Target-independent diversification and target-specific projection of chemically defined retinal ganglion cell subsets

In diverse vertebrate species, defined subsets of retinal ganglion cells (RGCs, the neurons that project from retina to brain) are distinguishable on the basis of their dendritic morphology, physiological properties, neurotransmitter content and synaptic targets. Little is known about when this diversity arises, whether diversification requires target-derived signals, and how subtype-specific projection patterns are established. Here, we have used markers for two chemically defined RGC subsets in chick retina to address these issues. Antibodies to substance P (SP) and the nicotine acetylcholine receptor (AChR) beta 2 subunit label two small ( < 10%), mutually exclusive groups of RGCs in mature retina. SP and AChRs accumulate in distinct RGCs before retinotectal synapses have formed. Moreover, both populations of RGCs form in retinae that develop following tectal ablation or transplantation to the coelomic cavity. Thus, RGC subsets acquire distinct neurotransmitter phenotypes in the absence of extraretinal cues. In the mature optic tectum, SP- and AChR-positive RGC axonal arbors are confined to distinct retinorecipient (synaptic) laminae. In the developing tectum, SP- and AChR-positive axons are initially intermingled in a superficial fiber layer, but then enter and arborize in appropriate laminae soon after those laminae form. Importantly, SP-positive axons, which synapse in a superficial lamina, never extend into the deeper, AChR-positive lamina. Tectal interneurons rich in SP receptors are concentrated in the lamina to which SP-positive RGC axons project, and a set of cholinergic (choline acetyltransferase-positive) tectal projection neurons elaborate dendrites in the lamina to which AChR-positive RGC axons project. These populations of tectal neurons, which are likely targets of the RGC subsets, form in tecta that develop following enucleation. Thus, RGCs and their targets can diversify in each others absence. Accordingly, we propose that the lamina-selective connectivity we observe reflects the presence of complementary cues on RGC subsets and their laminar targets.