This study describes the localization of γ-aminobutyric acid (GABA), glycine, and glutamate immunoreactive neurons, fibers, and terminal-like structures in the vestibular nuclear complex (VNC) of the frog by using a postembedding procedure with consecutive semithin sections at the light microscopic level. For purposes of this study, the VNC was divided into a medial and a lateral region. Immunoreactive cells were observed in all parts of the VNC. GABA-positive neurons, generally small in size, were predominantly located in the medial part of the VNC. Glycine-positive cells, more heterogeneous in size than GABA-positive cells, were scattered throughout the VNC. A quantitative analysis of the spatial distribution of GABA or glycine immunoreactive cells revealed a complementatory relation between the density of GABA and glycine immunoreactive neurons along the rostrocaudal extent of the VNC. In about 10% of the immunolabeled neurons, GABA and glycine were colocalized. Almost all vestibular neurons were, to a variable degree, glutamate immunoreactive, and colocalization of glutamate with GABA and/or glycine was typical. GABA, glycine, or glutamate immunoreactive puncta were found in close contact to somata and main dendrites of vestibular neurons. A quantitative analysis revealed a predominance of glutamate-positive terminal-like structures compared to glycine or GABA containing profiles. A small proportion of terminal-like structures expressed colocalization of GABA and glycine or glycine and glutamate. The results are compared with data from mammals and discussed in relation to vestibulo-ocular and vestibulo-spinal projection neurons, and vestibular interneurons. GABA and glycine are the major inhibitory transmitters of these neurons in frogs as well as in mammals. The differential distribution of GABA and glycine might reflect a compartmentalization of neurons that is preserved to some extent from the early embryogenetic segmentation of the hindbrain. J. Comp. Neurol. 377:149–164, 1997. © 1997 Wiley-Liss, Inc.

The neuronal organization of the horizontal vestibulo-ocular reflex (VOR) of frogs is reviewed and compared to results obtained from other vertebrates. Extensive similarities in the anatomical, physiological and pharmacological properties of the basic, three-neuronal reflex arc indicate the presence of a conservative, common plan for the VOR of vertebrates and its inhibitory control. To meet the requirements resulting from different locomotor patterns and from misalignments between semicircular canal and extraocular muscle planes, species-specific adaptations of the VOR performance must be present as well. Major differences in the performance of supplementary networks in different species provide a basis for this coadaptation.

Vestibular afferent fibers and their somata in the ganglion of Scarpa colocalize glutamate and glycine in a size-related manner. In this study tritiated aspartate, glycine or GABA was injected in the vestibular nuclear complex of frogs to investigate the uptake by afferent fibers and the retrograde transport of these amino acids to the cell bodies in the ganglion by autoradiographical methods. Ganglion cells were labeled by [3H]aspartate or [3H]glycine but not by [3H]GABA. The intensity of labeling with [3H]glycine increased and the intensity of labeling with [3H]aspartate decreased with cell size. On consecutive semithin sections the immunoreactivity of the same neurons was investigated with antibodies against glutamate or glycine. The results of this combined study showed that smaller, strongly glutamate immunopositive ganglion cells exhibited only weak or no labeling with [3H]glycine whereas larger, less strongly glutamate immunopositive ganglion cells were more intensely labeled with [3H]glycine. A similar size-related labeling pattern was observed in ganglion cells for [3H]aspartate and glycine-immunoreactivity. Both glycine uptake and glutamate immunoreactivity, as well as aspartate uptake and glycine-immunoreactivity, tended to be inversely correlated with the size of a given ganglion cell. These results provide evidence for a specific, size-related uptake of aspartate and glycine and are compatible with our hypothesis that the two amino acids are coreleased by thick but not by thin vestibular afferents. In an accompanying paper [Straka H. et al. (1995) Neuroscience 70, 697-707], we provide evidence for a size-related, monosynaptic activation of different glutamate receptors by vestibular afferent fibers.

Presence and distribution of glutamate, glycine, GABA and beta-alanine in VIIIth nerves of frogs and rats were investigated with postembedding immunocytochemical methods on serial semithin sections. In Scarpa's ganglion of the frog, all cell bodies were glutamate immunoreactive. About 17% of the cells per section were also glycine immunoreactive, but none were GABA or beta-alanine immunoreactive. The mean diameter of glycine-positive cell bodies (26.7 +/- 6.9 microns; N = 130) was significantly (P < 0.0001) larger than that of glycine-negative cell bodies (15.7 +/- 5.4 microns; N = 272). The intensity of glutamate immunostaining decreased with cell diameter, whereas the intensity of glycine immunostaining increased with cell diameter. As a result, the staining intensities for glutamate and glycine in a given cell were negatively correlated. Glycine immunoreactivity was also present in a size-related manner in distal and proximal afferent fibers. The majority of thin fibers (< 4 microns) was glycine negative, whereas most of the thick fibers (> 10 microns) were glycine positive. Glycine-positive fibers were observed in the sensory epithelial of all end organs in the inner ear. The saccular macula and its nerve, however, contained only few glycine immunoreactive structures. In Scarpa's ganglion of the rat, all cells were immunoreactive for glutamate, about 12% for colocalized glycine, and none for GABA or beta-alanine. Glycine-positive cell bodies were significantly (P < 0.0001) larger (32.2 +/- 5.2 microns; N = 82) than glycine-negative cell bodies (25.1 +/- 5.3 microns; N = 274). Cell bodies in the spiral ganglion were only glutamate immunoreactive, whereas staining for glutamate, glycine, and GABA was dense in the ventral cochlear nucleus. These results demonstrate that thicker vestibular afferent fibers represent a particular subpopulation that differs from the majority of thinner afferents due to their glycine immunoreactivity.

Horizontal eye movements in response to vestibular and optokinetic stimulation were investigated in 20 species of marine and fresh water teleosts. Most species performed spontaneous saccadic eye movements in light and in darkness. Successive saccades occurred either at irregular intervals and in different directions (few species), or in bursts of smaller saccades in one direction followed by saccades in the opposite direction (majority of species). Thus, in the latter group the lines of sight shifted cyclically from one side to the other about every 10-15 s. Eye movement responses in goldfish and toadfish were prototypical for two major functional groups. Optokinetic and vestibular nystagmus in the goldfish were much more regular and attained significantly higher gains than in the oyster toadfish. Optokinetic afternystagmus (OKAN) and per-/postrotatory nystagmus lasted about 10-12 s in the goldfish, but only 2-3 s in the oyster toadfish. Bilateral labyrinthectomy in goldfish resulted in optokinetic gains and OKAN durations that were very close to those of intact toadfish. These results indicate the presence (in goldfish) and the absence (in toadfish) of a functioning velocity storage network, respectively. Other non-ostariophysan teleosts failed to show a 'goldfish-like' response pattern as well. Ostariophysan species differed in their reflex performance as well. A 'goldfish-like' response pattern was observed in five of the seven cypriniform species and in two of the four characiform species but not in the four siluriform species. Positively correlated, combined response properties suggest that some species are better equipped for gaze stabilization at higher velocities than other species. These behavioral differences may parallel different specializations in locomotor pattern and mode of life. Absence of a functioning velocity storage network in bottom-dwelling teleosts (as in Amphibia) may be related to the sporadic, slow locomotion of these species and the resulting small requirements for continuous gaze stabilization during self-motion at higher velocities.

Immunocytochemistry was performed on consecutive semi-thin sections of frog cerebellum with antibodies raised against GABA and glycine. Somata of Purkinje cells and stellate cells were immunoreactive for GABA but not for glycine. In the granular layer some neurons were immunoreactive for GABA, fewer for glycine and even fewer for both GABA and glycine. Most of the GABA-positive neurons appear to be Golgi cells. Most of the glycine-positive neurons appear to be displaced cerebellar nucleus neurons. Compared with mammals true basket cells are missing, stellate cells are fewer in number and colocalization of GABA and glycine in Golgi neurons is encountered about ten times less often in the frog than in the rat.