• E A Thomas and J C Bornstein.
• Department of Physiology, University of Melbourne, Parkville 3010, Australia. e.thomas@unimelb.edu.au
• Neuroscience. 2003 Jan 1; 120 (2): 333-51.

AbstractRecurrent networks of neurons communicating via excitatory connections are common in the nervous system. In the absence of mechanisms to control firing (collectively termed negative feedback), these networks are likely to be bistable and unable to meaningfully encode input signals. In most recurrent circuits, negative feedback is provided by a specialized subpopulation of interneurons, but such neurons are absent from some systems, which therefore require other forms of negative feedback. One such circuit is found within the enteric nervous system of the intestine, where AH/Dogiel type II neurons are interconnected via excitatory synapses acting through metabotropic receptors to produce slow excitatory postsynaptic potentials (slow EPSPs). Negative feedback in this recurrent network may come from either inhibitory postsynaptic potentials arising from the terminals that produce slow EPSPs or from the after hyperpolarizing potentials (AHPs) characteristic of these neurons. We have examined these possibilities using mathematical analysis, based on the Wilson-Cowan model, and computer simulations. Analysis of steady states showed that, under appropriate conditions, both types of negative feedback can provide robust regulation of firing allowing the networks to encode input signals. Numerical simulations were performed using large, anatomically realistic networks with realistic models for metabotropic transmission and suppression of the AHP. In the presence of constant exogenous input, parameters controlling aspects of synaptic events were varied, confirming the analytical results for static stimuli. The simulated networks also responded to time varying inputs in a manner consistent with known physiology. In addition, simulation revealed that neurons in networks with inhibitory contransmission fired in erratic bursts, a phenomenon observed in neurons in unparalysed tissue. Thus, either inhibitory contransmission or AHPs, or both, can allow recurrent networks of AH/Dogiel type II neurons to encode ongoing inputs in a biologically useful way. These neurons appear to be intrinsic primary afferent neurons (IPANs), which implies that the IPANs in a region act in a coordinated fashion.

17 keywords...

hide…