The effects of presynaptic inhibition on quantal release of transmitter were investigated at neuromuscular junctions of the motor axon supplying one of the limb muscles of a crab (Pachygrapsus crassipes). 2. Binomial analysis of transmitter release recorded at selected neuromuscular junctions with an extracellular 'macro-patch' electrode indicated high probability of release (p) from a limited number of available sites (n). During presynaptic inhibition, both n and p were reduced. 3. The binomial model provided a good description of results from non-inhibited junctions. During presynaptic inhibition, results from some junctions could be described by the binomial model, while those from other junctions could not. An interpretation of this finding is that presynaptic inhibition differentially affects the probability of release at various release sites of the neuromuscular junctional complex. 4. A morphological study of the region of transmitter release under the macropatch electrode was made. Release-dependent uptake of horseradish peroxidase (HRP) into presynaptic terminals was restricted to the region under the recording electrode, by perfusing the preparation with calcium-free solution containing HRP. Transmitter release, and HRP uptake, occurred only at the site of the electrode, which was filled with a calcium-containing solution. Subsequently, serial sections were prepared for electron microscopy and the region of transmitter release was reconstructed. 5. Numerous axo-axonal synapses were found in the HRP-labelled region. Thus, the morphological prerequisite for presynaptic inhibition exists at the site of transmitter release, and not exclusively at a more remote region. 6. The number of morphologically identified excitatory neuromuscular synapses exceeded the 'release sites' estimated from the binomial model (n) by a wide margin. Morphological differences among synapses were observed. It is proposed that not all morphologically identified synapses participated in transmitter release under the experimental conditions employed. Thus, morphologically defined synapses are likely to be non-uniform in their response properties, including probability of transmitter release (p).
