Biophysical models of the striatum and basal ganglia
The systems level model of the basal ganglia (Gurney et al, 2001a, Gurney et al, 2001b) has proved successful in explaining the gross anatomy of the basal ganglia in terms of a new functional architecture that implements an action selection mechanism. However the model is restricted in the range of behavior it can capture to those behaviors that emerge as a result of the principal excitatory and inhibitory inter-nucleus interactions and, to a lesser degree, intra-nuclei interactions. In order to capture behaviors that emerge from lower level, biophysical and micro-anatomical mechanisms, it is necessary to move the model to a lower level of description. T he long term goal of this project is to implement a computational model of the basal ganglia, highly constrained by two things: the anatomical and physiological data available at this lower level of description, and the computational hypothesis that the primary function of these nuclei is to act as an action selection mechanism (Redgrave et al, 1999). As the model progresses we aim to answer two major questions. 1) is the functional architecture for action selection (discovered at the systems level) maintained at a lower level of description, and 2) what benefits are gained, in terms of the action selection mechanism, by including more anatomical and physiological detail in the model. The current focus of this project is to implement a detailed biophysical model of the striatum.
The striatum is the major source of input to the basal ganglia with most areas of the cortex and limbic systems provide collaterals to these nuclei. In relation to the other nuclei of the basal ganglia the striatum is by far the largest, comprising over 90% of the total cell count. Although there is not yet a complete description of striatal intranuclear connectivity, the basic structure is now well defined. The striatum consists principally of medium spiny cells which account for over 90% of the neurons within the nuclei of the rat (Gerfen and Wilson, 1996). These can be further subdivided into different populations based on their target nuclei, the predominant type of dopamine receptor expressed, and whether they fall into the patch of matrix compartments of the striatum. These cells also exhibit some novel physiological features giving rise to a dual state behavior know as the up-down state. The other 5-10% of the neural population within the striatum consists of interneurons. So far several different species have been identified including the fast spiking cells, large cholinergic aspiny cells, and low threshold spike cells (Kawaguchi, 1997).
Work is currently progressing on a compartmental model of the medium spiny neuron that includes a non-inactivating K current, an anomalous rectifier, fast and slow A currents, Hodgkin-Hulxey Na and K currents, and a calcium dependent K current. The intention is to use this model to explore the mechanisms that produce the dual state behavior, and to use it as a reference to implement simplified, quasi spiking, compartmental models for network simulations. We also intend to implement models of the fast spiking and large cholinergic interneurons using a similar approach.