The role of dimensionality in neuronal network dynamics.

Francesco Paolo Ulloa Severino,Guosheng Cheng,Vincent Torre,Qin Song,Ginestra Bianconi,Jelena Ban,Mingliang Tang

doi:10.1038/srep29640

Abstract

Recent results from network theory show that complexity affects several dynamical properties of networks that favor synchronization. Here we show that synchronization in 2D and 3D neuronal networks is significantly different. Using dissociated hippocampal neurons we compared properties of cultures grown on a flat 2D substrates with those formed on 3D graphene foam scaffolds. Both 2D and 3D cultures had comparable glia to neuron ratio and the percentage of GABAergic inhibitory neurons. 3D cultures because of their dimension have many connections among distant neurons leading to small-world networks and their characteristic dynamics. After one week, calcium imaging revealed moderately synchronous activity in 2D networks, but the degree of synchrony of 3D networks was higher and had two regimes: a highly synchronized (HS) and a moderately synchronized (MS) regime. The HS regime was never observed in 2D networks. During the MS regime, neuronal assemblies in synchrony changed with time as observed in mammalian brains. After two weeks, the degree of synchrony in 3D networks decreased, as observed in vivo. These results show that dimensionality determines properties of neuronal networks and that several features of brain dynamics are a consequence of its 3D topology.

Highlights

Figure 1. 3 dimensions (3D) network model. (a) Simulation of a 3D neuronal network which is modular and has short range connections and some long range connections
We show that 3D networks have dynamical properties that are quantifiably more similar to what is observed in the brain than 2D networks[4,5,18,19,20]
The present manuscript demonstrates that the dynamics of 3D neuronal networks differ from those of 2D neuronal networks and better recapitulate what is observed in vivo[19,40]

Summary

Introduction

Figure 1. 3D network model. (a) Simulation of a 3D neuronal network which is modular and has short range connections and some long range connections (small-word network). (a) Simulation of a 3D neuronal network which is modular and has short range connections and some long range connections (small-word network). The neurons are distributed along a fractal tree and primarily connected by short-range interactions; long-range interactions constitute a small proportion of the connections. The Kuramoto model of this network yields three dynamical regimes as a function of the strength of the coupling, K, between the oscillators. Large values of K result in a fully synchronized phase, whereas low values of K do not produce synchronization. Intermediate values of K produce a phase of frustrated synchronization. In panel (b), we plotted the order parameter, R, for the synchronization as a function of time, t, for different values of the coupling, K. As a function of K, the plots indicate three different synchronization phases for the simulated 3D network. We show that 3D networks have dynamical properties that are quantifiably more similar to what is observed in the brain than 2D networks[4,5,18,19,20]

Methods

Results

Conclusion