Molecular and cellular mechanisms of tissue interaction and invasion in medulloblastoma

The biological principles underlying tumor-host tissue interactions in medulloblastoma (MB), as well as the key molecular regulators of cellular processes and morphological changes that drive tumor growth and tissue invasion, remain poorly understood. In particular, how MB tumor cells “communicate” with cells in the cerebellar microenvironment — and how the microenvironment responds — is not yet fully elucidated. Beyond soluble chemical factors, tumor cells release extracellular vesicles (EVs) and extend tunneling nanotubes (TNTs), both of which may facilitate the transfer of chemical signals between different cellular entities.

The primary objective of this study is to characterize the molecular mechanisms governing EV biogenesis and TNT formation in tumor-tumor and tumor-astrocyte crosstalk. Our secondary objective is to investigate the functional significance of MB-derived EVs and TNTs in tumor-host tissue interactions and tissue invasion.

We anticipate that this work will provide crucial mechanistic insights, laying the foundation for the rational design of mechanism-based therapies aimed at targeting tumor-induced deregulation of host tissue functions that promote tumor growth and invasion.

 SNF Sinergia Project: 

Rational Small Molecule Inhibition of Dissemination (RaSMID) for pediatric brain tumors

The lack of effective therapeutics that precisely target the biological mechanisms driving tumor progression remains a obstacle to developing anti-metastatic treatments. Targeting protein-protein interactions (PPIs) is an emerging concept in drug discovery that holds promise for overcoming this challenge. When successfully implemented, it offers an innovative strategy to selectively inhibit or alter protein functions under pathological conditions.

Key prerequisites for discovering effective and safe small molecule PPI inhibitors include advanced computational (“in silico”) screening, robust functional validation technologies, and physiologically relevant model systems to assess the biological impact of compound activity and PPI disruption.

We apply this discovery approach to identify small bioactive molecules against medulloblastoma (MB), the most common malignant pediatric brain tumor. With the anticipated lead compounds targeting tumor dissemination and the establishment of novel methodologies for bioactive compound validation, we aim to contribute to ongoing efforts to develop effective and safe treatments for MB patients.

Our collaborators:

Prof. Gisbert Schneider, Computer-assisted Drug Design

Prof. Stephan Neuhauss, Neuhauss Group

The tumor in its tissue context: Mechanisms and consequences of tumor-tissue interaction

MB arises in the cerebellum, from where it can invade locally, and spread to the meninges of the brain and the spinal cord. With its unique cellular, structural and chemical composition, the neural tumor microenvironment (nTME) in the cerebellum determines MB tumor cell behavior. Conversely, through the direct interaction and the secretion of soluble and transmissible factors such as proteins, lipids, nucleic acids and metabolites, the growing tumor is continuously modulating the nTME. This reciprocal, functional interplay between the nTME and MB tumor cells and its consequences for tumor growth and progression is is incompletely understood.

In this project, we seek to understand how Sonic Hedgehog (SHH) MB tumor cell growth and invasion modify the nTME, and how consequent changes in the nTME reciprocally control tumor cell behavior. Our primary objective is to determine molecular mediators of dynamic reciprocity between SHH MB and the nTME and its functional implications for tumor growth and tissue invasion. Our secondary objective is to explore strategies targeting identified mediators tailored to repress growth and tissue invasion of MB. 



Image-based morphological and functional analysis of tumor cell behavior at single-cell level

Microscopic images of the F-actin cytoskeleton structures of tumor cells capture the underlying molecular alterations as morphological and dynamic features. These features are complex, and their quantification requires advanced image analysis tools, such as machine learning (ML) approaches. By harnessing the information embedded in F-actin cytoskeleton structures or phase images, treatment impact on phenotype-defining mechanisms can be assessed and compared across different treatments. With this, a more efficient functional screening for target agnostic low molecular weight compounds can be achieved, and compounds that repress tumor-promoting F-actin dynamics and invasiveness identified.

Simple image analysis methods do not allow quantifying changes in F-actin architecture that define motile and invasive cancer cell behavior. To capture and quantitatively assess such phenotype-defining F-actin features, we are integrating deep learning approaches into our current phenotype analyses. Specifically, image classifiers to predict treatment efficacy and generative models to construct an interpretable representation of the phenotype-defining information embedded in the F-actin cytoskeleton of control and treated cells are established. These methodologies are then used to assess the effects of de novo and established compounds in an unbiased manner.