Functional Analysis of the p53 Pathway in Neuroblastoma Cells Using the Small-Molecule MDM2 Antagonist Nutlin-3 Free

The p53 transcription factor plays a critical role in the cellular defense against malignant transformation by promoting cell-cycle arrest, DNA damage repair, apoptosis, and senescence in response to stress signals (1). Tumor cells therefore invariably acquire aberrations that permit them to escape from p53-mediated growth control. It is estimated that approximately 50% of all human cancers harbor inactivating mutations in the TP53 (p53) gene, whereas defects in upstream or downstream components of the p53 pathway are believed to account for loss of p53 activity in the other half of malignancies. Dissection of the p53 pathway defects in individual tumor types is important, because improved understanding of the mechanisms behind p53 inactivation may guide the development of targeted therapeutic strategies.

Neuroblastoma is an aggressive childhood tumor of neural crest origin that has a lethal outcome in the majority of high-risk patients (2). A remarkable feature is that p53 is rarely mutated at diagnosis and only in a minority of neuroblastoma tumors at relapse, as shown by a recent study that found mutation rates of 2% and 15%, respectively (3). Conflicting data exist about p53 pathway signaling in neuroblastoma cells. The DNA damage–induced G₁ checkpoint function and apoptotic activity of p53 have been reported to be impaired by cytoplasmic p53 sequestration (4–6), which may be caused by p53 hyperubiquitination (7). Furthermore, wild-type p53 in neuroblastoma cells may be in a conformation refractory to integration into transcriptional complexes, resulting in reduced transcriptional activity (8). In contrast, others have shown a normal DNA-binding and transactivation capacity of the p53 protein and an intact p53 signal transduction pathway in neuroblastoma cells with wild-type p53 (9–11). No study has yet systematically investigated the functional integrity of the p53 pathway in neuroblastoma cells on a larger scale, as the reports mentioned above relied on the use of only 1 to 5 neuroblastoma cell lines to judge on p53 functionality.

We have previously reported that a small-molecule MDM2 antagonist, nutlin-3, is capable of inducing potent antitumor effects against neuroblastoma cells and xenografts with wild-type p53, which may provide a new opportunity for targeted therapeutic intervention (12, 13). Nutlin-3 is designed to compete with p53 for binding into a hydrophobic pocket on the surface of MDM2 (14). The resulting disruption of the interaction between both proteins releases p53 from negative control by MDM2, which functions as an E3 ubiquitin ligase to promote p53 proteasomal degradation and as an inhibitor of p53 transcriptional activity. Treatment with nutlin-3 thus leads to stabilization and activation of p53 and, if downstream effectors are functionally intact, to a robust p53 response.

The availability of a direct and selective p53 activator makes it possible to systematically search for defects in p53 and its downstream signaling components. Here, we set out to examine the nature of p53 pathway defects in a large panel of neuroblastoma cell lines by using nutlin-3 as a tool for interrogating the functionality of the p53 pathway.

Human neuroblastoma cell lines were obtained between 1993 and 2010 from Peter Ambros (STA-NB-1.2, STA-NB-3, STA-NB-8, STA-NB-9, and STA-NB-10), Garrett Brodeur (NGP, NLF, and NMB), Susan Cohn (NBL-S, SHEP), Valérie Combaret (CLB-GA), Thomas Look (SJNB-1, SJNB-8, and SJNB-10), John Lunec [SK-N-BE(1n), SK-N-BE(2c)], Sven Påhlman (SH-SY5Y), Patrick Reynolds (CHP-134, CHP-901, CHP-902R, SMS-KAN, and SMS-KCNR), and Rogier Versteeg (GICIN-1, IMR-32, LA-N-1, LA-N-2, LA-N-5, LA-N-6, N-206, SK-N-AS, SK-N-FI, SK-N-SH, and TR-14), or established in our laboratory (UHG-NP). The authenticity of the cell lines was verified during this study by array comparative genomic hybridization and short tandem repeat genotyping. Cell culturing and treatment with nutlin-3 (Cayman Chemical) were conducted as previously described (12).

Sequencing of the p53 coding region was done as previously described (12).

Cells were seeded in duplicate or triplicate wells of a 96-well plate (10⁴ cells per well) and incubated for 6 hours before treatment was initiated. Treatment typically consisted of exposure to 0, 2, 4, 8, 16, and 32 μmol/L nutlin-3 for 24, 48, and 72 hours, except for experiments with inducible model systems, in which the inducing agent or a negative control was applied for 16 hours before incubation with nutlin-3. Cell viability was measured by using a luminescent ATP-based assay (CellTiter-Glo assay; Promega).

Cells were plated in duplicate or triplicate wells of a 96-well plate (10⁴ cells per well) and incubated for 6 hours before treatment, which was done in a similar way as described for the cell viability experiments. The combined activity of caspase-3 and caspase-7 was determined by using the Caspase-Glo 3/7 assay (Promega).

Measurements of cell-cycle phase distribution and hypodiploid (sub-G₁) DNA content were done as previously described (13).

Cells were treated with 0 or 8 μmol/L nutlin-3 for 24 hours (or, in the case of an inducible model system, with the inducing agent or a negative control for 16 hours and then with 0 or 8 μmol/L nutlin-3 for an additional 24 hours). Total RNA extraction, DNase treatment, cDNA synthesis, and SYBR Green I quantitative real-time reverse transcriptase PCR (qRT-PCR) were done as previously described (13). Primer sequences are available in the RTPrimerDB database (15): BAX (RTPrimerDB ID #814), BBC3 (PUMA; #3500), CDKN1A (p21^WAF1/CIP1; #631), GAPDH (#3), SDHA (#7), and UBC (#8). Expression levels of the p53 target genes BAX, PUMA, and p21^WAF1/CIP1 were calculated by using qbase^PLUS software version 1.5 (Biogazelle; ref. 16). Levels of GAPDH, SDHA, and UBC were used for normalization.

Western blotting was done as previously described (12) by using primary antibodies against p53 (mouse clone DO-1; Calbiochem), p21^WAF1/CIP1 (mouse clone SX118; BD Biosciences), and BAX (rabbit monoclonal antibody; Upstate). An anti–β-actin antibody (mouse clone AC-74; Sigma) was used to confirm equal loading.

See Supplementary Data.

See Supplementary Data.

The 34 human neuroblastoma cell lines used in this study were first characterized for mutations in the p53 gene. Sequencing of the entire coding region in 2 overlapping fragments showed wild-type p53 in 25 cell lines (74%) and various genetic defects in the other 9 cell lines (26%; Table 1). The most frequent aberrations were missense mutations, which were located in exon 5 [N-206, SK-N-BE (2c)], exon 6 (NLF), exon 7 (NMB, SK-N-FI), and exon 10 (LA-N-2) of p53. One cell line, LA-N-1, was characterized by a nonsense mutation, resulting in a stop codon at amino acid residue 182. SJNB-8 cells were found to possess an in-frame deletion that removes the coding sequence for amino acids 105 to 125. The PCR step before the sequencing did not produce an amplicon for the second part of the p53 coding region in SK-N-AS cells. It could be shown by quantitative real-time PCR (qPCR) that this was because of a homozygous deletion of the 3′ end of p53 (Supplementary Fig. S1), in line with previously published findings (17, 18).

We next used the selective MDM2 antagonist nutlin-3 to test whether the p53 pathway was functional in our series of neuroblastoma cell lines. As illustrated in Fig. 1A, determination of the nutlin-3 concentration that causes 50% reduction in cell viability (IC₅₀ value) provides a quantitative measure of the functional integrity of the p53 pathway. IC₅₀ values were established at 24, 48, and 72 hours of nutlin-3 treatment and correlated with the mutation status of p53 (Fig. 1B). Cell lines with wild-type p53 displayed highly significantly lower IC₅₀ values than lines harboring mutant p53 (P = 0.004 at 24 hours, P < 0.001 at 48 and 72 hours). All 9 cell lines with p53 mutation were characterized by high IC₅₀ values, indicating that the genetic aberration effectively impaired the function of the p53 protein. Pronounced reductions in cell viability after nutlin-3 treatment and corresponding low IC₅₀ values were observed in 23 of 25 cell lines with wild-type p53. This suggests that p53 downstream signaling pathways are not a major target for p53-inactivating lesions in neuroblastoma and lends support to the development of therapeutic strategies aimed at p53 reactivation.

Two cell lines, LA-N-6 and SHEP, were relatively resistant to nutlin-3 despite the presence of wild-type p53 (IC₅₀ values comparable with those observed in neuroblastoma lines with mutant p53, i.e., IC₅₀ values >32 μmol/L, >30 μmol/L, and >20 μmol/L nutlin-3 at 24, 48, and 72 hours of treatment, respectively; Fig. 1B). Of particular interest were SHEP cells, because their response to nutlin-3 was strikingly different from that of 2 closely related cell lines, SK-N-SH and SH-SY5Y. Cell line SK-N-SH was originally derived from bone marrow metastases of a patient with stage 4 neuroblastoma, and subcloning of these cells has generated several morphologically distinct sublines, including SHEP and SH-SY5Y (19). Figure 2A shows that nutlin-3 profoundly suppressed the viability of SK-N-SH and SH-SY5Y cells, whereas only mild effects were noted in SHEP cells. Further experiments were conducted to determine whether the poor nutlin-3 response of SHEP cells was caused by failure to enter apoptosis or by defective cell-cycle arrest. Analysis of caspase-3 and caspase-7 activity indicated that a 24-hour exposure to nutlin-3 induced a dose-dependent apoptotic response in SK-N-SH and SH-SY5Y cells (Fig. 2B). In contrast, no increase in caspase-3 and caspase-7 activity was observed in nutlin-3–treated SHEP cells. This was confirmed by flow cytometric analysis of sub-G₁ DNA content after treatment with vehicle control or 8 μmol/L nutlin-3 for 24 hours, which showed a nutlin-3–induced increase in the apoptotic sub-G₁ fraction in SK-N-SH and SH-SY5Y cells but not in SHEP cells (Fig. 2C). Flow cytometric cell-cycle profiling further showed a reduction in the percentage of cells in S phase 24 hours after treatment of SK-N-SH, SH-SY5Y, and SHEP cells with 8 μmol/L nutlin-3, indicative of cell-cycle arrest in all 3 cell lines (Fig. 2D). The phenotypic effects of nutlin-3 on apoptosis and cell-cycle progression were paralleled by similar changes in expression levels of p53 target genes. As shown in Fig. 2E, a 24-hour treatment of SK-N-SH and SH-SY5Y cells with 8 μmol/L nutlin-3 induced an increase in the mRNA levels of p53 target genes involved in apoptosis (BAX, PUMA) and cell-cycle arrest (p21^WAF1/CIP1). A large increase in p21^WAF1/CIP1 expression was also present in SHEP cells treated with 8 μmol/L nutlin-3 for 24 hours, but expression levels of the proapoptotic target genes BAX and PUMA remained considerably lower in nutlin-3–treated SHEP cells than in nutlin-3–treated SK-N-SH and SH-SY5Y cells. Similar findings were observed by Western blot analysis. Treatment with 8 μmol/L nutlin-3 for 24 hours induced p53 accumulation and increased expression of p21^WAF1/CIP1 in all 3 cell lines, whereas induction of BAX expression was observed only in nutlin-3–treated SK-N-SH and SH-SY5Y cells (Fig. 2F). Taken together, these data indicate that SHEP cells have an intact cell-cycle checkpoint control mechanism but fail to undergo apoptosis in response to treatment with nutlin-3.

Interestingly, SHEP cells have previously been reported to contain a homozygous deletion of the CDKN2A gene on chromosome 9p21, in contrast to SK-N-SH and SH-SY5Y cells (20). We confirmed the copy number status of CDKN2A in these 3 cell lines by quantitative PCR (Supplementary Fig. S2). The CDKN2A gene encodes 2 structurally distinct growth inhibitory proteins, p16^INK4a and p14^ARF, that are important regulators of the pRb and p53 tumor suppressor pathways, respectively (21). This raised the possibility that the homozygous CDKN2A deletion may underlie the nutlin-3–resistant phenotype of SHEP cells. Analysis of the entire panel of 25 neuroblastoma cell lines with wild-type p53 further revealed that the presence of homozygous CDKN2A deletion was strongly associated with a higher IC₅₀ value at 48 and 72 hours of nutlin-3 treatment (P = 0.009 and P < 0.001, respectively; Supplementary Table S1). Amplification of MDM2 did not have an impact on the IC₅₀ values of neuroblastoma cell lines with wild-type p53 (P > 0.05; Supplementary Table S1). MYCN-amplified neuroblastoma cell lines with wild-type p53 were characterized by a lower IC₅₀ value at 72 hours of nutlin-3 treatment than wild-type p53 neuroblastoma cell lines without MYCN amplification (P = 0.023), but this difference was not observed anymore after exclusion of cell lines with homozygous CDKN2A deletion (P > 0.05; Supplementary Table S1). Neither a significant difference in p53 mutation status nor in MDM2 and CDKN2A copy number status was found between MYCN-amplified and MYCN-nonamplified neuroblastoma cell lines (P > 0.05; Supplementary Table S2).

A possible involvement of p14^ARF and p16^INK4a in the nutlin-3 response was first tested by transient knockdown of the CDKN2A gene in IMR-32 and NGP cells, 2 easy-to-transfect neuroblastoma cell lines that have a good and previously well-characterized nutlin-3 response (12), by using a pool of small interfering RNAs (siRNA) directed against sequences common to both p14^ARF and p16^INK4a transcripts. The efficiency of CDKN2A knockdown, measured by qRT-PCR 24 hours posttransfection, is shown in Fig. 3A. Silencing of CDKN2A resulted in a moderate reduction in the sensitivity of IMR-32 and NGP cells to nutlin-3, as shown by cell viability assays done after 24, 48, and 72 hours of exposure to nutlin-3 (Fig. 3B and C).

To unravel whether this potentiating effect of CDKN2A expression on the response to nutlin-3 was mediated by p14^ARF or p16^INK4a, NGP cells were infected with lentiviruses encoding a p14^ARF-specific short hairpin RNA (shRNA), a p16^INK4a-specific shRNA, an shRNA directed simultaneously against both transcripts, or a negative control shRNA targeting firefly luciferase, and subsequently selected with puromycin to eliminate uninfected cells. qRT-PCR analysis of p14^ARF and p16^INK4a expression in the established sublines, termed NGP-LV-p14, NGP-LV-p16, NGP-LV-p14/p16, and NGP-LV-luc, respectively, showed successful transcript-specific knockdown (Fig. 4A). Treatment of these stable knockdown cell lines with nutlin-3, followed by cell viability assays, indicated that the influence of CDKN2A expression on the nutlin-3 response was primarily attributable to p14^ARF (Fig. 4B). These findings were confirmed by analysis of caspase-3 and caspase-7 activity, which showed that silencing of p14^ARF decreased the susceptibility of NGP cells to undergo apoptosis on nutlin-3 treatment (Fig. 4C). Quantification of p53 target gene expression showed that knockdown of p14^ARF, but not p16^INK4a, attenuated the p53 transcriptional response induced by a 24-hour exposure to 8 μmol/L nutlin-3 (Fig. 4D). This was accompanied by a marked p14^ARF shRNA-induced decrease in the basal mRNA levels of PUMA and p21^WAF1/CIP1 in vehicle-treated cells, whereas BAX expression was upregulated to a lesser extent by nutlin-3, rather than basically suppressed, when p14^ARF was silenced.

We next examined whether overexpression of CDKN2A could enhance the response of neuroblastoma cells to nutlin-3. We therefore generated stable transfectants of an IMR-32 subclone, IMR-5/75, in which transgenic expression of either p14^ARF or p16^INK4a or, as a negative control, lacZ was inducible by addition of tetracycline. Figure 5A shows the relative mRNA expression levels of p14^ARF and p16^INK4a in these sublines, designated as IMR-Tet-p14, IMR-Tet-p16, and IMR-Tet-lacZ, respectively, 24 hours after treatment with tetracycline or vehicle control. Overexpression of p14^ARF resulted in a more pronounced reduction in cell viability and stronger caspase-3 and caspase-7 activation following nutlin-3 treatment, whereas overexpression of p16^INK4a or lacZ had no appreciable effect on the nutlin-3 response (Fig. 5B and C). In line with these observations, incubation of IMR-Tet-p14 cells with 8 μmol/L nutlin-3 for 24 hours induced a more potent p53 transcriptional response when the cells had been exposed to tetracycline compared with vehicle control (Fig. 5D). The expression of PUMA and p21^WAF1/CIP1 in these cells in the absence of nutlin-3 was also considerably increased by the addition of tetracycline. In contrast, switching on transgene expression in IMR-Tet-p16 and IMR-Tet-lacZ cells did not affect basal nor nutlin-3–induced expression levels of p53-responsive genes.

Finally, similar CDKN2A overexpression experiments were undertaken in SHEP cells to investigate whether this manipulation could restore the sensitivity to nutlin-3. Despite successful generation of several sublines with tetracycline-inducible expression of p14^ARF and p16^INK4a, we did not observe a reversal or improvement of the nutlin-3–resistant phenotype of SHEP cells (Supplementary Fig. S3).

Taken together, our data provide evidence for a dosage effect of p14^ARF expression on the response of neuroblastoma cells to nutlin-3, but they also indicate that the homozygous CDKN2A deletion in SHEP cells is not responsible for the poor response of these cells to nutlin-3.

The p53 tumor suppressor protein is at the crossroads of cellular stress response pathways that control decisions between life and death. We used here the selective MDM2 antagonist nutlin-3 as a tool to gain insight into the mechanisms by which neuroblastoma cells escape from p53-mediated growth control. Mutation analysis showed a p53 gene alteration in 9 of 34 neuroblastoma cell lines, which rendered the p53 pathway nonfunctional in all cases. Three mutations were located outside the classic hot-spot region (exons 5–9), indicating that p53 mutations are best identified by sequencing the entire coding region. The observed mutation frequency in our cell line panel (26%) is considerably higher than the p53 mutation rate of approximately 1% that was found in early studies of neuroblastoma tumors (22–27). This may reflect the fact that cell lines are frequently derived from progressive or relapsed tumors, as p53 mutations can develop during chemotherapy and malignant progression of neuroblastoma (28). In addition, older studies may have underestimated to some extent the p53 mutation frequency in neuroblastoma tumors, because analysis was often confined to the mutational hot-spot region.

Treatment with nutlin-3 was capable of inducing potent antiproliferative and cytotoxic effects in 23 of 25 neuroblastoma cell lines with wild-type p53. These findings are particularly relevant in the light of an ongoing debate whether p53 is functional in neuroblastoma or not (4–11). Discrepancies between previous studies may be in part attributed to different treatment regimens (11) and to whether the p53-inducing stimulus directly interferes with potential restraints on p53 activity, such as p53 hyperubiquitination (7). Our data provide good evidence of almost uniform functionality of the p53 protein and its downstream effectors in neuroblastoma cells with wild-type p53, when the interaction between p53 and MDM2 is disrupted by nutlin-3. As a consequence, selective MDM2 inhibitors may prove beneficial for treating neuroblastoma patients, provided that wild-type p53 is present. Our findings of functional p53 effector pathways also suggest that circumvention of the p53-driven antitumor barrier in neuroblastoma cells relies primarily on defects upstream of p53. Cumulating evidence indicates that it is precisely an increased activity of MDM2 which serves as the predominant mode of p53 inactivation in neuroblastoma (28), but further study is warranted to identify the full spectrum of aberrations in regulators of p53 activity.

The presence of a homozygous CDKN2A deletion in the nutlin-3–refractory SHEP cells but not in the nutlin-3–sensitive SK-N-SH and SH-SY5Y cells prompted us to investigate the role of p14^ARF and p16^INK4a in the response to nutlin-3. The nutlin-3–resistant phenotype of SHEP cells could not be reversed by reintroduction of p14^ARF or p16^INK4a, but knockdown and overexpression experiments in other neuroblastoma cell lines pointed to a stimulatory effect of p14^ARF expression on the nutlin-3 response. Our data suggest that a p14^ARF-driven increase in basal expression levels of p53-responsive genes, such as PUMA and p21^WAF1/CIP1, contributes to this potentiating effect of p14^ARF, although other mechanisms cannot be excluded. High levels of the MDM2 inhibitory protein p14^ARF may result in a larger fraction of the nuclear pool of MDM2 molecules being inhibited after nutlin-3 treatment and thus in stronger activation of the p53 pathway. Alternatively, p14^ARF may provide a costimulatory signal for the p53 response independently of MDM2. For instance, p14^ARF may increase p53 protein synthesis (29), inhibit p53 turnover by repressing other components of the p53 degradation pathway than MDM2 (30), enhance p53 transcriptional activity (31), or regulate pathways that crosstalk with p53 signaling (32). We did not aim to identify the molecular basis of the cooperation between p14^ARF and nutlin-3 in this study, but rather wish to comment on the potential clinical implications. By using mouse models, previous studies have shown that Cdkn2a mutations induce chemoresistance by disabling p53 (33) and that loss of p19^ARF, the murine homolog of p14^ARF, limits the therapeutic response to the tyrosine kinase inhibitor imatinib (34). On the basis of our findings, it can be expected that tumor cells may also gain resistance to nutlin-3 treatment by suppressing p14^ARF. The likelihood of this scenario is corroborated by data from a switchable p53 knock-in mouse model of lymphoma showing that p53 reactivation strongly selects for the emergence of p53-resistant tumors through inactivation of either p53 or p19^ARF (35). Several early-phase clinical studies with selective MDM2 inhibitors and other p53-reactivating compounds have recently been initiated (36). Our data indicate that the occurrence of aberrations in p14^ARF should be monitored in these studies and provide an incentive for the development of strategies to counter p14^ARF lesions.

The lack of improvement in nutlin-3 sensitivity after reintroduction of p14^ARF into SHEP cells leaves us with the question of how to explain the resistant phenotype of these cells. We provided evidence of intact cell-cycle arrest but defective apoptosis following nutlin-3 treatment of SHEP cells. This cell line is also resistant to other apoptosis-inducing stimuli, including irradiation (37, 38) and adenoviral gene therapy (39). The poor sensitivity to death-inducing triggers might be related to the S-type (substrate-adherent/Schwannian/melanoblastic) morphology of SHEP cells, as S-type neuroblastoma cells seem to be more resistant to apoptosis than N-type (neuroblastic/neuroendocrine) neuroblastoma cells (40). Another notable feature is that SHEP cells have lost the capacity to form colonies in soft agar and tumors in nude mice (41). One could therefore wonder whether the loss of oncogenic signals, which often have a collateral proapoptotic effect, may result in desensitization to apoptosis. For instance, SHEP cells lack expression of the MYCN oncoprotein, and artificial induction of MYCN expression in these cells has been shown to slightly increase the sensitivity to nutlin-3 (42). Alternatively, SHEP cells may contain high levels of antiapoptotic proteins, as has been previously proposed (37). Further study is needed to pinpoint the exact mechanism underlying the nutlin-3–resistant phenotype of SHEP cells.

In conclusion, this study provides several insights into the spectrum of p53 pathway defects in neuroblastoma cells that may prove useful for designing new therapeutic approaches. The rarity of signaling defects downstream of p53 indicates that p53-reactivating strategies may represent an excellent therapeutic tool for treating neuroblastoma tumors with wild-type p53. Resistance to nutlin-3 is mostly attributable to the presence of p53 mutation, which is not uncommon in neuroblastoma cell lines. This highlights the need to search for effective p53-independent anticancer agents or mutant p53–targeting compounds as a complementary therapeutic modality. Finally, the finding that p14^ARF expression levels modulate the sensitivity of neuroblastoma cells to nutlin-3 raises the possibility that p14^ARF may contribute to the outcome of p53 activation in patients treated with selective MDM2 inhibitors. It remains to be determined whether clinical treatment failure with this new class of anticancer drugs may result from loss or suppression of p14^ARF.

No potential conflicts of interest were disclosed.

We thank Griet Van Lancker and Xiaoyang Zhang for technical assistance.

The study was financially supported by Research Foundation–Flanders (FWO), Concerted Research Actions–UGent (GOA), Interuniversity Attraction Poles–Belgium (IUAP), and Emmanuel van der Schueren Foundation. T. Van Maerken has conducted the study as PhD fellow of the FWO.

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.