In 1952 Fatt and Katz recorded at a frog neuromuscular junction while stimulating the nerve and found ''... that successive endplate potential responses varied in a step-like manner, corresponding to units of miniature endplate potentials'' (J Physiol 117, 109-128). This led them to propose that fast neuromuscular transmission is 'quantal'. Quantal release is now commonly ascribed to a vesicular form of neurosecretion since vesicles have routinely been visualized in presynaptic terminals. The vesicular hypothesis (Del Castillo and Katz, 1955) assumes that quanta, or 'transmitter packets of standard size', are assembled and stored in the numerous vesicles routinely identified in micrographs of virtually all central and peripheral presynaptic nerve terminals. Simply stated, this model predicts that each one of the miniature synaptic signals (MSSs) follows from the exocytosis of one vesicle's contents. However, the time required for membrane fusion preceding exocytosis (Almers and Tse, 1990) and the variability in MSS amplitude and time course (Vautrin et al, 1992a,b) cannot readily be reconciled by a simple, exocytotic model of quantal release from preloaded vesicles. These difficulties with the original model have led us to re-evaluate MSSs generated at the classical peripheral synapse. the cholinergic neuromuscular junction of the mouse diaphragm, as well as at central synapses between embryonic hippocampal neurons mediated by gamma-aminobutyric acid (GABA). At these synapses, the release of GABA is also assumed to have classical quantal properties like peripheral acetylcholine release (Edwards et al, 1990). Our results show that at both synapses, progressive alterations in elementary signal properties can be induced in a remarkably rapid manner. The original report of preferred amplitudes and intervals in the spontaneous miniature signals (Fatt and Katz, 1952) has repeatedly been confirmed and is here incorporated into a dynamic model of fast synaptic transmission. Although MSSs exhibit variable rise-times and peak amplitudes, they can both be described in terms of synchronization of transmitter release. We have reviewed many experimental findings, which together strongly suggest that the original interpretation of Fatt and Katz (1952) regarding MSSs as reflecting the non-propagated 'neurogenic' activity of 'terminal spots' may be a useful concept to pursue since it may help to explain part of the underlying molecular basis of quantal release. As a working hypothesis to explain MSS properties we propose that the local intra-terminal Ca2+ transients recently recorded with digital video microscopic recording techniques (Llinas et al, 1991 ; Melamed et al, personal communication) embody the regenerative activity that directly controls the secretion of transmitter to produce either evoked or spontaneous quantal release. Both extracellular and internally-stored Ca2+ would participate in the regenerative formation of the cytosolic free Ca2+ transients that would determine transmitter release kinetics. The relative rate of synchronization in a variable number of elementary Ca2+ transients may generate central and peripheral MSSs with preferred amplitudes and rise-time characteristics being set by frequency modulation of variable numbers of elementary transmitter liberations.
