Emerging Insights into the Ependymoma Epigenome

Abstract

Ependymoma is the third most common pediatric brain tumor, yet because of the paucity of effective therapeutic interventions, 45% of patients remain incurable. Recent transcriptional and copy number profiling of the disease has identified few driver genes and in fact points to a balanced genomic profile. Candidate gene approaches looking at hypermethylated promoters and genome‐wide epigenetic arrays suggest that DNA methylation may be critical to ependymoma pathogenesis. This review attempts to highlight existing and emerging evidence implicating the ependymoma epigenome as a key player and that epigenetic modifiers may offer new targeted therapeutic avenues for patients.

Transcriptional and Copy Number Profiling of Pediatric Ependymoma Identifies Limited Driver Events

Ependymoma is a central nervous system (CNS) tumor, occurring in both children and adults, and is incurable in nearly 45% of patients 14, 18. It can arise almost anywhere along the neuroaxis including the supratentorial region (comprising the cerebral hemispheres), the posterior fossa (encompassing the cerebellum and brain stem) and the spinal cord 15. Ninety percent of childhood ependymomas present intracranially, with about two‐thirds of cases arising within the posterior fossa 11. To date, surgery and adjuvant radiotherapy are the mainstays of ependymoma treatment, while chemotherapy for ependymoma has demonstrated little to no clear overall survival benefit 11. Transcriptional profiling studies have shown that, despite histological similarity, ependymomas from different regions of the CNS exhibit distinct gene expression signatures, and harbor numerous distinct subgroups 9, 34, 36, 37. Genomic characterization of these subgroups has led to the identification of molecular drivers of ependymoma such as EPHB2 amplification and INK4A deletion in supratentorial ependymoma 9. Copy number alterations have also been beneficial in the prognostic stratification of ependymoma, such as the risk classification scheme proposed by Korshonuv et al. 14, and further validation of gain of 1q25 as an indicator of poor clinical outcome 12. However, despite extensive characterization of ependymoma at a copy number level, and in contrast to other CNS neoplasms, few bona fide oncogenes and tumor suppressor genes have been identified in the form of recurrent focal genomic amplifications or deletions 16, 22. In the case of posterior fossa ependymoma, the most common location of occurrence in children, up to 50% of cases exhibit a balanced genomic profile 4, 6, 9, 14, 17. Furthermore, this rarity of copy number alterations is associated with aggressive forms of the disease, recently described as group A and group 1 posterior fossa ependymoma 36, 37. It remains uncertain whether somatic single nucleotide variants (SNVs), structural rearrangements, and/or epigenetic alterations contribute to ependymoma formation, particularly in the case of tumors, which do not appear to be copy number driven. The review to follow summarizes the current state of knowledge regarding the epigenetic basis of ependymoma and highlights areas where further extensive investigation is needed.

Candidate Gene Approaches Identify Hypermethylated Gene Promoters in the Pathogenesis of Ependymoma

Aberrant promoter methylation of CpG islands is a well‐recognized feature seen in numerous cancers 7. As such, early studies have focused on those candidate gene promoters in ependymoma, which have also been reported to be hypermethylated in other neoplasms. RASSF1A has been shown to be the most frequently hypermethylated tumor suppressor gene, reported in up to 100% of ependymomas and occurring in all clinical and pathological subtypes 8, 29. HIC1 has also been reported to be commonly methylated in up to 83% of ependymomas, with a higher incidence in intracranial tumors 35. Furthermore, the CDKN2A/INK4a locus, which is focally and recurrently deleted in supratentorial ependymoma 9, 14, has been shown to be hypermethylated in 21% of cases, followed by CDKN2B and p14ARF in 32% and 33% of tumors, respectively 30. To a lesser extent, putative tumor suppressor genes found to be hypermethylated in ependymoma include BLU, GSTP1, DAPK, FHIT, MGMT, MCJ, RARB, TIMP3, THBS1, TP73 and the TRAIL gene family CASP8, TFRSF10C, TFRSF10D 1, 2, 13, 19. Despite the frequency of methylation of these potential tumor suppressor genes in ependymoma, their role and significance in tumor formation remains unclear, requiring validation in independent cohorts and functional investigation in appropriate ependymoma models. However, this raises the possibility that other genes and pathways may be epigenetically altered in the pathogenesis of ependymoma, and suggests that epigenome‐wide examinations of DNA methylation in ependymoma might be warranted.

Genome‐Wide Epigenomic Profiling Identifies Pathways and Mechanisms Targeted by Aberrant DNA Methylation

In the last 5 years, the expansion of microarray and next‐generation sequencing technologies has allowed for genome‐wide investigations of DNA methylation and histone modifications at unprecedented resolution and throughput. Using the Illumina Golden Gate Methylation Cancer Panel 1 platform (1505 CpG sites), Rogers et al 29 profiled a series of 73 primary and 25 recurrent ependymomas. They reported that the DNA methylation profiles of ependymoma are distinguished largely according to their location in the CNS, supporting the notion that ependymomas arising from different anatomic compartments are molecularly distinct 9, 34, 36, 37. Furthermore, they demonstrated that supratentorial and spinal ependymomas together exhibited a larger number of hypermethylated and downregulated genes in comparison to posterior fossa tumors. These changes in DNA methylation were shown to be associated with alterations in gene expression of de novo and maintenance of DNA methyltransferases DNMT1, DNMT3A and DNMT3B. Interestingly, genes involved in immune cell response (NOD2, IRF7, IRAK3, OSM, PI3), cell growth and death (MAPK10, TP73), and the JNK pathway were found to be hypermethylated. Understanding the contribution of epigenetic alterations in these pathways may reveal mechanisms of ependymoma tumorigenesis, and potential actionable targets for therapeutic intervention.

In contrast to hypermethylation of CpG island promoters in cancer, global hypomethylation is a trend observed in numerous tumor types and is associated with cancer progression. A global decrease in methylation has been observed predominantly at repetitive elements such as LINEs, SINEs and LTRs, which are important for maintaining genomic stability 3. To elucidate the contribution of DNA methylation alterations at repetitive sequences in ependymoma, Xie et al 38 developed a novel genome‐wide approach to generate methylation profiles for thousands of Alu elements (the most abundant class of repetitive elements) and their flanking sequences. They demonstrated that while the majority of Alu elements and flanking sequences remain unaltered in ependymoma genesis, a small subset of Alu flanking sequences, with low CpG density, exhibited variable methylation patterns. These sequences tended to be hypermethylated in ependymoma at regions proximal to CpG islands and hypomethylated in intergenic regions. Importantly, several of these patterns were shown to be associated with aggressive primary ependymomas and tumor relapse. The impact that these epigenetic alterations have on genomic stability, and the pathways and biological processes that they impinge on, remain to be elucidated. These studies do however point to alterations in the epigenome that may play important roles in the pathogenesis of ependymoma and therefore warrant further study.

Potential Applications of Epigenetic Modifiers for Ependymoma Treatment

Characterizing the epigenome of ependymoma may hold therapeutic promise, as these marks, such as CpG methylation and histone modification, are generally reversible by pharmacologic inhibition. Importantly, inhibitors of DNA methylation (decitabine) and histone deacetylation (vorinostat) are approved by Food and Drug Administration and have shown efficacy in hematological malignancies 23, 32. This would make them rapidly translatable if effective in appropriate ependymoma pre‐clinical models. In a recent report, Milde et al 20 established a model reminiscent of high‐risk molecular group C supratentorial ependymoma by in vivo transplantation of primary tumor cells in NOD/SCID immunodeficient mice. Like primary ependymomas, which are highly chemo‐resistant, cell cultures from this model were refractory to temozolomide, vincristine and cisplatin. However, these cells were sensitive to a panel of histone deacetylase inhibitors (HDACi) including panobinostat, entinostat, valproic acid and vorinostat. Importantly, cell cultures treated with vorinostat at therapeutically achievable doses demonstrated decreased proliferation, cell cycle arrest, decreased self‐renewal capacity and increased neuronal differentiation. These findings were also supported by Rahman et al 27, demonstrating that the ependymoma cell line nEPN2 underwent apoptosis in response to treatment with the HDACi trichostatin‐A. Given the rapid development of novel pharmacologic inhibitors of epigenetic marks, it raises the intriguing question as to whether these, or at least some, epigenetic modifications are central to ependymoma pathogenesis and whether they may represent novel avenues for therapeutic intervention.

Future Steps Toward Characterizing the Ependymoma Epigenome

Although genomic and transcriptomic profiling efforts have identified distinct molecular subtypes of ependymoma and revealed potential drivers of the disease, the vast majority of ependymoma tumors are characterized by either gross chromosomal alterations, precluding the identification of driver events, or are characterized by very few genomic abnormalities, observed in some of the most aggressive tumors 9, 37. It remains to be seen whether recurrent somatic SNVs or structural rearrangements contribute to the pathogenesis of ependymoma, as reported in several other adult and pediatric CNS neoplasms 5, 10, 21, 24, 25, 26, 28, 31. Indeed, somatic SNVs identified in CNS cancers have been shown to potentially alter the entire epigenetic landscape during tumor formation, such as the IDH1 mutations leading to a CpG hypermethylator phenotype in glioblastoma multiforme (GBM) 24 (cross‐refer to Vennetti and Thompson's review), histone H3.3 and ATRX mutations in pediatric GBM 31 (cross‐refer to Jabado's review), and a collection of chromatin‐associated genes shown to be mutated in medulloblastoma such as MLL2, MLL3 and KDM6A 5, 10, 25, 26, 28 (cross‐refer to Jones' review). DNA methylation profiling efforts have also been important in the molecular stratification of CNS tumors 33, and have also shown promise in distinguishing the principle molecular subgroups of ependymoma and identifying pathways targeted by DNA hypermethylation. As a future step, expanding DNA methylation profiling to platforms with higher CpG coverage surrounding gene promoters, including the gene body, might be warranted, and could reveal novel targets, pathways and mechanisms of epigenetic alteration. Also, given the contributions of aberrant methylation near repeat elements in the ependymoma epigenome, more global investigations beyond gene promoters might be necessary, and could be readily examined with whole‐genome bisulfite sequencing. Ultimately, the elucidation of epigenetic changes in ependymoma pathogenesis will not only improve our understanding of the biology of this disease, but can also reveal actionable pathways that could be rapidly translated into the clinic. This is important especially in the case of ependymoma, a highly recurrent and chemo‐resistant brain tumor, for which the mechanisms and drivers of the disease remain largely unknown.