PALISADE PATTERN OF MORMYRID PURKINJE-CELLS - A CORRELATED LIGHT AND ELECTRON-MICROSCOPIC STUDY

The present study is devoted to a detailed analysis of the structural and synaptic organization of mormyrid Purkinje cells in order to evaluate the possible functional significance of their dendritic palisade pattern. For this purpose, the properties of Golgi-impregnated as well as unimpregnated Purkinje cells in lobe C1 and C3 of the cerebellum of Gnathonemus petersii were light and electron microscopically analyzed, quantified, reconstructed, and mutually compared. Special attention was paid to the degree of regularity of their dendritic trees, their relations with Bergmann glia, and the distribution and numerical properties of their synaptic connections with parallel fibers, stellate cells, "climbing" fibers, and Purkinje axonal boutons. The highest degree of palisade specialization was encountered in lobe C1, where Purkinje cells have on average 50 palisade dendrites with a very regular distribution in a sagittal plane. Their spine density decreases from superficial to deep (from 14 to 6 per mu-m dendritic length), a gradient correlated with a decreasing parallel fiber density but an increasing parallel fiber diameter. Each Purkinje cell makes on average 75,000 synaptic contacts with parallel fibers, some of which are rather coarse (0.45-mu-m), and provided with numerous short collaterals. Climbing fibers do not climb, since their synaptic contacts are restricted to the ganglionic layer (i.e., the layer of Purkinje and eurydendroid projection cells), where they make about 130 synaptic contacts per cell with 2 or 3 clusters of thorns on the proximal dendrites. These clusters contain also a type of "shunting" elements that make desmosome-like junctions with both the climbing fiber boutons and the necks of the thorns. The axons of Purkinje cells in lobe C1 make small terminal arborizations, with about 20 boutons, that may be substantially (up to 500-mu-m) displaced rostrally or caudally with respect to the soma. Purkinje axonal boutons were observed to make synaptic contacts with eurydendroid projection cells and with the proximal dendritic and somatic receptive surface of Purkinje cells, where about 15 randomly distributed boutons per neuron occur. The organization of Purkinje cells in lobe C3 differs markedly from that in lobe C1 and seems to be less regular and specialized, although the overall palisade pattern is even more regular than in lobe C1 because of the absence of large eurydendroid neurons. However, individual neurons have a less regular dendritic tree, there is no apical-basal gradient in spine density or parallel fiber density and diameter, and there are no "shunting" elements in the climbing fiber glomeruli. Purkinje axonal boutons are not substantially displaced and have more but smaller boutons (on average about 70), which are not only contacting eurydendroid and Purkinje cells (about 40 boutons per cell), but also deeply located stellate neurons. As discussed in this study, none of the parameters analyzed is specifically and indissolubly correlated with the dendritic palisade pattern, and its functional significance consequently cannot be explained on the basis of a specific synaptic connectivity pattern. We suggest that palisade dendrites have a similar functional significance as their spines and may be considered as super- or giant spines, subserving optimal tuning of mormyrid Purkinje cells for specific spatio-temporal patterns of parallel fiber activity. Comparison of different types of Purkinje cell organizations as encountered in vertebrates shows two extremes: on the one hand, the mammalian configuration, probably specialized for optimal interactions with climbing fibers, and, on the other hand, the mormyrid palisade pattern, probably specialized for optimal interactions with parallel fibers.