Skip to main content
European Commission logo print header

Molecularly defined models of human childhood brain tumors

Periodic Reporting for period 4 - Medulloblastoma (Molecularly defined models of human childhood brain tumors)

Reporting period: 2019-11-01 to 2020-04-30

The two main challenges in the clinical management of paediatric brain tumours are improving survival and reducing long-term detriments due to treatment toxicities, especially from craniospinal radiotherapy. The overall aim of this project is to improve survival of medulloblastoma, the most common paediatric brain malignancy, through better understanding of the biology underlying tumour initiation and progression.

Medulloblastoma is suggested to originate from specific cells in the small brain, cerebellum, and can be classified into four defined subgroups called: WNT, SHH, Group 3 and Group 4. MYC proteins such as MYC and MYCN are transcription factors that are mis-regulated in more than half of all types of human cancer including medulloblastoma. We have generated sophisticated regulatable transgenic animal models that overexpress MYC proteins. These models carefully resemble at least two of the four defined subgroups of human medulloblastoma. In this proposal we intend to use the models to identify the specific cell type these brain tumours originates from. We also aim to refine our medulloblastoma models and develop novel models to define and study cells involved in brain metastasis and tumour recurrence, the main causes of death in brain tumour patients.

We have managed to culture normal human cerebellar stem cells and we next plan to model human medulloblastoma development in these cells by overexpressing oncogenes, or silencing suppressor genes, that are defined as clinically relevant medulloblastoma drivers. We will use a forward genetics screen to identify novel drivers and specifiers of subtypes of medulloblastoma. We hope these combined efforts will help us better understand human medulloblastoma formation and we expect to generate tumours that correlate well, both pathologically and molecularly, with primary cell cultures derived from medulloblastoma patients. Our data will subsequently be used to provide novel targets for therapeutic intervention which will hopefully be benificial for patients affected by this devastating disease.
The main aim of this project is divided into two specific aims:

I. Characterise cells of origin and cells driving recurrence in MYCN-driven medulloblastoma.
II. Develop new models for the four medulloblastoma subgroups using human hindbrain stem cells.

We previously showed that stabilisation of MYCN is essential for brain tumour initiation (Swartling et al. Cancer Cell, 2012). MYCN stability is regulated by the ubiquitin ligase FBW7, which normally targets it for proteasomal degradation. FBW7 is a tumour suppressor gene mutated in various types of cancer.

Under Aim1 we published a report in EMBOJ showing that the stem cell factor SOX9 is regulated at the protein level in patients that have FBW7 mutations. We first showed that the ubiquitin kinase FBW7 mostly described in Aim 2 was found mutated in several medulloblastoma patients. We found that SOX9 had specific sites in its protein structure where it could be targeted by degradation from FBW7 binding. We mutated these sites and got a stabilised SOX9 model that when transduced into medulloblastoma cells promoted their metastatic capabilities. Our data provides evidence on why FBW7 mutations occur and cause increased malignancy in these patients.

In order to get a larger collection of expression data from human medulloblastoma samples for our analyses, we developed an algorithm method to batch-normalise data from various datasets from patients from several research centers/hospitals. The centers had performed microarray experiments and gathered transcription data on collected samples. From 23 transcription datasets, we got 1350 medulloblastoma and 291 normal brain samples. The data published in the journal Bioinformatics in 2019 can now be used to compare/analyse specific gene expression between normal brain and malignant brain tumors.

We next showed that SOX9-positive cells give rise to recurrences in children with medulloblastoma and made a regulatable mouse model where we can switch MYCN activity to SOX9-positive cells. Such cells divide slower that other SOX9-negative tumour cells. To show that this is a clinically relevant finding, we collected and studied paired samples of both primary and recurrent tumours from patients that developed relapses. It was clear that SOX9 accumulates in recurrences and upregulate MGMT which is involved in DNA repair and therapy resistance. We used MGMT inhibitors in combination with standard therapy to show how it specifically targets recurrent cells over primary tumour cells. The paper is currently being resubmitted to Nature Cell Biology.

Finally, we aimed to study where and how these brain tumours arise in the brain. We used lineage tracing as suggested in Aim 1 with Confetti animals and found out that tumours in the transgenic models are monoclonal, thus likely arise from a single cell. We had now started using single-cell sequencing to bioinformatically calculate and understand where in the brain the medulloblastoma develops. The process of analysing this data is ongoing and will continue in another project.

In Aim 2 we published a report in Cell Stem Cell (2019) presenting how molecularly defined humanised medulloblastoma can be developed from both embryonal and iPS-derived human stem cells. We used MYCN overexpression and found out that iPS-cell derived stem cells generate more malignant SHH-tumours than donated primary embryonic stem cells. We revealed that MYCN promotes epigenetic loss of methylation of the pluripotency factor OCT4 and promotes malignancy in the tumours by activating the mTOR pathway. By using mTOR inhibitors we found a novel therapy for these SHH-tumours that mostly affects very young children. However, as mTOR inhibitors, like everolimus, are approved and today can be used in infants we have a good rationale to provide a better therapy for these children.

We further completed a story on a forward genetic tool using NGS to find novel cancer-causing genes for brain tumours driven by the growth factor PDGF. The paper shows that retroviral insertional mutagenesis can promote tumour malignancy. It also validates some of the candidate genes identified in this forward genetic screen and the story will be submitted soon. Finally, we are working on generating a Group 4 model for MYCN but here we believe we first have to differentiate the human stem cells so they become more like unipolar brush cells that were recently described as the cell of origin for these tumours.
Identifying the true cell of origin of medulloblastoma has a huge impact for brain tumour biology and cancer research. Our work on brain tumour formation in this ERC project has shown which cells generate SHH infant medulloblastoma, the most common malignant brain tumour found in young children/infants. The identification of an OCT4/mTOR axis that promotes malignancy in these tumours was an essential finding and a way to understand how mTOR inhibitors can actually be a promising drug used to target this disease. We are now on our way to show how other molecular subgroups of these brain tumours arise.

Recurrence is what kills children with brain tumours. Good models for tumour recurrence are rare and we have generated the first genetically modified animal model for this. This will improve our understanding of cancer recurrence mechanisms which can identify novel drug targets for better cancer treatment for these patients.
Moleculary definded human models for medulloblastoma generated provides OCT4 as a malignancy driver
SOX9 is promoting metastatic spread of medulloblastoma and can be suppressed be mTOR inhibition.
The brain tumor recurrence process from dormant SOX9+ tumour cells.