Metabolic Reprogramming and Splicing in MYC-Driven Neuroblastoma

Introduction

Metabolic reprogramming is a hallmark of cancer which allows rapidly proliferating tumor cells to acquire nutrients to meet their bioenergetic, biosynthetic, and redox demands. One of the primary driving forces in reprograming cancer cell metabolism is the deregulated MYC family proto-oncogenes (C-MYC, MYCN, and MYCL), which are known to encode master transcriptional factors that regulate metabolic gene expression. MYC coordinates nutrient acquisition to produce ATP and key cellular building blocks that increase cell mass and promote DNA replication and cell division. The increase in total RNA and protein synthesis by overactive MYC signaling leads to dysregulation of macromolecular processing machineries including the spliceosome, and consequently pre-mRNA splicing, another hallmark of MYC-driven cancers, for the purpose of cellular stress adaptation. MYCN amplification is one of the most important biological features of high-risk neuroblastoma. Transgenic mouse and zebrafish models have demonstrated that MYCN is a neuroblastoma driver. In tumors without MYCN amplification, C-MYC is overexpressed, further indicating that neuroblastoma is a MYC-driven cancer. The metabolic dependency of neuroblastoma has been widely studied by us and others. A larger number of splicing changes have also been noticed in high-stage neuroblastomas. Splicing alterations lead to a spliceosomal vulnerability that provides a new opportunity to develop transformative therapies by disrupting aberrant pre-mRNA splicing. We and others have shown that targeting the splicing factor RBM39 by indisulam, a “molecular glue” that bridges RBM39 to E3 ubiquitin ligase DCAF15 for proteasomal degradation, achieved significant anti-tumor activity in neuroblastoma models. Disruption of spliceosome by Pladienolide B also resulted in significant anti-tumor effect in neuroblastoma models However, how the dysregulated pre-mRNA splicing machinery and metabolism are orchestrated in MYC-driven neuroblastoma has not been well elucidated. Whether metabolism modulates the anti-cancer effect of splicing inhibition remains to be answered.

Next-generation sequencing studies have revealed only a few recurrent somatic mutations in neuroblastoma at the time of diagnosis. However, copy number alterations of chromosomal segments such as 17q gain, 1p36 or 11q23 loss frequently occur in high-risk neuroblastoma. While attempts to understand the functions of individual genes in these chromosomal segments have been reported (i.e., BIRC5, PHB, PPM1D, TRIM37 in 17q; ARID1A, CAMTA1, CASZ1, CHD5, KIF1Bβ, miR-34a, RUNX3 in 1p36), the biological consequences of these genetic events in MYC-driven tumors still remain largely unknown. Gain of 17q is the most frequent genetic event in high-risk neuroblastoma and is associated with MYCN amplification. In addition, in the transgenic MYCN mouse model of neuroblastoma, the chromosomal locus syntenic to human 17q is partially amplified, indicating that chromosome 17q is needed for MYC-mediated tumorigenesis.

JMJD6 is a JmjC domain–containing nuclear protein with iron- and 2-oxoglutarate–dependent dioxygenase activity, whose coding gene is located on chromosome 17q25. While the histone arginine demethylase activity of JMJD6 that catalyzes demethylation of H4R3me1/me2 is controversial, JMJD6 is a lysyl-5-hydroxylase that catalyzes 5-hydroxylation on specific lysine residues of target proteins. JMJD6 has pleiotropic functions in normal physiology and in cancer. We previously found that JMJD6 is essential for the survival of neuroblastoma cells (including MYCN-amplified and C-MYC– overexpressed cells), which was further validated by an independent study, indicating that neuroblastoma has JMJD6 dependency. However, the exact mechanism of JMJD6 in MYC-driven cancers remains elusive. One study has shown that JMJD6 and BRD4 co-bind at antipause enhancers, regulating promoter-proximal pause release of a large subset of transcription units. By harnessing a similar mechanism, JMJD6 promotes cell survival of glioblastoma in vivo. These findings are particularly interesting because BRD4 occupies exceptionally large super-enhancers associated with genes, including C-MYC and MYCN, and the expression of those enhancers can be disrupted by BRD4 inhibitors, which have a potent antitumor effect. Here we show a new mechanism by which JMJD6 promotes tumorigenesis mediated by the MYC oncogene in that JMJD6 interacts with a subset of RNA binding proteins including RBM39 in neuroblastoma cells and regulates the alternative splicing of metabolic genes that are involved in mitochondrial metabolism. “Glutamine addiction” is one key feature of MYC-driven tumors. Glutaminase (GLS) is the enzyme responsible for conversion of glutamine to glutamate in the process of glutaminolysis to feed the tricarboxylic acid (TCA) cycle and has two splice isoforms, GAC (glutaminase C) and KGA (kidney-type glutaminase). We show that JMJD6 controls the alternative splicing of KGA and GAC, and, consequently, impacts the central carbon metabolism in neuroblastoma. Further we show that JMJD6 is correlated with the anti-cancer activity of indisulam, a “molecular glue” that degrades the splicing factor RBM39. The indisulam-mediated cancer cell killing is at least partly dependent on the glutamine-related metabolic pathway mediated by JMJD6. Our findings demonstrate a new mechanism by which JMJD6 coordinates metabolic programs and alternative pre-mRNA splicing, providing a rationale to target JMJD6 as a therapeutic target for MYC-driven cancers.

Results

The essential genes for neuroblastoma cell survival on chromosome 17q target pre-mRNA splicing and metabolism

An incomplete understanding of the biological consequences of chromosome 17q gain remains a barrier to the understanding of high-risk neuroblastoma. 1132 genes are located on 17q. We surmised that some of the 17q genes are particularly important for neuroblastoma cell survival. Analysis of the cancer dependency genes in neuroblastoma cell lines screened with the Avana sgRNA library revealed that 114 were essential to neuroblastoma (mean score <-0.4). Protein interaction network analysis followed by functional annotation revealed that proteins encoded by these 114 essential genes formed distinct but interconnected modules including RNA splicing (i.e., SRSF2, DDX5, DDX42, DHX8), mitochondrial metabolism (i.e., NDUFA8, COX11, SLC25A10, SLC35B1), protein homeostasis (i.e., UBE2O, PSMB3, PSMC5), DNA repair (i.e., BRIP1, BRCA1, RAD51C, RAD51D) and transcriptional regulation (i.e., PHF12, CBX1, SMARCE1, MED1), as well as endocytosis (i.e., CHMP6, CTLC, EPN3, HGS, SNF8, VPS25). Children aged ≥18 months with metastatic disease and patients with MYCN amplification tumors are classified as high-risk, which requires a multimodal therapy including induction chemotherapy, surgical resection of primary disease, consolidation with high-dose chemotherapy and stem cell rescue, radiotherapy, and post-consolidation treatment with cis-retinoic acid and immunotherapy. Using these 114 genes as a signature, we found that 81 of them were highly expressed in high-risk neuroblastomas, which were enriched with MYCN amplification. Correspondingly, neuroblastomas with high expression levels of this gene signature were associated with a poorer event-free (time from treatment until the cancer progresses) and overall survival (time from treatment to death) of patients in two large clinic cohorts. Interestingly, in low-risk neuroblastoma patients, high expression of the 114 essential genes was associated with poor event-free and overall survival, while no difference was observed in high-risk patients. These data demonstrate that 17q genes are involved in essential biological processes and highly expressed in high-risk neuroblastomas. Nevertheless, the 114 essential genes cannot further stratify the high-risk patients.

17q contains neuroblastoma dependency genes

a. CRISPR score for 17q genes in 10 neuroblastoma cell lines. Score <-0.4 is defined as neuroblastoma dependency genes. Data are derived from Avana sgRNA library screening.

b. STRING protein interaction network showing 17q essential genes with various biological functions.

c. Heatmap by K-means clustering analysis showing 17q essential genes are highly expressed in high-risk neuroblastomas based on RNA-seq data (SEQC dataset).

d. Kaplan-Meier survival curve showing 17q essential gene signature is correlated with worse event-free survival (SEQC dataset).

e. Kaplan-Meier survival curve showing 17q essential gene signature is correlated with worse overall survival (SEQC dataset).

JMJD6 is required for neuroblastoma growth

JMJD6 was among these 114 essential genes. To understand the role of JMJD6, we examined the genetic features of JMJD6 in neuroblastoma and other types of cancers. Among the genes encoding JmjC-domain containing proteins, JMJD6 was the only one that was frequently amplified in neuroblastoma. High JMJD6 expression was associated with poor event-free outcome, as shown by Kaplan-Meier analysis. Further analysis of JMJD6 expression in low-risk and high-risk patients showed that high levels of JMJD6 expression was associated with poor event-free and overall survival in both low-risk and high-risk patients, indicating that JMJD6 is high-risk factor regardless of disease status. To examine whether JMJD6 amplification is limited to specific tumor types, we explored genomic data from different cancers using the cBioportal program. JMJD6 was amplified across multiple types of adult cancers such as breast and liver cancer, and correlated with worse relapse-free survival. We further compared the RNA-seq expression of JMJD6 in 2337 samples across over 20 pediatric cancer subtypes and found that JMJD6 showed the highest expression levels in neuroblastoma, suggesting that JMJD6 might be particularly important in neuroblastoma. We validated this hypothesis using shRNA knockdown of JMJD6 in MYCN amplified cells (BE2C, SIMA, KELLY, IMR32) and non-MYCN amplified cells (SK-N-AS and CHLA20). The results showed that loss of JMJD6 greatly reduced the colony numbers in all tested cell lines, demonstrating that JMJD6 is essential to neuroblastoma cells regardless of MYCN amplification. Neuroblastic tumors comprise a histologic spectrum that ranges from less-differentiated neuroblastoma to well-differentiated ganglioneuroma. The extent of differentiation in the tumor cells is correlated with prognosti