We have developed a technique, called synaptic reconstruction, that permits nerve terminals of living vertebrate neuromuscular junctions (NMJs) to be isolated and then manually recombined with vacant endplate sites to form functional synapses. By reconstructing NMJs with various combinations of pre-and postsynaptic partners, or with varying degrees of pre-to postsynaptic alignment, the functional properties of the three anatomical components of the NMJnerve terminal, endplate, and the alignment between themmay be studied independently. Our experiments thus far indicate, surprisingly, that reconstructed NMJs function nearly normally.

No abstract

1. Motor nerve terminals innervating fibres in the transversus abdominis muscle of the garter snake comprise discrete boutons. Using a combination of enzymatic digestion and mechanical manipulation, individual boutons were removed from living terminals for study in isolation. 2. Boutons freed from terminals were usually allowed to remain in their original location on the endplate ('attached' one-bouton synapse). Alternatively, they were removed from the endplate, and then placed on the same or another vacant endplate site to form a 'reconstructed' one-bouton synapse. When removed from the endplate, boutons were 2-4 ,sm in diameter and nearly spherical in shape, in contrast to the variety of complex shapes seen among boutons still in contact with muscle fibre endplates. 3. Transmitter release was assessed by intracellular recording from the postsynaptic fibre.Boutons produced spontaneous miniature endplate potentials (MEPPs) of nearly normal amplitude; extracellular stimulation elicited endplate potentials (EPPs) which resembled MEPPs. Typical EPP amplitudes fluctuated between zero and five quanta per stimulus. For low-frequency stimulation under normal physiological conditions, mean quantal content, m, averaged 1-4; the binomial number of release sites, n, averaged 2-4; and the binomial probability of release, p, averaged 0 57. Statistics of the quantal fluctuations recorded from single boutons agreed only approximately with predictions of simple binomial theory, the discrepancy being that the apparent number of quanta released exceeded n in 5 % of the events.4. In separate experiments, activity-dependent probes were used to locate rare naturally occurring nerve terminals comprising a single bouton. Activation of these small synapses evoked quantal responses similar to those synapses described above.