Predicting Reactive Astrogliosis Propagation by Bayesian Computational Modeling: the Repeater Stations Model

Abstract

Reactive astrogliosis occurs upon focal brain injury and in neurodegenerative diseases. The mechanisms that propagate reactive astrogliosis to distal parts of the brain, in a rapid wave that activates astrocytes and other cell types along the way, are not completely understood. It is proposed that damage-associated molecular patterns (DAMP) released by necrotic cells from the injury core have a major role in the reactive astrogliosis initiation but whether they also participate in reactive astrogliosis propagation remains to be determined. We here developed a Bayesian computational model to define the most probable model for reactive astrogliosis propagation. Starting with experimental data from GFAP-immunostained reactive astrocytes, we defined five types of astrocytes based on morphometrical cues and registered the position of each reactive astrocyte cell type in the hemisphere ipsilateral to the injured site after 3 and 7 days post-ischemia. We developed equations for the changes in DAMP concentration (due to diffusion, binding to receptors or degradation), soluble mediators secretion, and for the evolution reactive astrogliosis. We tested four predefined models based on abovementioned previous hypothesis and modifications to it. Our results showed that DAMP diffusion alone has not justified the reactive astrogliosis propagation as previously assumed. Only two models succeeded in accurately reproducing the experimentally measured data and they highlighted the role of microglia and the glial secretion of soluble mediators to sustain the reactive signal and activating neighboring astrocytes. Thus, our in silico analysis proposes that glial cells behave as repeater stations of the injury signal in order to propagate reactive astrogliosis.