Connectivity and convergence of single corticostriatal axons

Abstract

The distribution of synapses formed by corticostriatal neurons was measured to determine the average connectivity and degree of convergence of these neurons and to search for spatial inhomogeneities. Two kinds of axonal fields, focal and extended, and two striatal tissue compartments, the patch (striosome) and matrix, were analyzed separately. Electron microscopic examination revealed that both kinds of corticostriatal axons made synapses at varicosities that could be identified in the light microscope, and each varicosity made a single synapse. Thus, the distribution of varicosities was a good estimate of the spatial distribution of synapses. The distance between axonal varicosities was measured to determine the density of synaptic connections formed by one axon within the volume occupied by a striatal neuron. Intersynaptic distances were distributed exponentially, except that synapses were rarely located <4 microm apart. The mean distance between synapses was approximately 10 microm, so axons made a maximum of 40 synapses within the dendritic volume of a spiny neuron. There are approximately 2840 spiny neurons located within the volume of the dendrites of one spiny cell (Oorschot, 1996), so each axon must contact </=1.4% of all cells in its axonal arborization. Within the same volume there are approximately 30.5 million asymmetric synapses (Ingham et al., 1996), approximately half of which are cortical in origin. Thus, approximately 380,000 cortical axons innervate the volume of the dendritic tree of one spiny cell. Striatal neurons with totally overlapping dendritic volumes have few presynaptic cortical axons in common, and cortical cells with overlapping axons have few striatal target neurons in common. These results explain the absence of redundancy in the responses of neurons located near each other in the striatum.

Fig. 1.

Synapses occur exclusively at varicosities on corticostriatal axons. A, Varicosities as they appear in the light microscope stained by anterograde transport of BDA. The diameters of the intervaricose regions cannot be measured accurately, but the varicosities were as large as 1 μm in diameter.B, High-voltage electron micrograph from a 4-μm-thick section through axons as in A. Intervaricose segments were resolved and measured to be ∼0.1 μm in diameter. Varicosities were between 0.2 and 1.0 μm. C, Thin-section conventional electron micrograph through a varicosity on a BDA-labeled corticostriatal axon. A single synapse was formed with an unlabeled dendritic spine. The diameter of the axon at the synaptic site is ∼0.5 μm. Scale as in D. D, Thin-section conventional electron micrograph through the intervaricose segment of a labeled corticostriatal axon. As was typical for this part of the axon, there were no vesicles present and no synaptic contact. The diameter of the axon is ∼0.1 μm. E, Histogram of the diameters of nonsynaptic vesicle-free profiles of BDA-labeled corticostriatal axons examined in conventional electron micrographs as in Cand D. Axonal diameter was measured on the short diameter of each profile. F, Histogram of diameters of vesicle-containing profiles in thin sections measured inE.

Fig. 2.

Distribution of boutons along axonal branches in the patch and matrix and in single axons arborizing in extended and focal patterns. A, Drawing of a focal arborization from a single corticostriatal neuron stained by intracellular injection of biocytin. In the drawing, the arborization was projected onto the sagittal plane. Twenty-two boutons were present in this arborization.B, Drawing of the axon of another corticostriatal neuron arborizing in the extended manner, shown at the same magnification and in the same plane. This axonal arborization contained 935 boutons.C–F, Distribution of distances between nearest neighbor boutons along corticostriatal axons. The straight-line distance between nearest neighbor boutons along axonal branches was measured to make these histograms. Individual axonal branches seldom approached each other (the appearance that they do in the drawings is caused by the projection onto two dimension), so measurements along branches accurately represent absolute nearest-neighbor relationships. For each histogram, an exponential curve has been fit to the tail of the histogram (ignoring the bins preceding the peak of the histogram) and is shown as a solid line. The decay constant for the best-fitting exponential is indicated for each histogram (λs). This would be the mean spacing between boutons if the distribution were truly exponential. The actual mean spacing is also indicated for each histogram ($\overline{X}$). The difference between these is primarily attributable to the deviation from an exponential distribution near zero. In all cases, there were fewer interbouton segments shorter than 1–4 μm than expected on the basis of the exponential distribution. C, Histogram for the focal axonal arborization shown in A.D, Histogram of interbouton spacing for the extended arborization shown in B. E, Histogram of spacing among 1425 boutons measured from corticostriatal axons arborizing in the matrix compartment (mostly extended type) in sections from BDA injections from four animals. F, Histogram of the patch compartment (mostly focal) calculated from four animals as inE.

Fig. 3.

Three different corticostriatal convergence patterns. A, In the patches (striosomes) and perhaps in the matrix focal arborizations (matrisomes), focal corticostriatal axonal fields and spiny neuron dendrites obey the patch boundaries and are approximately the same size as the patch cross-section. In this arrangement, all connectivity is discontinuous, and every cell in the patch has access to all the axons arborizing there. The number of possible synaptic connections on one spiny neuron is merely the number of synapses made in the patch, and there is a direct trade-off between that number and the degree of input sharing in the patch.B, The extended axonal fields of some corticostriatal neurons take a straight course through their large arborizations, with branches separated so that a single neuron is crossed by only one branch. The number of possible synaptic contacts is determined by the interval between synapses along the axon and the diameter of the dendritic tree of the spiny cell. C, Unlike the situation in the striosomes, the focal axonal arborizations of corticostriatal cells in the matrix may not be totally overlapping, and spiny cells may not observe boundaries of matrisomes. In this hypothetical case, a continuous topography is possible.