Our research is completely theoretical and uses computational methods to model the biochemical and electrophysiological behavior of synapses, astrocytes, neurons and microcircuits in cerebellum and hippocampus. These models span from nano to single cell to (connected) networks spatial scales. We also perform analysis of experimental data provided by external collaborators.
All our research is supported by in house development of dedicated software.
We use molecular modeling methods to study how signaling pathways and cell biology processes involved with synaptic plasticity are influenced by the detailed morphology of neurons and by the stochastic behavior of the reactions due to the small number of molecules. To support this work we develop a simulator program called STEPS (1) that implements the spatial Gillespie SSA in a tetrahedral grid and is meant to simulate molecular events in spines or small parts of a neuronal dendrite or axon. An extension of STEPS allows for calculation of the membrane potential on the same tetrahedral mesh so that voltage-gated ion channels can also be simulated. Recently we have succeeded in effectively parallelizing the spatial SSA (2) and STEPS (3), this allows us to simulate a complete neuron or astrocyte at the nanoscale. Recent versions of STEPS support distributed meshes (4) and spatial modeling of vesicles (5).
We have been using STEPS to study anomalous diffusion in Purkinje cell spiny dendrites (6) and the vesicle cycle during synaptic release in a hippocampal terminal (7). More recently we have developed an integrated molecular model of cerebellar long-term depression and long-term potentiation (8). Work on modeling AMPA receptor trafficking in hippocampal spines is ongoing.
Erik De Schutter is well known for the detailed model of the Purkinje cell he developed in the past (9). This model replicates the complete morphology and electrophysiology of the neuron and has demonstrated strong predictive power (10). We have now developed a new Purkinje cell model that incorporates recent data about voltage-gated channels and use it to investigate the complex spike (11). This model replicates the firing-rate dependence of the Purkinje cell Phase Response Curve, enabling firing-rate dependent oscillations (12) and parallel fiber evoked dendritic calcium spikes support multiplexed coding by the Purkinje cell (13).
We have extended our stochastic modeling to the cellular level and using the STEPS simulator (1) we demonstrated that the variability of dendritic calcium spikes, which has been observed experimentally, is caused by stochastic calcium mechanisms (14). This work links stochasticity at the molecular level with cellular properties. We are now extending this model to simulating a full Purkinje cell at the nanoscale level (15). The importance of morphological properties of astrocyte branches on calcium signaling was simulated (16).
More in general we are interested in the importance of neuronal morphology and excitability for function. We showed that the type of excitability a neuron expresses determines its type of network correlation (17,18), an important correction of the literature on the subject.
We are extending analysis of morphology to the network level: what are the properties of the forest of dendritic trees? This is a step towards modeling the development of neuronal morphology using environmental clues. To support such modeling we have developed the NeuroDevSim software (19), a more perfomant successor to NeuroMac (20). We are used NeuroDevSim to model the early development of cerebellar cortex (21).
We have a strong interest in cerebellar oscillations (22) which we continue to investigate in detailed 2D and 3D network models of cerebellar cortex (23). These network models comprise simplified models that capture the excitability of the neurons and detailed connectivity based on real anatomy.
The output of these network simulations are spike trains and we like to compare these to recordings of simple and complex spikes from Purkinje cells in rodents (24) and monkeys, which we analyze using sophisticated statistical methods. This has resulted in the discovery of a relationship between the duration of complex spikes in Purkinje cells and their firing rates (25) and of multiplexed coding (18) by simple spikes (26). Such multiplexed coding turns out to be also a property of our new Purkinje cell model (13). A recent study of manifolds in monkey cerebellar cortex demonstrated that Purkinje cells expand the variability of the mossy fiber input (27).
