Schwann cells in the regenerating fish optic nerve: evidence that CNS axons, not the glia, determine when myelin formation begins.

Fish optic nerve fibres quickly regenerate after injury, but the onset of remyelination is delayed until they reach the brain. This recapitulates the timetable of CNS myelinogenesis during development in vertebrate animals generally, and we have used the regenerating fish optic nerve to obtain evidence that it is the axons, not the myelinating glial cells, that determine when myelin formation begins. In fish, the site of an optic nerve injury becomes remyelinated by ectopic Schwann cells of unknown origin. We allowed these cells to become established and then used them as reporters to indicate the time course of pro-myelin signalling during a further round of axonal outgrowth following a second upstream lesion. Unlike in the mammalian PNS, the ectopic Schwann cells failed to respond to axotomy and to the initial outgrowth of new optic axons. They only began to divide after the axons had reached the brain. Shortly afterwards, small numbers of Schwann cells began to leave the dividing pool and form myelin sheaths. More followed gradually, so that by 3 months remyelination was almost completed and few dividing cells were left. Moreover, remyelination occurred synchronously throughout the optic nerve, with the same time course in the pre-existing Schwann cells, the new ones that colonised the second injury, and the CNS oligodendrocytes elsewhere. The optic axons are the only common structures that could synchronise myelin formation in these disparate glial populations. The responses of the ectopic Schwann cells suggest that they are controlled by the regenerating optic axons in two consecutive steps. First, they begin to proliferate when the growing axons reach the brain. Second, they leave the cell cycle to differentiate individually at widely different times during the ensuing 2 months, during the critical period when the initial rough pattern of axon terminals in the optic tectum becomes refined into an accurate map. We suggest that each axon signals individually for myelin ensheathment once it completes this process.