Fredrik Swartling's research on childhood brain tumours

In search for the cellular origin of childhood brain tumors

Miao Zhao, Oliver Mainwaring, Gabriela Rosén, Fredrik Swartling

Medulloblastoma is divided into four distinct molecular subtypes (WNT, SHH, Group 3 and Group 4). Group 3 medulloblastoma presents with a large proportion (15-20%) of MYC amplifications (Hovestadt et al. Nat Rev Cancer, 2020) and correlate with poor prognosis. MYC proteins are unstable oncoproteins with short half-lives.

We have shown that MYCN generates Group 3 tumours from a glutamate transporter (GLT1) promoter in a transgenic inducible model (GTML) of medulloblastoma (Swartling et al. Genes & Dev., 2010) and found that tumours originate from a GLT1-positive photoreceptor-positive progenitor cell.

Transgenic mouse model

There are currently no transgenic mouse models for MYC-driven pediatric brain tumours, so we developed an inducible transgenic model (called GMYC), which is driving MYC from the Glt1 promoter (Mainwaring et al. accepted in Nat. Comms., 2023). Aggressive Group 3 tumours are generated with good penetrance in the GMYC model. Tumours histologically and molecularly resemble human Group 3 medulloblastoma.

We also found that the CDKN2A tumour suppressor gene upstream of the p53 pathway is silenced in the MYC-driven model, as compared to our older MYCN-driven model. When CDKN2A was knocked out early, the model induced pediatric high grade gliomas (pHGGs) which resemble RTK-driven or G34 mutant pHGG subtypes.

In the search for the true cell of origin of this brain tumour, we perform analysis using cell fate tracking (using Confetti models) and single-cell RNA sequencing (scRNA-Seq) analysis of hindbrain regions at different developmental time point before and after tumour formation.

Novel models of medulloblastoma and high-grade glioma for use in targeted treatment studies

Karl Holmberg Olausson, Géraldine Giraud, Gabriela Rosén, Tobias Bergström, Fredrik Swartling

In this project we are transforming human iPS-derived cells and embryonic hindbrain neural stem cells in order to model the different subgroups of medulloblastoma using lentiviruses carrying clinically relevant cancer driver genes for the distinct tumour subgroups. We have recently shown that MYCN overexpression can induce infant SHH-driven tumours from these types of human stem cells in vivo (Čančer et al. Cell Stem Cell, 2019).

Depending on cellular origin and OCT4-reprogramming status, tumours of varying prognosis developed. When OCT4 was induced it promoted mTOR pathway signalling in tumour models and human PDXs. We therefore showed that tumours were dependent on this pathway (see Figure 1) and that mTORC inhibitors could be used to effectively treat these MYCN-driven infant brain tumours.

Schematic drawing of different ways to molecularly target MYC proteins in childhood brain tumours
 Figure 1. Different ways to molecularly target MYC proteins in childhood brain tumours. Figure from Borgenvik et al. Front. Oncol., 2021.

Modelling tumour development

We are also transforming brain stem-specific cells from humans and mice with MYC in order to model medulloblastoma or diffuse-intrinsic pontine glioma (DIPG) development. We will evaluate the relevance of using well-defined human hindbrain stem cells to generate these childhood brain tumours and we will compare them to subtype-specific cells similarly cultured from medulloblastoma or DIPG patients. We further develop and study patient-derived xenografts and evaluate their use in drug screens in combination with radiation therapies in co-culture systems with organoid (3D) cultures.

We hope we will understand what actually drives the initiation of medulloblastoma and DIPGs and if various subgroups match certain hindbrain cell types. We currently use genetic and epigenetic analyses to find prognosis markers. We hope to predict how some of these tumours could be treated or if they would be resistant to certain targeted therapies.

Targeting MYC in childhood brain tumours by using CDK2 suppression

Tina Lin, Anna Borgenvik, Fredrik Swartling

We have shown that MYCN levels and early proliferation of brain tumours could be reduced by specific inhibition of the bromodomain inhibitor JQ1, which targets MYC proteins epigenetically (Bandopadhayay et al. Clin Can Res., 2014). We also found good efficacy controlling MYCN stabilization by using a CDK2 inhibitor called Milciclib (Bolin et al. Oncogene, 2018) or by using Aurora Kinase inhibition in glioblastoma cell lines (Čančer et al. Cell Death & Disease, 2019) in combination with JQ1 (Figure 1).

We are currently evaluating the role of complete CDK2 inhibition during tumour initiation and downstream CDK treatment effects in our models, especially the GMYC model in vivo. Our goal is to understand the efficacy and underlying mechanisms of MYC inhibition in cells and further evaluate the potential of using combinations of promising MYC-targeting drugs in the clinic.

A new model for childhood brain tumour recurrence

Tina Lin, Karl Holmberg Olausson, Thale Kristin Olsen, Anders Sundström, Géraldine Giraud, Fredrik Swartling

Despite precise surgery, intensive irradiation treatment and various combinations of chemotherapy, medulloblastoma can relapse which is associated with very poor prognosis in children. In this project we are studying how MYCN interacts with SOX9, a transcription factor involved in glial fate determination in the brain. Few scattered SOX9-positive cells are found in GTML tumours that are similar to Group 3 human MB.

By using a combination of Tet-ON and Tet-OFF inducible systems we managed to target this rare population of SOX9-positive GTML tumour cells in vivo to show how they were capable of initiating tumour recurrence (Borgenvik et al. Cancer Res. 2022).

We also showed how FBW7 is regulating SOX9 stability and increases tumour cell migration and metastasis (Suryo Rahmanto et al. EMBOJ, 2016). By suppressing the mTOR/PI3K/AKT pathway we can obstruct this stabilisation (see Figure 2). Further characterisation of SOX9-positive tumour cells using expression profiling (including scRNA-Seq) and ATAC-Seq will help us understand the mechanisms behind metastatic medulloblastoma recurrence.

Schematic drawing of how analysis of PDXs and organoids from matched primary-recurrent patient samples is performed
Figure 2. Analysis of PDXs and organoids from matched primary-recurrent patient samples (in collaboration with Barntumörbanken, Karolinska Institutet).

Mechanisms of pediatric brain tumor relapse using standard radiation therapy in vivo

Miao Zhao, Tobias Bergström, Erika Dalmo, Gabriela Rosén, Holger Weishaupt, Géraldine Giraud, Fredrik Swartling

It is not known which types of tumour clones are surviving standard treatment and give rise to tumour relapse. In order to study the recurrence mechanism of medulloblastoma and high-grade glioma, we will use a barcoding technique using lentiviral vectors to study the clonal evolution of cells that survive radiation. Barcode labelling will be performed on transplanted cells/PDXs in animals that are undergoing standard radiation treatments.

In this project, we will compare primary tumours with recurring tumours that escape radiation therapies using scRNA-Seq to detect the specific barcodes. We will also collect fresh biopsies from matched pairs of primary-recurrent samples (focusing on high-grade glioma, medulloblastoma but also ependymoma) in collaboration with Barntumörbanken, Karolinska Institutet (Figure 2).

We aim to see if cells with particular expression of genes and distinct signalling pathways are enriched in recurrent tumour cells as compared to matched primary tumour cells. It would be important to see if minor or major clones from the primary tumour are driving relapse and we finally hope to identify distinct molecular therapies that can specifically target relapsing cells.

Using forward genetic tools to identify candidate cancer genes of virally-induced glioma and neural network analysis to predict brain tumour progression

Holger Weishaupt, Oliver Mainwaring, Anders Sundström, Fredrik Swartling

To study pathways involved in glioma development, we have used a retrovirus-driven PDGFB-induced murine glioma model that causes tumours that closely resemble human gliomas (Weishaupt et al. Neuro-Oncol., 2022). Retroviruses have a capacity to induce so called insertional mutagenesis and thereby promote brain tumour formation.

We have used whole genome sequencing methods and developed computational algorithms to identify several potential genes that, together with PDGFB, induce brain tumour development. We are currently functionally evaluating some of our top candidates and investigating their distinct roles in brain tumour biology.

Genetic defect involved in tumour development

Amplification of Chr. 17q or isochromosome 17q is the most common chromosomal defect in tumours that depend on MYC or MYCN, including neuroblastoma, certain leukaemia (e.g. CML) and medulloblastoma, to mention a few tumour types. Despite this region being so frequently (70 %) amplified with MYC genes, it has been impossible to identify the gene or those genes responsible for helping MYC in driving tumour development.

We have developed batch-normalization tools (Weishaupt et al. Bioinformatics, 2019) to combine larger cohorts of molecularly analysed brain tumours in order to perform a more statistically relevant analysis of Chr 17q genes. We have then employed advanced neural network analysis from multi-omics data from these larger patient cohorts and used computational models and machine learning

Last modified: 2023-02-20