Vesicle shape and amino acids in synaptic inputs to phrenic motoneurons: do all inputs contain either glutamate or GABA?

Varicosities that made synapses or direct contacts with retrogradely labelled rat phrenic motoneurons were examined for their content of immunoreactivity for either glutamate or glutamate decarboxylase, the enzyme involved in synthesis of gamma-aminobutyric acid (GABA). Phrenic motoneurons were identified by retrograde tracing from the diaphragm with cholera toxin B subunit conjugated to horseradish peroxidase. Cell bodies and medium-sized to large dendrites were labelled. Preembedding immunocytochemistry identified glutamate decarboxylase-immunoreactive nerve fibres; glutamate-immunoreactive nerve terminals were identified using postembedding immunogold labelling of ultrathin sections. The presence of glutamate- or glutamate decarboxylase immunoreactivity in nerve terminals was correlated with the morphology of the synaptic vesicles. Two major classes of nerve terminals were identified. Nerve terminals with round (presumably spherical) synaptic vesicles (S terminals) comprised 55% of synapses and contacts on phrenic motoneuron somata and 58% of synapses and direct contacts with dendrites. Nerve terminals with flattened synaptic vesicles (F terminals) comprised 42% of synapses direct contacts with somata and 41% of synapses and direct contacts with dendrites. Analysis of immunogold-labelled sections showed that S terminals contained statistically higher levels of glutamate immunoreactivity than F terminals. At the light microscope level, many glutamate decarboxylase-immunoreactive nerve terminals surrounded retrogradely labelled motoneurons. Varicosities with glutamate decarboxylase immunoreactivity made 33% of all synapses and direct contacts on somata, and 33% of synapses and direct contacts with dendrites of the retrogradely labelled phrenic motoneurons. Flattened synaptic vesicles were present in those glutamate decarboxylase-immunoreactive nerve terminals in which synaptic vesicle morphology could be judged. An additional 10% of all nerve terminals were of the F type, but were not glutamate decarboxylase-immunoreactive. Three percent of terminals on somata and 1% of nerve terminals on dendrites could not be classified as S or F types. These findings suggest that more than 90% of all inputs to phrenic motoneuron cell bodies and proximal dendrites could contain either GABA or glutamate. Some of these glutamatergic and GABAergic nerve fibres undoubtedly represent the source of inspiratory drive to, or expiratory inhibition of, phrenic motoneurons.