adding them to the samples prior to the binding reaction (30 min in ice).

Subcellular distribution assay.

NIH 3T3, COS-7, Saos-2, and H1299 cells (10⁵/35-mm plate) were grown on micro cover glasses (BDH) and transfected with the indicated vectors. At 24 h after transfection, cells were washed with cold phosphate-buffered saline (PBS) and fixed with 4% paraformaldehyde (Sigma-Aldrich) for 15 min at 4°C. After being washed with PBS, the cells were permeabilized with ice-cold 0.5% Triton X-100 (COS-7 and NIH 3T3 cells) or 0.1% Triton X-100 (Saos-2 and H1299 cells) for 10 min and then washed with PBS, incubated with DAPI (4′,6′-diamidino-2-phenylindole; 10 mg/ml [Sigma-Aldrich]) for 3 min, and washed again with PBS. Finally, the glasses were mounted with Mowiol (Sigma-Aldrich). The cells were examined under a fluorescence microscope (Nikon). All images were digitally processed by using Adobe Photoshop software. To analyze ARF localization, before incubation with DAPI, cells were blocked with 5% fetal bovine serum in PBS for 30 min, washed with PBS, and then incubated with anti-His polyclonal antibodies (Invitrogen) at 37°C for 1 h. After being washed with PBS three times, cells were incubated with a secondary antibody (Cy3-conjugated anti-mouse immunoglobulin G; ImmunoResearch Laboratory) at room temperature for 30 min.

CAT assay.

H1299 and Saos-2 cells (5 × 10⁵ cells/60-mm dish) were transiently cotransfected, as described above, with Apaf1CAT (0.5 μg), WAF1CAT (0.4 μg), or Hsp70CAT (0.25 μg) promoter reporter constructs and the indicated amounts of the expression plasmids encoding each p63 isoform or p53 with or without the indicated amount of pcDNA-p14^ARF. Cells were collected 48 h after transfection. Equal amounts of cell extracts (10 to 30 μg), determined by the Bradford method (Bio-Rad), were assayed for chloramphenicol acetyltransferase (CAT) activity by using 0.1 μCi of [¹⁴C]chloramphenicol and 4 mM acetyl coenzyme A. Separated products were detected and quantified by using a PhosphorImager (Molecular Dynamics) and ImageQuant software. The pCMV-βgal plasmid (1.5 μg) was used to normalize CAT values for transfection efficiency.

RESULTS

p14^ARF inhibits p63 transcriptional activity.

Under conditions of genotoxic stress or hyperproliferative stimuli, p53 stabilization leads to transcriptional activation and repression of different sets of genes involved in the control of cell proliferation and genomic-repair processes (5, 34). Some of them, such as p21, MDM2, BAX, GADD45, and 14-3-3σ (37) are also activated by p63. Depending on the specific p63 isoform and promoter sequence being tested, wild-type p63 can be either a transcriptional activator, a repressor, or a dominant-negative repressor of the transactivation function (11).

Among the known p53 target promoters, we found that the Apaf1 promoter is efficiently upregulated by all p63 isoforms, with the exception of ΔNp63α (Fig. 1A). The apoptosis protease-activating factor 1 (Apaf1) is a proapoptotic gene that has been demonstrated to be involved in several cell death pathways (13). Figure 1A shows the transcriptional activities and dose responses of the different p63 isoforms on the Apaf1 promoter. In this experiment, we used H1299 cells, a p53-null human lung carcinoma-derived cell line expressing undetectable levels of p63 and p73 (data not shown). Expression of p63 proteins, upon transfection into H1299, was verified by Western blotting and immunodetection with anti-myc antibodies that recognize the myc epitope at the N-terminal of all transfected proteins. Figure 1B shows that the relative abundance of transfected proteins, expressed from pcDNA3 vectors, varies to some extent, with TAα being the more abundant and TAγ and TAβ being the less abundant, in agreement with published results (11).

FIG. 1.

p63 transactivates the Apaf1 promoter. (A) H1299 cells were transiently cotransfected with 0.5 μg of the Apaf1CAT reporter plasmid/dish and the indicated amounts of each p63-expressing plasmid. After 48 h, cells were harvested, and the CAT activity was determined. The basal activity of the reporter was set to 1. The data are presented as the fold induction relative to the sample without effector. Each histogram bar represents the mean of three independent transfection duplicates. The standard deviations are indicated. (B) Expression of transfected proteins. H1299 cells were transiently transfected with 0.5 μg of the indicated plasmids. At 48 h after transfection cells were lysed and 50 μg of the lysates were immunoblotted with anti-myc antibodies.

To investigate the effects of ARF on p63-dependent transcription, we transiently cotransfected H1299 cells with the Apaf1CAT reporter construct, increasing the amounts of either TAp63γ, ΔNp63γ, TAp63α, TAp63β, or ΔNp63β expression vectors and a fixed amount of p14^ARF. Our results clearly indicate that ARF is able to inhibit both TA and ΔN-mediated transactivation (Fig. 2 and data not shown). In Fig. 2 are shown, as representative examples, results obtained with TA and ΔNp63γ. Similar experiments were performed with a different target promoter, p21^WAF, and the TAp63α and γ isoforms as transactivators, and similar results were obtained (data not shown).

FIG. 2.

p63 transactivation is inhibited upon p14^ARF transfection. H1299 cells were transiently cotransfected with the indicated combinations of the following expression plasmids (total DNA 2 μg): Apaf1CAT reporter plasmid (500 ng/dish; bars 1 to 12), a fixed amount of p14^ARF (200 ng/dish; gray hatched bars) or pcDNA (200 ng/dish, open bars) and increasing amounts (25 ng, bars 3 to 4; 50 ng, bars 5 to 6; 75 ng, bars 7 to 8; 100 ng, bars 9 to 10; 200 ng, bars 11 to 12) of TA (A)- or ΔN p63γ (B)-expressing plasmids. After 48 h, cells were harvested, and the CAT activity was determined. The basal activity of the reporter was set to 1. The data are presented as the fold induction relative to the sample without effector. Each histogram bar represents the mean of three independent transfection duplicates. The standard deviations are indicated.

Next, we decided to confirm the observed inhibitory effect of p14^ARF on p63-driven transcription on a different cellular context. We used Saos-2 cells, a human osteosarcoma-derived cell line, expressing neither p53 nor p63. We transiently cotransfected into Saos-2 cells the Apaf1 or p21^WAF CAT reporters and a fixed amount of TAp63γ (0.1 μg) alone or with increasing amounts of p14^ARF-expressing plasmid (0.2, 0.4, and 0.8 μg). As shown in Fig. 3A, increasing expression of ARF resulted in a progressive reduction of TAp63γ-driven transcription up to the background level. Similar results were obtained when ΔNp63γ was used as transactivator (Fig. 3B). On the other hand, when p14^ARF was cotransfected with p53 by the same experimental procedure, we did not observe reduction but a slight increase of p53-driven transcription (Fig. 3C), in agreement with the known function of ARF.

FIG. 3.

p14^ARF inhibits TAγ- and ΔNp63γ-dependent transactivation. (A) Saos-2 cells were transiently cotransfected with the Waf1CAT (0.4 μg/dish, open bars) or the Apaf1CAT (0.5 μg/dish, gray bars) reporter plasmids, a fixed amount of TAp63γ (0.1 μg/dish, bars 5 to 12) or increasing amounts of p14^ARF (0.2 μg/dish, bars 7 and 8; 0.4 μg/dish, bars 9 and 10; 0.8 μg/dish, bars 3, 4, 11, and 12) expressing plasmids. After 48 h, cells were harvested, and the CAT activity was determined. (B) Saos-2 cells were transiently cotransfected with 0.5 μg of Apaf1CAT reporter plasmid (bars 1 to 6), a fixed amount of ΔNp63γ (0.1 μg/dish, bars 3 to 6) and increasing amounts of p14^ARF (0.2 μg/dish, bar 4; 0.4 μg/dish, bar 5; 0.8 μg/dish, bars 2 and 6)-expressing plasmid. After 48 h, cells were harvested, and the CAT activity was determined. The basal activity of the reporter was set to 1. The data are presented as the fold induction relative to the sample without effector. Each histogram bar represents the mean of three independent transfection duplicates. The standard deviations are indicated. (C) Saos-2 cells were transiently cotransfected with the Waf1CAT (0.4 μg/dish, open bars) or the Apaf1CAT (0.5 μg/dish, gray bars) reporter plasmid, a fixed amount of p53 (0.1 μg/dish, bars 5 to 12), and increasing amounts of p14^ARF (0.2 μg/dish, bars 7 and 8; 0.4 μg/dish, bars 9 and 10; 0.8 μg/dish, bars 3, 4, 11, to 12)-expressing plasmids. After 48 h, cells were harvested, and the CAT activity was determined. The basal activity of the reporter was set to 1. The data are presented as the fold induction relative to the sample without effector. Each histogram bar represents the mean of three independent transfection duplicates. The standard deviations are indicated. (D) Saos-2 cells were transiently cotransfected with 0,5 μg of ΤΑp63γ (lanes 2 to 5) and increasing amount of p14^ARF-expressing plasmid (1.0 μg/dish, lane 4; 2.0 μg/dish, lane 5; 4.0 μg/dish, lanes 1 and 5). At 48 h after transfection the cells were lysed, and 25 μg of the lysates was immunoblotted with anti-p63 and anti-p14^ARF antibodies. (E) Saos-2 cells were transiently cotransfected with 0.5 μg of ΔNp63γ (bars 2 to 5) and increasing amounts of p14^ARF-expressing plasmids (1.0 μg/dish, bar 3; 2.0 μg/dish, bar 4; 4.0 μg/dish, bars 1 and 5). At 48 h after transfection cells were lysed, and 10 μg of the lysates was immunoblotted and revealed with anti-p63 and anti-p14^ARF antibodies.

At first, we reasoned that a decrease in the p63 protein level could account for the observed inhibition of p63 transcriptional activity by p14^ARF. On the other hand, it has been shown that ARF is able to promote polyubiquitination and degradation of proteins such as E2F1 (6) and B23 (15). Therefore, we decided to check whether or not p14^ARF affects the level of p63 exogenous protein. To this aim, we compared levels of p63 protein with or without ARF coexpression. We transfected into Saos-2 cells p63 alone or p63 with p14^ARF and examined the level of exogenous p63 protein by Western blotting. As shown in Fig. 3D and E, the abundance of transfected TAp63γ and ΔNp63γ proteins remained unchanged upon p14^ARF overexpression.

To validate our data, we felt it was important to show a correlation between the decrease of p63 transactivation potential by p14^ARF and modulation of endogenous target genes. For this reason, we checked the expression of p21^WAF and MDM2 endogenous genes upon transfection of TAp63γ alone or with increasing amounts of p14^ARF. For this experiment, we used the H1299 cells. In mock-transfected cells endogenous p21^WAF protein was detectable, whereas no MDM2 was observed (Fig. 4). As shown in Fig. 4, endogenous p21^WAF and MDM2 proteins were both induced by transfection of TAp63γ alone (compare lanes 1 and 3); the addition of increasing amounts of p14^ARF efficiently repressed this induction in a dose-dependent way (compare lane 3 with lanes 4 to 7). Again, the level of TAp63γ protein appeared to be unaffected by p14^ARF coexpression. These results are consistent with our data from CAT reporter assays.

FIG. 4.

Induction of endogenous MDM2 and p21WAF upon p63 transfection is counteracted by p14^ARF cotransfection. Immunoblot analysis of total protein lysates from H1299 cells cotransfected with a fixed amount of TAp63γ (0.5 μg/dish, lanes 3 to 7) and increasing amounts (0.5 μg/dish, bar 4; 1 μg/dish, bar 5; 2 μg/dish, bar 6; 4 μg/dish, bars 2 and 7) of p14^ARF-expressing plasmids was performed. An anti-actin antibody was used for normalization of cell lysate loading.

The p63 transcriptional activity in the presence of p14^ARF was also investigated in repression assays with the Hsp70 promoter. The Hsp70 promoter has a high level of intrinsic activity even without activation by heat shock and is efficiently repressed by p53 (1). Saos-2 cells were cotransfected with the Hsp70CAT construct, a fixed amount of each of the different p63 constructs, with or without a fixed amount of p14^ARF plasmid DNA. According to a previous report (11), all p63 isoforms, with the exception of ΔNp63α, are able to transrepress this promoter. As shown in Fig. 5, even though to a different extent, p14^ARF was able to decrease the ability of both TA and ΔNp63 isoforms to transrepress the Hsp70 promoter. It is noteworthy that p14^ARF alone causes a decrease of the basal activity of the Hsp70 promoter. Taken together, these data suggest that p14^ARF is able to inhibit both activation and repression of p53-target promoters by p63, leaving unaltered the intracellular p63 protein level.

FIG. 5.

p14^ARF counteracts p63-dependent transrepression. Saos-2 cells were cotransfected with a fixed amount of Hsp70 reporter plasmid (0.25 μg/dish), 0.1 μg of the indicated p63 expression plasmids, and 0.4 μg of p14^ARF-expressing plasmid (open bars) or 0.4 μg of pcDNA (gray bars). The basal activity of the reporter was set to 1. The data are presented as the fold repression relative to the sample without effector. Each histogram bar represents the mean of three independent transfection duplicates. The standard deviations are indicated.

Physical interaction of TA and ΔNp63 with p14^ARF in mammalian cells.

In searching for a mechanism for ARF inhibition of p63-driven transcription, we decided to examine whether ARF can associate with p63. Human ARF was expressed alone or together with TA or ΔNp63 into the ARF-null NIH 3T3 cells to perform coimmunoprecipitation assays. NIH 3T3 cells did not have detectable p63 endogenous protein. Using anti-human ARF polyclonal antibodies, coimmunoprecipitation of TA and ΔNp63γ (Fig. 6A), TA and ΔNp63β (Fig. 6B), or TA and ΔNp63α (Fig. 6C), occurred only when each of these proteins was coexpressed with p14^ARF; p63-p14^ARF complexes were not found when either protein alone was expressed in cells. Similar results were obtained when the cellular lysates were immunoprecipitated with anti-p63 antibodies (Fig. 6D and data not shown). The interaction between p14^ARF and TA or ΔNp63γ and α isoforms was also observed in H1299 cells in a similar experimental procedure (Fig. 6E and data not shown).

FIG. 6.

Coimmunoprecipitation of TA and ΔN p63 proteins with p14^ARF. NIH 3T3 cells were transfected with 1 μg of expression plasmids encoding TA or ΔNp63γ (A), TA or ΔNp63β (B), and TA or ΔNp63α (C and D) alone or together with 2 μg of p14^ARF-expressing plasmid. Cellular extracts were immunoprecipitated with anti-ARF antibodies (A, B, and C) or with anti-p63 antibodies (D). Immunocomplexes were analyzed with anti-p63 and anti-p14^ARF antibodies. (E) H1299 cells were transfected with 1 μg of expression plasmids encoding TA or ΔNp63α alone or together with 2 μg of p14^ARF-expressing plasmid. Cellular extracts were immunoprecipitated with anti-ARF antibodies and analyzed with anti-p63 and anti-p14^ARF antibodies. (F) HaCaT cells were transfected with 2 μg of p14^ARF-expressing plasmid. Cellular extracts were incubated with anti-ARF antibodies or, as control, with an unrelated antibody. Immunocomplexes were analyzed with anti-p63 and anti-p14^ARF antibodies.

Moreover, to confirm the observed interaction in a more physiological context, we performed an immunoprecipitation experiment into HaCaT cells, a spontaneously immortalized keratinocyte cell line expressing detectable levels of endogenous ΔNp63α. Since no endogenous p14^ARF protein was revealed, HaCaT cells were transiently transfected with a p14^ARF-expressing plasmid. Cellular extracts were incubated with anti-human ARF antibodies. Immunoprecipitates were blotted and probed with anti-p63 and anti-p14 antibodies. As shown in Fig. 6F, ΔNp63α was coimmunoprecipitated only when p14^ARF is expressed. This experiment confirmed that a complex between p14^ARF and p63 occurs in mammalian keratinocyte cells.

To identify the region of p63 essential for the interaction with p14^ARF, three deletion mutants of p63 were constructed starting from the TAp63α or ΔNp63γ wild-type constructs. A schematic representation of these mutants is shown in Fig. 7A. First, we verified by Western blotting that the mutant proteins were correctly expressed upon transfection into NIH 3T3 cells. As shown in Fig. 7B, with the exception of the Δ1-86 mutant, which appears to be relatively less abundant, the expression level of the tested mutants was comparable to that of wild-type TAp63γ. Each mutant was assayed in NIH 3T3 cells for its interaction with p14^ARF by coimmunoprecipitation experiments. Removal of the carboxy-terminal portion (Δ297-449 mutant) encompassing the entire TID, SAM, and OD domains of TAp63α does not impair p63-ARF interaction (Fig. 7C). On the other hand, deletion of either amino acids from 1 to 86 (Δ1-86 mutant), including both the TA2 and the PRD domains, or the first 26 amino acids of ΔNp63γ (Δ1-26 mutant), encompassing only the TA2 domain, completely abolished p63-p14^ARF interaction (Fig. 7D).

FIG. 7.

Identification of p63 domains involved in p14^ARF interaction. (A) Schematic representation of the p63 deletion constructs used in this experiment. (B and C) NIH 3T3 cells were transfected with 1 μg of expression plasmids encoding wild-type TAp63γ or the indicated p63 deleted constructs alone or together with 2 μg of p14^ARF-expressing plasmid. Cellular extracts were immunoprecipitated with anti-ARF antibodies. (D) Equal amounts of total protein extracts from cells transfected with the indicated plasmids were immunoblotted with anti-p63 antibodies.

To define which parts of the ARF protein are required for the physical interaction with p63, we performed coimmunoprecipitation of in vitro-cotranslated proteins. We assayed two deletion mutants: the first lacking amino acids from 66 to132 (p14^ARFΔ66-132) and the second lacking the N-terminal 1 to 38 amino acids (p14^ARFΔ1-38) (Fig. 8A). We were able to show that deletion of amino acids 1 to 38 (Fig. 8B, compare lanes 2 and 3) impairs the interaction with p63, whereas the C-terminal portion of the protein appears to be dispensable for the interaction (Fig. 8B, lane 4).

FIG. 8.

Immunoprecipitations of in vitro-translated proteins and mapping of the p14^ARF domain involved in p63 interaction. (A) Schematic representation of the p14^ARF deletion constructs used in this experiment. (B) A total of 40 μl of in vitro-translated ³⁵S-labeled TAp63γ alone (lane 1) or cotranslated with either p14^ARF (lane 2), p14^ARF(Δ1-38) (lane 3), or p14^ARF(Δ66-132) (lane 4) was immunoprecipitated with anti-His antibodies (recognizing the poly-His tag fused to the 5′ of the coding region of wild-type and mutant p14^ARF). The immunocomplexes were analyzed on an SDS-15% polyacrylamide gel and detected by autoradiography. (C) A total of 5 μl of in vitro-translated proteins used in the above-described immunoprecipitations was analyzed on an SDS-15% polyacrylamide gel and then detected by autoradiography.

ARF decreases the binding of p63 to a p53 DNA-binding site in the p21^WAF promoter.

Since we have found that ARF binds to p63 and inhibits p63-driven transcription, we decided to examine whether p14^ARF impairs the binding of p63 to a canonical p53 consensus sequence. We thus compared the binding of p63 in presence or absence of ARF. We performed an in vitro DNA-binding assay using, as a target DNA, a radiolabeled duplex oligonucleotide representing a p53-binding site previously identified in the p21^WAF promoter (44). Incubation of the radiolabeled oligonucleotide with in vitro-translated TAp63γ led to the formation of a specific protein-DNA complex (Fig. 9A, lane 1). The specificity of the TAp63γ-DNA complex was tested by a competition experiment: a 100× cold molar excess of the same oligonucleotide used as a probe completely abolished the binding, whereas a nonrelevant control oligonucleotide had no effect (Fig. 9A, lanes 1, 2, and 3). The identity of the TAp63γ-DNA complex was confirmed by a supershift experiment (Fig. 9A, lanes 4, 5, 6, and 7) in which the in vitro-translated TAp63γ protein was incubated prior to the binding reaction with an antibody recognizing the poly-His tag fused upstream of the coding region of TAp63γ, with a nonrelevant anti-p21 antibody as a control and with an antibody recognizing the p63 DNA-binding domain. Interestingly, when TAp63γ was cotranslated with p14^ARF, the binding was significantly reduced (Fig. 9A, lanes 11 and 12). A Western blot analysis of the in vitro-translated proteins showed no significant differences in the relative abundance of the TAp63γ protein translated alone or in presence of p14^ARF (Fig. 9B). These observations indicate that TAp63γ specifically binds to a p53 consensus in the p21^WAF promoter and that p14^ARF impairs this binding.

FIG. 9.

Protein DNA-binding assay. (A) A 40-bp radiolabeled oligonucleotide containing a p53-binding site was incubated with in vitro-translated TAp63γ. The specificity of the binding was assessed by competition with either a 100× cold molar excess of the same oligonucleotide used as a probe or of an unrelated oligonucleotide (compare lane 1 to lanes 2 and 3). Supershift (lanes 4 to 7) was carried out with anti-His antibodies (recognizing the poly-His tag fused to the 5′ of the coding region of TAp63γ) (lane 5), anti-p63 antibody (recognizing the DNA-binding domain of p63) (lane 7) or, as a control, anti-p21 antibody (lane 6). The binding reaction was also carried out, as a control, with the rabbit reticulocytes (lane 9) and with in vitro-translated p14^ARF (lane 8). Binding reactions with TAp63γ translated alone (lane 11) or cotranslated with p14^ARF (lane 12) are shown. (B) The same aliquots of in vitro-translated TAp63γ or TAp63γ/ARF used in the above-described binding reactions were analyzed by Western blotting with anti-His and anti-ARF antibodies.

p63 relocalizes ARF in the nucleoplasm.

The p14^ARF protein is mainly located into nucleoli. The nucleolar localization of ARF is even more predominant when it is transiently overexpressed or expressed in a subset of p53-deficient tumor cell lines (21). Moreover, ARF is able to sequester its binding partners, in particular MDM2 and E2F1, to this location (6, 42). We reasoned that a possible mechanism through which the p14^ARF protein could inhibit the transcriptional activity of p63 is by sequestering it into the nucleolus where it cannot perform its transcriptional functions. To investigate this hypothesis, we first monitored the subcellular localization of ectopically transfected p63 isoforms.

Therefore, TA and ΔNp63 isoforms were produced as green fluorescent protein (GFP) fusion proteins and expressed in H1299, COS-7, Saos-2, and NIH 3T3 cell lines. The α and β isoforms of both TA and ΔNp63 exhibited a very similar localization pattern in the various cell lines, i.e., they were uniformly distributed in the nucleus with nucleolar sparing. A different pattern was observed for the TAp63γ and ΔNp63γ proteins that appeared to be distributed both in the nucleus and cytoplasm. Again, we did not observe TAp63γ or ΔNp63γ proteins in the nucleoli. Moreover, most TAp63γ and ΔNp63γ expressing cells showed nuclear and cytoplasmic dots that were often located on the surface of the nuclear membrane (data not shown). Similar results were obtained when cells expressing TA and ΔNp63 isoforms lacking the GFP domain were revealed with anti-p63 antibodies (data not shown).

In agreement with previous observations, we found that transfected p14^ARF accumulates predominantly into the nucleoli of NIH 3T3, COS-7, Saos-2, and H1299 (39, 42). In the remaining cells it shows a diffuse nuclear distribution or a nuclear distribution with nucleolar sparing (Fig. 10A and Table 1). B23 anti-nucleolin antibody was used in these experiments as a control for nucleolar localization (data not shown). When p63 and ARF were cotransfected in a 1-to-1 ratio, the above-described subcellular distribution of p63 proteins remained unaltered (data not shown). Unexpectedly, a large percentage of cells expressing both TAp63 isoforms and ARF proteins exhibited a complete exclusion of p14^ARF from nucleoli. The results obtained with TAp63β are shown as a representative example (Fig. 10B). Conversely, no significant change of the typical subcellular distribution of p14^ARF was seen when p14^ARF was cotransfected either with ΔNp63 isoforms (see Table 1) or p53 (data not shown). According to the above results, the C-terminal deleted (Δ297-449) TAp63 protein efficiently induced p14^ARF nucleolar exclusion, whereas the Δ1-26 and the Δ1-86 proteins were ineffective, underlining the importance of the p63 amino-terminal region in this phenomenon (Table 1).

FIG. 10.

TAp63 impairs p14^ARF nucleolar localization. COS-7 cells were seeded on glass coverslips and transfected with 0.5 μg of expression vector encoding p14^ARF (A) or cotransfected with 0.5 μg of p14^ARF and 0.5 μg of GFP::TAp63β fusion protein (B). The cells were examined under a fluorescence microscope (Nikon). All images were digitally processed by using Adobe Photoshop. The subcellular localization of GFP::TAp63β and p14^ARF are shown. Numerical data indicate the percentage of cells showing different ARF localizations.

TABLE 1.

Percentage of cells showing nucleolar localization of p14^ARF upon cotransfection of p14^ARF (0.5 μg) with the indicated p63 isoforms (0.5 μg) in Saos-2, H1299, and NIH 3T3 cells and the percentage of cells showing nucleolar localization of p14^ARF upon cotransfection of p14^ARF (0.5 μg) with the indicated wild-type or mutant p63 isoforms (0.5 μg) in COS-7 cells^a

Cell line	% of cells (mean ± SD) showing localization of p14ARF upon cotransfection with:
	pcDNA	TAα	TAβ	TAγ	ΔNα	ΔNβ	ΔNγ	TAαΔ(297-449)	ΔNγΔ(1-26)	ΔNγΔ(1-86)
Saos-2	90 ± 3	39 ± 5	ND	26 ± 2	85 ± 2	ND	90 ± 5
H1299	93 ± 3	40 ± 7	ND	10 ± 5	86 ± 4	ND	88 ± 3
NIH 3T3	90 ± 5	40 ± 5	10 ± 5	10 ± 3	97 ± 2	90 ± 5	95 ± 3
COS-7	95 ± 2	85 ± 5	30 ± 3	35 ± 5	96 ± 3	94 ± 5	95 ± 3	27 ± 3	96 ± 2	95 ± 3

^aNumerical data represent the means of three independent experiments. ND, not determined.

Remarkably, the TAp63α isoform appears to be significantly less efficient than the other TA isoforms in mediating the exclusion of ARF from the nucleolus. However, when increasing amounts of TAp63α plasmid were cotransfected in COS-7 cells, with a fixed amount of ARF-expressing plasmid, the proportion of cells showing ARF nucleolar localization decreased in a dose-dependent manner up to 40% (Fig. 11).

FIG. 11.

TAp63α and ΔNp63α affect p14^ARF nucleolar localization. COS-7 cells were transfected with 0.5 μg of the expression vector encoding p14^ARF alone or in presence of increasing amounts (0.5, 1, and 1.5 μg) of either TAp63α- or ΔNp63α-expressing plasmids. The percentage of cells showing p14^ARF nucleolar localization is reported. The results are expressed as a mean value of three independent experiments.

DISCUSSION

We present here evidence indicating a physical and functional relationship between p63 and p14^ARF. To our knowledge, this is the first report that supports an intimate association between p14^ARF and a member of the p53 family. Our data indicate that p63 is able to associate with p14^ARF, both in the TA and ΔN versions, in different mammalian cell lines. Coimmunoprecipitation experiments of in vitro-translated proteins support the conclusion that a direct physical association between p14^ARF and p63 occurs. By deletion analysis, we have shown that this interaction is mediated by the N-terminal region from amino acids 1 to 26 of ΔNp63. Remarkably, only 12 amino acids of this region (encompassing amino acids 15 to 26 of ΔNp63 and amino acids 109 to 120 of TAp63) are in common between TA and ΔN isoforms. Since interaction between p63 and p14^ARF occurs with both TA and ΔN isoforms, we infer that this stretch of 12 amino acids contains residues that might be crucial for p63-p14^ARF association. On the other hand, the p63 C-terminal region seems to be dispensable for p14^ARF-p63 binding. Two distinct observations support this hypothesis; first, p63 proteins with either the α, β, or γ alternative type of carboxy-terminal equally associate with p14^ARF and, second, the (Δ297-449)TAp63 mutant, in which the entire TID, SAM and OD domains were removed, is still able to interact with p14^ARF.

Our results suggest that the ARF carboxy-terminal portion is dispensable for the ARF-p63 interaction, whereas the N-terminal region seems to be involved. It has been reported (22) that the first N-terminal 22 amino acids of p14^ARF retain the ability to bind MDM2, but a second, less-efficient, MDM2-binding site is present in the ARF carboxy-terminal region. However, a more accurate definition of the region of ARF interacting with p63 is necessary to verify whether the MDM2-ARF binding overlaps with the p63-ARF binding.

Normally, the ARF protein localizes in the nucleolus. However, we found that p63 proteins are all prevalently localized in the nucleoplasm and excluded from nucleolus. Interestingly, we have noticed a remarkable decrease of the nucleolar fraction and a corresponding increase in the nucleoplasmic fraction of p14^ARF upon TAp63 overexpression, findings that support a physical association between the two proteins. The importance of the amino-terminal TA domain in promoting p14^ARF nucleolar exclusion is emphasized by the observation that ΔN isoforms leave unaltered ARF nucleolar localization, whereas the TAp63α(Δ297-449) mutant shows the same behavior of the β and γ TAp63. The question remaining is why TAp63α appears to be less efficient than the other TAp63 isoforms in promoting ARF nucleolar exclusion. Concerning this point we recall that the extreme C-terminal domain (TID), unique to the α isoforms, binds to the N-terminal TA domain through an intramolecular interaction. This binding is both necessary and sufficient for protein stabilization and transcriptional inhibition of TAp63 (32). Thus, we suggest that such intramolecular association could, in a similar way, mask sequences located in the TA domain that are responsible for p14^ARF nucleolar exclusion. The mechanism by which coexpression of TAp63 isoforms and p14^ARF alters the subcellular localization of p14^ARF is under investigation; however, we think that residues located in the TA domain of p63 might increase the binding affinity between p63 and p14^ARF so that, once established, the complex keeps p14^ARF in the nucleoplasmic compartment. However, we cannot exclude that interaction of p14^ARF with the TA domain of p63 might hamper association with additional molecular partners that regulate p14^ARF nucleolar import.

The ARF tumor suppressor acts as a sensor of hyperproliferative signals emanating from oncoproteins and inducers of S-phase entry, such as Myc, E1A, Ras, and E2F-1 (reviewed in reference 33). ARF, in turn, triggers p53-dependent growth arrest in the G₁ and G₂ phases of the cell cycle or, in the presence of appropriate collateral signals, sensitizes cells to apoptosis. ARF binds directly to MDM2, enabling transcriptionally active p53 to accumulate in the nucleoplasm (43); (42). Emerging evidence suggests that ARF can, though less efficiently, suppress the proliferation of cells that express mutant p53 or lack both MDM2 and p53, implying the existence of p53/MDM2-independent functions of ARF working through interactions with other regulators (41).

Very recently, it has been reported that ARF may function in coordinating cell growth with proliferation through its interaction with B23, a nucleolar protein involved in ribosome biogenesis. Inducing B23 degradation, p14^ARF inhibits rRNA processing (15, 35). Thus, it is possible that the phenomenon of ARF nucleolar exclusion may interfere with the specific role of ARF in controlling ribosome biogenesis. Further investigations are necessary to clarify this point.

Other p53-independent functions of ARF include repression of E2F (9) and NF-κB (31) activity. Various proteins that associate with ARF have been identified, including Spinophilin (39), MdmX, Pex19, CARF, a novel serine-rich protein (30), and Tat binding protein 1 (27). However, the extent to which these ARF complexes modulate the cell cycle inhibitory action of ARF is not well established.

Interestingly, our data show that p14^ARF inhibits the ability of p63 to enhance the expression of various endogenous genes. Moreover, since we have shown the inhibitory function of ARF on both p63 transactivation and transrepression assays, it appears that p14^ARF antagonizes p63 whatever is the activity of p63 on that particular promoter. Actually, we could test this phenomenon only on p53 target promoters such as p21^WAF, Apaf1, and Hsp70 because no specific p63 targets have been identified thus far. If this phenomenon is also observed on p63-specific target promoters, it would indicate a more general role, one not necessarily involving p53.

We have previously reported that overexpression of MDM2 induces TAp63 protein stabilization and transcriptional activation and that both effects are counteracted by ARF coexpression (3). In principle, one could reasonably suggest that the observed inhibitory effect of p14^ARF on p63-dependent transcription might be due, at least in MDM2-expressing cells, to the well-described antagonistic effect that p14^ARF exerts on MDM2. Actually, two different considerations argue against this hypothesis. First, we observed that p14^ARF inhibition of p63-dependent transcription does not correlate with any decrease in p63 protein intracellular levels. Second, our data demonstrate a direct interaction between p14^ARF and p63, whereas the relationship between ARF and p53 is mediated by MDM2.

The discovery of the p53 homologs has sparked speculation on how surveillance of cellular integrity might be achieved through the network of p53-like proteins characterized by similar structural and biochemical properties. Actually, both p63 and p73 share several p53 transcriptional gene targets and can induce apoptosis and cell growth arrest. Increasing evidence points to highly tissue specific mechanisms and differential regulation as the principal factors accounting for the majority of the differences in biological function (3, 40).

Recent studies demonstrated that among the p63 proteins TAp63 isoforms are the first to be expressed during embryogenesis and are required for commitment to an epithelial stratification program. ΔNp63α is the predominant isoform expressed in mature epidermis (20). Since TAp63 isoforms seem to inhibit terminal differentiation, they must be counterbalanced by ΔNp63 to allow cells to respond to signals required for the maturation of embryonic epidermis (18). On the other hand, UVB-induced DNA damage decreases levels of ΔNp63α, whereas the levels of the TAp63 isoforms increase. This mechanism is a prerequisite for UV-induced apoptosis in the skin (20, 45).

Interestingly, it has been shown that, in epidermis, p53 is expressed in the basal layer, where it plays a surveillance role in progenitor and/or stem cell renewal (24). The p53 protein acts not only to keep stem cells quiescent but also to ensure the correct control of cell cycle and cell division as keratinocytes proliferate. In fact, p53-deficient epidermal keratinocytes differentiate normally, but they are affected in their growth control and underwent malignant transformation (7). On the other hand, it is still unclear whether ARF is regulated under physiological conditions, and little is understood concering its role in adult tissue homeostasis. Our observation that p14^ARF associates with p63-inhibiting p63 transcriptional activity suggests, that under p14^ARF overexpression, the pool of p63 proteins might be kept inactive in a p63-p14^ARF complex.

Hence, we speculate that under mitogenic stimuli, p14^ARF physically associate with TA and perhaps dominant-negative p63 isoforms, removing them from p53/p63-responsive promoters. This process might turn on p53 transcriptional functions, activating the p53-dependent checkpoint control. Actually, our EMSA experiments show that p14^ARF is able to affect the binding of TAp63γ to a canonical p53 consensus sequence, lending support to this hypothesis. More specific experimental in vivo approaches, such as chromium immunoprecipitation assays, will elucidate the real contribution of ARF in the regulation of transcription of p53 target genes in their natural setting, given the important role of chromatin structure in the regulation of gene expression. Studies to address this question are under way.