Mechanism for the formation of synaptic projections in the arthropod visual system

Abstract

THE axons of photoreceptors in arthropods project in a precisely ordered manner from the retina to the lamina. As retina and lamina consist of two-dimensional arrays of structural units (the ommatidia and the cartridges, respectively), a particular unit of origin and a specific target unit can be defined uniquely for each axon. In a number of species, such as water fleas¹, crabs² and locusts³, all axons originating from an ommatidium terminate in the same cartridge, where they form synapses onto laminar neurones. How is a particular target cartridge selected by the retinular axons as they grow into the laminar anlage during embryogenesis? The adult pattern of projections could arise in several ways. For example, each ommatidium and each cartridge could have unique chemical labels varying with or according to their positions in their respective domains. Input and target could then be matched by a chemoaffinity mechanism such as that originally proposed by Sperry⁴ for the formation of vertebrate retinotectal connections. A second possibility is that retinular fibres do not recognise any specific labels in the laminar cells, but grow into this region in the appropriate spatio–temporal order to sequentially fill available sites in a pre-formed laminar array. In a variation of this second mechanism the lamina would be unstructured at the time of arrival of retinular axons and the axons would impose a pattern on the lamina by sequentially recruiting immature laminar neurones to form the regular array of cartridges. These possibilities have been tested by deleting ommatidia at specified developmental stages in Daphnia magna, a small branchiopod crustacean with a total of 22 ommatidia and 22 cartridges¹. The observations reported here support the last hypothesis, that the order of retinular axon ingrowth determines the retino–laminar projections in this organism.