Neuroblastoma is an embryonic solid tumor of neural crest origin and accounts for 11% of all cancer-related deaths in children. Novel therapeutic strategies are therefore urgently required. MYCN oncogene amplification, which occurs in 20% of neuroblastomas, is a hallmark of high risk. Here, we aimed to exploit molecular mechanisms that can be pharmacologically addressed with epigenetically modifying drugs, such as histone deacetylase (HDAC) inhibitors. Grainyhead-like 1 (GRHL1), a gene critical for Drosophila neural development, belonged to the genes most strongly responding to HDAC inhibitor treatment of neuroblastoma cells in a genome-wide screen. An increase in the histone H4 pan-acetylation associated with its promoter preceded transcriptional activation. Physically adjacent, HDAC3 and MYCN colocalized to the GRHL1 promoter and repressed its transcription. High-level GRHL1 expression in primary neuroblastomas correlated on transcriptional and translational levels with favorable patient survival and established clinical and molecular markers for favorable tumor biology, including lack of MYCN amplification. Enforced GRHL1 expression in MYCN-amplified neuroblastoma cells with low endogenous GRHL1 levels abrogated anchorage-independent colony formation, inhibited proliferation, and retarded xenograft growth in mice. GRHL1 knockdown in MYCN single-copy cells with high endogenous GRHL1 levels promoted colony formation. GRHL1 regulated 170 genes genome-wide, and most were involved in pathways regulated during neuroblastomagenesis, including nervous system development, proliferation, cell–cell adhesion, cell spreading, and cellular differentiation. In summary, the data presented here indicate a significant role of HDAC3 in the MYCN-mediated repression of GRHL1 and suggest drugs that block HDAC3 activity and suppress MYCN expression as promising candidates for novel treatment strategies of high-risk neuroblastoma. Cancer Res; 74(9); 2604–16. ©2014 AACR.
Neuroblastoma, an embryonic tumor of neuroectodermal origin, is the most common extracranial solid malignancy of childhood. It is characterized by a heterogeneous tumor biology and, hence, a clinical variability ranging from spontaneous regression or localized, stable disease to rapid metastasizing progression with fatal outcome (1). Treatment of patients with neuroblastoma varies with this broad prognostic range, and spans observation or surgical tumor removal only to multimodal therapy in patients at high risk (2). The number of long-term survivors of high-risk disease has remained unsatisfactorily poor with survival rates as low as 20% to 40% despite considerable efforts over the last decades to improve the outcome (3). Major obstacles still to be overcome include managing of highly chemotherapy-resistant relapses frequently occurring years after initial diagnosis and managing resistance to induction therapy leading to early death by tumor progression in ultra high-risk patients (2). More than half of survivors suffer hearing loss, endocrine dysfunction, or other irreversible treatment-related side effects that limit their quality of life (4). The development of more effective and less toxic novel targeted therapeutics is urgently needed. Such new agents will likely have to be used in combination with conventional chemotherapeutic strategies to minimize the onset of resistance and to potentially enable a dose reduction of chemotherapy to limit treatment-related side effects.
The basic helix-loop-helix transcription factor, MYCN, regulates migration, proliferation, and differentiation of neural crest progenitor cells (5). MYCN amplification occurs in approximately 20% of primary human neuroblastomas (1, 6) and increases the rate of DNA synthesis, promotes cell-cycle progression and suppresses differentiation (7, 8). We and others have demonstrated that histone deacetylase (HDAC) inhibitors trigger signaling pathways involved in cell-cycle arrest, differentiation, and/or cell death of neuroblastoma cells and slow tumor growth in xenotransplantation (9–12). On a molecular level, HDAC inhibition results in MYCN protein expression to be diminished (11). Different HDACs exert nonredundant functions in neuroblastoma biology. HDAC2 cooperates with MYCN to suppress miR-183–mediated apoptosis (12), and HDAC8 inhibits differentiation of neuroblastoma cells (13). Recently, we showed that HDAC10 promotes autophagy-mediated neuroblastoma cell survival (14). Further advances in understanding the role of HDACs are a prerequisite to fully exploiting the level of plasticity that can be therapeutically addressed with acetylome-modifying drugs such as HDAC inhibitors. Here, we aimed to assess the significance of grainyhead-like 1 (GRHL1) expression for neuroblastoma biology because HDAC inhibitor treatment strongly induces its expression in neuroblastoma cells. GRHL1 is an evolutionary conserved transcription factor with a critical role in Drosophila nervous system development (15). We assessed the expression of GRHL1 in primary neuroblastomas, and unraveled its upstream repressive control and the effect of GRHL1 reexpression on phenotype in neuroblastoma cell models.
Materials and Methods
Characteristics of patients and tumor samples
GRHL1 mRNA expression values were derived from existing whole-genome expression profiles of 476 primary neuroblastomas samples (16). Tumor samples were collected internationally between 1989 and 2007 and had at least 60% tumor cell counts. The expression profiles of 400 tumors were previously classified using Prediction Analysis for Microarrays (PAM; ref. 17). A second independent dataset for GRHL1 mRNA expression in primary neuroblastomas was derived from publicly accessible whole-genome expression profiles of 88 neuroblastomas (18) via the R2 microarray analysis and visualization platform (http://r2.amc.nl). The GRHL1 probeset with the highest average signal was selected for analysis. Formalin-fixed, paraffin-embedded samples from 15 primary neuroblastomas (10 localized or stage IV-S and 5 stage IV, two of which harbored MYCN amplifications) were selected for GRHL1 immunohistochemistry.
The BE(2)-C cell line was obtained from European Collection of Animal Cell Cultures, and the Kelly and IMR-32 cell lines from the DSMZ (Braunschweig, Germany). SH-EP, SK-N-AS, and the synthetic MYCN-inducible SH-EP Tet-21/N cell model (7) were kindly provided by Manfred Schwab [Department of Tumor Genetics, German Cancer Research Center (DKFZ), Heidelberg, Germany]. Cell lines were monitored for infections by high-throughput multiplex cell contamination testing (19). Cell line authenticity was validated by high-throughput single-nucleotide polymorphism-based assays (20). BE(2)-C, IMR-32, and SK-N-AS cell lines were cultured in Dulbecco's Modified Eagle Medium (Lonza), supplemented with 10% fetal calf serum (FCS; Sigma-Aldrich) and 1% nonessential amino acids (NEAA; Lonza) at 37°C and 5% CO2. Kelly and SH-EP cell lines were cultured in RPMI-1640 medium (Lonza) supplemented with 10% FCS and 1% NEAA at 37°C and 5% CO2. SH-EP Tet-21/N cells were cultured in RPMI-1640 medium with HEPES supplemented with 10% FCS, 1% NEAA, 200 μg/mL G418 (Merck Millipore), and 100 μg/mL hygromycin B (Sigma-Aldrich), and treated with 1 μg/mL tetracycline (AppliChem) to suppress synthetic MYCN expression. Stock solutions of panobinostat (1 mmol/L; Selleck Chemicals), abexinostat (10 mmol/L; Pharmacyclics), and vorinostat (1 mmol/L; Chemos) were prepared in dimethyl sulfoxide (DMSO), the valproic acid stock solution (1 mol/L; Sigma-Aldrich) in D-PBS (Cambrex), and the Helminthosporium carbonum (HC)-toxin stock solution (0.1 mmol/L; Sigma-Aldrich) in methanol. Actinomycin D (stock solution 5 mg/mL in DMSO; Sigma-Aldrich) was added at a final concentration of 1 μg/mL to the medium.
RNA extraction and quantitative real-time PCR
Total RNA was isolated from cell lines or snap-frozen xenograft tissue using the RNeasy Mini Kit (Qiagen). The Thermo Scientific First-Strand cDNA Synthesis Kit was used to transcribe cDNAs for quantitative real-time PCR (qRT-PCR) analysis. Relative gene expression was measured using SYBR Green dye (Eurogentec) on an ABI Prism 7500 thermal cycler (Perkin-Elmer Applied Biosystems; ref. 11), and normalized to the averaged expression of ACTB and 18S rRNA or the averaged expression of SDHA and HPRT1, a gene pair that is consistently expressed in stage IV and IV-S neuroblastomas (21). Primers used in qRT-PCR are listed in the Supplementary Table S1. Data were analyzed using Applied Biosystems 7500 software v2.0.5, and changes in expression were calculated using the ΔΔCt method for cell lines and using a standard curve for xenografts.
BE(2)-C cells were transfected for 24 hours with empty-, LacZ-, or GRHL1 vectors in culture, then 2 × 106 viable cells suspended in 200 μL Matrigel (BD Biosciences) were subcutaneously implanted in the flanks of 6-week-old female CB17-SCID mice (n = 12 per study group). Animals were sacrificed 8 days after grafting. Tumor size was daily measured with a caliper. Tumor volume was calculated by π/6(w1 × w2 × w2), where w1 was the largest tumor diameter and w2 was the smallest tumor diameter. Experiments conformed to the regulatory standards and were approved by the local ethics committee.
Kaplan–Meier curves relating GRHL1 expression to patient survival in the 476 neuroblastoma cohort were plotted using the Survival package for the R programming language. The optimal cutoff for both overall and event-free survival analyses was chosen as the expression value where the log-rank statistic for the separation of both survival curves reached a maximum. This value was calculated using the Maxstat package for the R programming language, and was 151.109. The reported P values were calculated using the P value approximation method according to Lausen and colleagues (22). Mann–Whitney U test was used for the comparison of GRHL1 expression between different patient subgroups. Kaplan–Meier analyses and comparison of GRHL1 expression between different patient subgroups for the 88 neuroblastoma cohort were performed online in the R2 platform (http://r2.amc.nl) and the resulting survival curves, box plots and P values (log-rank test) were downloaded. Results of cell culture experiments were compared using the one-sample t test in GraphPad Prism version 5.0 (GraphPad Software Inc.) unless otherwise indicated. A mixed linear model with fixed-factor treatment and random intercept for each siRNA was used to compare GRHL1 mRNA expression in cell groups transfected with either of the two negative controls or either of the two HDAC-specific siRNAs. The same model was used to analyze the effect of GRHL1 knockdown in MYCN single-copy cell lines on anchorage-dependent and -independent colony formation. The mixed linear model analyses were performed using SAS PROC MIXED, SAS Version 9.2 (SAS Institute Inc.). In vivo effects of enforced GRHL1 expression on tumor growth were analyzed using SAS PROC MIXED, SAS Version 9.2 (SAS Institute Inc.) by a mixed linear model with a fixed slope for each group and random intercept for each mouse after log transformation of the tumor volumes. P values below 0.05 were considered significant. Additional methods are described in the Supplementary Materials and Methods.
GRHL1 is an early response gene to HDAC inhibitor treatment
To decipher molecular mechanisms controlled by HDACs that have remained enigmatic, transcriptome changes were monitored in time-course experiments via gene expression profiling in SH-EP neuroblastoma cells following treatment with either valproic acid or HC-toxin, two structurally divergent HDAC inhibitors (23). In a genome-wide comparison, GRHL1 was the second most strongly induced gene (∼30-fold) after 6 hours of treatment with each of the HDAC inhibitors individually (Fig. 1A). GRHL1 expression remained elevated throughout the 120-hour time-course (Fig. 1A). In qRT-PCR analysis, GRHL1 induction reached significance at 1.5 hours of treatment (P < 0.01), peaked at 5- to 7-fold above the control at 6 hours, and then declined to 2- to 3-fold above the control, which was maintained at all the later time points investigated (Fig. 1B). Treatment with valproic acid, HC-toxin, or three other HDAC inhibitors that are either clinically approved or being evaluated in phase II or III clinical trials (vorinostat, abexinostat, and panobinostat) similarly induced GRHL1 by 2.7- to 9-fold in a panel of five neuroblastoma cell lines, supporting GRHL1 induction as a common event of HDAC inhibition in neuroblastoma cells (Fig. 1C). To differentiate between de novo transcript synthesis and increased stability, we cotreated BE(2)-C cells with actinomycin D and panobinostat or solvent control for up to 8 hours. Blocking de novo RNA synthesis with actinomycin D did not alter GRHL1 levels between HDAC inhibitor-treated and control cells, confirming transcriptional activation (Fig. 1D). To test for epigenetic changes before GRHL1 transcriptional activation, we performed chromatin immunoprecipitation (ChIP) in BE(2)-C cells treated for 5 hours with panobinostat or solvent control using an antibody against pan-acetylated histone H4, and looked for changes in the GRHL1 promoter (24). We detected an increase in pan-acetylated histone H4 associated with a DNA sequence approximately 500 bp proximal to the transcriptional start site of the GRHL1 gene after HDAC inhibitor treatment, indicating that epigenetic changes preceded transcriptional activation (Fig. 1E and F). We performed Western blot analyses on whole-cell lysates from selected neuroblastoma cell lines to investigate whether the GRHL1 induction by HDAC inhibitor treatment is translated to the protein level. HDAC inhibitor treatment for 48 hours increased GRHL1 expression by approximately 10-fold (Fig. 1G). We have previously shown that treating CB17-SCID mice carrying subcutaneous BE(2)-C xenografts with panobinostat arrested tumor growth by treatment day 14 (12). Here, we assessed GRHL1 expression in these xenograft tumors using qRT-PCR, and detected elevated levels of GRHL1 expression (Fig. 1H). Altogether, our data indicate that HDAC inhibitors, including those in or being tested for clinical use, induce GRHL1 transcription in neuroblastoma cells in vitro and in vivo.
High-level GRHL1 expression in neuroblastomas predicts favorable survival
Having shown that GRHL1 expression can be pharmacologically triggered in neuroblastoma cells, we assessed whether GRHL1 is differentially expressed in primary tumors. We reanalyzed microarray expression data from a cohort of 476 neuroblastomas (16). Kaplan–Meier analysis showed that high-level GRHL1 expression in tumors correlated both with favorable overall and event-free patient survival (Fig. 2A and B). This correlation was confirmed in microarray expression data from a second independent cohort of 88 neuroblastomas (Supplementary Fig. S1A and S1B; ref. 18). High-level GRHL1 expression in neuroblastomas from the 476 tumor cohort also significantly correlated with established clinical and molecular markers for favorable tumor biology, including International Neuroblastoma Staging System (INSS) localized or IV-S stage disease, age at diagnosis ≤18 months, favorable Shimada/International Neuroblastoma Pathology Classification (INPC) tumor histology, lack of MYCN amplification or 1p aberrations, and a low-risk tumor transcriptional profile defined by PAM analysis (Table 1; ref. 17). Correlations between elevated GRHL1 tumor expression and all available clinical and molecular parameters for favorable tumor biology were confirmed in the 88 tumor cohort. These included INSS stage IV-S, age at diagnosis <12 months, and single-copy MYCN status (Supplementary Fig. S1C–S1E). Taken together, high-level GRHL1 expression in primary neuroblastomas signals a favorable tumor biology and patient prognosis. To assess whether the differential GRHL1 expression in primary neuroblastomas is mirrored at the protein expression level, we assessed GRHL1 expression immunohistochemically in 15 selected samples from tumors with the most divergent characteristics and semiquantified its expression (25). Nearly all INSS stage I, IIa, and IV-S tumors strongly expressed GRHL1, and expression was localized to the nuclei, whereas no GRHL1 expression was detected in any of the INSS stage IV tumors with or without MYCN amplifications (Supplementary Fig. S2A and S2B). These results confirmed that the differential pattern of GRHL1 expression is translated to the level of protein expression in neuroblastomas having the most divergent tumor biologies.
|Factor .||n .||Mean rank .||P .|
|I, II, III, IV-S||328||248.15||<0.05|
|Age at diagnosis|
|Factor .||n .||Mean rank .||P .|
|I, II, III, IV-S||328||248.15||<0.05|
|Age at diagnosis|
NOTE: GRHL1 mRNA expression values were derived from existing whole-genome expression profiles of 476 neuroblastoma samples (16). The expression profiles of 400 tumors were previously classified using PAM (17). Mann–Whitney U test was used for the comparison of GRHL1 expression between different subgroups.
MYCN is a transcriptional repressor of GRHL1
A major factor influencing tumor biology is the MYCN status in the tumor (6), which inversely correlated with GRHL1 expression in primary neuroblastomas (Table 1). We used Western blotting to assess the GRHL1 expression in the panel of cell lines derived either from neuroblastomas lacking (SH-EP, SK-N-AS) or harboring (BE(2)-C, Kelly, and IMR-32) MYCN amplifications. The GRHL1 expression pattern in cell lines reflected that in primary tumors, well demonstrated by the higher GRHL1 expression in MYCN single-copy cell lines (Fig. 3A). We next used several different cellular models for neuroblastoma to investigate the nature of this relationship and the influence of MYCN on GRHL1 expression. We analyzed changes in GRHL1 expression occurring within 4 to 12 hours in the synthetic MYCN-inducible system, SH-EP Tet-21/N. Conditional MYCN expression downregulated GRHL1 expression on mRNA level by approximately 40% and on the protein level by approximately 30% within 4 hours (Fig. 3B). GRHL1 expression remained suppressed up to 12 hours after conditional MYCN expression (Fig. 3B). Depletion of endogenous MYCN expression in BE(2)-C cells by transient expression of a short hairpin RNA (shRNA) plasmid directed against MYCN for 48 and 72 hours increased GRHL1 expression on mRNA level by 1.5-fold and on the protein level by 1.8-fold at both time points investigated (Fig. 3C). Results from these two models suggest that MYCN suppresses GRHL1 expression. We performed ChIP-PCR experiments to test whether MYCN associates with the GRHL1 promoter. Not only was MYCN enriched in the GRHL1 promoter (Fig. 3D), but also this enrichment was confined to the region previously shown to be enriched for pan-acetylated histone H4 following HDAC inhibitor treatment (Fig. 1F). Taken together, our data argue for a role of MYCN in the transcriptional repression of GRHL1 in neuroblastoma cells.
HDAC3 depletion recapitulates the GRHL1 induction by HDAC inhibitor treatment
The HDAC inhibitors promoting GRHL1 transcription (Fig. 1) block the enzymatic activity of multiple HDAC family members. To identify HDACs involved in the transcriptional repression of GRHL1, we transiently knocked down each of the eleven HDACs belonging to classes I, IIa/b, and IV in BE(2)-C cells, then assessed GRHL1 expression. We used two different siRNAs for each HDAC (Supplementary Table S2) to control for unspecific and off-target effects. Only HDAC3 depletion induced GRHL1 expression 2.5- to 4-fold 96 hours after knockdown (Fig. 4A). HDAC3 depletion elevated GRHL1 expression in whole-cell lysates examined by Western blotting by approximately 2.5-fold 96 hours after knockdown (Fig. 4B). Enforced HDAC3 expression reduced endogenous GRHL1 mRNA and protein levels by 11% and 37%, respectively (Fig. 4C and D), and partly reverted the induction caused by HDAC3 depletion (Fig. 4E). These data indicated that (i) HDAC inhibitor treatment induced epigenetic changes before GRHL1 transcriptional activation (Fig. 1F), (ii) both HDAC3 and MYCN negatively regulated GRHL1 (Figs. 4A–E and Fig. 3B and C), and (iii) MYCN was enriched at the GRHL1 promoter site (Fig. 3D). To test whether HDAC3 is also recruited to this GRHL1 promoter region, we performed ChIP-PCR. HDAC3 was found to be recruited to this region (Fig. 4F). To analyze whether MYCN is required for HDAC3 recruitment to the GRHL1 promoter, we performed ChIP-qRT-PCR experiments. Transient MYCN knockdown in BE(2)-C cells with a shRNA plasmid directed against MYCN for 72 hours decreased recruitment of HDAC3 to the GRHL1 promoter by approximately 20% (Fig. 4G), indicating that MYCN contributes to the recruitment of HDAC3 to the GRHL1 promoter. Because MYCN and HDAC3 both associate with the same GRHL1 promoter region, we tested whether MYCN and HDAC3 proteins physically cooperate in coimmunoprecipitation experiments in BE(2)-C cells. In complexes immunoprecipitated with an antibody against HDAC3, MYCN was 1.6-fold enriched above the IgG control immunoprecipitate (Fig. 4H). Immunoprecipitation using an anti-MYCN antibody demonstrated a 1.4-fold enrichment of HDAC3 in complexes above the control (Fig. 4I). Taken together, HDAC3 is physically adjacent to MYCN and negatively regulates GRHL1.
GRHL1 inhibits colony formation, proliferation, and xenograft growth
Data obtained from primary tumors suggested that high-level GRHL1 expression is associated with favorable tumor biology. We functionally analyzed the phenotypic consequences of a modified GRHL1 expression in neuroblastoma cells. MYCN-amplified neuroblastoma cells endogenously express low GRHL1 levels (Fig. 3A). Hence, we transiently transfected the MYCN-amplified cell lines BE(2)-C, Kelly, and IMR-32 cells with empty-, LacZ- or GRHL1 vector, resulting in a 200- to 400-fold increase in GRHL1 mRNA expression (Fig. 5A). GRHL1 protein levels reached or were below those detected in MYCN single-copy cell lines, indicating in the transfected cells an expression at the range of the physiologic level of GRHL1 (Fig. 5B and C). Enforced GRHL1 expression reduced anchorage-independent colony formation capacity of all three cell lines in soft agar by up to 85% of control cells transfected with empty- or LacZ vector (Fig. 5D and E). We used semiautomated Trypan blue staining to assess proliferation in these cell lines, and demonstrated that enforced GRHL1 expression inhibited BE(2)-C, Kelly, and IMR-32 cell proliferation by up to 25% compared with control cells (Fig. 5F). Moreover, enforced GRHL1 expression in BE(2)-C cells markedly suppressed growth of xenografts of these cells in CB17-SCID mice (Fig. 5G). MYCN single-copy cells endogenously express high GRHL1 levels (Fig. 3A). Hence, we knocked down GRHL1 in SH-EP and SK-N-AS cells with two different siRNAs (Fig. 5H) and investigated its effect on anchorage-independent and -dependent colony formation. GRHL1 knockdown triggered anchorage-independent colony formation of SK-N-AS cells in soft agar by approximately 14-fold (Fig. 5I and J). SH-EP cells, which are not clonogenic in soft agar, were not investigated. GRHL1 knockdown increased colony formation capacity of SH-EP and SK-N-AS cells in the anchorage-dependent assay by approximately 2- to 21-fold (Fig. 5K and L). Taken together, GRHL1 exerts tumor-suppressive effects in MYCN-amplified cell lines.
GRHL1 favorably influences neuroblastoma biology at molecular level
To explore the effect of high-level GRHL1 expression on gene regulation, we performed time-resolved whole-genome expression analysis on BE(2)-C cells transfected with empty- or GRHL1 vector. Enforced GRHL1 expression resulted in differential regulation of 170 genes over time (24, 48, and 72 hours), with the majority of genes being induced (Supplementary Table S3). We performed hypergeometrical testing for overrepresented biologic processes to gain insight into the molecular mechanisms that are influenced by GRHL1. Terms associated with nervous system development, negative regulation of cell proliferation, negative regulation of the extracellular signal-regulated kinase (ERK1) and ERK2 cascade, substrate adhesion-dependent cell spreading, signal transduction, cell–cell adhesion, and cell differentiation were enriched (Fig. 6A). Differential expression of 10 genes representing the biologic processes enriched with the highest significance was confirmed by qRT-PCR (Fig. 6B). We assessed ERBB3 and protein kinase c-Δ (PRKCD) protein expression as representatives for the most enriched biologic process, signal transduction. Low-level ERBB3 expression has previously been shown to correlate with MYCN amplification in neuroblastomas and poor patient survival (26). PRKCD facilitates the induction of apoptosis in neuroblastoma cells (27). We assessed ERBB3 and PRKCD expression in whole-cell lysates of BE(2)-C cells 48 hours after transfection with empty- or GRHL1 vector on Western blot analyses to test whether the induction observed on the mRNA level is translated to the protein level. ERBB3 and PRKCD proteins were 2.5- to 3-fold induced (Fig. 6C), confirming a regulatory influence of GRHL1 on these genes. In summary, high-level GRHL1 expression favorably influences neuroblastoma biology both at molecular and phenotypic levels (summarized as a model in Supplementary Fig. S3).
This study was undertaken to decipher the molecular mechanisms regulated by HDACs in neuroblastoma. As an important finding, we show that GRHL1 is epigenetically and transcriptionally repressed by the chromatin-modifying enzyme, HDAC3, in association with MYCN. We show that high-level GRHL1 expression in primary neuroblastomas signals favorable tumor biology and patient prognosis in two independent cohorts. The differential pattern of GRHL1 expression was translated to the level of protein expression in neuroblastomas having the most divergent tumor biologies. Reexpression of GRHL1 in MYCN-amplified neuroblastoma cells abrogated anchorage-independent colony formation, inhibited proliferation, slowed growth in xenografts, and favorably influenced major pathways associated with tumor biology on a molecular level. Knockdown of GRHL1 in MYCN single-copy cells triggered colony formation. GRHL1 transcription was robustly induced by treatment of neuroblastoma cells with different clinically relevant small-molecule HDAC inhibitors, which block HDAC3 activity and suppress MYCN expression, thus, demonstrating that GRHL1 expression can be triggered via pharmacologic intervention.
The three mammalian GRHL genes (GRHL1, -2, and -3) constitute an evolutionarily highly conserved family of β-scaffold transcription factors (reviewed in ref. 28). GRHL1 maps to chromosome 2p25.1, a genomic region not frequently involved in aberrations in neuroblastomas (1). Previous sequencing projects on primary neuroblastomas that included this chromosomal region did not identify mutations in GRHL1 (18, 29–31). Grainyhead (grh) was first described as an embryonic lethal locus in Drosophila (32). The Grh protein acts as a transcriptional activator or repressor, and contains both a DNA-binding and a dimerization domain, with dimerization stabilizing the DNA–protein complex (reviewed in ref. 28). Grh is an important downstream effector of a temporal series of transcription factors including castor, which together orchestrate cell-cycle exit or apoptosis in neural progenitor cells (33). Beyond its critical role in Drosophila neurogenesis, Grh orthologs regulate other developmental processes in ectodermally derived tissues, including cuticle formation, tracheal elongation, dorsal closure, and wound repair (reviewed in ref. 28). CASZ1, the mammalian castor homolog, induces neuroblastoma cell differentiation, promotes adhesion, inhibits motility, and slows xenograft tumor growth (34). High-level CASZ1 expression in primary neuroblastomas correlates with favorable prognosis and is induced in neuroblastoma cells by treatment with retinoic acid or the HDAC inhibitor, depsipeptide (34). Genome-wide analysis of CASZ1 target genes carried out in three CASZ1-inducible human neuroblastoma cell lines revealed no evidence that GRHL1 expression was induced by CASZ1 (34). Taken together, the data on GRHL1 presented here identify yet another unbalanced developmental network in neuroblastoma cells contributing to tumor aggressiveness. This new network is interesting, because its upstream regulatory control composed of MYCN and HDAC3 is druggable by HDAC inhibitors, possibly in combination with other drugs reducing MYCN protein expression such as small-molecule Aurora-A inhibitors or BET bromodomain inhibitors (35–37).
We recently showed that MYCN and HDAC2 cooperate to repress miR-183, which has tumor-suppressive properties in neuroblastoma cells, via colocalization to the miR-183 promoter (12). Corecruitment of MYCN and HDAC2 to the CCNG2 promoter, resulting in transcriptional repression of cyclin G2, has also been reported (38). Furthermore, HDAC2 plays a role in the autoregulatory circuit of MYCN by associating with the MYCN promoter (39). MYCN also utilizes HDAC1 to repress transcription of genes involved in neuroblastoma cell differentiation (NTRK1, NGFR; ref. 40) or apoptosis (TP2; ref. 41). Here, we demonstrate that MYCN and HDAC3, another abundantly expressed class I HDAC family member, colocalize to the GRHL1 promoter in MYCN-amplified neuroblastoma cells and repress its transcription. The Myc box III in the MYC gene, another member of the MYC family of oncogenes, has been shown to contribute to transcriptional target gene repression via recruitment of HDAC3 to the chromatin of the Id2 and Gadd153 promoters (42). MYC was shown to cooperate with HDAC3 and colocalize to the two promoters of the miR-15a/16-1 cluster gene, DLEU2, in mantle cell lymphoma cell lines, and cause transcriptional repression (43). MYC was also reported to repress miR-29 via a corepressor complex that contained HDAC3 and EZH2, where restoration of miR-29 expression mediated suppression of lymphoma cell proliferation in vitro and in vivo (44). Collectively, the findings reported here support a role for HDAC3 in MYCN-mediated repression mechanisms, and highlight the significance of pediatric cancer epigenetics (45).
Here, we report that high-level GRHL1 expression in primary neuroblastomas correlates with clinical and molecular characteristics of a favorable tumor biology and signals a favorable patient prognosis in independent neuroblastoma cohorts. GRHL1 inhibited anchorage-independent growth and proliferation of MYCN-amplified neuroblastoma cell lines as well as xenografts in mice, demonstrating that GRHL1 suppresses critical processes of malignancy. GRHL1 was also in the group of genes responding most strongly to treatment with HDAC inhibitors in genome-wide expression studies, indicating that the MYCN/HDAC3-mediated transcriptional repression of GRHL1 can be pharmacologically reversed. This observation may present a new avenue for therapeutic intervention of MYCN-amplified high-risk neuroblastoma disease by blocking HDAC3 activity and suppressing MYCN protein expression with broad spectrum HDAC inhibitors that have already entered the clinic.
Disclosure of Potential Conflicts of Interest
O. Witt has a commercial research grant from Bayer Healthcare and Pharmacyclics. No potential conflicts of interest were disclosed by the other authors.
Conception and design: J. Fabian, O. Popanda, H.E. Deubzer
Development of methodology: J. Fabian, M. Lodrini, M. Schier
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): T. Thole, D. Capper, A. von Deimling, T. Milde, F. Westermann, B. Hero, F. Berthold
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Fabian, M. Lodrini, M. Schier, T. Hielscher, A. Kopp-Schneider, L. Opitz, D. Capper, A. von Deimling, F. Westermann, F. Roels, B. Hero, M. Fischer, H.E. Deubzer
Writing, review, and/or revision of the manuscript: J. Fabian, M. Lodrini, I. Oehme, A. Kopp-Schneider, L. Opitz, D. Capper, A. von Deimling, O. Popanda, A. Kulozik, O. Witt, H.E. Deubzer
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Fabian, I. Wiegand, U. Mahlknecht, F. Berthold, M. Fischer
Study supervision: O. Popanda, O. Witt, H.E. Deubzer
The authors thank J. Wünschel and M. Sohn for excellent technical assistance, Kathy Astrahantseff for critical reading of the manuscript, the German Neuroblastoma Tumor Bank (Cologne) for providing tumor samples, and both the Imaging and Cytometry Core Facility and the Genomics and Proteomics Core Facility of the DKFZ for their valuable services.
This work was supported by the BMBF through NGFNplus (F. Berthold, M. Fischer, F. Westermann, O. Witt, H.E. Deubzer), by the BMWi through SME (I. Oehme), and by the University of Heidelberg through the OLYMPIA MORATA and FRONTIER programs (H.E. Deubzer).