Invertebrate presynaptic inhibition and motor control

ConclusionPresynaptic inhibition is a widespread mechanism among invertebrates and its study has developed quite independently from that which has occurred in vertebrates. However, it is striking that recent studies on presynaptic inhibition in sensory afferents of Arthropod have shown great similarities with presynaptic control exerted in the mammalian spinal cord primary afferents (Gossard et al. 1989; Rudomin et al. 1991): in both cases, presynaptic inhibition of sensory terminals involves inhibitory PADs associated with GABA, operating via Cl- ion channels. Sensory modulation exerted in this way may affect the reflex gain. Paired recordings on invertebrate preparations have made it possible to explain some of these mechanisms in greater detail, as was the case in lamprey (Alford et al. 1991).Presynaptic mechanisms and particularly presynaptic inhibition have been studied mainly in neurons where the spike-initiating zone (initial segment of cell body in classic vertebrate neuron or the equivalent in the neurite of invertebrate neuron) is far from the output synapse. In such neurons spikes are actively conveyed in axons before they reach the output synapse. This is the case for primary afferents and for descending INs. In such cases presynaptic inhibition exerts a clear-cut effect onto spike-triggered synaptic transmission. However, a given axon generally divides into several branches, and therefore presynaptic inhibition activity onto some branches may not affect others. This possibility is likely to exist in local INs and is important to consider, in view of our understanding of how neuronal networks achieve their observed behavior. In the past, such networks were considered as input-output reflex chains; more recently, active properties that are essential for pattern generation (i.e., voltage-dependent and calcium-dependent conductances responsible for nonlinear properties of the membrane potential such as pacemaker potential and plateau properties), have been shown to exist in the component neurons; even more recently, connectivity as well as active properties on neurons have been demonstrated to be controlled by INs that could de novo rebuild new networks with specific functional role (Meyrand et al. 1994); the time has now come to consider that each neuron is not behaving as a single unit, but is rather made up of different compartments subjected to different local controls and local active properties (Coleman and Nusbaum 1994). This idea has already been supported by the demonstration that some neurons had several spike-initiating zones; it is reinforced by the finding that target cells are capable of presynaptically modulating their input fibers (Nusbaum et al. 1992). These findings seriously complicate the functional schema we had drawn of neuronal networks. However, this is certainly one of the most important challenges for the next decade, in view of understanding how the brain works.