Combining inhibitors of Brd4 and cyclin-dependent kinase can decrease tumor growth in neuroblastoma with MYCN amplification

Abstract

Introduction:

High-risk neuroblastoma is a deadly disease; poor prognosticators are MYCN-amplification and TERT-overexpression. We hypothesized that Gene Set Enrichment Analysis (GSEA) could identify pathways associated with MYCN-amplification and that inhibition of these pathways could decrease tumor growth.

Methods:

We analyzed the Neuroblastoma-Kocak dataset (GSE45547, n=649) and identified pathways associated with MYCN-amplification. Inhibitors were selected from upregulated gene sets for in vitro cytotoxicity testing using ST16-patient-derived primary neuroblastoma cells and in vivo testing using orthotopic ST16-patient-derived xenografts (PDX) in mice. Tumor volume was measured with ultrasound and tumor sections examined after H&E staining.

Results:

GSEA identified significantly overexpressed gene sets in MYCN-amplified tumors including MYC targets, cell cycle mitotic genes, TERT associated genes, loss of RB1 gene sets, and E2Fs targets. Several genes were potential Bromodomain-containing protein 4 (Brd4) targets, making Brd4 inhibitors - JQ1, AZD5153 - and cyclin-dependent kinase (Brd4’s binding partner) inhibitors – dinaciclib - potential therapeutic agents. JQ1 and dinaciclib were synergistic in inducing cytotoxicity in vitro. Dinaciclib-AZD5153 in vivo decreased tumor size compared to control, and increased tumor lymphocyte infiltration and necrosis on histology.

Conclusions:

GSEA is a powerful approach to identify upregulated genes and potential therapeutic targets. Dinaciclib-AZD5153 combination therapy can be effective against MYCN-amplified and TERT-overexpressing neuroblastoma tumors.

1. INTRODUCTION

Neuroblastoma is the most common solid extracranial tumor diagnosed in childhood, accounting for approximately 8% of childhood cancer. The cases are risk-stratified according to age at time of diagnosis, tumor stage based on imaging, and pathologic markers. High risk neuroblastoma accounts for 40% of all cases and prognosis for high risk neuroblastoma patients remains poor, with a 5-year event-free survival rate of only 40% [1].

MYCN amplification has been found in approximately 20% of neuroblastoma cases and is a significant adverse prognostic factor. A member of the MYC oncogene family, MYCN protein is necessary for embryonal development, drives pathways related to pluripotency and block underlying differentiation. MYCN amplification leads to genetic instability, driving neuroblastoma growth and progression [2].

Telomerase reverse transcriptase (TERT) is involved in telomere maintenance, and high TERT activity is associated with poor prognosis [3]. TERT activation can be due to rearrangement placing a super-enhancer sequence upstream of the TERT promoter and epigenetic changes causing chromatin remodeling. TERT overexpression has been thought to be mutually exclusive to MYCN overexpression but recent literature suggests that they can be co-expressed and portend worse prognosis [4].

GSEA is an analytic tool for interpreting gene expression data a priori. The method focuses on groups of genes, or gene sets, that share common biological function, chromosomal location, or regulation. In comparison to traditional single-gene analysis methods, GSEA is an unbiased modality that can identify interrelated pathways [5]. Used across many different biological phenotypes, GSEA has been used in a number of cancer models to identify putative pathways associated with prognosis and treatment outcomes. In neuroblastoma specifically, it has been used to identify gene expression levels (i.e. high versus low) associated with known phenotypes, such as MYCN amplification, to identify novel pathways for targeted therapy [4].

In this study, we hypothesize that GSEA can identify new targets of key pathways associated with MYCN amplification as potential novel therapies in high risk neuroblastoma.

2. METHODS

2.1. Gene Set Enrichment Analysis (GSEA)

GSEA was performed to identify top pathways associated with MYCN overexpression utilizing the Molecular Signatures Database (MSigDB, Broad Institute, University of California, San Diego). We analyzed the Neuroblastoma-Kocak dataset (GSEA45547, https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE45547) containing 649 patients using the R2 platform (Genomics Analysis and Visualization Platform). Six patients were excluded given unknown MYCN status. Default parameters for analysis and students t-test were used; gene sets with a false discovery rate <25% were considered significant.

2.2. In vitro cytotoxicity assay

The use of human tissue was approved by the Institutional Review Board of Stanford University (protocol #46426). Primary culture of ST16 patient-derived neuroblastoma xenograft cells were exposed to single or double agent chemotherapy or serum vehicle for 72 hours. Dose of dinaciclib was varied between 0 and 50 nM; dose of JQ1 when added in combination was held constant at 0.5 μM, a dose known to be effective [6]. After the incubation period, CellTiter-AQueous MTS colorimetric assay (Promega Corporation, Madison, WI) and microplate reader was used to determine the viability of neuroblastoma cells exposed to drug.

2.3. Murine orthotopic neuroblastoma patient-derived xenograft model and treatment

Animal use was approved by Stanford University Institutional Animal Care and Use Committee (protocol #32942). Orthotopic neuroblastoma patient-derived xenograft (PDX) mouse models were established as previously described [4]. In brief, seven-week old female NCr mice (Envigo, Indianapolis, IN) was anesthetized before a transverse left flank incision was made. The exposed left adrenal gland was injected with 2μL of phosphate buffered saline containing 1 million ST16 cells. Female mice were used given their docile nature. Tumor size was assessed bi-weekly using ultrasonography. When the tumor size reached >50mm³, mice received vehicle control, dinaciclib intraperitoneal (40mg/kg/day), AZD5153 oral gavage (5mg/kg/day) or combination dinaciclib-AZD5153. Tumor size analysis was stopped at day 14 to determine drug effect.

2.4. Tumor histopathology

At 14 days or when tumor size >1000mm³, the tumor was harvested, fixed in 10% neutral buffered formalin, paraffin-embedded, sectioned, and stained with hematoxylin and eosin (H&E).

2.5. Statistics

One-way ANOVA and student’s t-tests were performed to compare tumor volumes and simple linear regression used to compare best-fit trendlines for tumor growth over 14 days using GraphPad Prism (San Diego, CA). * p<0.05; ** p<0.01; ***, ### p<0.001.

3. RESULTS

3.1. Identification of Brd4 target genes in MYCN overexpressing neuroblastoma tumors

GSEA transcriptome analysis identified gene sets that were significantly enriched in patient neuroblastoma tumors with high MYCN expression (MYCN-amplified, n=93 versus MYCN-non-amplified, n=550, Figures 1A, 1B). GSEA identified a number of significantly overexpressed gene sets including MYC targets, cell cycle mitotic genes, HOXA9 targets, TERT associated genes, ß-Catenin (CTNNB1) targets, loss of RB1 gene sets, and E2Fs targets in MYCN-amplified tumors. These gene sets are known to regulate the cell cycle, and their transcriptional targets include cyclin-dependent kinases (CDK), cyclins, and telomerase reverse transcriptase (TERT), all of which contain binding sites for E2F and MYC [7]. Brd4 is also involved in driving telomerase elongation via Cdk9 and cyclin T1 which combine to form the positive transcriptional elongation factor b (p-TEFb) [5]. Given these findings and prior work suggesting the effectiveness of Brd4 and CDK inhibition in preventing cell proliferation, we hypothesized that the combination of a CDK inhibitor (dinaciclib) and a Brd4 inhibitor (JQ1 or AZD5153) would better decrease neuroblastoma cell growth than either drug alone in tumors with increased MYCN and TERT expression [4].

Figure 1.

Enriched gene sets using Tumor Neuroblastoma-Kocak-649 dataset. 649 patients were assigned into subgroups MYCN amplified (n=93) vs. non-amplified (n=550) within the R2 genomics analysis platform. Significantly enriched gene sets with MYCN amplification were listed in A (database:geneset_broad_2019_oncogenic), B (detabase:geneset_broad_2019_curated).

3.2. In vitro testing of target drugs

To test the efficacy of single versus dual agent CDK and Brd4 inhibitors, wecompared dinaciclib or JQ1 alone and dinaciclib-JQ1 in combination on primary ST16 patient-derived neuroblastoma cells. We have previously reported a combination index (CI) of 1.23, 0.88, 0.80, 0.58 with increasing dosages of dinaciclib (1.562, 3.125, 6.25, 12.5 nM respectively) and a fixed dosage of JQ1 (0.5 μM), suggesting a synergistic effect with increasing dinaciclib dosages [4]. While dinaciclib or JQ1 alone was able to significantly inhibit cell viability at increasing drug dosages, addition of JQ1 demonstrated additional growth inhibition (Figures 2A, 2B).

Figure 2.

(A, B) In vitro cytotoxicity of uncultured ST16 patient derived xenograft cells when incubated varying concentrations of dinaciclib or JQ1 alone and in combination. * p<0.05; ** p<0.01; ***, ### p<0.001.

3.3. In vivo testing of target drugs

To understand the role of CDK and Brd4 inhibitors in vivo, the effects of dinaciclib alone, AZD5153 alone, and combination dinaciclib-AZD5153 on orthotopic xenografts were analyzed. AZD5153 was tested in mice instead of JQ1 due to its availability in oral formulation and currently being used in clinical trials [4,8]. In a cohort of treated versus control mice over a 14-day period, we found that the growth curves were significantly different from one another (p=0.03). The growth curve of the control animals was significantly steeper (faster tumor growth) compared to animals treated with dinaciclib or dinaciclib-AZD5153 (p=0.03 and 0.02, respectively), and nearly significantly steeper than AZD5153 alone (p=0.05). Tumor volumes at the start of treatment were comparable in size amongst treatment groups (p=0.48). Tumor volumes were significantly lower in animals treated with combination therapy (dinaciclib-AZD5153, n=4, 25.9 mm³±5.8 mm³) as compared to control animals (n=3, 41.0 mm³±10.5 mm³, p=0.02) (Figure 3). Tumors treated with AZD5153 alone (n=3, 30.9 mm³±6.5 mm³) and dinaciclib alone (n=3, 30.2 mm³±5.8 mm³) had a smaller tumor volume compared to control, but this was not statistically significant (p=0.10 and 0.08, respectively).

Figure 3.

Linear regression demonstrating ST16-PDX tumor growth trendlines over 14 days of treatment with vehicle control, dinaciclib, AZD5153, or combination dinaciclib-AZD. * significantly steeper growth (faster tumor growth) curve of control compared to all drug modalities (p≤0.05); † significantly smaller tumor volume of dinaciclib-AZD5153 compared to control (n=4, 25.9 mm³±5.8 mm³ versus n=3, 41.0 mm³±10.5 mm³; p=0.02).

3.4. Histologic analysis

H&E staining of all tumors treated with a combination of dinaciclib and AZD5153 showed inflammation and necrosis within the tumor. In particular, all the tumors treated with a Brd4 inhibitor had increased lymphocytic infiltration (Figures 4A, 4B).

Figure 4.

Photomicrographs of treated ST16-PDX tumor sections stained with H&E: (A) areas of lymphocytic infiltration (white arrow). (B) tumor necrosis (black arrow). White scale bars indicate 50 μM.

4. DISCUSSION

This study identifies the presence of poor prognosis TERT over-expressing and MYCN-amplified neuroblastomas and highlights potential novel therapies by targeting bromodomains, or acetyl-lysine recognition motifs, and cyclin-dependent kinases identified using GSEA. We focused on this subset of tumors given the poor outcome of neuroblastomas with these mutations; approximately 25% of patients have MYCN amplification, 20% have TERT over-expression, and many tumors have both [9]. Our GSEA identified a number of gene sets related to cell cycle regulation and the transcription of activated chromatin and telomere lengthening in MYCN-amplified tumors [10]. Given that CDK is involved with Brd4 binding thus driving downstream action, we utilized the CDK inhibitor dinaciclib alone and in combination with Brd4 inhibitors JQ1 or AZD5153 to determine the effects on a MYCN-amplified PDX. We demonstrated a synergistic effect between dinaciclib and JQ1 in vitro and showed that these drugs in combination slowed tumor growth in patient-derived xenografts with significant lymphocytic tumor infiltration and tumor necrosis.

Dinaciclib is a CDK inhibitor, stopping cell cycle progression by allowing for activation of retinoblastoma (Rb) tumor suppressor which in turn inhibits activity of the E2F transcription factor. CDK is known to bind to p-TEFb complex, a transcription complex that drives transcription via RNA polymerase II; mutations in CDK are tumorigenic. Specifically related to neuroblastoma, dinaciclib has been shown to induce cell cycle arrest in a number of cell lines in vitro and knock down CDK2 and CDK9 activity with tumor regression in vivo [11]. Our group has demonstrated dinaciclib inhibition of Brd4 recruitment using ChIP-seq and ChIP-qPCR by either directly preventing Brd4 binding to p-TEFb or inhibiting Cdk9 [4]. Our results here suggest that in ST16 (MYCN-amplified, TERT-overexpressing) cells in vitro and in vivo, dinaciclib alone can inhibit cell growth, with a near significant effect in vivo (Figure 4, p=0.078 compared to control).

AZD5153 is a small lysine-mimetic BET bromodomain inhibitor which has been shown in hematologic cancer models to cause tumor regression via bivalent binding of both BRD4 bromodomains which does not allow binding to chromatin for transcription [12]. In a leukemia model, inhibiting Brd4 deregulates MYC and E2F resulting in stability or regression; a similar strategy was identified in the use of SF1126 which is a dual inhibitor of BRD4 and PI-3K in the treatment of MYCN-amplified neuroblastoma in vitro and in vivo [12,13]. JQ1 which we used in our in vitro model in combination with dinaciclib is known to competitively and monovalently bind BRD4, thus displacing this oncoprotein from chromatin, and has also been shown to successfully halt cell cycle progression in MYCN-amplified neuroblastoma cell lines [6].

Future study includes analyzing longer-term survival of animals treated with combination therapy to determine the effect of slowed tumor kinetics on overall survival, and ultimately the clinical application of this combination therapy.

5. CONCLUSIONS

As an unbiased approach, GSEA can identify pathways that can be targeted to suppress tumor growth. Targeted inhibition of the Brd4 pathway utilizing dinaciclib (CDK inhibitor) and AZD5153 (Brd4 inhibitor) is a potential treatment for MYCN-amplified neuroblastoma.