Interrogating the enhancer landscape of intracranial ependymomas: perspectives for precision medicine

KEYWORDS:

• Drug targets
• enhancers
• ependymoma
• epigenomics
• precision medicine

Ependymomas are chemotherapy-resistant brain tumors, which, despite intensive genomic sequencing efforts during recent years, lack effective molecular targets. To discover rationale precision-based treatments, alternative approaches to the identification of therapeutic leads against ependymoma are necessary. The integration of epigenomic data revealing functional transcriptional regulatory elements, i.e. enhancers/super-enhancers, may help to define oncogenic dependencies and to delineate transcription factor (TF) networks. In this editorial, we describe the current progress in investigating the enhancer landscape of intracranial ependymoma and pinpoint the possible directions for future mechanism-of-action-based treatments using epigenomics-based approaches.

Ependymoma comprises distinct molecularly defined groups of tumors that arise along the neuraxis, either within the supratentorial region of the brain (ST), the posterior fossa (PF) or the spinal cord (SP) and occur in both children and adults [1]. The extent of surgical resection is the most consistent prognostic marker, and both surgery and radiotherapy are the current therapeutic mainstays in treating ependymoma, while the role of chemotherapy in the management of these tumors remains unproven. Pediatric patients are mostly affected by intracranial ependymomas (WHO Grade II/III) for which treatment is challenging, which is being reflected in relatively low 5-year survival rates around 69% [2].

Since the prognostic ability of histopathological grading criteria to risk-stratify patients is inconclusive and contentious [3], a molecular classification scheme based on genome-wide DNA methylation (Illumina 450K arrays) that outperforms the classic WHO grading has been established [4]. To date, only two out of nine molecular groups oncogenic drivers could be identified, a C11ORF95-RELA fusion gene associated with chromothripsis as a genetic hallmark in ST-EPN-RELA [5] and fusions involving the HIPPO regulator YAP1 as a defining feature in ST-EPN-YAP1 [6]. However, for other molecular groups such as for the most common high-risk pediatric ependymoma, PF-EPN-A, there is no clear evidence of somatic genomic alterations determined to date. In order to better understand the underlying mechanisms of these aggressive subtypes, and to identify potential druggable targets, alternative strategies are urgently needed. A perspective area to achieve this goal is epigenomics focusing on understanding the functionality of chromatin organization and the elements controlling the connection between genome and transcriptome.

The importance of the epigenome for cellular maintenance has been strongly underlined by the discovery that binding of TFs is not restricted to promoters but also includes cis-regulatory elements, designated enhancers, which are active intronic or intergenic regions marked by histone acetylation (H3K27ac) typically in distances larger than ±2500 basepairs from transcription start sites [7]. Enhancers play a critical role for the regulation of cell type-specific gene expression [8]. Large clusters of enhancers, termed super-enhancers, control cellular key pathways as well as complex TF networks thought to be responsible for lineage determination and may also influence oncogenes [9]. In the pediatric brain tumor medulloblastoma, the impact of regulatory genomic regions was highlighted through discovery of a process called enhancer hijacking which describes the recruitment of enhancers leading to activation of oncogenes [10]. This oncogenic mechanism was more recently also identified in neuroblastoma, another embryonal tumor in children [11]. In addition, active medulloblastoma enhancers also reveal subgroup-specific cellular origins [12]. Since enhancers were also found to regulate oncogenic dependencies in other brain tumors, such as glioblastoma [13], we reasoned that leveraging the active chromatin landscapes in ependymomas may provide essential knowledge about super-enhancer-associated genes on which tumor cells depend.

Our study focused on the global exploration of enhancers in intracranial ependymoma in order to reveal putative oncogenes, molecular targets, or druggable pathways [14]. Pan-ependymoma as well as molecular group-specific enhancers were identified applying H3K27ac ChIP sequencing to two independent cohorts comprising 42 primary ependymomas representing all six intracranial molecular groups as defined by DNA methylation profiling (Table 1).

Table 1. EPN molecular subgroups selected for investigation of enhancers.

Subgroup	Tumor location	Genetic markers	Age	Risk
ST-EPN-RELA	Supratentorial	Chromothripsis; fusion gene RELA-C11Orf95	Child/adult	Poor
ST-EPN-YAP1		Fusion of YAP1 gene with MAMLD1 or FAM118B	Child	Normal
ST-EPN-SE		Stable	Adult	Normal
PF-EPN-A	Posterior fossa	Stable	Child	Poor
PF-EPN-B		Chromosomal instability	Child/adult	Normal
PF-EPN-SE		Stable	Adult	Normal

Unsupervised clustering of the most variable active enhancer loci revealed strong differences between ependymoma and other tissue types included in the Roadmap Epigenomics and ENCODE databases with acquisition and loss of hundreds of regulatory elements demonstrating the uniqueness of these ependymoma enhancer profiles. In a discovery phase, molecular group-specific enhancers were identified and integrated with data from RNA sequencing to identify genes within surrounding topological associated domains that are most likely regulated by the respective enhancers. Notably, super-enhancer-associated genes represent a subset of highly expressed genes, and thus potential targets for therapy, but not necessarily genes with the highest expression that would be prioritized using RNA-seq profiling alone. To determine whether super-enhancers reveal pathways and dependency genes for ependymoma cells, and which could be actionable by targeted therapy, top-ranking ependymoma super-enhancer genes were validated in a series of RNA interference studies. This approach demonstrated that the majority of ependymoma super-enhancer-associated genes is required for cellular proliferation. Furthermore, super-enhancers are maintained by a network of specific TFs across ependymoma and within specific groups. Taking into account both transcriptional data and enrichment of TF-binding motifs within enhancer and super-enhancer regions allowed us to construct these regulatory TF networks. The loss-of-function control experiments demonstrated that distinct TFs are also required for cell proliferation, further increasing the set of putative actionable targets.

To translate findings from this study into novel therapeutic leads, targets were prioritized for which small molecules are available by integrating tumor-specific super-enhancer-regulated genes with drug–gene interaction databases [15]. Indeed, pharmacologic blockage with mibefradil of the calcium channel encoded by CACNA1H, for which a proximal super-enhancer was observed in ST-EPN-RELA tumors only, specifically reduced proliferation in an ST-EPN-RELA cell line. As an additional proof-of-concept example, inhibition of FGFR1, found to be associated with a common ependymoma-specific regulatory element, using the small-molecule AZD4547 in preclinical in vitro and in vivo studies could successfully validate this target in both ST-EPN-RELA and PF-EPN-PFA cells, suggesting that chromatin landscapes may also inform therapeutic paradigms applicable to more than one molecular group.

Our inspection of active chromatin landscapes within ependymomas identified tumor- and molecular group-specific super-enhancer-driven genes and identified potential leads for further preclinical testing. We found that knowledge of crucial regulatory elements can be used to dissect the molecular differences between histologically similar tumors and may inform precision therapies against an entity that lacks known genetic drivers.

Declaration of interest

K Okonechnikov, KW Pajtler, M Kool and SM Pfister were supported by Hopp-Children’s Cancer Center at the NCT Heidelberg (KiTZ), Heidelberg, Germany; Division of Pediatric Neurooncology, German Cancer Research Center (DKFZ) and German Cancer Consortium (DKTK), Heidelberg, Germany. SC Mack was supported by the Department of Pediatrics, Baylor College of Medicine, Houston, Texas, USA The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed. Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Additional information

Funding

This article was funded by Hopp-Children’s Cancer Center at the NCT Heidelberg (KiTZ), German Cancer Research Center (DKFZ).