Alberto Broniscer, MD

Genomic Analysis of Brain Tumors

Brain tumors are the most common solid malignancies of childhood and a leading cause of cancer-related death in children. Ongoing efforts by the Neurobiology & Brain Tumor Program of the St. Jude Comprehensive Cancer Center are building upon key discoveries such as the ones described here to determine how specific genetic mutations contribute to brain tumorigenesis and how those mechanisms can be subverted through therapeutic intervention.

The first brain tumors studied by the Pediatric Cancer Genome Project (PCGP) are diffuse intrinsic pontine glioma (DIPG) and medulloblastoma, both of which cause devastating morbidity and mortality.


Toward Understanding Diffuse Intrinsic Pontine Glioma

There are currently no effective therapies for DIPG, a disease that occurs almost exclusively in children and results in the death of 90% of patients within 2 years of diagnosis. DIPGs are high-grade gliomas that arise in the brainstem, which controls many vital functions, including breathing, heart rate, and consciousness. Because the tumor infiltrates a brain region that is required for life, it cannot be surgically removed, which drastically limits therapeutic options. The lack of tumor tissue available for research also substantially impedes progress toward improving treatment.

Alberto Broniscer, MD (Oncology), and colleagues developed a clinical protocol for families to donate DIPG tissue at the time of autopsy for basic research. David W. Ellison, MD, PhD (Pathology), conducted a thorough neuropathologic assessment of samples of DIPG, as well as pediatric high-grade gliomas from outside the brainstem. Suzanne J. Baker, PhD (Developmental Neurobiology), and her laboratory analyzed the DNA copy number in the tumors and found some evidence that specific alterations occurred at different frequencies between the two groups.

However, it remained unclear whether DIPGs and glioblastomas located outside the brainstem represent the same disease arising at two different locations or whether unique genetic mutations underlie the development of DIPG. Based on these unanswered questions and the dire need to identify novel therapeutic approaches, DIPG was selected for in-depth study by the PCGP.


Discovery of Histone H3 Mutations in High-Grade Pediatric Gliomas

A team of investigators led by Dr. Baker, Jinghui Zhang, PhD (Computational Biology), and Gang Wu, PhD (Computational Biology), conducted a genomic analysis of DIPG. As reported in the journal Nature Genetics, they identified recurrent somatic mutations in histone H3 at high frequency in DIPG and at lower frequency in pediatric glioblastomas outside the brainstem. Whole-genome sequencing of seven DIPG samples identified five samples carrying same missense mutation in H3F3A or HIST1H3B, two closely related genes encoding nearly identical isoforms of histone H3.

Focused sequence analysis of a larger collection of 43 DIPG tumor samples and 36 nonbrainstem pediatric glioblastoma samples showed that 78% of DIPGs and 22% of nonbrainstem glioblastomas had the same recurrent mutation––the lysine at residue 27 was replaced by methionine. This mutation was found in either H3F3A or HIST1H3B. Also, an additional 14% of the nonbrainstem glioblastomas carried an H3F3A mutation in which arginine was substituted for glycine at residue 34. This mutation was not found in any of the DIPG samples tested. These three different mutations were mutually exclusive.

A further investigation of whether alterations in histone H3 genes occur in other cancers revealed that the frequent presence of this genetic anomaly may be exclusive to pediatric DIPG and high-grade nonbrainstem glioblastomas. The investigators reviewed genomic data from several other pediatric brain tumors and pediatric cancers that occur outside of the central nervous system and uncovered evidence of mutations in some of the 16 histone H3 genes in those diseases.

Histone H3 is one of four histone proteins that form the protein core of the nucleosome, which is the fundamental structural unit of chromatin that allows effective packaging of DNA in the nucleus. Chromatin structure influences cellular function by modulating the accessibility of DNA to proteins that are responsible for DNA replication, repair, and transcription.

Both mutations occurred in the unstructured tail of histone H3, which is extensively altered by posttranslational modifications that influence the recruitment of transcriptional activators or repressors. In particular, the modification of lysine-27 by methylation is associated with transcriptional repression, and that by acetylation is associated with transcriptional activation.

This work is the first to identify histone mutations in human cancer. Thus far, the mutations have been found at high frequency only in DIPG and pediatric nonbrainstem high-grade gliomas. In addition, no histone H3 lysine-27 mutations have been reported in adult glioblastomas, and glycine-34 mutations are only very rarely seen in glioblastomas of young adults; thus, these mutations represent a substantial and previously unappreciated distinction between the molecular mechanisms that drive glioma tumorigenesis in children and adults. Furthermore, the histone mutations represent a critical connection between chromatin regulation and cancer and may provide a much-needed novel therapeutic target for treating children who have this devastating disease.


Deciphering the Genomic Landscapes of Medulloblastoma Subgroups

Medulloblastoma originates in the cerebellum and is the most common malignant brain tumor in children. Four subtypes of medulloblastoma—the WNT subgroup, sonic hedgehog (SHH) subgroup, subgroup 3, and subgroup 4––have been defined based on gene expression profiling. These subgroups show differences in DNA copy number and prognosis. Although substantial differences in long-term survival distinguish the subgroups, all pediatric patients with medulloblastoma currently receive the same multimodality treatment comprising surgery, chemotherapy, and radiation therapy.

Thus, patients with WNT-subgroup medulloblastoma, the less severe form of the disease, may be overtreated and as a result endure long-term treatment-related effects on cognitive and endocrine functions. Therefore, the development of medulloblastoma subgroup–specific treatment approaches that are more effective and less toxic is warranted.

To identify gene mutations that may help in the design of novel therapeutic approaches that target the unique molecular features and clinical challenges of each medulloblastoma subgroup, Dr. Zhang worked with Richard J. Gilbertson, MD, PhD (Developmental Neurobiology, Oncology), Giles Robinson, MD (Oncology), and colleagues to sequence the entire genomes of 37 medulloblastomas through the PCGP. Samples from each tumor subgroup were included in this whole-genome sequencing analysis. Follow-up studies (e.g., PCR analysis and Sanger sequencing) were performed on an additional 56 medulloblastomas to validate the putative somatic alterations.

As reported in the journal Nature, whole-genome sequencing of DNA from the discovery group of 37 medulloblastomas revealed nearly 23,000 somatic mutations. The investigators identified recurrent mutations in 41 genes that had not been previously implicated in medulloblastoma. Novel mutations in the DDX3X gene, which encodes RNA helicase, were enriched in the WNT subgroup. DDX3X regulates several essential cellular processes, including the segregation of chromosomes, progression through the cell cycle, and transcription and translation of genes.

In mouse models of WNT-subgroup medulloblastoma, knocking down the expression of Ddx3x hindered the development of lower rhombic lip progenitor cells, which give rise to this subgroup of tumors. This finding suggests that Ddx3x is central to the proliferation of cells from which the WNT subgroup originates. Mutations in WNT-subgroup medulloblastoma were also identified that target chromatin-remodeling proteins that influence higher-order structure and function of chromatin.

The investigators were intrigued to discover that many of the mutations in DNA from medulloblastoma subgroups 3 and 4, which include the most severe forms of the disease, targeted genes encoding chromatin regulators, including enzymes that modulate specific posttranslational modifications of histone H3. This result indicates that a common epigenetic pathway is targeted through different mechanisms in medulloblastoma and DIPG. Thus, drug development efforts focused on inhibitors of the epigenetic machinery may be useful for treating patients with subgroup-3 or subgroup-4 disease.

In conclusion, whole-genome sequencing of medulloblastomas through the PCGP has identified novel therapeutic targets in the disease and highlighted potential roles for drugs that modulate the epigenetic machinery. Drugs targeting the mutations found in different medulloblastoma subgroups may facilitate the development of more specific and less toxic treatments for each distinct subgroup.