Development of synapses between identified sensory neurones and giant interneurones in the cockroach Periplaneta americana.

The cercal afferent, giant interneurone pathway in Periplaneta americana was used as a model for synapse formation. The morphology of the two identified filiform hair sensory neurones (FHSNs) and of two giant interneurones (GI2 and GI3) was followed throughout embryogenesis by cobalt injection. The FHSN axons enter the CNS at the 45% stage of embryogenesis, branch at 50% and form complete arborizations by 70%. The giant interneurones send out a primary dendrite at 45%. Secondary branches form between 50% and 60% and elaboration of the branching pattern takes place until 80% embryogenesis. At early stages the FHSN axons are within filopodial range of GI dendrites which may use these sensory processes as guidance cues. Synapse formation between the main FHSN axon shafts and GI dendrites was investigated by injection of the latter with HRP. From 55% to 65% the process is initiated by desmosome-like filopodial contacts, with subsequent vesicle clustering and formation of a small synaptic density. Numbers of contacts did not significantly increase after about 70%, but the number of synapses doubled between 65% and 75%, with each GI process becoming postsynaptic to two FHSN synapses and the presynaptic densities lengthening to become bars. From 75% embryogenesis to hatching there is a further small increase in synaptic bar length. In the first instar GI3 is postsynaptic to both FHSN axons, whereas GI2 forms very few synapses with the axon of the lateral FHSN (LFHSN). This imbalance of contacts is present throughout synaptogenesis, apart from some early filopodial contacts. GI3 forms synapses with the lateral side of the LFHSN axon from 60% embryogenesis but these are totally absent at hatching. The growth of glia along this side of the axon during the last 30% of development appears to be associated with degeneration of synapses in this region. Thus, as the dendrites of the GIs grow to form a miniature version of the adult without loss of branches, there is little evidence of an initial overproduction of FHSN-GI synapses. Similarly there is no evidence that GI2 forms 'incorrect' synapses with the axon of LFHSN. However, GI3 contacts are removed from an inappropriate region of a correct synaptic partner, LFHSN.