The Bromodomain Inhibitor JQ1 and the Histone Deacetylase Inhibitor Panobinostat Synergistically Reduce N-Myc Expression and Induce Anticancer Effects

MYCN oncogene amplification occurs in approximately 25% to 33% of neuroblastoma cases, and results in MYCN gene and N-Myc protein overexpression (1, 2). Patients with neuroblastoma associated with MYCN amplification usually undergo intensive chemotherapy and have poor overall and event-free survival.

Similar to other Myc family homologs, N-Myc modulates gene transcription and macromolecular biogenesis, blocks cell differentiation, stimulates cell-cycle progression as well as cell proliferation (2). To date, therapeutic strategies directly targeting the N-Myc or c-Myc protein have been elusive.

The BET bromodomain proteins, such as BRD3 and BRD4, bind to acetylated lysine residues on histone proteins as chromatin readers, and play essential roles in the transcription of oncogenes such as C-MYC, MYCN, BCL2, and FOSL1 (3–8). Small-molecule BET bromodomain inhibitors, such as JQ1 and I-BET151, competitively bind to acetyl–lysine recognition pockets, displace BET bromodomain proteins from chromatin, and reduce the expression of oncogenes, leading to cancer cell growth inhibition and apoptosis. BET bromodomain inhibitors have shown promising in vitro and in vivo anticancer effects against NUT-midline carcinoma, multiple myeloma, lymphoma, leukemia, medulloblastoma, and neuroblastoma (3–10). However, treatment with BET bromodomain inhibitors alone does not cause cancer remission in neuroblastoma-bearing mice.

Histone deacetylases (HDAC) remove the acetyl group from the lysine residue on the histone tails, leading to chromatin compaction and transcriptional repression, mainly of tumor suppressor genes (11). HDAC inhibitors, such as panobinostat, induce cancer cell growth inhibition, differentiation, and apoptosis in vitro and suppress tumor progression in mice (11). HDAC inhibitors have therefore emerged as promising anticancer agents, with more than a dozen HDAC inhibitors (such as vorinostat and panobinostat) currently being evaluated in the clinic and/or clinical trials (11).

In this study, we investigated the therapeutic efficacy of combination therapy with the BET bromodomain inhibitor JQ1 and the HDAC inhibitor panobinostat against neuroblastoma. JQ1 and panobinostat synergistically reduced the transcription of the LIN28B oncogene, leading to synergistic posttranscriptional reduction in N-Myc oncoprotein expression, and exerted synergistic anticancer effects in vitro and in vivo.

Human SK-N-BE(2), SK-N-AS, and Kelly neuroblastoma (from European Collection of Cell Cultures in 2010) and WI38 fetal lung fibroblast (from ATCC in 2010) cells were cultured in DMEM, RPMI1640, and minimum essential medium (MEM), respectively. All cell lines were authenticated by small tandem repeat profiling conducted at Garvan Institute of Medical Research (Darlinghurst, New South Wales, Australia) or Cellbank Australia in 2015.

Cells were transfected with 20 nmol/L siRNAs using Lipofectamine 2000 reagent (Invitrogen) or transfected with 50 ng plasmid using Lipofectamine 2000 reagent, 100 μL opti-MEM and 900 μL serum-free medium as described previously (12, 13). Validated control siRNA and siRNAs specifically targeting N-Myc, BRD3, BRD4, BRD2, or LIN28B were purchased from Qiagen.

RNA was extracted from cells using PureLink RNA Mini Kit (Invitrogen). After cDNA synthesis, real-time reverse transcriptase (RT)-PCR was performed as described previously (12, 13). Primer sequences used for PCR amplifications were as following: 5′-CGACCACAAGGCCCTCAGTA-3′ (forward) and 5′-CAGCCTTGGTGTTGGAGGAG-3′ (reverse) for N-Myc; 5′-GGAGCCCCTGTTTAGGAAGT-3′ (forward) and 5′-GCACTTCTTTGGCTGAGGAG-3′ (reverse) for LIN28B; 5′- AGCCGTAGCAGTGAGGAGAG-3′ (forward) and 5′-GTTTGTTTACTGGGGCCTGA-3′ (reverse) for BRD3; 5′-ATGGCAGAAGCTCTGGAAAA-3′ (forward) and 5′-TGTGTTTGGTACCGTGGAAA-3′ (reverse) for BRD4; 5′-GCACTTGGCCTGCATGACTA-3′ (forward) and 5′-TCTCCATCTTCCGCTTGACAGT-3′ (reverse) for BRD2; 5′-ATGTGTGTGGAGAGCGTCAA-3′ (forward) and 5′-ACAGTTCCACAAAGGCATCC-3′ (reverse) for BCL2; and 5′-CACCATGTACCCTGGCATT-3′ and 5′-ACGGAGTACTTGCGCTCAG-3′ for β-actin.

Cells were lysed, protein extracted, and separated by gel electrophoresis. After Western transfer, membranes were probed with an anti-N-Myc (1:1000, SC-53993, Santa Cruz Biotechnology), anti-BRD3 (1:1,000, catalog no. 61489, Active Motif), anti-BRD4 (1:1,000, catalog no. 39909, Active Motif), anti-LIN28B (1:1,000, 4196S, Cell Signaling Technology), or anti-actin (1:10,000, A1978, Sigma) antibody.

Cells were incubated with Alamar blue and read on a microplate reader at 570/595 nm. Results were calculated according to the optical density absorbance units, and expressed as percentage change in the number of cells relative to controls (12, 13).

Cells were harvested, washed, and then stained with FITC-conjugated Annexin V and 7-AAD (BD Biosciences). Detection of the cells positively stained with Annexin V and/or 7-AAD were performed using a FACSCanto Flow Cytometer (BD Biosciences) and analyzed with FlowJo software (TreeStar Inc).

Chromatin immunoprecipitation (ChIP) assays were performed with an anti-BRD3, anti-BRD4, or control antibody, and PCR carried out with primers targeting a negative control or the intron 1 region of the LIN28B gene promoter (14).

A LIN28B gene promoter construct was kindly provided by Dr. Joshua T. Mendell (University of Texas Southwestern Medical Center) (14). Kelly neuroblastoma cells were transfected with control siRNA, BRD3 siRNA-1, or BRD4 siRNA-2, followed by cotransfection with Cypridina TK control construct plus the pGL3 construct expressing empty vector or the LIN28B gene promoter (−1414 bp to +1403 bp) for 48 hours. Luciferase activity was measured and expressed as percentage change relative to control siRNA–transfected samples as described previously (13).

SK-N-BE(2) cells were treated with control solvent, JQ1 (1 μmol/L; prepared in house by the Bradner's Laboratory, Dana-Farber Cancer Institute, Boston, MA), panobinostat (10 nmol/L) (Selleckchem), or JQ1 plus panobinostat. RNA was harvested 6 hours after treatment and subjected to Affymetrix microarray analysis of differential gene expression. Gene expression was normalized with global normalization, the LimmaGP module (Gene Pattern, Broad Institute, Cambridge, MA) was used to identify genes differentially expressed between samples, and LimmaGP generated .rnk files were used for the gene set enrichment analysis (GSEA). An enrichment score was calculated for each gene set, and the enrichment plots were generated for each gene set. Gene expression data are available from Gene Expression Omnibus Expression (GEO; accession number GSE68690).

Animal experiments were approved by the Animal Care and Ethics Committee of the University of New South Wales (New South Wales, Australia). Female Balb/c mice, aged 5 to 6 weeks, were injected subcutaneously with 5 × 10⁶ SK-N-BE(2) cells into the flank. When the engrafted tumors reached 0.05 g, the mice were divided into 4 groups and treated via intraperitoneal injections with control solvent, JQ1 (50 mg/kg daily), panobinostat (7.5 mg/kg once every 3 days), or JQ1 plus panobinostat for 14 days. While we were not blinded to mouse group allocation, every effort was made to generate reliable data. Mice were euthanized at the end of the treatments. Tumor tissues were collected, formalin-fixed, and paraffin-embedded.

Mouse tissues were deparaffinized, rehydrated, antigen retrieved, and incubated with rabbit anti-N-Myc (1:1,000) antibody overnight. The tissue sections were then incubated with biotinylated antirabbit (1:500) antibody, streptavidin peroxidase, and diaminobenzidine. Slides were scanned, images taken, and positively stained cells quantified using ImageJ software (NIH, Bethesda, MD; ref, 12).

All experiments were replicated at least 3 times for statistical analysis. Differences were tested for significance using ANOVA among groups or unpaired t test for two groups. All statistical tests were two-sided. Synergy or additivity was calculated by combination index (CI) method for combinations of multiple doses of drugs (15), or by the fractional product (R) method for combinations of a single dose of two drugs (16, 17). Synergism is defined as a more than additive effect (CI < 1 or R < 1).

The BET bromodomain protein BRD4 has been shown to modulate MYC and MYCN gene transcription (3, 4, 7, 8, 18). We examined whether BRD3 also regulated MYCN gene expression, and whether BRD3 and BRD4 cooperatively modulated MYCN gene expression. RT-PCR and immunoblot analyses of N-Myc mRNA and protein expression were performed in MYCN-amplified SK-N-BE(2) and Kelly neuroblastoma cells after transfection with scrambled control siRNA, BRD3 siRNA (BRD3 siRNA-1 or BRD3 siRNA-2), BRD4 siRNA (BRD4 siRNA-1 or BRD4 siRNA-2), or a combination of BRD3 siRNA and BRD4 siRNA. As shown in Fig. 1A and B and Supplementary Fig. S1, both BRD4 siRNA-1 and BRD4 siRNA-2 significantly reduced BRD4 and N-Myc mRNA and protein expression, and both BRD3 siRNA-1 and BRD3 siRNA-2 significantly reduced BRD3 and N-Myc mRNA and protein expression in SK-N-BE(2) and Kelly cells. While simultaneously knocking down BRD3 and BRD4 mRNA and protein expression, cotransfection with BRD3 siRNA and BRD4 siRNA did not induce greater effects on N-Myc mRNA or protein reduction, compared with BRD3 siRNA or BRD4 siRNA alone, in SK-N-BE(2) and Kelly cells (Fig. 1A and B and Supplementary Fig. S1). In comparison, knocking down BRD2 gene expression with BRD2 siRNA showed no effect on N-Myc mRNA and protein expression in SK-N-BE(2) and Kelly cells (Supplementary Fig. S2), suggesting that BRD2 does not regulate N-Myc expression.

Next, we examined whether BRD3 and BRD4 cooperatively induce neuroblastoma cell proliferation. SK-N-BE(2) and Kelly cells were transfected with control siRNA, BRD3 siRNA-1 and/or BRD4 siRNA-1, BRD3 siRNA-2, and/or BRD4 siRNA-2, followed by Alamar blue assays. As shown in Fig. 1C, knocking down BRD3 or BRD4 expression significantly reduced the numbers of SK-N-BE(2) and Kelly cells. However simultaneous knocking down BRD3 and BRD4, compared with knocking down BRD3 or BRD4 alone, did not cause any further reduction in the numbers of SK-N-BE(2) and Kelly cells. Therefore, it can be concluded that BRD3 and BRD4 induce N-Myc gene expression and neuroblastoma cell proliferation, but do not function synergistically.

As BET bromodomain inhibitors reduce the expression of oncogenes (3–8), and HDAC inhibitors activate the expression of tumor suppressor genes (11), we investigated whether the BET bromodomain inhibitor JQ1 and the HDAC inhibitor panobinostat exerted synergistic anticancer effects. Treatment with JQ1 in combination with panobinostat for 72 hours significantly decreased the numbers of SK-N-BE(2), Kelly, CHP134, and LAN1 cells, compared with treatment with JQ1 or panobinostat alone (Fig. 2A and Supplementary Fig. S3). The combination index (CI) method showed that JQ1 and panobinostat exerted synergistic anticancer effects (Supplementary Table S1 and Supplementary Fig. S3). Consistently, treatment with various concentrations of JQ1 in combination with 10 nmol/L panobinostat synergistically reduced the numbers of SK-N-BE(2) and Kelly cells (Fig. 2B).

To determine whether JQ1 and panobinostat synergistically induce apoptosis, SK-N-BE(2) and Kelly cells were treated with control solvent, 1 μmol/L JQ1, 10 nmol/L panobinostat, or the combination of JQ1 and panobinostat for 48 hours, followed by staining with Annexin V and 7-AAD and flow cytometry analysis. While treatment with JQ1 or panobinostat alone slightly increased the percentage of cells positively stained with Annexin V alone or Annexin V plus 7-AAD, combination therapy with JQ1 and panobinostat synergistically increased the percentage of cells positively stained with Annexin V alone or Annexin V plus 7-AAD to 70% for SK-N-BE(2) cells and 61% for Kelly cells, respectively (Fig. 2C and D), with combination R values of 0.15 and 0.41, respectively (fractional product method). Importantly, treatment of the normal embryonic fibroblast WI38 cells with JQ1 and panobinostat at different concentrations did not induce apoptosis (Fig. 2C and D and Supplementary Fig. S4). The data confirm that JQ1 and panobinostat synergistically induce apoptosis in neuroblastoma but not normal cells.

As BET bromodomain inhibitors and HDAC inhibitors exert anticancer effects by modulating gene expression, genome-wide differential gene expression studies were performed with Affymetrix microarray in SK-N-BE(2) cells 6 hours after treatment with control solvent, 1 μmol/L JQ1, 10 nmol/L panobinostat, or the combination of JQ1 and panobinostat. Using 2-fold change as the cutoff, analysis of the microarray data with GenePattern software showed that JQ1 upregulated the expression of 14.5% (51/352) genes activated by panobinostat, and panobinostat upregulated the expression of 29.8% (51/171) genes activated by JQ1 (Fig. 3A). Similarly, JQ1 downregulated the expression of 27.5% (192/511) genes suppressed by panobinostat, and panobinostat downregulated the expression of 54.9% (192/350) genes suppressed by JQ1 (Fig. 3A). The data suggest that JQ1 and panobinostat commonly activate, and more considerably commonly repress, target gene expression in neuroblastoma cells.

Importantly, the microarray data showed that the JQ1 and panobinostat combination synergistically modulated target gene expression (Fig. 3A). For example, using the fold change cutoff of 4-fold, LimmaGP analysis showed that JQ1 alone and panobinostat alone regulated the expression of 10 and 20 target genes, respectively, and JQ1 and panobinostat combination modulated the expression of 109 genes (Fig. 3A). In addition, GSEA analysis revealed significant enrichment of MYC/MYCN amplification and HDAC target gene sets as targets of JQ1 and panobinostat combination treatment (Fig. 3B; Supplementary Table S2).

Among the genes most significantly downregulated by the JQ1 and panobinostat combination were the LIN28B and BCL2 oncogenes (Fig. 3C; Supplementary Table S3). RT-PCR analysis confirmed that JQ1 and panobinostat synergistically reduced LIN28B and BCL2 mRNA expression in SK-N-BE(2) cells 6 hours after treatment (R < 1.0 in both cases, fractional product method; Fig. 3D). Taken together, the data suggest that JQ1 and panobinostat synergistically modulate target gene expression.

LIN28B has recently been shown to upregulate N-Myc mRNA translation into N-Myc protein by suppressing the miRNA let-7 (19). As our microarray differential gene expression experiment showed that JQ1 and panobinostat cooperatively reduced LIN28B gene expression (Fig. 3C), we examined whether JQ1 and panobinostat synergistically reduced LIN28B and N-Myc expression.

SK-N-BE(2) and Kelly cells were treated with control solvent, 1 μmol/L JQ1, 10 nmol/L panobinostat, or the combination of JQ1 and panobinostat. As shown in Fig. 4A, while JQ1 or panobinostat alone reduced LIN28B mRNA expression, the combination synergistically reduced LIN28B mRNA expression with R values of 0.22 and 0.57 (fractional product method) in SK-N-BE(2) cells and R values of 0.51 and 0.52 in Kelly cells at 12 and 36 hours posttreatment, respectively. Consistently, immunoblot analysis showed that JQ1 and panobinostat synergistically reduced the expression of the long as well as the short isoforms of LIN28B protein, 36 and 48 hours posttreatment (Fig. 4B). In comparison, N-Myc mRNA expression was reduced by JQ1, and the JQ1/panobinostat combination did not synergize in reducing N-Myc mRNA expression (Fig. 4A). Importantly, combination treatment with JQ1 and panobinostat synergistically and significantly reduced N-Myc protein expression in both of the cell lines at 36 and 48 hours posttreatment (Fig. 4B).

GSK3β protein phosphorylation and Aurora A protein overexpression are well known to promote and block N-Myc protein degradation, respectively (2). We therefore examined whether JQ1 and panobinostat cooperatively increased GSK3β protein phosphorylation and reduced Aurora A protein expression. As shown in Supplementary Fig. S5A, combination therapy with JQ1 and panobinostat synergistically reduced N-Myc protein expression but did not show consistent effects on GSK3β protein phosphorylation and Aurora A protein expression. We then transfected SK-N-BE(2) and Kelly cells with control siRNA, LIN28B siRNA-1, or LIN28B siRNA-2. RT-PCR and immunoblot analyses showed that knocking down LIN28B did not affect N-Myc mRNA expression (Fig. 4C), but significantly downregulated N-Myc protein expression (Fig. 4D). In addition, immunoblot analysis showed that transfection with LIN28B siRNAs did not have an effect on Aurora A protein expression in SK-N-BE(2) and Kelly cells (Supplementary Fig. S5B).

We next examined whether JQ1 and panobinostat cooperatively regulated LIN28B and N-Myc protein expression in other neuroblastoma cell lines. Immunoblot analysis confirmed that JQ1 and panobinostat synergistically reduced the expression of both LIN28B and N-Myc proteins in CHP134 and LAN1 neuroblastoma cells (Fig. 4E). Taken together, the data suggest that JQ1 and panobinostat synergistically reduce N-Myc protein expression by suppressing LIN28B gene expression.

As the BET bromodomain inhibitor JQ1 significantly downregulates LIN28B mRNA and protein expression (Fig. 4A and B), we investigated whether the BET bromodomain proteins, BRD3 and BRD4, modulate LIN28B expression. SK-N-BE(2) and Kelly neuroblastoma cells were transfected with control siRNA, BRD3 siRNA-1, BRD3 siRNA-2, BRD4 siRNA-1, or BRD4 siRNA-2. RT-PCR and immunoblot analyses showed that BRD3 siRNAs and BRD4 siRNAs reduced LIN28B mRNA and protein expression (Fig. 5A and B).

To determine whether LIN28B is a direct BRD3 and BRD4 target gene, ChIP assays were performed to assess whether BRD3 or BRD4 associate with the LIN28B gene promoter using PCR primers targeting a negative control and the intron 1 regions of the LIN28B gene (Fig. 5C), as described by Chang and colleagues (14). ChIP assays showed that immunoprecipitation with the anti-BRD3 and the anti-BRD4 antibodies significantly enriched the intron 1 region of the LIN28B gene, compared with the negative control region (Fig. 5D). Importantly, enrichment of the LIN28B gene promoter (intron 1) by the anti-BRD3 and the anti-BRD4 antibodies was dramatically reduced after the cells were treated with JQ1 (Fig. 5D).

To demonstrate activation of the LIN28B gene promoter by BRD3 or BRD4, we employed the pGL3 luciferase reporter construct containing empty vector or the LIN28B gene promoter (−1414–+1403 bp; ref. 14). Kelly cells were transfected with control siRNA, BRD3 siRNA-1, or BRD4 siRNA-2, followed by cotransfection with Cypridina TK control construct plus the pGL3-empty vector construct or the pGL3-LIN28B promoter construct. In separate experiments, Kelly cells were cotransfected with Cypridina TK control construct plus the pGL3-empty vector construct or the pGL3-LIN28B promoter construct for 24 hours, followed by treatment with control solvent, 1 μmol/L JQ1, 10 nmol/L panobinostat, or combination of 1 μmol/L JQ1 and 10 nmol/L panobinostat for another 6 hours. Luciferase assays showed that knocking down BRD3 or BRD4 expression with siRNA significantly reduced luciferase activity of the LIN28B promoter construct (Fig. 5E); that treatment with JQ1 alone, but not panobinostat alone, significantly reduced luciferase activity of the LIN28B promoter construct; and that combination therapy with JQ1 and panobinostat synergistically reduced luciferase activity of the LIN28B promoter construct (Fig. 5F). Taken together, the data suggest that BRD3 and BRD4 upregulate LIN28B gene expression by binding to its gene promoter and activating its transcription, and that JQ1 and panobinostat synergistically reduce LIN28B gene expression by reducing its promoter activity.

To evaluate the potential clinical application of JQ1 and panobinostat combination therapy, it is essential to investigate the effect of the combination therapy on neuroblastoma progression in vivo. Five-week-old nude mice were xenografted with SK-N-BE(2) cells. When tumors reached 0.05 g, the mice were treated with vehicle control, JQ1, panobinostat, or the combination of JQ1 and panobinostat for 14 days. As shown in Fig. 6A, treatment with JQ1 or panobinostat alone suppressed tumor progression by approximately 40%, and treatment with JQ1 plus panobinostat substantially and synergistically blocked tumor progression (R = 0.00043, fractional product method).

We next evaluated N-Myc protein expression in neuroblastoma tissues from the mice. IHC analysis revealed that panobinostat did not have a significant effect on N-Myc protein expression. While JQ1 reduced N-Myc protein expression, the combination therapy with JQ1 and panobinostat synergistically reduced N-Myc protein expression (R = 0.27, fractional product method; Fig. 6B and C). The data therefore demonstrate that JQ1 and panobinostat combination therapy synergistically reduces N-Myc protein expression and synergistically blocks tumor progression in vivo.

The BET bromodomain protein BRD4 has been extensively shown to modulate MYC and MYCN gene transcription in leukemia, lymphoma, myeloma, and neuroblastoma cells (3, 4, 7, 8, 18). While BRD3 has also been shown to regulate MYC gene transcription (8), it is unknown whether BRD3 modulates MYCN gene expression and whether BRD3 and BRD4 cooperatively modulate MYC and MYCN gene expression. In this study, we have found that knocking down BRD3 or BRD4 with siRNAs reduces N-Myc mRNA and protein expression in neuroblastoma cells. However, simultaneous suppression of BRD3 and BRD4 is only as effective as suppression of BRD3 or BRD4 alone in reducing N-Myc mRNA or protein expression. We propose that blocking BRD3 and BRD4 does not show synergy in modulating MYCN gene expression in neuroblastoma cells, because BRD3 and BRD4 induce MYCN gene transcription by recruiting the same transcriptional regulatory complexes to acetylated chromatin at the same sites.

Both BET bromodomain inhibitors and HDAC inhibitors are promising novel anticancer agents. Currently, several BET bromodomain inhibitors, including GSK525762, CPI-0610, OTX015, and TEN-010 are in multiple clinical trials in patients with solid tumors or hematologic malignancies. Furthermore, the recent FDA approval of panobinostat in combination with bortezomib in advanced multiple myeloma establishes the feasibility of clinical investigation of a BET bromodomain inhibitor–HDAC inhibitor combination strategy. In this study, we have confirmed that JQ1 and panobinostat synergistically reduce cell viability and induce apoptosis in neuroblastoma but not normal nonmalignant cells. Importantly, in a mouse model of neuroblastoma, we have confirmed that treatment with JQ1 or panobinostat alone suppresses neuroblastoma progression, and that combination therapy with JQ1 and panobinostat substantially and synergistically blocks neuroblastoma progression. Our study therefore suggests that when neuroblastoma patients are treated with BET bromodomain inhibitors in the future, they may be cotreated with the HDAC inhibitor panobinostat. The finding is consistent with recent reports that the BET bromodomain inhibitor RVX2135 and the HDAC inhibitor vorinostat synergistically kill Myc-induced murine lymphoma cells (20), and that JQ1 and panobinostat synergistically induce leukemia cell apoptosis (6).

BET bromodomain inhibitors exert biologic effects by dislodging the acetylated histone readers BRD3 and BRD4 from chromatin, leading to transcriptional repression of oncogenes (7, 18, 21). In comparison, HDAC inhibitors exert biologic effects by blocking the function of HDACs, leading to transcriptional activation of tumor suppressor genes (11). Surprisingly, a recent study showed that treatment with RVX2135, vorinostat, or panobinostat for 24 hours commonly upregulated the expression of a subset of genes, but did not commonly downregulate target gene expression (20). In the current study, we have found that JQ1 upregulates the expression of 14.5% genes activated by panobinostat, that panobinostat upregulates the expression of 29.8% genes activated by JQ1, that JQ1 downregulates the expression of 27.5% genes suppressed by panobinostat, and that panobinostat downregulates the expression of 54.9% genes suppressed by JQ1. The data suggest that BET bromodomain inhibitors and HDAC inhibitors commonly activate, and more considerably commonly reduce, target gene expression in neuroblastoma cells. Importantly, while the majority of genes activated or suppressed by JQ1 or panobinostat, such as BCL2, are more considerably up or downregulated by JQ1 and panobinostat combination treatment, the combination also modulates the expression of a large number of genes not regulated by JQ1 or panobinostat alone. In addition, GSEA analysis of the microarray data reveals that Myc target gene sets are among the most significantly enriched by the combination therapy. On the basis of these data, it is concluded that JQ1 and panobinostat synergistically modulate gene expression. We hypothesize that acetylation of histone lysine residues by panobinostat increases the affinity of BET bromodomain proteins to chromatin, therefore increasing transcription machinery accumulation at the promoters of target genes such as BCL2 and LIN28B and enhancing responses to the BET bromodomain inhibitor.

This study identified LIN28B as one of the genes most considerably and synergistically reduced by JQ1 and panobinostat combination therapy. In comparison, the JQ1 and panobinostat combination did not significantly synergize in reducing N-Myc mRNA expression. Importantly, immunoblot analysis showed that JQ1 and panobinostat considerably and synergistically reduced both LIN28B and N-Myc protein expression. In the literature, LIN28B has been demonstrated to reduce N-Myc mRNA translation into N-Myc protein via let-7 suppression (19). Consistent with this, our RT-PCR and immunoblot analyses also show that knocking down LIN28B reduces N-Myc protein but not N-Myc mRNA expression. We therefore conclude that JQ1 and panobinostat combination treatment synergistically reduces N-Myc protein expression due to JQ1 and panobinostat-mediated synergistic reduction in LIN28B gene transcription, leading to suppression in N-Myc mRNA translation to N-Myc protein.

BRD3 and BRD4 induce gene transcription by binding to target gene promoters or super-enhancers (4, 18, 22). In this study, we have found that knocking down BRD3 or BRD4 expression reduces LIN28B mRNA and protein expression. ChIP assays demonstrate that BRD3 and BRD4 bind to the LIN28B gene promoter, and luciferase assays show that knocking down BRD3 or BRD4 reduces LIN28B gene promoter activity, and that JQ1 and panobinostat synergistically reduce LIN28B gene promoter activity. We therefore conclude that BRD3 and BRD4 induce LIN28B gene expression by direct binding to its gene promoter, and that JQ1 and panobinostat synergistically reduce LIN28B gene expression by blocking its promoter activity.

In summary, this study has demonstrated that BRD3 and BRD4 directly activate LIN28B gene transcription, and that JQ1 and panobinostat synergistically reduce LIN28B gene and N-Myc protein expression. Importantly, JQ1 and panobinostat synergistically induce neuroblastoma cell growth inhibition and apoptosis in vitro, reduce N-Myc protein expression in tumor tissues, and block tumor progression in neuroblastoma-bearing mice. Our findings have therefore identified a novel strategy to reduce N-Myc oncoprotein expression and a novel potential therapy for aggressive neuroblastoma.

J.E. Bradner has ownership interest (including patents) in Tensha Therapeutics. No potential conflicts of interest were disclosed by the other authors.

Conception and design: J. Shahbazi, T. Liu

Development of methodology: J. Shahbazi, R.B. Lock, T. Liu

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Shahbazi, P.Y. Liu, B. Atmadibrata, G.M. Marshall, T. Liu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Shahbazi, P.Y. Liu, B. Atmadibrata, R.B. Lock, T. Liu

Writing, review, and/or revision of the manuscript: J. Shahbazi, G.M. Marshall, R.B. Lock, T. Liu

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Shahbazi, J.E. Bradner, G.M. Marshall, R.B. Lock, T. Liu

Study supervision: T. Liu

The authors thank Drs. J.T. Mendell and T.-C. Chang for providing the LIN28B gene promoter construct.

The authors were supported by National Health and Medical Research Council, Cancer Institute NSW and Cancer Council NSW. T. Liu is the recipient of an Australian Research Council Future Fellowship. R.B. Lock is the recipient of a National Health and Medical Research Council Fellowship.

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.