Evaluation of a novel in vitro model of sonic hedgehog medulloblastoma by single cell transcriptomics

CONTEXT OF THE RESEARCH: Brain cells that originate in the cerebellum give rise to the commonest childhood malignant brain tumour, termed medulloblastoma. Although cure rates of up to 70% have been reported, this is typically a devastating disease that is accompanied by considerable adverse effects resulting from multi-modal treatments including growth retardation, seizures and strokes. Like most other cancers, treatment advances crucially depend on tumour models that re-capitulate human disease, so that promising experimental treatments have a high chance of success in patients. Current medulloblastoma models are inadequate for several reasons. First, medulloblastoma cells from patients are hard to grow outside their normal environment, which prevents testing of experimental therapies. For the few tumour cell lines that have been grown ex vivo, they are deprived of their normal environment, which is an important factor in cancer progression as it influences the types of genes expressed by the tumour. Tumour gene expression in turn determines the response to treatments which explains why experimental treatments mostly fail in patients. Second, attempts to re-create the tumour environment using in vivo models of medulloblastoma, either based on genetic mouse models, or patient-derived tumour grafting into the mouse (termed PDX models) also have limitations. These models are costly, species differences limit the translation of therapies to the clinic, and in the case of PDX models, genetic divergence from the original tumour still occurs. Lastly, tumours differ genetically between patients and tumour heterogeneity is also evident within an individual. The latter discoveries have arisen from new technologies which can detect differences in the types of genes expressed at single cell resolution. Previous genetic assays of tumour cells in bulk were incapable of detecting these differences, which determine tumour growth patterns and influence treatment success. Consequently, treatment responses can be patchy across patients. Together, these shortcomings are responsible for the lack of new treatments for the last 30 years.

We hypothesise that the presence of the tumour environment better simulates in vivo tumour growth conditions. To test this, we have previously coerced a type of human stem cell, called human induced pluripotent stem cells (hiPSCs), which can make any cell type in the body, to grow into miniature cerebellar structures, termed cerebellar organoids, in vitro. In a novel approach, we are currently growing human medulloblastoma cell lines within human cerebellar organoids to mimic the tumour environment.

AIMS AND OBJECTIVES: I will investigate whether for one of the most prevalent genetic subtypes of medulloblastoma, termed sonic hedgehog medulloblastoma, our new model promotes tumour characteristics that better resemble human tumours. To this end, I will use the expression of genes by tumour cells as a proxy for tumour behaviour. By determining gene expression across the genome of thousands of medulloblastoma cells in our in vitro model individually, I can build a picture of their heterogeneity and determine how closely aligned our in vitro model is with its in vivo counterpart.

POTENTIAL APPLICATIONS AND BENEFITS: Growing freshly obtained medulloblastoma cells on cerebellar organoids could overcome current medulloblastoma modelling difficulties and provide a new improved platform to test patient-specific therapies, including drug toxicity on normal cerebellar cells. Single cell analysis can be used to generate a cellular atlas of sonic hedgehog medulloblastoma diversity, which in turn can be used to identify patient-specific druggable molecular targets.

RESEARCH OBJECTIVES: The childhood malignant brain tumour, medulloblastoma (MB), which originates in the cerebellum has devastating consequences. The lack of tumour models that capture treatment-relevant tumour heterogeneity and model tumour microenvironment has hampered treatment advances. I will develop a novel human model of a major MB subtype, sonic hedgehog (SHH)-MB, whose prognosis is negatively affected by additional mutations, in particular in the tumour suppressor gene, TP53. I will test whether TP53 wild-type (WT) and TP53 mutant SHH-MB can be better modelled by co-culture with wild type (WT) cerebellar organoids (CBOs), to mimic the tumour micro-environment. Conversely, I will also determine whether tumour cells induce permissive changes in the CBO.
METHODOLOGY: I willl analyse the transcriptomes of tumour spheroids from the human SHH-MB TP53 WT cell line (ONS76) and a TP53 mutant cell line (DAOY) in co-culture with WT CBOs, at single cell level using the 10X Genomics Chromium platform. The R toolkit, Seurat will be used to visualise cell identities, and identify differentially expressed genes. Tumour heterogeneity will be determined by identifying cell subsets defined by distinct 'gene signatures'. I will compare the two cell lines to each other and to published single cell and bulk transcriptome profiles of human MB biopsies to establish similarities and differences. Pathway analysis will be applied to test whether specific biochemical pathways are upregulated in tumour cells and/or the WT CBOs.
APPLICATION/EXPLOITATION: Patient-specific tumour models could be developed by growing biopsy samples ex vivo in WT CBOs to test personalised therapies. The effects of drug treatments on tumour and wild-type cerebellar cells can be concurrently analysed to assess treatment toxicity.