Tonic and phasic overall activities in biologically plausible excitatory neural networks

Abstract

AbstracfThis paper deals with the modelling of excitatory networks. The conditions in which to& or phasic activities arise in such networks were investigated. These behaviors a re iniplicated in the generation of the brainstem spontaneous activity or in epaeptic-like synchronous discharge patterns. It was shown that a physiologically relevant parameter couId control the switclling from one type of activity to the other. INTRODUCTION Excitatory synapses play N) important role in the amplification and synchronization of the electrical activity in living neural networks. The following examples show that excitatory connections may lead to both phasic and tonic electrical activities in living neural networks: 1 ) Direct application of penicillin to the brain's surface decreases the activity of inhibitory GABAergic synapses and elicits seizures similar IO epileptic seizures [I]. Withdrawal of GABA in neocorticd neurons [2 ] , and the injection of a minute dose of tetanus toxin f3] both produce similar epileptic syndromes. In fact a decrease in inhibition has been associated with epilepsy [4], [ 5 ] . The epileptic seizures are characterized by a synchronized overall (phasic) activity in the network. Therefore the lack of inhibitory connections in the network contributes to the emergence of an overall synchronized panem. 2 ) The Solitary Complex is an area of the brainstem which has been implicated in various functions such as cardiovascular regulation, respiration and feeding functions. In vitro studies of the Solitary Complex have revealed that in the absence of afferent sensory inputs, fully excitatory networks within the Solitary Complex generate a tonic background synaptic activity [6 ] . Such spontaneous activities in the brainstem are important in the control of locomotion (71, respiration [8], sleep w'aking cycles [9], and cardiovascular parameters [lo]. Therefore other networks of excitatory synaptic connections may also be implicated in generating brainstem spontaneous activities. These excitatory networks would constitute re-excitatory loops which conmbute to the generation of tonic background synaptic activities. Quantitative and qualitative modelling aided by computer technology have made important contributions to the understanding of the nervous system. Simulations submitted to biological constrainls of structural and dynamical plausibility can shed light on how the assembly of complex neural units behaves. This led us to simulate fully excitatoly neural networks. We investigated the conditions which led to tonic or phasic activities in excitatory networks. Special attention was paid to the possibility to transform one type of behavior to the other through the modification of a single control parameter. METHODS The modelNeural networks were simulated using a connectionist model with high biological plausibility called Neuro-Bio-Clusters (NBC) [ 1 I]. Simulations were performed using a phenomenological model of the temporal evolution of the neural membrane potential. The NBC neumn behaves according to an integrate and fire model. It takes into account the absolute and the relative refractory period, the post-synaptic potential characteristics, the accomodation and the membrane shunt. Each neuron receives a background Gaussian noise which represents the synaptic noise. The simulationsIn the networks simulated, nll synaptic connections were excitatory. At the beginning of each simulation. the initial conditions were set as if the network had had a random activity for some time. Simulations were done using excitatory networks consisting of 2 to IO00 neurons, and lasted from hundreds of milliseconds to 30 seconds of' simulated time. AnaIysis of the simulation resultsIn this study the overall network activiy was analyzed using the number of f ~ n g neuruns at a given time. This observable was chosen because it represents a relevant physiological parameter. It is also a relevant feature in the study of synchronization. The overall activity represents the ability of the neurons in a network to organize their spike trains into a coherent pattem. Other observabfes such as membrane potential of individual neurons and point process analysis of the spike trains were also available. RESULTS The activity of excitatory networks can be divided into the four following classes: I) Each neuron fires periodically and the overall activity is synchronized (synchronized oscillatory overall activity). 2) Each neuron fires periodically. but the overall activlty is not synchronized (uasynchronized oscillatory overdl activity). 3) Not all neurons Ere periodically, and the overall activity is synchronized (synchronized irregular overall activity). 4) Not all neurons Ere periodically. and the overall activity is not synchronized (unsynchronized irregular overall activity). Ail four classes were observed in the simulations. The type