Modeling of bursting mechanisms and coordination in a spinal central pattern generator

Abstract: Mechanisms underlying lotal bursting as well as coordinationbetween different levels of a spinal CPG generating locomotionhave been investigated using computer simulations. A"primitive" jawless vertebrate, the lamprey, is used a.s aprototype model. Most simulations have been conducted using abiophysical neu ron model built on the Hodgkin-Huxley formalismand equipped with Nu+, K+,Ca²+, Kca, LVACa²+ and NMDA activated channels. Inhibitory andexcitatory AMPA/kainate and NMDA synapses are modeled as timedependent conductances with appropriate reversal potentials.For tomparison, Morris-Letar oscillators as well as adaptingleaky integrator-like units are also used.The basic identified building blocks of the CPG, generatingalternating left right burst activity, tonsist of ipsilaterallyprojecting excitatory neurons (E) and contralaterallyprojecting inhibitory neurons (C). The model neurons are connected in the same way ss has been established experimentally.Sinte several complementary mechanisms may play a role, thepotential of two different neural mechanisms have been exploredwhich can provide burst activity at the segmen tal level, andintersegmental coordination. When alternating left-rightactivity is produced through an escape-like mechanism the quietside is able to become ac tive despite ongoing inhibition fromthe contralateral side. Reciprocal inhibition is then a crucialburst terminating factor. Burst frequency is strongly affectedby the effective inhibition and the drive to escape fromongoing inhibition. Several factors influence this process. Kcacurrents control spike frequency on the active sideand also a post-burst hyperpolarization on the inactive side.Postin hibitory rebound properties, carried by e.g. low voltageactivatedCa²+ currents further can promote escape. Phasicipsilateral excitation and NMDA membrane properties stabilizethe rhythm, especially in the lower frequency range. Severalexperimental observations can be explained based on the effectthese different factors have on effective inhibition andtendency for escape.Bursting can, however, also be produced by a networkdeprived of inhibition, showing that powerful burst terminatingmechanisms not requiring inhibition exist. In the model withbiophysically detailed neurons such a mechanism could beactivation ofKcacurrents due to accumulation ofCa²+ during the active phase. As shown innon-spiking, as well as biophysically detailed models, aconstant burst proportion over a wide frequency range can beachieved by modulation of the rel ative strength of adaptationin such networks. The left-right inhibition causes left-rightalternation but may not affect the frequency of bursting.When both types of lotal oscillatory networks are extendedlongitudinally, a rostral to caudal phase delay is producedwhen caudal projections are extended further than the rostralenes. However, the excitatory versus inhibitory projec tionsmay have different roles in the two alternative models. Thisrelative phase delay expressed as % of cycle duration,increases in general with frequency. The simulations suggestthat the conditions at the ends of the simulated chain arecritical for the resulting phase lag. The capability ofbuffering against frequency variations and rapid adjustmentsfollowing perturbations is discussed and com pared with chainsof relaxation oscillators and phase-coupled oscillators.Keywords:adaptation, central pattern generator,computer simulation, inter segmental coordination, lamprey,locomotion, neural network, rhythmogenesis