Probabilistic determination of synaptic strength.

This work was carried on to analyze the presynaptic components of synaptic efficacy, which is designated by the term of synaptic strength. To assess the relations between synaptic strength and innervation density, the properties of unitary Cl(-)-dependent inhibitory postsynaptic potentials (IPSPs) evoked in the potentials (IPSPs) evoked in the goldfish Mauthner (M-) cell by single impulses in individual presynaptic cells, including quantal release parameters, were compared with the histological features of the same neurons, determined from their reconstructions after intracellular injection with horseradish peroxidase (HRP). Because the M-cell is a stereotyped target neuron, comparison of the synaptic strength from different experiments was accomplished by using as a quantitative measure of this parameter the mean unitary IPSP amplitude normalized with respect to the reversal potential for Cl- (i.e., the driving force). In 108 experiments at low stimulus frequency, the majority of the normalized responses (63%) were grouped in a rather narrow range, varying about fourfold, or from 1.5 to 6% of the driving force, with results from stained (n = 46) and unstained (n = 62) cells being the same. In contrast, for the same restricted set of responses, the number of presynaptic terminals (histological n) encompassed a larger range, varying from 3 to 52. Impulses in neurons with quite different complements of terminal boutons could evoke similarly sized, normalized IPSPs, and these two parameters were poorly correlated, with there being, at most, a tendency for the responses to increase with histological n for small values of the latter. Quantal fluctuations in IPSP amplitudes were analyzed according to a binomial model having three parameters, p, which is the probability of release, n, which is the number of releasing units and was previously shown to equal the number of presynaptic boutons or active sites, and q, or the quantal size. The normalized quantal size varied randomly, with a mean value of 0.51% (SD = 0.20) and was relatively independent of n. In contrast, the distribution of p, which ranged from 0.17 to 0.74 (mean = 0.40, SD = 0.155), was skewed to the right; this parameter tended to decrease as a function of increasing n. The normalized unitary inhibitory conductance (g'IPSP) underlying an IPSP is equal to the product of npg'q, where g'q is the normalized quantal conductance.(ABSTRACT TRUNCATED AT 400 WORDS)