The tumour suppressor ARF (alternative reading frame) is one of the most important oncogenic stress sensors. ARF provides an ‘oncogenic checkpoint’ function through both p53-dependent and p53-independent mechanisms. In the present study, we demonstrate a novel p53-independent interaction between p14ARF and the adenovirus oncoprotein E1A. p14ARF inhibits E1A transcriptional function and promotes ubiquitination-dependent degradation of E1A. p14ARF overexpression relocalizes E1A into the nucleolus and inhibits E1A-induced cellular DNA replication independent of p53. Knockdown of endogenous p14ARF increases E1A transactivation. In addition, E1A can competitively inhibit ARF–Mdm2 (murine double minute 2) complex formation. These results identify a novel binding partner of p14ARF and reveal a mutually inhibitory interaction between p14ARF and E1A. We speculate that the ARF–E1A interaction may represent an additional host defence mechanism to limit viral replication. Alternatively, the interaction may allow adenovirus to sense the functional state of p53 in host cells, and fine-tune its own replication activity to prevent the triggering of a detrimental host response.
The INK4a [inhibitor of CDK (cyclin-dependent kinase) 4a]/ARF (alternative reading frame) tumour-suppressor locus encodes two functionally distinct proteins, p16INK4a and ARF (p14ARF in humans and p19ARF in mice) . As an inhibitor of CDKs, p16INK4a inhibits an E2F-dependent transcriptional programme to decelerate the cell cycle by hyperphosphorylation of pRb (retinoblastoma protein) via the CDK4–CDK6 complex. The ARF protein predominantly localizes in the nucleolus and its expression levels are very low in both developing and adult tissues. The principal function of the ARF tumour suppressor is to suppress abnormal cell proliferation in response to high levels of oncogenic signalling from factors such as Myc, Ras, E1A, Abl and E2F1, resulting in cell-cycle arrest and/or apoptosis .
Previous studies have indicated that ARF provides a tumour-suppressive function through both a p53-dependent pathway and/or a p53-independent pathway. The classical p53-dependent pathway model indicates that ARF responds to oncogenic signals, recruits Mdm2 (murine double minute 2) into the nucleolus and inhibits Mdm2-mediated ubiquitination of p53, resulting in p53 stabilization, cell-cycle arrest or apoptosis [3,4]. However, other studies have demonstrated that ARF can inhibit Mdm2 and activate p53 without sequestering Mdm2 into the nucleolus. It was proposed that ARF forms a ternary complex with Mdm2 and p53 in the nucleoplasm and blocks p53 nuclear export and degradation . In addition to Mdm2, ARF-BP1 (ARF-binding protein 1)/Mule is another key E3 ubiquitin ligase for p53 and regulates the p53 cell-cycle-checkpoint function in an Mdm2-independent manner .
Several lines of evidence indicate that ARF also exerts its function independent of p53. Over 25 ARF-binding proteins have been identified to date . For example, ARF interacts with and negatively regulates Myc via a p53-independent negative-feedback mechanism [7,8]. ARF interacts with p120E4F and exhibits enhanced cell-cycle-inhibitory activity . Furthermore, other studies have implicated a role for ARF in regulating ribosomal RNA transcription and ribosome biogenesis [10–13]. Other partners of ARF include the E2F transcription factors [14,15], ANCO1 (ankyrin repeats co-factor 1) , Bcl6 , CARF (collaborator of ARF) , LZAP (LXXLL/leucine-zipper-containing ARF-binding protein) , CtBP (C-terminal-binding protein) , p63 , PXF (peroxisomal farnesylated protein), spinophilin , TBP1 (TATA-box-binding protein 1) , Tip60 [Tat (transactivator of transcription)-interactive protein 60 kDa] , HIF-1 (hypoxia-inducible factor 1) , WRN (Werner's syndrome protein) , YY1 (Yin Yang 1)  and HPV (human papillomavirus) 16-E7 . Identifying novel ARF-binding proteins is critical for understanding the p53-independent function of ARF.
Exploring the molecular mechanisms regulating the oncogenic activities of adenovirus has contributed significantly to the understanding of cancer biology. The adenovirus E1A oncoprotein exerts a range of activities to facilitate virus propagation and cellular transformation by targeting several key cell-cycle regulators including members of the pRb family, the p300/CBP [CREB (cAMP-response-element-binding protein)-binding protein] transcriptional co-activators, and the p400–TRRAP (transformation/transcription-domain-associated protein) chromatin-remodelling complex . In addition to its oncogenic characteristics, E1A has been reported to be useful in cancer gene therapy for its apoptosis-promoting function . A series of studies have shown that E1A can stabilize p53 by inducing ARF , which antagonizes the E3 ubiquitin ligase activity of Mdm2. It has also been reported that E1A stabilizes p53 by targeting Mdmx (Mdm4 homologue) in an ARF-independent manner . On the other hand, ARF expression is also induced after expression of certain viral proteins or in virus-infected cells, suggesting a potential role for ARF in the surveillance of viral infection [29,34]. Whether ARF exerts its antiviral function responding to adenovirus E1A oncoprotein remains unclear.
Previous studies have shown that E1A has the ability to induce p19ARF expression . In the present study, we show that p14ARF interacts directly with E1A and inhibits E1A transcriptional activity via a p53-independent mechanism. p14ARF overexpression relocalizes E1A into the nucleolus. Furthermore, we demonstrate that p14ARF promotes ubiquitination and proteasomal degradation of E1A, and blocks E1A-induced cellular DNA replication. We propose that there is a direct negative-feedback mechanism between p14ARF and E1A as a result of their interaction.
MATERIALS AND METHODS
Cell culture and reagents
Human cell lines HeLa (p53-defective), H1299 (ARF wild-type, p53-null), U2OS (ARF-null, p53 wild-type), HEK (human embryonic kidney)-293T (expressing E1A, p53 wild-type), HCT116 (ARF-mutant, p53 wild-type), SaoS2 (p53-null, Mdm2-undetectable), HaCat (ARF-undetectable, p53-mutant), H1299-E2F1–ER [an H1299 cell line stably expressing E2F1–ER (oestrogen receptor) complex] were maintained in DMEM (Dulbecco's modified Eagle's medium) with 10% (v/v) fetal bovine serum. H1299-E1A, an H1299 stable cell line expressing adenoviral E1A, was established by transfecting adenoviral E1A DNA into H1299 cells and selection of G418-resistant colonies. CHX (cycloheximide), MG132, 4-OHT (4-hydroxytamoxifen) and BrdU (bromodeoxyuridine) were obtained from Sigma.
Plasmid constructs and antibodies
The Mdm2 and Myc–ARF expression plasmids have been described previously . FLAG–ARF, GFP (green fluorescent protein)–ARF and its deletion mutant plasmids were constructed by PCR from the Myc–ARF plasmid template and cloning into the pCDNA3.1-GFP vector. The E1A13S expression plasmid was a gift from Dr F.A. Dick . Adenovirus E3–luciferase reporter and E4–luciferase reporter plasmids were a gift from Dr Peter Pelka and Dr Joe S. Mymryk . Lenti-FLAG–ARF (lentivirus expressing FLAG–ARF) and Lenti-E1A (lentivirus expressing E1A) were cloned from pCDNA3.1-FLAG-ARF and pCMV-E1A into pLVX-IRES-Zs Green1 vector (Clontech Laboratories) respectively. The EGFP (enhanced GFP)–E4F1 expression plasmid was a gift from Dr Laurent Le Cam . The following antibodies were used: mouse anti-FLAG M2 antibody (Sigma), rabbit anti-FLAG antibody (Santa Cruz Biotechnology), rabbit anti-Myc antibody (Cell Signaling Technology), rabbit anti-E1A antibody (Santa Cruz Biotechnology), mouse anti-β-actin antibody (Sigma), rabbit anti-Mdm2 antibody (Santa Cruz Biotechnology), rabbit anti-GFP antibody (Santa Cruz Biotechnology), goat anti-ARF antibody (Santa Cruz Biotechnology) and mouse anti-BrdU antibody (Sigma).
Transfection and lentivirus infection
Transient transfections were carried out using the calcium phosphate method or Lipofectamine™ 2000 (Invitrogen) according to the manufacturer's instructions. For lentivirus infection, the virus was packaged in HEK-293T, H1299, H1299-E1A and HaCat cells, which were incubated with the virus for the times indicated. An MOI (multiplicity of infection) of 20 PFU (plaque-forming units)/cell was used for each virus.
Gene silencing by siRNA (small interfering RNA)
siRNA duplexes were transfected into cells using Oligofectamine™ (Invitrogen) according to the manufacturer's instructions. The siRNA oligonucleotide sequence for p14ARF was 5′-CGGGAAACUUAGAUCAUCATT-3′. A 21 nt siRNA duplex corresponding to a non-relevant gene (lacZ) was used as a control.
Luciferase reporter assay
Cells (2 × 105/well) were plated in six-well plates and transfected with a mixture of 0.5 μg of the adenovirus E4–luciferase or E3–luciferase reporter plasmid (pGL3-E4 and pGL3-E3 respectively) and the indicated amounts of plasmids expressing ARF or its deletion mutants. For normalization of transfection efficiency, 0.2 μg of pRL-TK (Promega) was co-transfected which expressed Renilla reniformis luciferase under the regulation of the HSV-TK (herpes simplex virus thymidine kinase) promoter. The activity of firefly luciferase was measured with the dual-luciferase assay kit (Promega) and results were obtained from at least three independent experiments, each run in triplicate.
Immunofluorescence and BrdU incorporation
Cells were seeded on coverslips in six-well plates and transfected with 0.5–2 μg of plasmids 12 h later using the Lipofectamine™ 2000 (Invitrogen) method. At 42 h post-transfection, the cells were washed in PBS and fixed in 2% (w/v) formaldehyde. Cells were immunostained for 2 h with primary antibodies followed by a 1 h exposure to Alexa Fluor® 488- or Alexa Fluor® 594-conjugated secondary IgG (Molecular Probes). Immunofluorescence was visualized by confocal microscopy. For BrdU incorporation, synchronized cells on coverslips were infected with lentivirus. At 24 h post-infection, BrdU (10 μM) was added to the culture medium and incubated for another 5 h. The level of BrdU incorporation was determined using a protocol described previously .
For cell-cycle analysis, cells were fixed in 70% ethanol at 4 °C for 2 h, washed in PBS and stained with propidium iodide (50 ng/μl) containing RNase A (50 ng/μl) at 37 °C for 1 h. Thereafter, cellular DNA content was measured with a BD FACSAria flow cytometer.
Immunoprecipitation and Western blotting
H1299 and HeLa cells were transfected as indicated; 36 h later, cell lysates were incubated with FLAG M2 affinity gels for 5 h at 4 °C. For endogenous IP (immunoprecipitation), HEK-293T cells were lysed and incubated with anti-E1A antibody and Protein A–agarose beads for 8 h at 4 °C. The beads were washed five times with IP assay buffer [1% Triton X-100, 150 mM NaCl, 50 mM Tris/HCl (pH 7.5), 10% glycerol, 10 μM NaF, 10 μM Na3VO4, 1 mM PMSF and 1% (w/v) aprotinin], boiled in SDS sample buffer, and subjected to Western blot analysis.
H1299 cells seeded on 100-mm-diameter dishes were transfected as indicated. At approx. 48 h post-transfection, 20% of the cells were harvested and were lysed in RIPA buffer (50 mM Tris/HCL, 150 mM NaCl, 0.1% SDS, 0.5% sodium deoxycholate and 1% Triton X-100, pH 8) for Western blotting to confirm expression of transfected plasmids. To purify ubiquitinated proteins, we used a protocol described previously 
p14ARF represses E1A transactivation and relocalizes E1A to the nucleolus
It has been shown previously that p14ARF could exert potential antiviral function and protects cells exposed to oncogenic virus including HIV-1  and HPV 16 . In a first attempt to determine whether p14ARF affects adenovirus infection, the adenovirus oncoprotein E1A transactivational activity was investigated. HCT116 cells which lack endogenous ARF were transfected with a luciferase reporter construct driven by the promoter of one of the E1A downstream targets, the adenovirus E4 gene. Expression vector for E1A, p14ARF or E1A plus p14ARF were also co-transfected as indicated in Figure 1(a). In this assay, E1A expression stimulated transcription from the E4 reporter gene as expected. Transfection of the p14ARF vector inhibited E1A-mediated activation of E4 promoter in a dose-dependent manner, whereas no effect was observed on basal expression of the E4 promoter. When the same experiment was carried out in p53-null H1299 cells, similar results were obtained (not shown).
p14ARF associates with E1A, inhibits E1A transactivation and relocalizes E1A into the nucleolus
Next, we sought to address whether ARF has the ability to affect the localization of E1A. Because ARF has been described to relocalize HPV E7 protein to the nucleolus, U2OS cells which lack endogenous ARF were transfected with E1A expression vector alone or in combination with the ARF expression vector. As shown in Figure 1(b), E1A protein was found primarily in the nuclear compartment (90%) as reported previously . ARF, when expressed alone, showed predominantly nucleolar localization (99%). E1A localization was significantly altered from primarily nucleoplasm to nucleolus when co-transfected with p14ARF (86%), whereas ARF protein retained the nucleolar localization. Similar results were observed in other experiments using p53-null H1299 cells (not shown). All results presented were derived from at least three independent experiments. Collectively, these data suggest that ARF functions to block the transactivation of E1A and relocalize E1A to the nucleolus.
p14ARF associates with E1A in mammalian cells
On the basis of the co-localization of ARF and E1A in the nucleolus, we reasoned that p14ARF interacts physically with E1A. To test this hypothesis, H1299 cells were transfected with GFP–E1A13S, FLAG–ARF or both. Western blotting of immunoprecipitated FLAG–ARF complexes using anti-GFP antibody revealed the presence of GFP–E1A13S (Figure 1c). In a reverse IP, the H1299 cells were co-transfected with FLAG–E1A13S and Myc–ARF plasmids. The results showed that Myc–ARF was co-precipitated only with FLAG–E1A, but not with the vector control (Figure 1d). To confirm binding between endogenous ARF and E1A, the HEK-293T cell line (containing relatively high levels of ARF and wild-type E1A) was immunoprecipitated with anti-E1A antibody. ARF was also detected in the E1A complex, but not in mock IP samples (Figure 1e). These results demonstrate that ARF interacts with E1A in mammalian cells.
p14ARF promotes E1A ubiquitination-dependent proteasome degradation
Previously, a number of studies reported that ARF could affect the turnover of E2Fs, HIV-Tat and c-Myc, which may contribute to its transcriptional inhibitory function. In order to evaluate the influence of p14ARF on E1A stability, we transfected HEK-293T cells (which constitutively express E1A protein) with increasing amounts of p14ARF and examined the levels of endogenous E1A by Western blotting. As shown in Figure 2(a), in the presence of p14ARF, E1A protein levels decreased in an ARF dose-dependent manner. Analysis of the E1A mRNA level isolated from ARF-transfected HEK-293T cells showed that the reduction of E1A by p14ARF was not due to a decrease in transcription (Figure 2b). Confirming this observation, a similar level of E1A reduction was detected in the experiment carried out using a p53-null H1299-E1A stable cell line (Figure 2c). After Lenti-FLAG–ARF infection for 48 h, the E1A protein level in this stable cell line was dramatically reduced compared with the control.
p14ARF promotes ubiquitination-dependent degradation of E1A
Next, the effect of oncogene-induced endogenous ARF expression on E1A degradation was examined. In this assay, E1A plasmids were transfected into H1299-E2F1–ER cells, in which ARF expression could be induced with 4-OHT through activation of the E2F1 fusion protein . As shown in Figure 2(d), left-hand panels, E1A expression levels were not influenced in normal H1299 cells with 4-OHT treatment, but decreased in H1299-E2F1–ER cells treated with 4-OHT. When we knocked down endogenous ARF of H1299-E2F1–ER cells by siRNAs, E1A stability remained stable with 4-OHT treatment, indicating that endogenous ARF is responsible for E1A degradation. Meanwhile, we detected E1A subcellular localization with 4-OHT treatment in H1299-E2F1–ER cells. In Figure 2(d), right-hand panels, localization of E1A was dramatically altered from primarily nucleoplasm (the control) to nucleolus when treated with 4-OHT, indicating that activated endogenous ARF was capable of relocalizing E1A.
Then we carried out an endogenous E1A protein stability assay using CHX block in HEK-293T cells (Figure 2e). As expected, endogenous E1A in cells transfected with FLAG–ARF showed a much higher rate of degradation compared with the control. Given that ubiquitination is essential for the degradation of E1A , we next determined whether p14ARF affects E1A stability by regulating its ubiquitination. As shown in Figure 2(f), ubiquitination of E1A was stimulated by exogenous transfected (left-hand panel) or endogenous induced (right-hand panel) ARF. Considering the close relationship between ARF, Mdm2 and p53, especially Mdm2 as an ubiquitin E3 ligase and also a bona fide binding protein of ARF, we sought to explore further whether p53 and Mdm2 could affect the stability of the E1A protein. H1299 cells were transfected with exogenous p53 and Mdm2 plasmids. As shown in Figure 2(g), left-hand panel, neither p53 nor Mdm2 induces E1A degradation compared with ARF. Similar results were obtained by using p53-null and Mdm2-undetectable SaoS2 cells (Figure 2g, right-hand panel). These results suggest that E1A is destabilized in the presence of p14ARF independently of p53 and Mdm2. Previous studies found that ubiquitination of E1A leads to the degradation via a proteasome mechanism, hence we attempted to determine whether proteasome inhibitors block ARF-induced E1A degradation. FLAG–ARF or vector-transfected HEK-293T cells were incubated with MG132, a potent proteasome inhibitor. Results showed that endogenous E1A was remarkably increased and the degradation induced by p14ARF was also blocked in the presence of MG132 (Figure 2h, left-hand panel), demonstrating that ARF promotes the degradation of E1A through the proteasome. Next we sought to investigate the cellular location of E1A in cells treated with proteasome inhibitor. As shown in Figure 2(h), right-hand panels, in H1299 cells transfected with a low dose of E1A expression plasmids, MG132 treatment resulted in obvious nucleolar accumulation of E1A protein. These results indicate that ARF is capable of inducing E1A loss via a p53-independent ubiquitin–proteasome mechanism within the nucleolus.
The N-terminal region of p14ARF is sufficient to bind E1A and block E1A transactivation
To delineate the region of ARF involved in the interaction with E1A, we carried out co-IP analysis using protein extracts from H1299 cells transiently co-transfected with FLAG–E1A and GFP–ARF deletion mutants: N-terminal half (amino acids 1–64) and C-terminal half (amino acids 65–132). As indicated in Figure 3(a), the N-terminal half of p14ARF was required for binding E1A. It has been shown that the N-terminal half of p14ARF is responsible for most of the biological effects of p14ARF . To validate whether the N-terminal half of p14ARF was sufficient to repress E1A transactivation, an luciferase assay analogous to that shown in Figure 1(a) was carried out using HCT116 cells. In Figure 3(b), upper panel, the E4 reporter activation by E1A was strongly inhibited when p14ARF or the N-terminal half of p14ARF were co-expressed. However, co-transfection of the C-terminal half of p14ARF did not block E1A transactivation. Similar results were obtained using the adenoviral E3 promoter construct (Figure 3b, lower panel). The above results prompted us to examine whether the N-terminal half of p14ARF is sufficient to induce E1A degradation and relocalization, which may well contribute to inhibition of E1A transactivation. As shown in Figure 3(c), full-length ARF promoted E1A to loss as observed previously, whereas expression of ARF N- or C-terminal deletion mutants did not result in E1A reduction. In addition, the N-terminal half of p14ARF was sufficient to relocalize E1A to the nucleolus in a similar manner to the full-length (Figure 3d). Taken together, these findings indicate that the repression function of ARF on E1A depends on their physical interaction. Nevertheless, we believe that relocalization and degradation of E1A by ARF may serve as two different mechanisms, but to repress E1A collectively. The depletion of the C-terminal half may impair ARF to form a complex with some other unknown factors which is responsible for E1A degradation.
The N-terminal region of p14ARF is sufficient to bind E1A and block E1A transactivation
The central domain of E1A is indispensable for interacting and co-localizing with ARF
To identify E1A sequences involved in the E1A–ARF interaction, we carried out co-IP analysis using protein extracts from H1299 cells transiently co-transfected with FLAG–ARF and GFP-tagged E1A deletion mutants (Figure 4a). We found that E1A-(136–289) did not associate with p14ARF despite expression at a high level. In contrast, E1A-(1–139) or E1A-(25–289) retained ARF-binding ability, suggesting a crucial role for the central domain (amino acids 25–139) of E1A in ARF binding. Then, we studied the effects of ARF on degradation of E1A deletion mutants. As shown in Figure 4(b), in the presence of ARF expression, E1A-(1–139) and E1A-(25–289) became unstable compared with E1A-(136–289) which lack a binding site for ARF, indicating that ARF promotes rapid degradation of E1A depending on their physical interaction. In addition, we analysed the effect of transient expression of ARF on the subcellular distribution of GFP–E1A deletion mutants in U2OS cells (Figure 4c). All three GFP–E1A deletion mutants were predominantly localized in the nucleoplasm. When co-expressed with FLAG–ARF, both GFP–E1A-(1–139) and GFP–E1A-(25–289) were translocated to the nucleolus (>90%). In contrast, ARF did not alter subcellular distribution of GFP–E1A-(136–289). This was consistent with co-IP results showing that GFP–E1A-(136–289) lacks the ability to bind ARF. Collectively, these results demonstrate that co-localization of ARF and E1A requires their physical interaction.
The central domain of E1A is indispensable for interacting and co-localizing with ARF
Mdm2 and E4F are capable of interrupting the ARF–E1A interaction
The N-terminus of p14ARF is required for the interaction with Mdm2. The results described above showed that this region is also involved in binding to E1A. We therefore asked whether E1A and Mdm2 binding to ARF are mutually exclusive. In Figure 5(a), Mdm2 was found to co-precipitate with ARF, and overexpression of E1A significantly reduced ARF–Mdm2 binding. To confirm further our hypothesis that Mdm2 competes with E1A for ARF binding, immunofluorescence analysis was carried out using ARF-null U2OS cells (Figure 5b). As expected, GFP–E1A translocated to the nucleolus in the presence of exogenous p14ARF (90%). Interestingly, co-expression of Mdm2 caused GFP–E1A to exhibit a diffused pattern, with only 34% of the cells retaining the exclusive nucleolus localization of E1A. These results indicated that E1A and Mdm2 compete for binding to ARF.
Mdm2 and E4F are capable of interrupting the ARF–E1A interaction
Previous studies showed that p120E4F is targeted by E1A . ARF also interacts with E4F . Therefore we tested whether E4F as an ubiquitin E3 ligase forms a complex with ARF to degrade E1A. Although the results did not support this hypothesis (not shown), we found that E4F was also capable of disrupting the ARF–E1A complex. Co-expression of E4F strongly blocked ARF–E1A co-precipitation (Figure 5c). Further immunofluorescence analysis showed that E4F also abrogated the ARF-mediated nucleolar translocation of E1A (Figure 5d). These data suggest that E4F is capable of interrupting the ARF–E1A interaction similar to Mdm2.
Regulation of E1A transactivation by endogenous ARF
To address the role of endogenous p14ARF in regulating E1A activity, we used siRNAs to selectively reduce ARF expression in H1299 cells which are known to express a higher level of endogenous ARF (Figure 6a). In this assay, luciferase expression was analysed 48 h after siRNA transfection. As shown in Figure 6(b), E1A transactivation to the E3–luciferase reporter was compared in p14ARF-knockdown (ARF Si) and control (Ctrl Si) cells. In contrast, activation of E3–luciferase reporters was stimulated further in p14ARF-depleted cells compared with control. In addition, similar results were obtained in HeLa cells that also express abundant amounts of ARF (not shown), indicating that endogenous p14ARF plays a role in counteracting E1A transcriptional activity.
Regulation of E1A transactivation by endogenous p14ARF
ARF inhibits DNA replication triggered by E1A
It is widely accepted that adenoviral E1A reprogrammes the cell cycle and stimulates quiescent cells to enter S-phase. Therefore we tested whether ARF inhibits E1A-induced DNA replication. p53-null H1299 cells and H1299-E1A stable cells were infected with Lenti-FLAG–ARF. At 36 h post-infection, cell-cycle distribution was assayed by FACS. We observed an increase in S-phase population in H1299-E1A cells (57.3%) compared with vector control cells (48.6%), confirming that E1A is capable of inducing G1–S-phase progression in the absence of p53 (Figure 7a). The distribution of H1299 cells in S-phase was not altered after ARF infection. However, a decrease in the S-phase population was observed in ARF-infected H1299-E1A cells.
p14ARF inhibits DNA replication triggered by E1A
To determine further the effects of ARF on E1A-induced DNA replication, human HaCat cells (no endogenous p14ARF protein was revealed, express mutant p53) were used as a model for the BrdU-incorporation assay. Since the ARF–p53 pathway of cell-cycle inhibition is inactive in these cells, we can analyse the effect of ARF on E1A-induced DNA replication in a p53-independent context. In this assay, HaCat cells were serum-starved for 70 h to synchronize in G0/G1-phase, and the growth-arrested cells were infected with Lenti-vec (pLVX-IRES-ZsGreen Vector), Lenti-E1A, Lenti-FLAG–ARF or Lenti-FLAG–ARF plus LentiE1A. At 24 h post-infection, cells were incubated with BrdU and subjected to immunofluorescence staining (Figure 7b). We found that infection of serum-starved HaCat with E1A significantly stimulated DNA replication (~43% BrdU-positive), consistent with previous reports that E1A induces S-phase progression [30,32]. In contrast, co-transfection of ARF with E1A caused an observable diminution in the BrdU-positive cells (24%). In this experiment, ARF alone was not capable of influencing the G1–S-phase progression of HaCat cells, possibly due to the p53-defective context . Then, to test effects of endogenous ARF on E1A-induced DNA replication, an analogous assay was carried out with expression of ARF siRNAs in H1299 cells before serum starvation for 48 h. As shown in Figure 7(c), knockdown of endogenous ARF led to an increase in E1A-stimulated DNA replication compared with control. These data suggest that endogenous ARF is capable of inhibiting the DNA replication triggered by E1A.
Adenovirus E1A targets cellular pRb family members and p300/CBP to drive early viral transcription and induce cell-cycle progression. Previous studies have shown that ARF was induced in oncogenic virus-infected cells, implicating a potential antiviral response by ARF . In the present study, we found that p14ARF binds to E1A and inhibits the transactivation activity of E1A through translocalization and degradation of E1A. Furthermore, E1A can competitively inhibit ARF–Mdm2 binding which is critical for oncogenic stress signalling to p53.
p14ARF does not possess an intrinsic transcription repression activity [1,2]; it is likely that binding to E1A alters the formation of active E1A transcriptional complexes. p14ARF when induced causes E1A to translocalize to the nucleolar compartment, which may play a role in transcriptional inhibition. In addition, the degradation of E1A by p14ARF may also contribute to transcriptional inhibition. Given that a proteasome inhibitor caused accumulation of E1A in the nucleolus, the N-terminal half of p14ARF was sufficient to relocalize E1A to the nucleolus, but insufficient to induce E1A degradation, we believe that E1A activity is negatively regulated by both sequestration and proteasome-mediated degradation at the nucleolus. The depletion of the C-terminal half may influence the complex of ARF with some other unknown factors which also participate in the E1A-degradation process. It is possible that E1A translocated into the nucleolus and led to subsequent degradation by ARF through a ubiquitin–proteasome mechanism in the nucleolus. Actually, some previous studies have reported that several proteins, including B23 and c-Myc, could be degraded through a proteasome degradation mechanism in the nucleolus [45,46]. They found that a portion of proteasomes are recruited to the nucleoli, where a subset of their substrates, including transcriptional factor c-Myc, is sequestered. We speculated that E1A normally locates with the nucleolus, but is hardly observed due to its rapid turnover in the nucleolus. ARF is capable of promoting E1A nucleolar localization and degradation. However, we still do not exclude other mechanisms for E1A degradation in cells which were infected with adenovirus. Whether there exists a specific E3 ubiquitin ligase complex for E1A degradation and whether p14ARF acts as a linker to bring E1A and the E3 complex together remains to be determined. Some studies have indicated that ARF interacts with the proteasome . It is possible that ARF forms a complex with E1A and brings E1A in contact with the ubiquitin–proteasome machinery for degradation. To obtain a more comprehensive insight requires further experimentation.
Our results showed that E1A can be polyubiquitinated and degraded by the proteasome, and p14ARF promotes this process independently of the ubiquitin E3 ligase Mdm2. The biological relevance of these results can be interpreted from two opposing perspectives. From the perspective of the host cell, expression of ARF and binding/inhibition of E1A may be part of an antiviral defence by the host. Adenoviruses need to trigger quiescent host cells to enter a proliferative state in order to mobilize host resources for viral replication. E1A expression is critical for initiating cell-cycle entry by inhibiting the pRb family proteins. Furthermore, expression of the E1B-55K protein is critical for inactivation of p53 to prevent cell-cycle arrest and an apoptotic response . ARF induction by E1A is at least partly responsible for p53 activation during adenovirus infection. This is mediated by ARF–Mdm2 binding and inhibition of Mdm2 E3 ligase function for p53. Our findings in the present study suggest that, in addition to its role in signalling a p53 response, ARF may also directly suppress adenovirus replication by binding E1A.
From the perspective of the virus, inhibition of the p53 response is needed to prevent cell-cycle arrest and premature cell death that can result in unproductive infection. Therefore, in addition to using the E1B-55K viral protein to suppress p53 directly, E1A may have also evolved the ability to disrupt ARF–Mdm2 binding, thus preventing ARF signalling to p53. This activity may co-operate with the E1B-55K protein to overcome a p53-dependent host antiviral response, and increase the chance of successful viral replication.
Another potential function of the E1A–ARF interaction is to allow the virus to sense the functional state of the p53 pathway, and fine-tune the level of E1A activity in order to avoid overstimulation of the host response. To be a successful parasite, the virus needs to co-ordinate its own replication activity based on the metabolic capacity of the host to achieve a level of sustainable balance. By detecting negative-feedback signalling from ARF, the virus can produce a dose of E1A activity suitable for viral replication without triggering a detrimental response from p53 or exhausting host cell resources. This self-censoring mechanism by the virus may be important for maximizing viral replication during each infection cycle. Alternatively, self-censoring by the virus may help to create a persistent infection state without killing the host, resulting in better chances of long-term survival and spreading of the virus.
alternative reading frame
CREB (cAMP-response-element-binding protein)-binding protein
green fluorescent protein
human embryonic kidney
inhibitor of CDK4a
lentivirus expressing E1A
lentivirus expressing FLAG–ARF
murine double minute 2
small interfering RNA
Shengping Zhang and Chuangui Wang designed experiments. Jia Shan, Yang Li, Wen Zhang, Ting Wang, Ling Jiang and Xiuqun Zou performed experiments. Jiemin Wong, Xiaotao Li, Yongping Cui and Chuangui Wang analysed results. Jiandong Chen, Shengping Zhang and Chuangui Wang wrote the paper.
This work was supported by the National Natural Science Foundation of China [grant number 30772523], the National Basic Research Program of China [grant number 2009CB918401], Shanghai projects [grant numbers 07SG28, 08ZZ23 and 08PJ14042], and the Program for New Century Excellent Talents in University [grant number NCET-08-0194].