Ultrastructural organization of regenerated adult dorsal root axons within transplants of fetal spinal cord

It has previously been demonstrated that the severed central branches of adult mammalian dorsal root ganglion cells regenerate into transplants of fetal spinal cord. The aim of this study was to determine whether these regenerating axons form synapses, and, if they do, to characterize them morphologically. Embryonic day 14 or 15 spinal cord was transplanted into the limbar enlargement of adult Sprague‐Dawley rats, and the L4 or L5 dorsal root was cut and then juxtaposed to the transplant. One to 3 months later the regenerated dorsal roots were labeled by anterograde filling with wheat germ agglutinin‐horseradish peroxidase (WGA‐HRP) or by immunocytochemistry for calcitonin gene‐related peptide (CGRP). Dorsal root labeling with WGA‐HRP demonstrated that regenerated axon terminals made synaptic contacts within transplants, and stereological electron microscopic analysis demonstrated that CGRP‐immunorective axon terminals occupied an average of 9% of the neuropil within 2 mm of the dorsal root‐transplant interface. The majority of synapses were axodendritic, but a significant percentage were axosomatic or axoaxonic. Since axoaxonic synapses were observed in tranplants in which both pre‐ and postsynaptic profiles of axoaxonic synapses were labeled for CGRP, some regenerated axons apparently form synapses with each other. Approximately 90% of synaptic contacts were simple, 9% were complex, and 25% of the complex terminals were immunopositive for CGRP. Glia occupied 25% of the neuropil within 1 mm of the dorsal root‐transplant interface, but only 6% of the neuropil 1–2 mm from the interface. We also performed a stereological analysis of the neuropil in lamina I. The area fractions of neuropil occupied by myelinated axons, perikarya, and dendrites were similar in transplants and in lamina I. However, the area fraction occupied by unmyelinated axons was significantly smaller in transplants, and the area fraction occupied byu axon terminals was significantly larger in transplants compared with lamina I. Regenerated CGRP‐immunoreactive synaptic terminals in transplants were significantly larger than in normal lamina I, and their synaptic contact length was also increased, suggesting that a compensatory mechanism for increasing synaptic efficiency might occur within the transplants. Synaptic density, however, was significantly reduced in transplants, indicating a smaller number of synaptic terminals per unit area. In lamina I, as in the transplant, most synapses were axodendritic, but the percentage of axosomatic and axoaxonic terminals was lower in lamina I than in the transplants. The area occupied by glia in lamina I was similar to that observed 1–2 mm from the dorsal root‐transplant interface, but lower than that observed 0–1 mm from the interface. The results of this study show that regenerated primary afferent axons form synapses within transplants. These synapses retain many of the characteristics of primary afferent synapses in normal dorsal horn, suggesting that transplants may provide a strategy to restore some of the properties of damaged neural circuits.