Uncontrolled cell cycle entry, resulting from deregulated CDK-RB1-E2F pathway activity, is a crucial determinant of neuroblastoma cell malignancy. Here we identify neuroblastoma-suppressive functions of the p19-INK4d CDK inhibitor and uncover mechanisms of its repression in high-risk neuroblastomas. Reduced p19-INK4d expression was associated with poor event-free and overall survival and neuroblastoma risk factors including amplified MYCN in a set of 478 primary neuroblastomas. High MYCN expression repressed p19-INK4d mRNA and protein levels in different neuroblastoma cell models with conditional MYCN expression. MassARRAY and 450K methylation analyses of 105 primary neuroblastomas uncovered a differentially methylated region within p19-INK4d. Hypermethylation of this region was associated with reduced p19-INK4d expression. In accordance, p19-INK4d expression was activated upon treatment with the demethylating agent, 2′-deoxy-5-azacytidine, in neuroblastoma cell lines. Ectopic p19-INK4d expression decreased viability, clonogenicity and the capacity for anchorage-independent growth of neuroblastoma cells, and shifted the cell cycle towards the G1/0 phase. p19-INK4d also induced neurite-like processes and markers of neuronal differentiation. Moreover, neuroblastoma cell differentiation, induced by all-trans retinoic acid or NGF-NTRK1-signaling, activated p19-INK4d expression. Our findings pinpoint p19-INK4d as a neuroblastoma suppressor and provide evidence for MYCN-mediated repression and for epigenetic silencing of p19-INK4d by DNA hypermethylation in high-risk neuroblastomas.
A remarkable quality of neuroblastoma, the most common extracranial tumor of childhood, is the broad spectrum of clinical behavior, which ranges from spontaneous regression without any cytoreductive treatment to malignant progression despite intensive multimodal therapy. Although predicting outcome at diagnosis has improved during the past few decades, the outcome of children with a high-risk clinical profile, often characterized by MYCN oncogene amplification in the tumor, has only modestly improved, with long-term survival still being <40%. An increasing number of studies report on deregulation of the retinoblastoma 1 (RB1) tumor suppressor pathway as an essential element of neuroblastoma malignancy (1–7).
In non-malignant cells, the initial step of cell cycle entry from G1 to S phase is tightly controlled by D-type cyclins in complex with cyclin-dependent kinases 4 or 6 (CDK4/6) and by the corresponding CDK inhibitors. Their concerted interplay regulates RB1 phosphorylation that in turn controls free E2F levels and cell cycle progression (8). Several RB1 pathway aberrations have been described potentially shifting the G1-S phase balance towards unrestricted cell cycle entry of neuroblastoma cells. Aberrations occur on both the genetic and epigenetic levels and comprise CDK4 or CCND1 amplification, CDK6 mutation, p16-INK4a deletion or promoter hypermethylation of RB1 or p16-INK4a (1,2,9–12). In support of these findings, a recent study determined the overall frequency of copy number aberrations affecting G1-regulating genes to be 30% in 82 investigated neuroblastomas (13).
p19-INK4d belongs to the INK4 family of CDK4/6 kinase inhibitors together with p16-INK4a, p15-INK4b and p18-INK4c and blocks G1-S transition by CDK4 inhibition (14). p19-INK4d and p18-INK4c are the only INK4 family members expressed during embryogenesis and early development (15). p19-INK4d is expressed in proliferating and differentiating neurons in the brain paralleling the onset of neurogenesis and its sustained expression is involved in establishing and maintaining the quiescent state in adult neurons (16–18). The CDK inhibitory function designates p19-INK4d as a candidate tumor suppressor; however, only a few studies have directly linked p19-INK4d to tumor biology. In human head and neck squamous carcinoma cells, p19-INK4d is induced by vitamin D and retinoic acid and supports restoration of G1 arrest in cooperation with the CDK inhibitor, p27Kip1 (19). Two studies on T-cell leukemia revealed that inhibiting either histone deacetylases or deregulated NOTCH-signaling induces p19-INK4d and, subsequently, G1 arrest (20,21). Repressed p19-INK4d in lung cancer cells could be reactivated by treatment with the demethylating agent, 2′-deoxy-5-azacytidine (DAC), despite the absence of promoter hypermethylation (22). In neuroblastoma tumors, it has previously been shown that low p19-INK4d expression is associated with advanced disease stage (23). Furthermore, irradiation- or chemically induced DNA damage upregulates p19-INK4d in neuroblastoma cells, and p19-INK4d exerts a CDK4-independent cytoprotective effect by enhancing DNA repair mechanisms in response to these genotoxic treatments (24,25).
In the present study, we addressed the regulation and functional role of p19-INK4d in both primary neuroblastomas and neuroblastoma cell models. Our data identify p19-INK4d as a central effector controlling neuroblastoma cell growth and provide evidence for MYCN- and DNA hypermethylation-dependent p19-INK4d repression in high-risk neuroblastoma.
p19-INK4d expression is associated with clinico-biological variables of neuroblastoma and is repressed by MYCN
Mining of the GeneSapiens database (26) revealed that p19-INK4d is expressed at above average levels in the central and peripheral nervous system when compared across 31 human tissues (Fig. 1A). Importantly, p19-INK4d showed a broad range of expression in neuroblastomas and median expression was highest among 59 tumor tissues available in the GeneSapiens database (Fig. 1B). Analyses of p19-INK4d expression in a large cohort of 478 primary neuroblastomas revealed significant association of low p19-INK4d expression with advanced disease stage (Stage 4, P < 0.001), higher age at diagnosis (≥1.5 years, P < 0.001) and amplified MYCN (P < 0.001) (Fig. 1C). Maximally selected log-rank tests revealed a significant association between low p19-INK4d expression and shorter event-free and overall survival of neuroblastoma patients (both P < 0.001; Fig. 1D). We tested p19-INK4d expression in several neuroblastoma model systems allowing modulation of MYCN. p19-INK4d expression was reduced in SH-SY5Y cells (single-copy MYCN) upon induction of a tetracycline-inducible MYCN transgene as determined via mRNA-sequencing (Fig. 2A). The degree of MYCN-mediated p19-INK4d repression in this model was comparable with that observed for the established MYCN-repressed genes, NRTK1, DKK3 and c-MYC, which were analyzed in parallel (Supplementary Material, Fig. S1A). In complementary experiments, small hairpin RNA (shRNA)-mediated MYCN knockdown in MYCN-amplified Be2-C and IMR5-75 cells induced p19-INK4d in comparison to control cells (Fig. 2B and C). Together, these data suggest that p19-INK4d expression serves relevant functions for halting unlimited neuronal cell growth, its repression may be important in neuroblastoma pathogenesis and MYCN is involved in p19-INK4d repression.
p19-INK4d CpG hypermethylation is associated with p19-INK4d repression and correlates with amplified MYCN
Analyses of 450K methylation array data from 105 neuroblastoma patients revealed that differential methylation was present at two CpG sites (cg17491564 and cg06865036) located ∼1.6 kb downstream of the p19-INK4d transcription start site (TSS). Hypermethylation of these CpG sites was significantly associated with low p19-INK4d expression (cutpoint determined by maximally selected Wilcoxon statistic: P < 0.001 for both CpGs; Fig. 3A, left panel) and was associated with high-risk disease (Fisher's exact test: P < 0.001 for both CpGs; Fig. 3A, left panel) and amplified MYCN (Wilcoxon rank-sum test P < 0.001 for both CpGs, Fig. 3A, right panel). To validate these findings, a region spanning 1.4–2 kb downstream of the TSS and containing 14 CpGs including the two CpG sites analyzed on the 450K array was investigated in the same set of 105 neuroblastomas on a MassARRAY (Fig. 3B). MassARRAY data were highly correlated with 450K methylation array data (Pearson's correlation coefficient = 0.85 and 0.89, 95% CI: 0.79–0.90 and 0.84–0.92 for cg1749156 and cg06865036, respectively, P < 0.001 for both; Supplementary Material, Fig. S2) and confirmed the negative correlation of p19-INK4d expression with hypermethylation of the investigated region, as well as the association of hypermethylation with high-risk neuroblastoma and amplified MYCN for all CpGs analyzed (Fig. 3B and Table 1). Neuroblastoma cell lines IMR32, IMR5-75 shMYCN (transgenic for a tetracycline-inducible MYCN shRNA), Be(2)-C, SH-EP and SH-SY5Y were analyzed in parallel by MassARRAY and showed high CpG methylation levels throughout the whole investigated region (Fig. 3B). These methylation levels exceeded those observed in tumors, which is in line with reduced p19-INK4d expression in the cell lines when compared with neuroblastoma tumors with low-level p19-INK4d methylation (Supplementary Material, Fig. S3). IMR32, IMR5-75, Be(2)-C, SH-EP and SH-SY5Y cells were treated with the DAC demethylating agent for 72 h to test for methylation-dependent expression of p19-INK4d. DAC treatment significantly induced p19-INK4d expression in the MYCN-amplified IMR32 (P = 0.01), IMR5-75 shMYCN (MYCN shRNA uninduced, P = 0.002) and Be(2)-C (P = 0.02) cell lines, while no effect was observed in the non-amplified SH-EP and SH-SY5Y cell lines (Fig. 3C). Knockdown of MYCN in IMR5-75 shMYCN cells by activating the shRNA elevated p19-INK4d levels, and combined treatment with DAC did not further elevate p19-INK4d expression (Fig. 3C). The effect of DAC on p19-INK4d CpG methylation was confirmed in Be(2)-C cells using 450K methylation array analysis, which showed that 72-h DAC treatment decreased methylation of both the cg1749156 and cg06865036 CpG sites in two independent replicates (Supplementary Material, Fig. S4). Collectively, these data show that p19-INK4d hypermethylation in primary neuroblastomas is associated with amplified MYCN and high-risk disease, and suggest that this hypermethylation is involved in p19-INK4d repression.
|MassARRAY CpG||Maximally selected Wilcoxon statistics: CpG methylation and expression||Wilcoxon sum rank test: CpG methylation and MYCN status (amplified)||Fisher's exact test: CpG methylation and Risk group (high risk)|
|9 and 10||0.86||<0.001||<0.001||<0.001|
|MassARRAY CpG||Maximally selected Wilcoxon statistics: CpG methylation and expression||Wilcoxon sum rank test: CpG methylation and MYCN status (amplified)||Fisher's exact test: CpG methylation and Risk group (high risk)|
|9 and 10||0.86||<0.001||<0.001||<0.001|
p19-INK4d exerts oncosuppressive functions in neuroblastoma cells
To explore p19-INK4d function in neuroblastoma cells, stable clones of SH-EP, SH-SY5Y and IMR5-75 cells were generated allowing tetracycline-inducible expression of p19-INK4d in a physiological range, as verified by comparison with p19-INK4d expression levels in primary low-risk neuroblastomas (Supplementary Material, Fig. S5). In alamar blue reduction assays, ectopic p19-INK4d expression reduced SH-EP cell viability to 24.6 ± 5.5% SD of uninduced controls after 72 h, and IMR5-75 cell viability to 76.6 ± 5.5% SD of uninduced controls after 96 h (Fig. 4B). The slow growth rate of SH-SY5Y cells did not allow alamar blue assay analysis, but clonogenicity was assessed using an anchorage-dependent colony formation assay. SH-SY5Y cell clonogenicity was reduced to 20.8 ± 4.3% SD of uninduced controls after 96 h of ectopic p19-INK4d expression (Fig. 4C). Similarly, p19-INK4d ectopic expression reduced clonogenicity of SH-EP cells to 24.2 ± 8.8% SD of uninduced controls after 72 h (Fig. 4C). A p19-INK4d-dependent reduction in clonogenicity was also observed for IMR5-75 cells, but was difficult to quantify because of the weak adherence of these cells to the tissue culture plastic during fixation and staining. However, IMR5-75 cells are capable of forming colonies in soft agar, and the effect of ectopic p19-INK4d expression on this capacity was tested. The capacity for anchorage-independent growth was reduced to 52.5 ± 2.4% SD of uninduced controls after 2 weeks of ectopic p19-INK4d expression (Fig. 4D). To assess the effect of p19-INK4d expression on CDK4 activity, expression of RB1 pathway members was analyzed after p19-INK4d induction in SH-EP, SH-SY5Y and IMR5-75 cells. Ectopic expression of p19-INK4d in either SH-EP or SH-SY5Y cells reduced CDK4-specific RB1 phosphorylation at serine 780 (pRB1Ser-780, Fig. 4E). Protein levels of the E2F targets, SKP2 and E2F1, also decreased upon ectopic p19-INK4d expression (Fig. 4E). In IMR5-75 cells, ectopic p19-INK4d expression did not affect pRB1Ser-780 levels or expression of RB1 pathway components (data not shown). The effect of p19-INK4d expression on the cell cycle in SH-EP, SH-SY5Y and IMR5-75 cells was determined using flow cytometry (Fig. 4F). Ectopic p19-INK4d expression in SH-EP cells enriched the fraction of cells in the G1/0 phase from 56.8 ± 1.4% SD to 80.7 ± 1.9% SD, while the fractions of cells in the S and G2/M phases were reduced from 26.5 ± 1.4% SD to 12.4 ± 1.4% SD and 16.7 ± 0.9% SD to 6.9 ± 0.5% SD, respectively. Similarly, ectopic p19-INK4d expression in SH-SY5Y cells enriched the G1/0 fraction from 56.0 ± 0.3% SD to 79.0 ± 1.5% SD and reduced the fraction in S phase from 36.4 ± 0.3% SD to 13.5 ± 1.3% SD. IMR5-75 were least affected by ectopic p19-INK4d expression, which only mildly enriched the G1/0 fraction from 43.3 ± 0.3% SD to 49.0 ± 0.5% SD, and mildly reduced the fraction in S phase from 43.0 ± 0.3% SD to 38.8 ± 0.6% SD and the fraction in G2/M from 13.6 ± 0.2% SD to 12.2 ± 0.6% SD. The sub-G1 fraction, as a measure of cell death, was analyzed in parallel to cell cycle distribution, and remained unchanged upon ectopic p19-INK4d expression (data not shown). Collectively, these data suggest that p19-INK4d has an oncosuppressive function in cellular models for neuroblastoma, as it inhibits anchorage-dependent and -independent growth and triggers G1/0 cell cycle arrest.
p19-INK4d is involved in inducing senescence and differentiation of neuroblastoma cells
Ectopic p19-INK4d expression in SH-EP cells for 72 h shifted cellular morphology towards a senescent phenotype with enlarged flattened cells (Fig. 5A). In line, senescence-activated β-galactosidase activity was elevated by ectopically expressing p19-INK4d in SH-EP cells (Fig. 5B). Ectopic p19-INK4d expression in SH-SY5Y cells increased the degree of morphological differentiation, including the formation of neurite-like structures (Fig. 5C). To validate the effect of p19-INK4d expression on differentiation processes at the molecular level, the expression of neuronal differentiation markers was determined in p19-INK4d-induced and uninduced cell culture pairs. Ectopic p19-INK4d expression significantly increased the expression of neurofilament, light polypeptide (NEFL) and microtubule-associated protein 2 (MAP2) after 96 h (P = 0.005 and 0.003, respectively; Fig. 5D). In accordance, knockdown of endogenous p19-INK4d reduced the expression of NEFL (siRNA #1: P = 0.03, siRNA #2: P = 0.01) and MAP2 (siRNA #1: P = 0.02, siRNA #2: P = 0.09) in SH-SY5Y cells (Fig. 5E). We further asked whether p19-INK4d could be activated by agents used to induce differentiation in primary neuroblastomas or by natural inducers of neuronal differentiation. We used two established neuroblastoma differentiation models to answer this question: (i) all-trans retinoic acid (ATRA)-induced differentiation in Be(2)-C cells and (ii) differentiation induced by the combination of neurotrophic tyrosine kinase receptor type 1 (NTRK1) overexpression and treatment with its natural ligand, nerve growth factor (NGF) in SH-SY5Y cells. ATRA significantly induced both the expression of the neuronal differentiation marker, NEFL, as well as p19-INK4d in Be(2)-C cells (144 h ATRA: P = 0.03 and 0.01, respectively; Fig. 5F). Similarly, the combination of NTRK1 overexpression and treatment with NGF significantly induced both NEFL and p19-INK4d expression in SH-SY5Y cells (P < 0.001 and 0.001, respectively; Fig. 5G). In line with this, NTRK1 and p19-INK4d expression were positively correlated in a set of 478 primary neuroblastomas (Pearson correlation coefficient = 0.51, 95% CI: 0.44–0.57, P < 0.001; data not shown). Together, these results identify p19-INK4d as an inducer of neuroblastoma senescence and a sensor and effector of differentiation signals.
INK4 family proteins govern the G1- to S-phase cell cycle checkpoint, thus operating as crucial control units of the RB1 tumor suppressor pathway, which is frequently impaired in neuroblastomas (1–7). In the present study, we elucidated the oncosuppressive role of the INK4 family member p19-INK4d in neuroblastoma and identified novel mechanisms leading to its deregulation in this disease. Association of p19-INK4d with neuroblastoma was initially described in a small tumor cohort, where primary high-risk neuroblastomas expressed reduced levels of p19-INK4d (23). Here, we extend this finding in a large cohort of 478 primary tumors, and describe the correlation of reduced p19-INK4d expression in the primary tumors with poor patient survival and known neuroblastoma risk factors, such as MYCN amplification. Using conditional MYCN overexpression or knockdown, we provide functional evidence for MYCN-mediated repression of p19-INK4d expression in neuroblastoma cells. We further identified epigenetic repression by CpG methylation as a potential mechanism for p19-INK4d repression in neuroblastoma cells. Treatment with the demethylating agent, DAC, has previously been reported to reactivate repressed p19-INK4d expression in lung cancer cells, even though no hypermethylation of the p19-INK4d promoter was detected in the investigated cells (22). In line with these findings, initial methylation-specific PCR analyses failed to detect p19-INK4d promoter CpG methylation upstream of the TSS in neuroblastomas (data not shown). However, 450K array and MassARRAY methylation analysis identified a region ∼1.5 kb downstream of the p19-INK4d TSS which was differentially methylated in neuroblastomas. Recent studies have challenged the dogma that expression-associated CpG methylation is restricted to the regulatory regions upstream of the TSS by showing that intragenic methylation is frequently the event correlated with expression of the respective gene (27–29). Our in-depth analysis revealed that hypermethylation at this intragenic site of p19-INK4d was significantly associated with decreased p19-INK4d expression, high-risk disease and amplified MYCN in the tumors. In line with this observation, treatment with the demethylating agent, DAC, increased p19-INK4d expression in three of the five investigated cell lines, all of which harbored MYCN amplifications. Intriguingly, induction of an MYCN shRNA in the MYCN-amplified IMR5-75 cell line phenocopied the effect of DAC in terms of p19-INK4d activation, and no additional effect of DAC treatment was observed in this setting. This suggests that MYCN-mediated p19-INK4d repression may involve DNA methylation, which would be interesting to address in more detail in future analyses. To validate demethylation-associated p19-INK4d activation, we analyzed p19-INK4d expression in publicly available high-throughput expression data from DAC-treated neuroblastoma cell lines provided by Decock and colleagues (30). The data confirm DAC-dependent p19-INK4d activation in three (two harboring MYCN amplifications) of eight cell lines investigated. The DNA methylation data we present are in line with established concepts (i) associating aberrant DNA methylation with poor patient outcome and MYCN amplification in neuroblastomas (31,32) and (ii) showing that epigenetic silencing affects crucial tumor suppressor genes in neuroblastomas (33–36).
Repression of p19-INK4d, as seen in aggressive neuroblastomas, may confer a selective advantage to malignant neuroblastoma cells. We hypothesized that p19-INK4d (re-) expression would diminish features of neuroblastoma malignancy. In accordance, ectopic p19-INK4d expression reduced the viability and clonogenicity of SH-EP, SH-SY5Y and IMR5-75 neuroblastoma cells, as well as the capability of IMR5-75 cells to grow anchorage-independently. Ectopic p19-INK4d expression enriched the fraction of SH-EP and SH-SY5Y cells in the G1/0 phase while reducing the cell fraction proceeding to S-phase. This is in line with p19-INK4d functioning as a cell cycle constraint in the RB1 pathway (14) and with the decreased pRB1Ser-780 phosphorylation and E2F target gene expression in SH-EP and SH-SY5Y cells that we observed after ectopic p19-INK4d expression. Currently, it is unclear how p19-INK4d-dependent growth suppression and G1 phase enrichment were mediated in IMR5-75 cells, since p19-INK4d did not influence RB1 pathway targets in this cell line. Ectopic expression of p19-INK4d may not have been sufficient to completely repress CDK4 activity in IMR5-75 cells due to the high CDK4 levels in MYCN-mplified neuroblastoma cells (37). Hence, the observed phenotypic effects may have resulted from CDK4-independent functions of p19-INK4d. p19-INK4d has previously been described to confer chemo- or UV-resistance, independently of CDK4, through enhancing DNA repair capacity in neuroblastoma cells (24,25). These data show that p19-INK4d function extends beyond CDK4 inhibition and indicate that further, as yet unexplored, p19-INK4d functions may underlie the observed effects in IMR5-75 cells. Independent of the mode of action in IMR5-75 cells, the functional data we present strongly suggest that p19-INK4d expression is oncosuppressive in the context of neuroblastoma. Our findings that p19-INK4d induces senescence and is involved as a sensor and effector of differentiation signals in neuroblastoma cells is another line of evidence supporting an oncosuppressive role for p19-INK4d. Specifically, p19-INK4d was upregulated by ATRA-induced neuroblastoma cell differentiation, which is in accordance with studies reporting p19-INK4d induction by retinoids in other tumor entities (19,38). We also associated p19-INK4d expression with NTRK1 signaling, a central effector of neuroblastoma differentiation (39), in both neuroblastoma cell lines and tumors. Combined with the insight that ectopic p19-INK4d expression was sufficient to trigger processes of neuronal differentiation in neuroblastoma cells, p19-INK4d is a potential suppressor of neuroblastoma.
We conclude that p19-INK4d is a crucial component of neuroblastoma cell growth control that is transcriptionally repressed in aggressive subtypes. We provide functional evidence that MYCN is involved in p19-INK4d repression and identify a region within the gene whose hypermethylation is associated with p19-INK4d repression and amplified MYCN. In view of the oncosuppressive effects of p19-INK4d, its targeted reactivation may offer an attractive treatment option to induce growth arrest and/or differentiation in aggressive MYCN-amplified neuroblastomas. Here we provide the foundation for understanding direct and epigenetic regulation of p19-INK4d in neuroblastoma.
MATERIALS AND METHODS
Patient samples for methylome analysis
DNA was isolated from snap-frozen neuroblastoma tissue. All 105 patients were enrolled in the German Neuroblastoma Trial and diagnosed between 1998 and 2011. Informed consent was obtained from the patient's parents. Patients were treated according to the guidelines established by the German Neuroblastoma Trial, with risk stratification criteria as described in the German Trial Protocol for Risk Adapted Treatment of Children with Neuroblastoma. Clinical disease stage was evaluated according to the International Neuroblastoma Staging System (INSS): Stage 1, n = 10; Stage 2, n = 9; Stage 3, n = 10; Stage 4, n = 56; Stage 4S, n = 20. Patient's age at diagnosis ranged from 0 to 24.6 years (median age, 1 year). MYCN amplification was observed in 33 tumors and was absent in 72 tumors. Risk estimation of the patients was performed according to NB2004 (low risk, n = 40; intermediate risk, n = 9; high risk, n = 56).
Cell culture, transfections and treatments
Be(2)-C, IMR32, IMR5-75, SH-EP and SH-SY5Y neuroblastoma cell lines were cultured as described previously (6). All lines were kindly provided by Dr Larissa Savelyeva (German Cancer Research Center) and authenticated by multiplex-FISH karyotyping and short tandem repeat DNA typing at the DSMZ-German Collection of Mircoorganisms and Cell Cultures before the project start. ATRA treatment was performed as previously described (40). NTRK1 was activated using 100 ng murine NGF per milliliter culture medium (2.5S mNGF; Promega, Mannheim, Germany). siRNA transfections were performed using HiPerFect transfection reagent (Qiagen, Hilden, Germany) according to the manufacturers' description. 50 000 cells were seeded in 6-well plates 24 h prior to transfection. Final siRNA concentrations in wells were 10 nm. siRNAs used in this study are listed in Supplementary Material, Table S1. Transient transfections of plasmid coding a shRNA against MYCN were performed with Effectene transfection reagent (Qiagen) according to the manufacturers' description. 1 000 000 cells were plated on 10 cm dishes and were transfected 24 h later using 2 μg of plasmid DNA. Stable cell clones expressing p19-INK4d (IMR5-75, SH-EP, SH-SY5Y) NTRK1 (SH-SY5Y), MYCN (SH-SY5Y) or shRNA against MYCN (IMR5-75) under the control of tetracycline repressor were generated using Gateway Technology (Invitrogen, Carlsbad, CA, USA) (40). Expression was induced using 1 μg tetracycline/ml culture medium, and uninduced controls were treated with 0.07% solvent (70% ethanol). DAC treatment was performed for 72 h at a final concentration of 3 μm with medium changes and re-substitution of the drug every 24 h.
Cell cycle analysis, cell viability, clonogenicity, senescence and soft agar assays
Cell cycle distribution was determined using flow cytometry as previously described (41). Cell viability was assessed in triplicates in 96-well plates (1500 cells/well) using alamarBlue® assays (AbD serotec, Düsseldorf, Germany) according to the manufacturers' instructions. Clonogenicity was assessed in triplicates in 6-well plates (1500 cells/well) after 1–2 weeks of culture. Cultures were fixed, Giemsa stained and colonies were counted using Quantity One software (Bio-Rad, Munich, Germany). Soft agar assays were performed as described previously (40). Senescence was analyzed using the senescence-activated β-galactosidase staining kit (Cell Signaling Technology, Danvers, MA, USA) according to the manufacturers' instructions. Prior to fixing and staining, 10 000 cells per well were cultured for 48 h in 24-well plates.
Immunoblotting was performed using antibodies against p19-INK4d (DCS-100, Santa Cruz, Santa Cruz, CA, USA), CDK4 (DCS-31, Dianova, Hamburg, Germany), Ser780-RB [#9307, Millipore (Cell Signaling), Billerica, MA, USA], E2F1 (Medac, Wedel, Germany), SKP2 (Invitrogen) MYCN (sc-53993, Santa Cruz) and β-actin (Sigma-Aldrich, St. Louis, MO, USA) as previously described (42).
p19-INK4d mRNA expression in primary neuroblastomas was determined using existing data from oligonucleotide microarray analysis. Sample set composition, sample preparation and generation of single-color gene-expression profiles from 478 primary neuroblastomas were described previously (43). Raw and normalized microarray data are available at the ArrayExpress database (http://www.ebi.ac.uk/arrayexpress/; accession: E-TABM-38, E-MTAB-161 and E-MTAB-179).
Quantitative Real-time RT-PCR
RNA was isolated from neuroblastoma cell lines using TRIzol-reagent (Invitrogen) or the RNeasy Mini Kit (Qiagen) according to the manufacturers' instructions. cDNA synthesis, quantitative Real-Time RT-PCR (qRT-PCR) and data normalization were done as previously described (44). Primer sequences are listed in Supplementary Material, Table S2.
mRNA-sequencing and bioinformatics analysis
RNA from un-induced SY5Y-MYCN cells and cells induced to overexpress MYCN with 1 μg/ml doxycycline (Sigma-Aldrich) for 1, 4, 24 and 48 h were used for mRNA-sequencing. Biological duplicates for each sample were generated. RNA was isolated, libraries prepared, sequenced on a Genome Analyser IIx (Illumina, San Diego, CA, USA) and bioinformatics performed as previously described (45).
Genome-wide DNA methylation preprocessing
p19-INK4d DNA methylation status was derived from genome-wide DNA methylation data assessed in 105 neuroblastomas using the Infinium HumanMethylation450 BeadChip (Illumina) according to the manufacturers' instructions. Probes were removed if (i) proportion of non-detectable beta values >0.3 (N = 379), (ii) containing a single-nucleotide polymorphism at/near the targeted CpG site according to R-Forge package IMA (46) (https://rforge.net/IMA/) (N = 92 600), (iii) control probes (N = 65), (iv) mapping to the X or Y chromosome (N = 10 351). Together, N = 382 182 probes were kept for further analysis. The k-nearest neighbors method was used to impute missing values (47). Finally, subset quantile normalization was applied (48).
Bisulfite conversion, primer design strategy and MassARRAY methylation analysis
Genomic DNA was converted by bisulfite treatment using EZ DNA methylation kit (Zymo Research, Orange, CA, USA). After bisulfite conversion and PCR amplification of the DNA region of interest, MassARRAY technology (Sequenom, San Diego, CA, USA) was used to quantify the methylation status of the specific region in a set of 105 primary neuroblastomas as described (49). Three PCR primer assays were designed to cover a region spanning from 1.4 to 2 kb downstream of the p19-INK4d TSS. For CpGs present in more than one amplicon the mean percent methylation value was calculated. To illustrate MassARRAY data a heatmap was computed.
Student's t-test was used in cell lines experiments. Tests were computed for groups consisting of at least three measurements each. Overall and event-free survival distributions were estimated using the method of Kaplan and Meier. Maximally selected log-rank statistics were used to estimate cutpoints of p19-INK4d expression with respect to survival endpoints. Maximally selected Wilcoxon statistics were used to estimate CpG methylation cutpoints that separate patient groups with differential expression of the associated gene (50). This approach does not aim to identify global methylation-expression correlations across all patients. The method is designed to discover differences in gene-expression distribution with respect to differential methylation. The association of hypermethylation and high-risk neuroblastoma was tested by Fisher's exact test. The Wilcoxon rank-sum test was used to test for association between p19-INK4d expression or methylation with established prognostic factors. Pearson's correlation coefficient was computed to estimate the correlation between expression of NTRK1 and p19-INK4d as well as between p19-INK4d methylation and expression. The result of a statistical analysis was considered significant when the P-value of the corresponding statistical test was <5%. All statistical analyses were done using the R software package version 3.0.2 (51).
This work was supported with grants from the European Union [#037260 (E.E.T. Pipeline; FP6), #259348 (ASSET; FP7)] and the German Bundesministerium für Bildung und Forschung [#0316076A (MYC-NET; CancerSys), #01GS0896 (NGFNPlus), #01GS0895 (NGFNPlus)].
We thank the tumor bank team at the University Hospital of Cologne (Germany) for providing primary neuroblastoma samples, Klara Zwadlo, Gabriele Becker and Yvonne Kahlert for excellent technical assistance and Kathy Astrahantseff and Lena Brückner for proofreading the manuscript.
Conflict of Interest statement. None declared.