Computational Models of Ictogenesis: Synaptic Depression, Recovery, and Connectivity
Abstract number :
3.057
Submission category :
1. Translational Research: 1B. Models
Year :
2016
Submission ID :
195650
Source :
www.aesnet.org
Presentation date :
12/5/2016 12:00:00 AM
Published date :
Nov 21, 2016, 18:00 PM
Authors :
Theju Jacob, Massachusetts General Hospital & Harvard Medical School, Charlestown and Kevin J. Staley, Massachusetts General Hospital & Harvard Medical School, Charlestown
Rationale: The chain of events that lead to seizure onset remains a mystery, despite many decades of investigation. The parameters that may elucidate ictogenesis are not easy to evaluate experimentally. In our work, we have created a neural network model of the CA3 region of the hippocampus. Our network displays spontaneous transitions between epochs of interictal spiking and spontaneous seizures. We have characterized how different network parameters influence the generation of spikes and seizures in our model. Methods: Our model is comprised of 100 x 100 pyramidal cells, interspersed with a 20 x 20 array of interneurons. We use the MacGregor (integrate and fire) model for our neurons, adding synapses that undergo activity-dependent short-term depression and recovery. The amount of glutamate released at a synapse depends on: 1) probability of release of a glutamate vesicle and 2) the number of releasable glutamate vesicles currently available at the synapse. Both spontaneous release and activity dependent release of glutamate are supported by the model. We also model release of GABA at inhibitory synapses. Every neuron is connected to a neighborhood of surrounding neurons, using several different strategies for synaptic connectivity. Keeping the total number of inhibitory and excitatory synapses constant, we tested 3 models of network connectivity. In the first model, called the uniform connectivity model, every neuron is synaptically connected to neurons in its local neighborhood, with the connection probability falling exponentially with distance. In the second model, called the small world network model, the majority of connections are local. Additionally, a small percentage of connections are long range connections. The third category is scale free networks. In scale free networks, the number of synaptic connections of each neurons follows a power law, so that some neurons have a great many connections, but most neurons have few. The networks were constructed with stochastic synaptic connections that followed one of these three strategies prior to each simulation. Standard network-connectivity algorithms were employed for all three synaptic connectivity strategies. In this work, we are investigating 1) the role that network connectivity strategies play in seizure generation 2) the role of spontaneous glutamate release probability in seizure and spike frequency. Results: Our algorithms reliably create neural networks capable of generating population spikes as well as seizures, and spontaneously transitions between the two states. In 25 simulations of 60 sec duration for a total of 25 minutes of simulation for each strategy, we found that 1) small world connectivity was conducive for seizure generation (8 seizures observed). We have not observed spontaneous seizures in networks connected uniformly or with scale free connectivity. Uniform networks could develop brief ictal events with properly timed and targeted exogenous activation. 2) higher spontaneous glutamate release probability led to higher number of seizures and interictal spikes. Conclusions: Small world network connectivity is most conducive to seizure generation over the range of connectivity parameters and transmitter release probabilities that we have analyzed to date. Analyzing connectivity in real neural networks is important for understanding ictogenesis. We will provide measures of the accuracy of current connectivity estimates based on correlated neuronal activity as demonstrated in In a companion poster by Kyle Lillis. Funding: NIH/NINDS 5R01NS086364-03 NIH/NINDS 2R37NS077908-05A1
Translational Research