Epigenetic dysregulation: a novel pathway of oncogenesis in pediatric brain tumors
- First Online:
- Cite this article as:
- Fontebasso, A.M., Gayden, T., Nikbakht, H. et al. Acta Neuropathol (2014) 128: 615. doi:10.1007/s00401-014-1325-8
- 1.5k Downloads
A remarkably large number of “epigenetic regulators” have been recently identified to be altered in cancers and a rapidly expanding body of literature points to “epigenetic addiction” (an aberrant epigenetic state to which a tumor is addicted) as a new previously unsuspected mechanism of oncogenesis. Although mutations are also found in canonical signaling pathway genes, we and others identified chromatin-associated proteins to be more commonly altered by somatic alterations than any other class of oncoprotein in several subgroups of childhood high-grade brain tumors. Furthermore, as these childhood malignancies carry fewer non-synonymous somatic mutations per case in contrast to most adult cancers, these mutations are likely drivers in these tumors. Herein, we will use as examples of this novel hallmark of oncogenesis high-grade astrocytomas, including glioblastoma, and a subgroup of embryonal tumors, embryonal tumor with multilayered rosettes (ETMR) to describe the novel molecular defects uncovered in these deadly tumors. We will further discuss evidence for their profound effects on the epigenome. The relative genetic simplicity of these tumors promises general insights into how mutations in the chromatin machinery modify downstream epigenetic signatures to drive transformation, and how to target this plastic genetic/epigenetic interface.
Current challenges in subgroups of high-grade pediatric brain tumors
High-grade pediatric and young adult astrocytomas: an epigenetic defect of the developing brain?
Defects in H3K27 at the core of midline and hindbrain tumorigenesis
Defects affecting H3K36 define a unique set of tumors with a variety of clinical and biologic parameters
A recent study identified recurrent H3.3 mutations in the vast majority of chondroblastoma and giant cell tumors of the bone, two tumors affecting soft tissues and younger patients (adolescents and young adults primarily) . H3F3B K36M mutations were identified in 68/77 of chrondroblastoma samples and H3F3A G34W mutations were identified in 48/53 giant cell tumors of bone (GCT); in addition to cases with rare variants H3F3A K36M or G34L in chondroblastoma or GCT respectively . No additional genetic alterations were identified to be associated with these H3.3 mutations that appeared to be the sole drivers of these tumors. To date, these are the only group of tumors other than pediatric HGA where a histone gene is recurrently affected in cancer. They illustrate that the residue and histone 3 isoform targeted is specific to age, tumor type and tumor location.
Mechanisms for telomere lengthening vary with age and tumor type
ATRX is a critical member of a multiprotein complex that includes DAXX and plays a role in regulating chromatin remodeling, nucleosome assembly, telomere maintenance and deposition of histone H3.3. The H3.3 chaperone HIRA loads H3.3 at active and repressed genes and at several transcription factor binding sites, while the ATRX–DAXX complex mediates H3.3 deposition in silent chromatin at telomeres, where the presence of H3.3 is correlated with the repression of telomeric RNA transcription, and near certain specific active genes . Hypomorphic germline mutations in ATRX lead to the α-thalassemia/mental retardation X-linked syndrome. Conversely, complete loss-of-function mutations have recently been identified in cancers, including pancreatic neuroendocrine tumors (PanNETs), neuroblastoma and alpha-thalassemia myelodysplasia syndrome. We and others showed ATRX to be mutated in pediatric HGA [28, 63] and adult IDH-mutant astrocytomas [32, 44] and showed alternative lengthening of telomeres (ALT) to be associated with ATRX mutations [28, 63]. These mutations are mutually exclusive with TERT promoter mutations responsible for telomere elongation that seem to specify primary GBM and oligodendroglial IDH-CIC/FUBP1 mutants (Fig. 2). Interestingly, younger patients with HGA, mainly patients with DIPG did not harbor TERT promoter mutations possibly reflecting age and the effect of the cell of origin [9, 19, 76].
DNA methylation and its role in pediatric brain tumorigenesis
In addition to histone code alterations observed in the pediatric and young adult form of HGA, studies utilizing high-density methylation arrays have revealed striking associations between histone code alterations and global DNA methylation patterns. Sturm and colleagues identify six subgroups of GBM that vary in both clinical and mutational variables associated with individual methylation subgroups; three of which are delineated by mutations shown to affect the histone code, namely H3.3 K27M, H3.3 G34R/V and IDH1 R132-mutated tumors . Additional subgroups are comprised largely of adult tumors with classical oncogenic alterations such as EGFR, PDGFRA amplification and mesenchymal profiles , suggesting that epigenomic dysregulation in the form of histone code alterations may be at the core of pediatric tumorigenesis specifically. Herein we have detailed the mutations and alterations leading to histone methylation defects in pediatric brain tumors, and have indicated that these are in addition, associated with novel DNA methylation patterns. Utilizing an adult glioma dataset, Noushmehr and colleagues  were able to subgroup tumors via DNA methylation profiling and identify a prominent G-CIMP subgroup defined by IDH1 mutation, which was later shown as detailed above, to be sufficient to create this phenotype . Expanding global DNA methylation analyses to incorporate a significant pediatric subset, H3.3 K27M, H3.3 G34R/V and IDH1 mutant tumors were shown to specifically map three epigenetic subgroups of GBM that were comprised largely of pediatric and young adult tumors . As a technology to classify pediatric brain tumors, DNA methylation profiling constitutes quite a robust method, with recent reports demonstrating this for not only gliomas, but also medulloblastoma, pilocytic astrocytoma, ependymoma and primitive neuroectodermal tumors (PNETs) as well as embryonal tumors with multilayered rosettes (ETMRs) [4, 29, 39, 41, 47, 62]. Although in medulloblastoma, DNA methylation derived subgrouping corresponds very well with subgrouping performed via gene expression microarrays , for GBM this has proven more difficult, with only supervised analyses demonstrating unique gene expression associated with the epigenetic subgroups defined by H3.3 K27M and G34R/V . Bender and colleagues demonstrate however that H3K27me3 and DNA hypomethylation areas may lie at the core of gene expression programs driven by K27M mutations . Their data also argue for caution in isogenic cell lines expressing H3.3 mutations, as these may not recapitulate H3K27me3 and DNA methylation profiles present in human tumor tissue . Recent multidimensional studies of DIPG tumors confirmed a global landscape of DNA hypomethylation seen in K27M-mutant DIPG tumors, and showed distinct subgrouping of tumors with activated Hedgehog (Hh) or N-Myc (MYCN) seen by corroborative transcriptomic and proteomic studies . Differences in DNA methylation profiles of DIPG tumors associated with MYCN activation were first shown as delineating one of three epigenetic subgroups across 28 DIPG samples analyzed by methylation arrays . MYCN alterations occurred independently of H3 K27M and ACVR1 mutations, although only a small sample size of MYCN-group tumors (n = 2) were included in methylation analysis necessitating future investigation of this MYCN subgroup . Boot-strapping assessments of DNA methylation data of DIPG and other pediatric HGA tumors strongly suggest that K27M mutations across H3.3 and H3.1 govern distinct epigenomic profiles . Subgroup-specific modeling of HGA and DIPG associated with these particular alterations presents a challenge and will undoubtedly represent a critical step forward in the study of this multifaceted group of diseases. Accurate recapitulation of the striking DNA methylation signatures we see robustly in tumors will form a driving force for progress in understanding the biology of individual subgroups of HGAs.
A recent example by our group in demonstrating the complex interplay between genomic alterations, DNA methylation and gene expression has resulted from the study of pediatric embryonal tumors (Fig. 1). While high-grade astrocytomas are more rare in children, embryonal brain tumors are very specific to the pediatric years and are rarely, if ever, seen in adults. These are aggressive high-grade malignant tumors and include medulloblastoma (neuronal high-grade neoplasms in the cerebellum), primitive neuroectodermal tumors (PNETs), atypical teratoid/rhabdoid tumors (ATRT) and a newly described variant embryonal tumor with multilayered rosettes (ETMR) [40, 45]. Recent re-classification of many diverse histological entities into ETMRs prompted a view into the molecular characteristics underlying these aggressive tumors of the early pediatric years . DNA methylation profiling revealed a very distinct global profile for a series of ETMRs when compared to a diverse set of gliomas, PNETs and other brain tumors . RNA sequencing revealed a recurrent fusion between TTYH1 and the C19MC microRNA cluster, which is primate-specific . When assessing significant genes up- or down-regulated in ETMRs specifically, a fetal-specific isoform of the de novo DNA methyltransferase DNMT3B (isoform 1b) was shown to be increased specifically in ETMRs, notably when assessed across a large dataset of a variety of tumors . This isoform is known to be uniquely expressed in early post-conceptional fetal brain and may underlie the unique cellular differentiation of ETMR tumors. Specific members of the C19MC cluster were able to upregulate the 1b isoform associated with a decrease in retinoblastoma-like 2 (RBL2), a gene which regulates the expression of DNMT3B . Taken together, these data point to a fusion between a gene and the C19MC cluster driving expression of the microRNA. This leads to subsequent downstream upregulation of a fetal isoform of a DNA methyltransferase, influencing the global epigenomic signature of ETMR tumors. Herein once more, data points to the epigenome as a potential driver for pediatric brain tumorigenesis. Defining the function of these unique methylation patterns across tumor subtypes will be important to understand how alterations in both the histone code and DNA methylation alter the genome in such a way as to directly mediate tumorigenesis, or act as a permissive environment for transformation.
“Epigenetic addiction” in pediatric high-grade brain tumors
Adult HGA is characterized by intra and inter-tumoral heterogeneity. Strikingly, our studies in pediatric HGA and ETMR unravel a previously unsuspected level of homogeneity within molecular subgroups of these tumors. Indeed, separate biopsies from H3 mutant HGA showed similar mutational profiles (including H3.3 K27M mutation) and close to identical global DNA methylation patterns, with heterogeneity seen largely for copy number variants in growth regulatory genes, such as PDGFRA amplification . Moreover, recurrences following complete global resection of ETMR were identical for both these features in the cases where the primary and recurrent samples were available despite high-dose alkylating agents and/or radiation therapy . These results mirror recent findings in IDH-mutant glioma where IDH mutation is present universally, independent of grade or recurrence, whereas non-IDH mutant gliomas showed strikingly different mutational patterns at recurrence including growth regulatory gene mutations such as BRAF V600E . Although further experimentation focused on animal and cell modeling and inhibition of growth following blockade H3 K27M in mutant cells is required, these findings suggest that these tumors are “addicted” to specific forms of epigenetic dysregulation. To meet the criteria of oncogene addiction further experimentation is needed; however, if proven correct, this offers an unprecedented therapeutic opportunity unique to pediatric brain tumors contrary to the landscape of complex heterogeneity seen in epithelial and other cancers.
Recent work by our lab, other independent groups as well as large consortia including TCGA and the International Cancer Genome Consortium (ICGC) have shown epigenetic defects to be present in a large proportion of pediatric brain tumors. They call for the advent of molecular pathology, a needed change in the WHO classification to integrate data that takes molecular defects into account when classifying/stratifying a brain tumor. As patients continue to consent to participation in sequencing and molecular studies, the research community will be able to continue to explore the novel epigenetic mechanisms at play underlying their formation. Recent discoveries of histone variant mutations at critical residues have placed defects leading to H3 deposition and post-translational modifications at the center of pediatric high-grade gliomagenesis. Future work modeling the impact of these mutations, the potential need for additional hits or associated mutations and the careful interpretation of the interplay between genomic and epigenomic data may allow us to reconcile major mechanisms ongoing in either dimension. Examples such as ETMR tumors support an integrated approach to studying driving forces governing tumorigenesis. With such an approach, we hope to inform clinical trials with biomarker development and drug discovery fueled by molecular subgrouping of this deadly group of pediatric tumors. Uncovering the untapped biology in this young field in pediatric brain tumors may aid in laying the foundation we need to tackle these tumors clinically to improve outcome for patients.
This work was performed as part of the I-CHANGE consortium (International Childhood Astrocytoma iNntegrated Genomics and Epigenomics consortium) and supported by funding from Genome Canada, Genome Quebec, The Institute for Cancer Research of the Canadian Institutes for Health Research (CIHR) McGill University and the Montreal Children’s Hospital Foundation. N. Jabado is a member of the Penny Cole lab and the recipient of a Chercheur Clinicien Senior Award. J. Majewski holds a Canada Chair Tier 2. A. M. Fontebasso is supported by a studentship from the CIHR, as well as an award from the CIHR Systems Biology Training Program at McGill University. T. Gayden holds a fellowship from the Canadian Gene Cure Foundation.
Conflict of interest
The authors declare that they have no competing interests.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.