Personalized diagnosis of medulloblastoma subtypes across patients and model systems

Exploring the impact of intra-tumoural heterogeneity on liquid biopsy cell-free DNA methylation and copy number in head and neck squamous cell carcinoma

Liquid biopsy profiling is gaining increasing promise towards biomarker-led identification and disease stratification of tumours, particularly for tumours displaying significant intra-tumoural heterogeneity (ITH). For head and neck squamous cell carcinoma (HNSCC), which display high levels of genetic ITH, identification of epigenetic modifications and methylation signatures has shown multiple uses in stratification of HNSCC for prognosis, treatment, and HPV status. In this study, we investigated the potential of liquid biopsy methylomics and genomic copy number to profile HNSCC. We conducted multi-region sampling of tumour core, tumour margin and normal adjacent mucosa, as well as plasma cell-free DNA (cfDNA) across 9 HNSCC patients. Collectively, our work highlights the prevalence of methylomic ITH in HNSCC, and demonstrates the potential of cfDNA methylation as a tool for ITH assessment and serial sampling.

A temporal in vivo catalog of chromatin accessibility and expression profiles in pineoblastoma reveals a prevalent role for repressor elements

Pediatric pineoblastomas (PBs) are rare and aggressive tumors of grade IV histology. Although some oncogenic drivers are characterized, including germline mutations in RB1 and DICER1, the role of epigenetic deregulation and cis-regulatory regions in PB pathogenesis and progression is largely unknown. Here, we generated genome-wide gene expression, chromatin accessibility, and H3K27ac profiles covering key time points of PB initiation and progression from pineal tissues of a mouse model of CCND1-driven PB. We identified PB-specific enhancers and super-enhancers, and found that in some cases, the accessible genome dynamics precede transcriptomic changes, a characteristic that is underexplored in tumor progression. During progression of PB, newly acquired open chromatin regions lacking H3K27ac signal become enriched for repressive state elements and harbor motifs of repressor transcription factors like HINFP, GLI2, and YY1. Copy number variant analysis identified deletion events specific to the tumorigenic stage, affecting, among others, the histone gene cluster and Gas1, the growth arrest specific gene. Gene set enrichment analysis and gene expression signatures positioned the model used here close to human PB samples, showing the potential of our findings for exploring new avenues in PB management and therapy. Overall, this study reports the first temporal and in vivo cis-regulatory, expression, and accessibility maps in PB.

A transcriptome-based classifier to determine molecular subtypes in medulloblastoma

Medulloblastoma is a highly heterogeneous pediatric brain tumor with five molecular subtypes, Sonic Hedgehog TP53-mutant, Sonic Hedgehog TP53-wildtype, WNT, Group 3, and Group 4, defined by the World Health Organization. The current mechanism for classification into these molecular subtypes is through the use of immunostaining, methylation, and/or genetics. We surveyed the literature and identified a number of RNA-Seq and microarray datasets in order to develop, train, test, and validate a robust classifier to identify medulloblastoma molecular subtypes through the use of transcriptomic profiling data. We have developed a GPL-3 licensed R package and a Shiny Application to enable users to quickly and robustly classify medulloblastoma samples using transcriptomic data. The classifier utilizes a large composite microarray dataset (15 individual datasets), an individual microarray study, and an RNA-Seq dataset, using gene ratios instead of gene expression measures as features for the model. Discriminating features were identified using the limma R package and samples were classified using an unweighted mean of normalized scores. We utilized two training datasets and applied the classifier in 15 separate datasets. We observed a minimum accuracy of 85.71% in the smallest dataset and a maximum of 100% accuracy in four datasets with an overall median accuracy of 97.8% across the 15 datasets, with the majority of misclassification occurring between the heterogeneous Group 3 and Group 4 subtypes. We anticipate this medulloblastoma transcriptomic subtype classifier will be broadly applicable to the cancer research and clinical communities.

Modeling germline mutations in pineoblastoma uncovers lysosome disruption-based therapy

Pineoblastoma is a rare pediatric cancer induced by germline mutations in the tumor suppressors RB1 or DICER1. Presence of leptomeningeal metastases is indicative of poor prognosis. Here we report that inactivation of Rb plus p53 via a WAP-Cre transgene, commonly used to target the mammary gland during pregnancy, induces metastatic pineoblastoma resembling the human disease with 100% penetrance. A stabilizing mutation rather than deletion of p53 accelerates metastatic dissemination. Deletion of Dicer1 plus p53 via WAP-Cre also predisposes to pineoblastoma, albeit with lower penetrance. In silico analysis predicts tricyclic antidepressants such as nortriptyline as potential therapeutics for both pineoblastoma models. Nortriptyline disrupts the lysosome, leading to accumulation of non-functional autophagosome, cathepsin B release and pineoblastoma cell death. Nortriptyline further synergizes with the antineoplastic drug gemcitabine to effectively suppress pineoblastoma in our preclinical models, offering new modality for this lethal childhood malignancy.

Preclinical Models of Pediatric Brain Tumors-Forging Ahead

Approximately five out of 100,000 children from 0 to 19 years old are diagnosed with a brain tumor. These children are treated with medication designed for adults that are highly toxic to a developing brain. Those that survive are at high risk for a lifetime of limited physical, psychological, and cognitive abilities. Despite much effort, not one drug exists that was designed specifically for pediatric patients. Stagnant government funding and the lack of economic incentives for the pharmaceutical industry greatly limits preclinical research and the development of clinically applicable pediatric brain tumor models. As more data are collected, the recognition of disease sub-groups based on molecular heterogeneity increases the need for designing specific models suitable for predictive drug screening. To overcome these challenges, preclinical approaches will need continual enhancement. In this review, we examine the advantages and shortcomings of in vitro and in vivo preclinical pediatric brain tumor models and explore potential solutions based on past, present, and future strategies for improving their clinical relevancy.

Development of zebrafish medulloblastoma-like PNET model by TALEN-mediated somatic gene inactivation

Genetically engineered animal tumor models have traditionally been generated by the gain of single or multiple oncogenes or the loss of tumor suppressor genes; however, the development of live animal models has been difficult given that cancer phenotypes are generally induced by somatic mutation rather than by germline genetic inactivation. In this study, we developed somatically mutated tumor models using TALEN-mediated somatic gene inactivation of cdkn2a/b or rb1 tumor suppressor genes in zebrafish. One-cell stage injection of cdkn2a/b-TALEN mRNA resulted in malignant peripheral nerve sheath tumors with high frequency (about 39%) and early onset (about 35 weeks of age) in F0 tp53^e7/e7 mutant zebrafish. Injection of rb1-TALEN mRNA also led to the formation of brain tumors at high frequency (58%, 31 weeks of age) in F0 tp53^e7/e7 mutant zebrafish. Analysis of each tumor induced by somatic inactivation showed that the targeted genes had bi-allelic mutations. Tumors induced by rb1 somatic inactivation were characterized as medulloblastoma-like primitive neuroectodermal tumors based on incidence location, histopathological features, and immunohistochemical tests. In addition, 3' mRNA Quanti-Seq analysis showed differential activation of genes involved in cell cycle, DNA replication, and protein synthesis; especially, genes involved in neuronal development were up-regulated.

In vitro models of medulloblastoma: Choosing the right tool for the job

The recently-defined four molecular subgroups of medulloblastoma have required updating of our understanding of in vitro models to include molecular classification and risk stratification features from clinical practice. This review seeks to build a more comprehensive picture of the in vitro systems available for modelling medulloblastoma. The subtype classification and molecular characterisation for over 40 medulloblastoma cell-lines has been compiled, making it possible to identify the strengths and weaknesses in current model systems. Less than half (18/44) of established medulloblastoma cell-lines have been subgrouped. The majority of the subgrouped cell-lines (11/18) are Group 3 with MYC-amplification. SHH cell-lines are the next most common (4/18), half of which exhibit TP53 mutation. WNT and Group 4 subgroups, accounting for 50% of patients, remain underrepresented with 1 and 2 cell-lines respectively. In vitro modelling relies not only on incorporating appropriate tumour cells, but also on using systems with the relevant tissue architecture and phenotype as well as normal tissues. Novel ways of improving the clinical relevance of in vitro models are reviewed, focusing on 3D cell culture, extracellular matrix, co-cultures with normal cells and organotypic slices. This paper champions the establishment of a collaborative online-database and linked cell-bank to catalyse preclinical medulloblastoma research.

MM2S: personalized diagnosis of medulloblastoma patients and model systems

Background

Medulloblastoma (MB) is a highly malignant and heterogeneous brain tumour that is the most common cause of cancer-related deaths in children. Increasing availability of genomic data over the last decade had resulted in improvement of human subtype classification methods, and the parallel development of MB mouse models towards identification of subtype-specific disease origins and signaling pathways. Despite these advances, MB classification schemes remained inadequate for personalized prediction of MB subtypes for individual patient samples and across model systems. To address this issue, we developed the Medullo-Model to Subtypes (MM2S) classifier, a new method enabling classification of individual gene expression profiles from MB samples (patient samples, mouse models, and cell lines) against well-established molecular subtypes [Genomics 106:96-106, 2015]. We demonstrated the accuracy and flexibility of MM2S in the largest meta-analysis of human patients and mouse models to date. Here, we present a new functional package that provides an easy-to-use and fully documented implementation of the MM2S method, with additional functionalities that allow users to obtain graphical and tabular summaries of MB subtype predictions for single samples and across sample replicates. The flexibility of the MM2S package promotes incorporation of MB predictions into large Medulloblastoma-driven analysis pipelines, making this tool suitable for use by researchers.

Results

The MM2S package is applied in two case studies involving human primary patient samples, as well as sample replicates of the GTML mouse model. We highlight functions that are of use for species-specific MB classification, across individual samples and sample replicates. We emphasize on the range of functions that can be used to derive both singular and meta-centric views of MB predictions, across samples and across MB subtypes.

Conclusions

Our MM2S package can be used to generate predictions without having to rely on an external web server or additional sources. Our open-source package facilitates and extends the MM2S algorithm in diverse computational and bioinformatics contexts. The package is available on CRAN, at the following URL: https://cran.r-project.org/web/packages/MM2S/, as well as on Github at the following URLs: https://github.com/DGendoo and https://github.com/bhklab.

Electronic supplementary material

The online version of this article (doi:10.1186/s13029-016-0053-y) contains supplementary material, which is available to authorized users.