Abstract
We have used computer simulations as one way to test the hypothesis that locomotor rhythm production for swimming in frog embryo spinal cord depends on rebound from inhibition and is sustained by mutual re-excitation among spinal excitatory interneurons. All simulations were based on physiological and anatomical data on the neurons and circuitry of Xenopus embryo spinal cord. Model 'neurons' had resistively coupled axon, soma, and dendrite compartments. Membrane properties were based on Hodgkin - Huxley equations with resting potential at - 75 mV and where soma and dendrite had reduced K+ and Na+ conductance and slowed K+ conductance. These 'neurons' fired a single non-overshooting spike both to depolarizing current and after hyperpolarizing current given during imposed depolarization. Synapses were made on to the dendrite. Inhibitory and excitatory synaptic channels had Nernst potentials of - 80 and 0 mV, time constants for opening of 1 ms, and closing of 6 and 75 ms. When the short inhibitory postsynaptic potential occurred on the long (N-methyl-D-aspartate-type) excitatory postsynaptic potential, it led to rebound firing. A four 'neuron' symmetrical network was built with reciprocal inhibition and where excitatory 'neurons' re-excited themselves and the inhibitory 'neuron' on their own side. The rhythmic alternating activity with one spike per cycle produced reliably by this network was self-sustaining, initiated by a brief synaptic input, and closely resembled the spinal cord motor pattern during swimming. The robustness of this activity pattern was investigated by varying cellular and synaptic parameters, initiating inputs, and network connectivity. We conclude that cellular, synaptic, and network properties are all important and that mutual re-excitation, a form of positive feedback, could sustain motor rhythm production in the Xenopus embryo spinal cord.