Selective retinal reinnervation of a surgically created tectal island in goldfish. I. Light microscopic analysis.

Through anatomical and physiological studies of the regenerating retinotectal projection of goldfish, we sought to determine whether the establishment of a topographic projection is attained through a refinement of an initially less precise pattern of innervation. A 1-mm-wide mediolateral strip of caudal tectum was removed so that a small island of tectal tissue was spared at the caudal pole, and the contralateral nerve was either crushed (TIX) or left intact (TI). The presence of regenerated axons in the ablated zone and the reinnervation of the caudal island were assessed with anterograde and retrograde labeling methods in the following postoperative intervals: early, 20-50 days; middle, 50-110 days; and late, more than 170 days. The anterograde radioautographic method revealed that the appropriate layers of the tectal island became reinnervated by optic axons during the early period. During the middle and late periods, one to several large, discrete bundles bridging the lesion zone along the surface of exposed subtectal structures were readily identified both by radioautography and by anterograde or retrograde labeling following application of horseradish peroxidase to the transected optic nerve or tectal island, respectively. In contrast, the anterograde horseradish peroxidase method did not reveal axon bundles extending caudal to the half-tectum in the absence of a tectal island. Among TIX cases, retrograde horseradish peroxidase labeling of the contralateral nasal retina was more widespread in the middle period than in the late period, a result we interpret as reflecting an improvement in topographical precision with time. The area of retinal labeling among TIX cases in the late period was similar to that following caudal tectal injection in cases with simple nerve crush, although it was still elevated above normal control values. Physiological maps indicated a focal representation of the nasal retina in the tectal island in both periods and did not reveal a transient extreme convergence of retinal input. These findings are discussed in relation to Sperry's chemoaffinity theory.