Nutlin-3a Activates p53 to Both Down-regulate Inhibitor of Growth 2 and Up-regulate mir-34a, mir-34b, and mir-34c Expression, and Induce Senescence

p53 has many functions in its role as a tumor suppressor gene (1). It is well-documented that p53 is a critical mediator of the senescence response to several stimuli, such as DNA damage, oncogene activation, oxidative stress, and overexpression of tumor suppressor genes (2, 3). Cellular senescence is mainly classified into two types based on the mechanism. One is replicative senescence, which is involved with telomere-shortening and found in human fibroblast strains (4). The other is telomere-independent cellular senescence, called premature senescence and is induced by cellular stress. p53 plays a pivotal role for both types of cellular senescence. Recent reports show that the senescence response by p53 activation is considered to be one of the mechanisms protecting against malignant transformation (5, 6).

Under normal physiologic conditions, p53 expression is maintained at low levels in proliferating cells by degradation via MDM2 functioning as an E3 ubiquitin ligase (7). MDM2 is overexpressed in several human cancers, especially in sarcomas (8). ADP ribosylation factor (ARF) functions in part to activate p53 through MDM2 inhibition (9, 10). Because MDM2 inhibition could be effective for p53 stabilization and accumulation, several MDM2 inhibitors have been recently developed for potential cancer therapy (11). Nutlin-3, which is one of the MDM2 inhibitors, is a small molecule that binds to MDM2 so that it functions as an activator of p53 expression due to the interference of binding between p53 and MDM2 (11–13). Recent reports have shown that nutlin-3 induces p53-dependent apoptosis and cell cycle arrest in several kinds of cancer cell lines that contain wild-type p53 (14, 15). The effect was especially dramatic in cancer cells that overexpress MDM2, such as certain sarcomas. Although phosphorylation of various serine resides on p53 NH₂ terminal domain was not detected after nutlin-3 treatment, p53 seems to be fully active as a transcription factor and apoptotic inducer (16). Moreover, nutlin-3 also prevented the association of MDM2 with both hypoxia-inducible factor 1α, so that the induction of vascular endothelial growth factor was attenuated (17), and E2F1, which allowed transcriptional activation of p73α and NOXA resulting in apoptosis (18). Nutlin-3 has the potential to perform various functions through MDM2 inhibition. In the present study, our initial aim was to investigate if nutlin-3 induced senescence in normal human fibroblasts.

When p53 is activated by several stimuli, many genes are regulated positively and negatively. Previous reports have examined changes in p53-related genes after different treatments that activate p53 (19–21). Although p53 induces cellular senescence, the mRNA expression pattern exhibited some differences between replicative senescence and premature senescence (22, 23). In addition to p53-related mRNA expression, recent reports have shown that microRNA-34s (mir-34a, mir-34b, and mir-34c) are downstream effectors of p53-dependent senescent and apoptotic pathways (24–27).

In this study, we further focused on the inhibitor of growth 2 (ING2) gene, a potential nutlin-3–induced p53-responsive gene, which was down-regulated by the treatment of nutlin-3 in the microarray experiments. ING2 is a member of the ING gene family, which contains a plant homeodomain finger (28). ING2 may be a tumor suppressor gene associated with p53 (29). Recent reports have shown that ING2 forms complexes with mSin3a and histone deacetylase 1 (HDAC1; ref. 30) and binds to the histone H3 trimethylated at lysine 4 (H3K4me3; refs. 31–33). This evidence indicates that ING2 is involved in chromatin remodeling to regulate gene activation or suppression. Specifically, it has been reported that H3K4me3 is associated with gene activation (34, 35). Chromatin structure consists of many forms of modification that regulate gene transcription; thus, the impairment of these epigenetic events may affect tumor development, cell proliferation, and senescence (36).

Cell culture. GM08402 primary human fibroblasts were obtained from the Coriell Cell Repositories. The hTERT-immortalized fibroblast cell line NHF-hTERT and the E6-expressing version of this cell line NHF-hTERT E6 were derived by infecting the primary cell strain GM07532 (Coriell Cell Repositories), as previously published (37). The NHF-hTERT p53 short hairpin RNA (shRNA) cell line was derived by transfecting p53 shRNA into the NHF-hTERT cells. Cultured human cancer cell lines, including RKO, LS174T, SW620, WiDr, A549, NCI-H157, NCI-H1299, Calu-6, MCF7, and U-2OS, as well as the primary fibroblasts, IMR-90, MRC-5, and WI-38, were originally obtained from American Type Culture Collection. MDAH 041 p53^−/− were derived from fibroblasts of a patient with Li-Fraumeni syndrome and were kindly provided by Michael Tainsky (Case Western Reserve University). The HCT116 human colon cancer cell lines (p53^+/+ and p53^−/−) were kindly provided by Dr. Bert Vogelstein (Johns Hopkins University School of Medicine). All cells were grown at 37°C in the presence of 5% CO₂ in the recommended media.

Drug treatment and senescence-associated β-galactosidase assays. Nutlin-3a and its enantiomer, nutlin-3b, were generous gifts from Dr. Lyubomir Vassilev (Hoffmann-La Roche, Inc.). Cells were plated at varying cell densities in either six-well dishes or 100 mm² dishes. Nutlin-3a or nutlin-3b was added at varying concentrations (0–10 μmol/L) after 24 h. Cells were then fixed in 2% formaldehyde/0.2% glutaraldehyde in PBS for 5 min and incubated at 37°C for 12 to 18 h with fresh senescence-associated β-galactosidase (SA-β-gal) stain solution containing 1.0 mg/mL X-gal (37). Doxorubicin and trichostatin A (Sigma-Aldrich) were used at the indicated concentrations.

Immunoprecipitation and Western blotting. Protein lysates were prepared as described previously (19). p53 immunoprecipitation was done by incubating with agarose-conjugated DO-1 and Pab 1801 (Santa Cruz) antibodies for 2 h. Proteins were electrophoresed on SDS-PAGE gels and transferred to Immobilon-P membrane (Millipore). Protein detection was done by using the following antibodies: p53 (DO-1; Oncogene Science), ser15 p53 and ac382 p53 (Cell Signaling), threonine 18 p53 (gift from Dr. Ettore Appella), p21 (Oncogene Science), β-actin (Sigma), and rabbit polyclonal anti-ING2, which was generated against the COOH terminal 14mer amino acid of ING2. Bound antibodies were detected using enhanced chemiluminescence (ECL) detection reagents (Amersham Pharmacia Biotech) and visualized by autoradiography.

Microarray analysis. Microarray experiments were performed as previously described (19). Samples for the microarray experiments were prepared as follows: both NHF-hTERT and NHF-hTERT p53 shRNA cells were treated with 10 μmol/L nutlin-3a or the same amount of DMSO. Cells were harvested at 24 and 36 h after the treatment. The data analysis was performed using the BRB Array Tools.⁶

Real-time reverse transcription–PCR analysis of microRNA and mRNA expression. Total RNA from each cell line was harvested using TRIzol (Invitrogen) according to the manufacturer's protocol. For detecting microRNA expression, reverse reactions were performed using TaqMan MicroRNA Assay kit (Applied Biosystems). Five micrograms of total RNA were used for the synthesis of first-strand cDNA using the SuperScript III First Strand cDNA Synthesis kit (Invitrogen) following the manufacturer's instructions. Real-time PCR analysis was performed using ABI prism 7500 (Applied Biosystems) with a TaqMan probe provided by the manufacturer. The TaqMan probes used were ING2 (Id Hs00357543_m1), p21 (Id Hs00355782_m1), glyceraldehyde-3-phosphate dehydrogenase (GAPDH; Id Hs99999905_m1), hsa-mir-34a, hsa-mir-34b, hsa-mir-34c, and RNU66. The relative amounts of the mRNA and microRNA targeted gene were normalized by the amount of GAPDH and RNU66 transcript, respectively.

Chromatin immunoprecipitation. Chromatin immunoprecipitation (ChIP) was performed using the ChIP Assay kit (Upstate) according to the manufacturer's protocol. PCR reactions contained 4 μL of immunoprecipitate or diluted total input, 500 nmol/L of each primer, 200 nmol/L deoxynucleotide triphosphate, 500 nmol/L MgSO₄, and 1 unit KOD plus (Toyobo) in a total volume of 50 μL. The primers for detecting the expected p53 DNA binding sequence were designed as follows: ING2 site 1, forward 5′-AGATTCCAGCGTAGGGGAAG-3′ and reverse 5′-CTTTTCGGACGACCTGGAG-3′; ING2 site 2, forward 5′-AGGCTAGCGGGAGGCTCT-3′ and reverse 5′-CTCCAGGACCGGAGCAGT-3′. After 32 to 35 cycles of amplification, PCR products were run on a 1.5% agarose gel and analyzed by ethidium bromide staining.

Electrophoretic mobility shift analysis. Electrophoretic mobility shift assays (EMSA) were carried out using the LightShift Chemiluminescent EMSA kit (Pierce). The oligonucleotide with the p53 binding site, from −890 to −851 of the ING2 promoter, called ING2 site 1, was 5′-TCCCTGATCATCAGCCCCGCGGTGCCGGGCGAGCCCGAGGGC-3. Another oligonucleotide with the p53 binding site of the ING2 promoter (from −268 to −229), named ING2 site 2, was 5′-GACTGGGGACAGGCGGGCGGCGACGGGCTTGTCATGGGGA-3. All oligonucleotides were hybridized with complementary-synthesized oligonucleotides and used in double-stranded form. The oligonucleotides were labeled with biotin. The DNA binding reaction of biotinated double-stranded oligonucleotides with 10 ng of recombinant p53 (active motif) was done at room temperature for 20 min according to the manufacturer's protocol in 1× binding buffer, containing 2.5% glycerol, 5 mmol/L MgCl₂, 0.05% NP40, and 50 ng/μL poly(deoxyadenylate-deoxyTMP). Anti-p53 (DO-1) antibody and anti-Sp1 antibody (Santa Cruz) were used for detecting a super-shifted band. Unlabeled oligonucleotides were used as competitors. DNA-protein complexes were detected by the ECL method.

Luciferase assay. The reporter construct for ING2(A) containing −1251 to +140 bases from the putative transcription start site and ING2(C) containing −413 to +140 bases from the putative transcription start site were prepared by PCR amplification from the human genomic DNA of Beas2B, a normal human lung epithelial cell line. The products were cloned into the MluI and BglII sites of the pGL-3 basic vector (Promega). The construct for ING2(B) was generated from the ING2(A) by digesting the region from −560 to −162 using smaI. The construct for ING2(D) containing −162 to +140 bases was further modified from the ING2(B).

A dual-luciferase assay was performed in triplicate according to the manufacturer's instructions (Promega). 1 × 10⁵ cells were plated on 12-well plates 1 d before transfection. A pGL-3 luciferase reporter (1.6 μg) and 0.16 μg of the Renilla luciferase assay vector pRL (Promega) were cotransfected into HCT116 and RKO cells using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. At 24 h posttransfection, the cells were treated with nutlin-3a or were left untreated. Cell lysates were obtained by adding 250 μL of cell lysis buffer per well. Luciferase activity was measured by using 20 μL of cell lysate per assay tube in a single-photon channel of a scintillation counter (Beckman). The level of firefly luciferase activity was normalized by that of the Renilla luciferase activity in each experiment.

Small interfering RNA, overexpression, and antisense experiments. The ING-2 small interfering RNAs (siRNA; Invitrogen) were designed as follows: ING2-1, 5-ACACAAAUGCUCGAAUUGGUGGAAA-3′ and ING2-2, 5′-ACAUACUGCUUAUGCAACCAAGUGU-3′. The scrambled siRNA sequences were 5-ACAAAGUUCGCAAUUGGGUGACAAA-3. These were transfected into the NHF-hTERT cells using the Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer's protocol. For Western blotting, cells were treated 24 h with nutlin-3a, nutlin-3b, or no treatment at 72 h posttransfection. For the SA-β-gal assay, cells were treated with the nutlins or left untreated for 3 d at 24 h posttransfection. The ING2 complete cDNA expression vector was transferred into an adenovirus-packing cell line. The high-titer purified preparations were generated by Gene Expression Laboratory of National Cancer Institute-Frederick. GFP gene expression in IMR-90, MRC-5, and WI-38 cells treated with the GFP inserted adenovirus (adeno-GFP) at a multiplicity of infection (MOI) of 100 was observed in ∼90% of the cells. The cells were treated with 100 MOI adeno-GFP or ING2 inserted adenovirus (adeno-ING2) for 72 h. MSCV retroviral vectors encoded for mir-34a were kindly provided by Dr. Gregory L. Hannon (Watson School of Biological Sciences, Cold Spring Harbor Laboratory).

Nutlin-3a induced cellular senescence in normal human fibroblasts. The normal human skin fibroblast strain GM08402 and the hTERT immortalized cell line NHF-hTERT were treated with nutlin-3a for 7 days. Nutlin-3b, which has a 150-fold lower affinity to MDM2 (13), was used as a negative control. After the treatment, the cells were fixed and stained for the low pH SA-β-gal activity (Fig. 1A). A high percentage of cells that had been treated with nutlin-3a were stained positive for the low pH β-gal, indicating cells had undergone senescence. Nutlin-3b–treated cells exhibited similar levels as the nontreated cells. Next, we determined the time and dose dependency of nutlin-3 in NHF-hTERT cells. In the time course, 50% of the cells were SA-β-gal positive at 3 days and almost 100% showed positive staining at 7 days after nutlin-3a treatment, whereas few positive cells were detected by treatment with nutlin-3b (Supplementary Fig. S1A). Cells were also treated with indicated concentrations of nutlin-3 for 7 days. Ninety percent of the cells were stained with SA-β-gal at 3 μmol/L nutlin-3a, whereas 1 μmol/L nutlin-3a showed a weak effect. All of the cells showed positive staining at 10 μmol/L, the reported maximal dose without toxicity (Supplementary Fig. S1B).

We further examined the effect of nutlin-3 treatment for other normal human lung fibroblast strains, IMR-90, MRC-5, and GM08402, and analyzed the number of positive cells by SA-β-gal staining. All three cell strains showed a morphologically senescent phenotype after nutlin-3a treatment. Again, almost 100% of the cells showed positive staining (Fig. 1B). There is a low percentage of β-gal–positive cells after nutlin-3b treatment; however, it was not a significant change compared with the untreated cells.

The hallmark of cellular senescence is an essentially irreversible arrest of cell division. To determine if the nutlin-induced senescence was reversible, colony-forming efficiencies (CFE) and population doublings (PD) were determined for cells after removal of the nutlin drugs. CFEs were determined by allowing the cells to be grown in the presence of 10 μmol/L nutlin-3a or nutlin-3b or left untreated for 7 days at subconfluence. Cells were then trypsinized, washed, and replated at densities of either 1 × 10² or 1 × 10³ per 100 mm² dish in normal growth media for an additional 7 days. The cells were fixed and stained with crystal violet, and the number of colonies was counted. At both densities, the nutlin-3a–treated cells were unable to form a significant number of colonies compared with the untreated or nutlin-3b–treated cells. There was a 10-fold difference in the number of colonies formed (Supplementary Fig. S2A).

Next, we calculated the PDs for these cells after the same protocol as above, except that cells were plated at densities of either 1 × 10³ or 1 × 10⁴ per dish. Cells were trypsinized after 7 days and counted to determine population doublings. Results showed the nutlin-3a–treated cells could not divide after being replated because there was <0.1 PDs for either density (Supplementary Fig. S2B). On the other hand, the nutlin-3b or untreated cells underwent over 2.5 PDs, depending on the number of cells plated. Assuming that ∼90% of the nutlin-3a–treated cells underwent senescence, these data support the hypothesis that these nutlin-3a–treated cells have undergone a permanent senescence.

Nutlin-3a–induced senescence is dependent on p53. To determine if the senescence is induced in a p53-dependent manner, p53-deficient cell lines, NHF-hTERT E6 and NHF-hTERT p53 shRNA, were exposed to nutlin-3a or nutlin-3b for 7 days. The parental cells, NHF-hTERT cells, exhibited the senescent phenotype after treatment with nutlin-3a, with over 80% of the cells stained by SA-β-gal staining. However, we did not observe any changes in both NHF-hTERT E6 and NHF-hTERT p53 shRNA cells when they were treated with nutlin-3a or nutlin-3b (Fig. 1C).

To verify the accumulation of p53 by nutlin-3a, Western blot analysis was performed. In NHF-hTERT cells, the total amount of p53 expression was increased by nutlin-3a, but not by nutlin-3b (Fig. 1D). Although phosphorylation was not seen on residue serine 15 of p53 in the nutlin-3a–treated cells in agreement with a previous report (16), acetylation of p53 residue 382 was present in the nutlin-3a–treated cells. p21 expression was also increased with nutlin-3a treatment. In contrast, there was no increase of p53 expression with nutlin-3a treatment in the p53 null cells. To confirm these observations, we examined p53 expression with nutlin-3 treatment using the p53 isogenic-paired cell lines, HCT116 p53^+/+ and HCT116 p53^−/− cells. The result was very similar with the previous data from studies using human fibroblasts (Supplementary Fig. S3A).

According to a previous report, phosphorylation of threonine 18 is mediated by casein kinases (38), requiring the previous phosphorylation of serine 15, and this phosphorylation interferes with the p53 binding to MDM2 (39). Therefore, we examined the status of threonine 18 with nutlin-3 treatment. Phosphorylation on residue threonine 18 of p53 was not present in nutlin-3–treated NHF-hTERT cells (Supplementary Fig. S3B).

It has been reported that U-2OS, an osteosarcoma cell line which possesses a functional p53, did not exhibit the apoptotic phenotype after nutlin-3a (14). We hypothesized that nutlin-3a could induce senescence instead of apoptosis even in this cancer cell line. The levels of senescence were minimal at 7 days, but there was a significant increase seen at 14 days (Supplementary Fig. S4). According to the literature, the subset of apoptotic-associated genes regulated by p53 was lacking in U-2OS cells compared with cells that were strongly induced to undergo apoptosis by nutlin-3a treatment (14). The difference in cellular response by nutlin-3a can be explained from the defects downstream of p53.

Nutlin-3a–induced p53-responsive genes. The spectrum of p53-responsive genes differs depending on the types of stimuli (19). Therefore, we next examined the contribution of p53 activation to gene expression after treatment with nutlin-3a using microarray technology. We compared the gene expression in NHF-hTERT and NHF-hTERT p53 shRNA cells with treatment of nutlin-3a or DMSO at 24 and 36 h using the univariate two-sample t test (P < 0.05) with randomized variance. A total of 2,942 genes were differentially expressed after nutlin-3a treatment, of which 1,737 genes (59%) were up-regulated and 1,205 genes (41%) were down-regulated (Supplementary Tables S1 and S2). Among the up-regulated genes, the expression of BTG2, GDF15, TP53I3/PIG3, SULF2, NINJ1, PLTP, and SESN1 was strongly increased in a p53-dependent manner (Table 1). The well-known targets of p53, BAX, GADD45A, TP53INP1, TP53AP1, CCNG1, and IGFBP3 were also detected as up-regulated genes. On the other hand, the expression of CCNB1, CENPF, CCNA2, CDKN3, SPAG5, CKS2, CASC5, KIFC1, CDC20, and KIF2C was remarkably decreased (Table 1). Additionally, the expression of minichromosome maintenance complex component (MCM) genes, including MCM2 to MCM8, which are associated with DNA replication, and ING2, which is involved in chromatin remodeling and regulating gene expression, were down-regulated in NHF-hTERT cells treated with nutlin-3a.

p53 activates mir-34 expression and suppresses ING2 expression. In addition to mRNA expression, recent reports have shown that mir-34a, mir-34b, and mir-34c are regulated by p53 activation, suggesting that they have a potential function in the senescence pathway in normal human fibroblasts (24–27). We found that the expression of all three mir-34s were up-regulated by nutlin-3a treatment and that the response of mir-34a was earlier than mir-34b and mir-34c (Fig. 2A). mir34a, mir34b, and mir34c levels were shown to be up-regulated to varying amounts in various human tumor cell lines that were wild type for p53 after treatment with nutlin-3a after 24 hours (Fig. 2B–D). The tumor cell lines that were either p53 null or have a mutant p53 did not show this up-regulation.

We further focused on ING2 as one of the p53-responsive genes from the microarray data because of previous studies of the ING family (28–32). We examined the ING2 expression when p53 was activated using nutlin-3a in NHF-hTERT cells. Interestingly, ING2 expression in the cells treated with nutlin-3a was remarkably decreased compared with nontreated cells while ING2 expression was not changed in nutlin-3b–treated cells (Fig. 3A).

To further show the association between p53 status and ING2 expression, we examined other cell lines. p53 wild-type cells, including HCT116, RKO, LS174T, A549, MCF-7, and NHF-hTERT, were treated with 10 μmol/L of nutlin-3a or DMSO as a control. ING2 mRNA transcripts were consistently decreased by nutlin-3a treatment in all cell lines using a real-time PCR method, whereas ING2 mRNA expression was not changed significantly in p53 mutant and null cells (Fig. 3B).

Next, we investigated whether the expression of ING2 mRNA was altered in a dose-dependent and time-dependent manner by nutlin-3a treatment. In both HCT116 and RKO cells, the ING2 mRNA level was down-regulated, depending on the dose, reaching a maximum level at 5 μmol/L at 24 h. Furthermore, the decrease in ING2 expression was time dependent. p21 mRNA expression was examined as a positive control of p53 activation and was up-regulated under the same conditions (Fig. 3C).

ING2 protein expression was examined using the p53 isogenic cell lines, HCT116 p53^+/+ and HCT116 p53^−/−, NHF-hTERT and NHF-hTERT p53 shRNA, to see whether the suppression occurs in a p53-dependent manner. The accumulation of p53 in HCT116 p53^+/+ was seen earlier than in NHF-hTERT (Fig. 3D). In p53 wild-type cells, ING2 protein expression was consistently decreased by the treatment with nutlin-3a in a p53-dependent manner. ING2 protein expression correlated with its mRNA expression level. There was no remarkable change in p53 null cells.

p53 binds to the ING2 promoter region and suppresses ING2 activity. We next analyzed the ING2 promoter region for a p53-DNA consensus binding sequence by computational analysis.⁷

To examine if wild-type p53 binds to the ING2 promoter, we first performed ChIP assays. The primer pairs were constructed to detect each sequence for ING2, site 1 and site 2. As a positive control, the p53-DNA consensus binding sequence on the p21 promoter region was also examined. Cells were left untreated or treated with nutlin-3a for 24 h. Lysates were harvested and cross-linked, and then immunoprecipitations were performed using an antibody directed against p53. Promoter fragments of ING2 site 1, site 2, and p21 were amplified by PCR and a representative agarose gel of the PCR products is shown in Fig. 4A. The PCR bands were detected at both sites 1 and 2 under treatment with nutlin-3a in p53 wild-type cells, whereas no bands were detected in the control cells and p53 null cells.

The EMSA was performed using biotinylated double-stranded oligonucleotides corresponding to the sequence of ING2 sites 1 and 2. Recombinant p53 protein was used for these experiments. The shifted bands were detected in both ING2 sites 1 and 2 (Fig. 4B). An antibody to p53 generated a super-shifted complex for both sites 1 and 2 of ING2, further confirming that the binding is specific. Moreover, when the unlabeled double-stranded oligonucleotides corresponding to ING2 sites 1 and 2 were used as competitors, the formation of complexes was inhibited in each case, supporting the specificity of the DNA-p53 complex for these putative sites.

To determine whether the decrease in ING2 mRNA is dependent on the function of p53 as a regulator of transcription, we analyzed the regulation of the ING2 promoter. For this purpose, four kinds of constructs were generated and cloned into the pGL3 luciferase vector. The longest construct (A) from −1251 to +140 included both ING2 site 1 and site 2. The construct (B) lacked the site 2, and the construct (C) had only the site 2. The construct (D) was deleted of both sites 1 and 2 as a negative control. These constructs were transiently cotransfected with Renilla luciferase vector into the p53 isogenic pairs of HCT116 and RKO cell lines. After transfection, the cells were treated with nutlin-3a for 12 hours. In p53 wild-type cells, the luciferase activity in nutlin-3a–treated cells was clearly down-regulated in the constructs (A), (B), and (C) of the ING2 promoter when compared with the nontreated cells, whereas there was no remarkable change in the construct (D) between nutlin-3a–treated and nontreated cells (Fig. 4C). No significant alteration of the luciferase activity was detected in nutlin-3a–treated and nontreated p53 deficient cells (Fig. 4C).

Both ING2 site 1 and site 2 share similar sequences with the Sp1 consensus sequence (Supplementary Fig. S5A). Previous reports showed Sp1 was involved in the repression mechanism by p53 (40–43). To investigate whether Sp1 can bind to these sites, EMSA was performed using recombinant Sp1 protein. The labeled oligonucleotides of a typical Sp1 consensus sequence were used as a positive control. A shifted band was detected at ING2 site 2, but not detected at site 1 (Supplementary Fig. S5A). Moreover, we examined the binding affinity at ING2 site 2 with the competition of Sp1 and p53. When both Sp1 and p53 were added, p53 dominantly bound to ING2 site 2 (Supplementary Fig. S5B). It has previously been shown that negative regulation of transcription by p53 may involve p53-mediated recruitment of HDAC and mSin3A (44); therefore, we tested whether this is also true for the ING2 promoter. HCT116 cells were treated with trichostatin A to inhibit HDAC activity. ING2 expression was not attenuated by a combined treatment with nutlin-3a and trichostatin A (Supplementary Fig. S5C). In the luciferase assay, the activity was not repressed by the same treatment (Supplementary Fig. S5D).

Disregulation of ING2 induces senescence. We have previously shown that overexpression of ING2 can induce p53-dependent senescence (45). We have confirmed this result (Supplementary Fig. S6). We also have shown that down-regulation of ING2 by siRNA technology induces cellular senescence (Fig. 5A and B) that is p53 independent (Fig. 5C and D).

We found that nutlin-3a induced senescence in normal human fibroblasts in a time-dependent, dose-dependent, and p53-dependent manner. A complimentary report showed that nutlin-3a induces cellular senescence in murine fibroblasts in a p53-dependent manner (46). Although phosphorylation was not seen on residue serine 15 of p53 after nutlin-3a treatment, acetylation of p53 residue 382 was detected. Previous studies have implicated ING1 and 2 in acetylation of p53 at lysine 382 (46, 47) and acetylation of p53 K382 increases during cell senescence (3). In addition, we examined the phosphorylation status at threonine 18 with nutlin-3 treatment, because this modification was previously found to be associated with MDM2 (39). Although we could not detect this p53 phosphorylation with nutlin-3 treatment, unlike doxorubicin treatment, p53 could still function as a transcriptional factor because p21 and mir-34a, mir-34b, and mir-34c expression was increased. It has been reported that the phosphorylation on the NH₂ terminal of p53 was not required for p53-dependent transcription based on the result that ARF-induced p53 was not phosphorylated on the NH₂ terminal of p53, including serine 6, 9, 15, 20, and 37, but showed the phosphorylation at serine 392 (48). Therefore, phosphorylation on the NH₂ terminal of p53 seems not be essential for induction of p53-dependent senescence.

Previously, we reported that p53 activation by treatment with several stimuli regulated various genes positively and negatively depending on the stimuli (19). To better understand the association with nutlin-3a–induced p53 and senescence in human fibroblasts, microarray experiments were performed. We found that the genes up-regulated by nutlin-3a treatment were associated with antiproliferative and proapoptotic functions. Plasminogen activator inhibitor-1 was one of these genes and is a potential p53 target and involved in cellular senescence (3, 49). As to the down-regulated genes, cell cycle and DNA replication-related genes, such as CCNB1, CCNB2, CCNA2, CKN3, and MCM family genes, were significantly decreased.

According to recent reports, mir-34a, mir-34b, and mir-34c, which are induced by p53 activation, are downstream effectors of p53 (24–27). Overexpression of mir-34a, mir-34b, and mir-34c has been shown to induce senescence in normal human fibroblasts (24). We confirmed that all three of the mir-34s were increased in nutlin-3a–treated NHF-hTERT cells. Whereas mir-34a, mir-34b, and mir-34c were not induced by nutlin-3a in the p53 null or mutant cell lines, the level of induction varied among the cancer cell lines with wild-type p53. These data support the hypothesis that mir-34a, mir-34b, and mir-34c are involved in the p53-dependent senescence pathway.

We found that ING2 mRNA transcripts were suppressed by nutlin-induced p53 activation from the microarray data. We also detected a decrease of ING2 protein expression in nutlin-3a–treated cells and ING2 mRNA levels were consistently down-regulated with nutlin-3a treatment in p53 wild-type cells, whereas there was no remarkable change in p53-mutant and p53-deficient cells. Recently, there is evidence that a large number of genes are suppressed by p53 (19–21). It has been reported that p53 consensus DNA-binding sequences have been found among genes repressed in a p53-dependent manner (20, 21). Recent reports have indicated that some of these down-regulated genes might be associated with microRNA expression. In fact, mir-34a was determined to regulate a subset of p53 related genes (24). We investigated whether overexpression of mir-34a could regulate ING2 expression. ING2 expression in mir-34a overexpressed cells was not altered (Supplementary Fig. S7).

We found candidates of p53 consensus DNA-binding sequences on the ING2 regulatory region within 1.5 kb and confirmed the p53 binding ability to the expected sites from ChIP assay and EMSA. Furthermore, the data from the luciferase assays showed p53 could repress ING2 promoter activity. There was sufficient evidence that p53 repressed ING2 and this phenomenon might be occurring through p53 binding directly to the consensus DNA-binding sequences on the ING2 regulatory region. At least, three kinds of mechanisms have been proposed (51). First, p53 may interact with DNA-binding transcriptional activators and interfering with the function of activators. This mechanism includes the competition between p53 and activators through the overlapping DNA binding site. Second, p53 may interact with components of the basal transcriptional machinery. Finally, p53 may recruit chromatin modifying factors, such as HDAC, so that gene expression could be suppressed. These three mechanisms of p53 transcriptional repression can also act in combination (40). Based on these previous studies, we first investigated the possibility that other transcriptional factors could bind to the p53 consensus DNA-binding sequences on the ING2 promoter. Both ING2 sites 1 and 2 share similar sequences with the Sp1 consensus sequence. One of the mechanisms of ING2 suppression by p53 could be explained that p53 binding affinity to the promoter region was stronger than Sp1, which might regulate ING2 expression endogenously.

Recent reports indicate that ING2 has the ability to bind the histone H3 trimethylated at lysine 4 (31–33). Consequently, ING2 is involved in chromatin remodeling to regulate gene activation or suppression. Chromatin structure consists of many forms of modification that drastically regulate gene transcription; thus, the impairment of these epigenetic events may affect tumor development, cell proliferation, and senescence (36). A previous report shows ING2 suppresses cyclin D1 mRNA expression in response to DNA damage (32). It is well known that cyclin D1 has diverse effects on cells depending on its level of expression and cell type (51). For example, cyclin D1 expression is increased in senescent cells compared with young cells (52, 53). Based on this evidence, there is the possibility that the down-regulation of ING2 results in the enhancement of cyclin D1 to modulate senescence.

In addition to the induction of senescence by down-regulation of ING2, overexpression of ING2 can also induce senescence. The senescence induced by knockdown of ING2 is p53 independent, whereas the senescence induced by overexpression of ING2 is p53 dependent. These results suggest that both different mechanisms and the physiologic range of ING2 levels may be important to maintain cellular homeostasis. Further studies with either ING2 transgenic or knockout mice will be interesting to pursue.

Grant support: Intramural Research Program of NIH, National Cancer Institute and Center for Cancer Research.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Dorothea Dudek-Creaven for editorial assistance and Karen MacPherson for bibliographic assistance.