Neurorosettes: a novel computational modelling framework to investigate the Homer-Wright rosette formation in neuroblastoma

Abstract

AbstractCancer deregulates the interactions between cells and their microenvironment, leading to disrupted architectures. Homer-Wright rosettes, observed in neuroblastoma, comprise radial arrangements of neurons surrounding a meshwork of fibres. Currently, scientists believe that the presence of Homer-Wright rosettes reflects aberrant neuronal differentiation. Nonetheless, additional understanding of how these structures develop is required since neither experimental nor computational research has characterised this mechanism properly. In this work, we propose a mechanics-based computational framework to investigate Homer-Wright rosette formation. Our model depicts neurons as a combination of spherical (cell bodies) and cylindrical (neurites) agents, and it includes proliferation, neuronal differentiation, and adhesion/repulsion dynamics between neurons. We implemented our framework as an open-source user-friendly Python package called neurorosettes that provides real-time rendering of simulation results, making it adequate for general researchers to test and visualize hypotheses of Homer-Wright rosette formation. Furthermore, we present three example use-cases to replicate the emergence of this rosette subtype and investigate how mechanical interactions between neurons and neuronal differentiation may regulate its architecture. Due to the spare amount of experimental data on the formation of these histological patterns, our applications serve primarily as preliminary examples of how our tool can be used and extended. Although our preliminary results show the relevance of mechanical interactions and poor neuronal differentiation to Homer-Wright rosette formation, these factors appear to only predict the initial stages of rosette formation. Overall, our tool can improve the theoretical knowledge on this process and drive the design of new experimental studies to validate model results.