Retinotopic refinement of the regenerating goldfish optic tract is not linked to activity-dependent refinement of the retinotectal map.

Abstract

The cut optic nerve of a goldfish can regenerate, restoring an orderly projection from the retina to the optic tectum. At first, regenerating axons make transient connections, many of them in inappropriate tectal locations. Later, their arrangement is gradually refined into an accurate retinotectal map by a process that depends on afferent activity. On their way to the tectum, many regenerating axons make erroneous choices between the two arms (brachia) of the optic tract. However, since they commonly possess divergent collateral branches, a secondary refinement of the brachial pattern can occur by selective collateral elimination. How or why a particular collateral is lost is not known, but we have previously suggested that sibling branches might compete to form stable tectal synapses, implying that there might be a causal link between refinement of the brachial pattern and refinement of the retinotectal map. In this paper, we have tested directly for such a link, blocking map refinement with tetrodotoxin (TTX) or stroboscopic light, verifying the effectiveness of the block and measuring the extent of brachial refinement by standard methods in experimental and control fish. Both TTX and stroboscopic light reliably prevented map refinement, their results being indistinguishable. However, neither had even the slightest detrimental effect on brachial refinement, either 42 days or 70 days after nerve cut. Evidently, neither activity nor a sharp retinotectal projection is necessary for brachial refinement. Theory and experiment both dictate that the basic projection pattern be controlled by a mechanism (such as chemoaffinity) that is independent of activity, and it would seem that selective collateral loss must depend on the same mechanism.