Relation between trains of action potentials across an inhibitory synapse. Influence of presynaptic irregularity

The correspondence between pre- and postsynaptic fluctuations in generation of action potentials was studied using the inhibitory synapse on the crayfish stretch receptor neuron. The presynaptic discharges were imposed by an appropriately controlled stimulator modulated with triangular or sine waves at 1/60 to 10 cps by±1 to 10 ips and around an average rate of 5 to 10 ips. The rate of the regular postsynaptic discharge was set, in the absence of IPSPs, at 5 to 20 ips by the degree of constant stretch on the receptor organ. Cycle histograms were constructed, i.e. discharges were averaged across several cycles, presynaptically on the one hand and postsynaptically on the other. This exploration led to the recognition of some of the rules whereby the correspondence between spike trains is established across a synapse with IPSPs. The mapping or coding from the presynaptic to the postsynaptic cycle histogram was complex, and its main facets were: i) the possibility of the transfer of a broad spectrum of frequencies, at least from 1/60 to 10 cps; ii) a general tendency for faster presynaptic discharges to be associated with slower postsynaptic ones (i.e. a negative overall slope in the presynaptic rate-postsynaptic rate graph); iii) interspersed consistently located segments of appreciable size where inhibitory accelerations led to faster postsynaptic discharges (i.e. positively-sloped “paradoxical” segments); iv) a sensitivity to whether the presynaptic cell was accelerating or decelerating (hysteresis); and, finally, v) minimal consequences of changes around extremes where the receptor was either hardly affected or halted by very low or high rates, respectively (saturation). The type, magnitude and ubiquity of the deviations (iii, iv, v) from a simple and negatively-sloped linearity (suggested by ii) makes them integral and practically important facets of inhibitory transfer. The form of the correspondence depended on such issues as the overall pre- and postsynaptic rates (e.g. 5 versus 10 ips), the modulation frequency (e.g. 1/60 versus 2 cps) and depth (e.g.±1 versus 10 ips), etc. When restrictions were placed upon these variables, the correspondence adopted special forms whose particular descriptions (e.g. linear in- or out-of-phase, rectifier-like, etc.) were acceptable only if conditional to those restrictions. A rapid presynaptic irregularity which increased the varibility of individual cycles without altering the average cycle reduced, sometimes markedly, deviations from linearity. This demonstrates the applicability to the living synapse, represented here by a system whose input and output spike trains are assimilated to point processes, of a concept developed for system whose input or output are continuous: namely, that the addition at the input of a high frequency “dither” reduces the complexities of the element. This effect has interesting physiological possibilities and explains apparent discrepancies between earlier publications.