Experimental and Computational Methods

The underlying strategy in the lab is to combine computer modeling and experimental investigations of neural properties to address the question of how computation in specific brain circuits might occur.  We construct neuron models with the GENESIS software package under Linux to obtain computer simulations that build directly on experimental measurements of neurons.  The underlying theory is called compartmental modeling and is largely built on the seminal cable theory by Wilfried Rall.  The models we build are designed to match the dynamics of the original neuron, and the membrane potential of the model in various states can be directly compared with experimental results.

To build a single neuron model we first obtain whole cell recordings from rat brain slices to characterize basic properties the model needs to show. Neurons are visualized with infrared optics using a water immersion objective.  We use glass pipettes to obtain intracellular electrical recordings from visualized cells in the whole cell configuration.  Different experimental manipulations we use include application of specific pharmacological blockers to isolate specific voltage-gated currents in our neurons, and electrical stimulation of input pathways to examine synaptic properties.  Most important, we use the technique of dynamic current clamping to apply artificial synaptic inputs in vitro.  This technique allows us to study the input/output processing of neurons in controlled conditions, while we mimic the input condition as it might occur in the behaving animal.  Synaptic conductance waveforms are typically derived from computational models and dynamic current clamping allows us to directly compare model neurons and neurons recorded in vitro.

The compartmental models we make use the full morphology of recorded cells.  Thus, we first need to histologically stain and reconstruct recorded neurons. The Neurolucida software (MicroBrightField, Inc) allows us to obtain an accurate computer representation of stained cells.  We then transform the Neurolucida morphology files into a format suitable for GENESIS simulations using the cvapp software.

We also study the electrical behavior of neurons in anesthetized rats both with intracellular and extracellular recordings.  These recordings allow us to examine the actual activity patterns of neurons in vivo, and provide the goal state of our simulations.  In addition, we can use the recorded spike trains as input sequences to our models and to our dynamic clamp studies