Research Areas

Basal Ganglia Modeling/Software Hippocampus Hermissenda Learning

Theoretical and Experimental Investigations of the Basal Ganglia

This work is a result of an ongoing collaboration with Dietmar Plenz, Chief, Unit of Neural Network Physiology, Laboratory of Systems Neuroscience at the NIMH and Jeanette Hellgren Kotaleski, Assistant Professor at the Royal Institue of Technology, Stockholm. Our goal is to discover mechanisms whereby dopamine modulates striatal activity. This is of critical importance for understanding the pathophysiology of Parkinson's disease, as well as reinforcement learning.

How does striatal circuitry contribute to action selection? We have determined that synaptic inputs during the down-state average 20-30 per second, and increases to 800 per second, for 0.2-1 sec, during the up-state. The number of inputs is the same for both fast spiking interneurons and spiny projection neurons, though it is likely that many of the glutamatergic synaptic inputs to spiny projection neurons were not detected. Approximately half of the inputs detected were GABAergic, showing a strong contribution by local circuit interneurons. We developed a multi-compartmental model of a fast spiking interneuron, and evaluated the role of the transient potassium (KA) current in signal processing. We showed that signal to noise ratio is enhanced by the KA current, though too much KA current suppresses up-state spikes excessively. Dopamine, by depolarizing the fast spiking interneuron, produces an increase in up-state spikes. We are investigating the role of gap junctions on spike synchronization during up-states.

Is Long Term Potentiation (LTP) the mechanism whereby striatal neurons learn the best motor action? If so, then LTP should be sensitive to the temporal interval of glutamate and dopamine. We are investigating how dopamine activated second messenger pathways interact to produce synaptic plasticity, and how such plasticity depends on temporal pattern of synaptic inputs. We developed a model of intracellular second messenger pathways activated by dopamine and glutamate. Our results show that transient activation of both inputs produces a larger increase in phosphorylated DARPP32 and protein kinase A than either input alone.

Software Development for Modeling Signaling Pathways

We have developed new software for modeling stochastic diffusion. It is analogous to Gillespie's Tau-Leap algorithm. Dendrites and spines are subdivided into sub-volumes. Pre-calculate the probability that one molecule leaves the compartment The probability that k out of N molecules leave a compartment is determined using the binomial distribution. These values are stored in a look-up table, and then the number of molecules leaving the compartment is determined with a single uniform random number. This algorithm is being integrated with the tau-leap algorithm, to create Rapid, Approximate, Stochastic reaction-diffusion software. The computational efficiency of the software will allow modeling second messenger pathways in the dendritic spines of an entire neuron. The figure shows diffusion of IP3 or cAMP in a 20 um x 2 um dendrite with 5 evenly spaced spines. The left panels show the concentration versus time when molecules are initialized in the spine head. On the right, molecules are initialized in the dendrite.

Biological Basis of Memory                   

I am interested in the biophysical and biochemical mechanisms of long term memory storage. The seaslug Hermissenda crassicornis is a valuable animal model because it can be classically conditioned (similar to Pavlov's dogs). Hermissenda learns to associate light with turbulance; the association is stored in the type B photoreceptor as an increase in membrane resistance mediated by a reduction in potassium channel conductance. We use experimental and modeling techniques to investigate critical issues in Hermissenda classical conditioning , such as which intracellular signals are contributed by conditioned and unconditioned stimuli, and which biochemical reactions require that signals are presented in temporal proximity.

Detailed models of the type B photoreceptor soma, rhabdomere and terminal branches have been developed using the GENESIS software. The models includes the biochemical reactions underlying phototransduction, regulation of intracellular calcium concentration and the second messenger cascade leading from GABAB receptor activation to IP3 production. The model also includes synaptic channels, ligand-gated channels, and voltage-dependent ionic channels. Diffusion, calcium release, and biochemical reactions are implemented using Chemesis libraries written in this laboratory. Calcium-gated and ligand-gated channels also are implemented using Chemesis libraries.

Other useful neural modeling software:

NEURON Home Page

XPP (Bard Home Page)

Intracellular recordings in current clamp and voltage clamp mode have been performed to quantitatively characterize the interactions between the light induced signal and the turbulence induced signal that lead to memory storage when these stimuli are presented in temporal proximity. One experiment demonstrated that calcium release through the ryanodine receptor was essential for in vitro classical conditioning. Another experiment measured the light-induced currents, which have a major effect on the excitability of the cell.

Research

Personnel

Publications

Software

Positions

Avrama Blackwell