regulation of their postsynaptic receptors? Do ganglion cells exhibit physiological alterations in synaptic efficacy and, if so, by which mechanisms?
What are the identities of the guidance molecules that lead ganglion cell axons across the retinal surface to the optic nerve, determine which axons cross the midline at the optic chiasm, direct different axons to the midbrain or thalamus (as well as to other destinations), and produce the precise arrangement of synaptic contacts within the midbrain and LGN?
What genetic regulatory circuits distinguish retinal ganglion cell types and how are these set up during development? How are the numbers of different ganglion cell types determined, and what are the mechanisms by which these differ between species? How are the numbers and morphologies of each type of ganglion cell programmed to vary as a function of retinal eccentricity?
Many of the questions posed above are under active investigation. As an illustration of one area in which some progress has been made, we discuss below current work on the identification and characterization of transcription factors that are likely to be involved in controlling ganglion cell development.
The specification of a final differentiated cellular phenotype consists, in large part, of the selective transcriptional activation of particular genes. Work on myoblast differentiation in the mouse (18) and on early embryonic development in Drosophila (19) suggests that this is accomplished by a combinatorial network of interacting transcription factors. These act both to stably set the cell along a particular pathway of differentiation and to activate a battery of downstream genes, the products of which are the structural proteins, enzymes, etc., that functionally distinguish one cell type from another.
A number of transcription factors have been localized to the retina; most are also present in a variety of neural, and in some cases nonneural, tissues. Pax6 is the best characterized of these factors. It contains both a PAX domain and a homeodomain and is expressed in all or nearly all ocular tissues including the lens, iris, and retina (20). In mice, homozygous Pax6 mutants lack eyes and nasal primordia (21). In the heterozygous condition, mutations in the murine Pax6 gene cause a small eye phenotype, and mutations in the human PAX6 gene cause aniridia (22,23). SOHo-1, a homeodomain gene identified in chickens, is expressed in all layers of the developing retina as well as in other sensory organs including the otocyst and dorsal root and facial ganglia (24). Several homeodomain genes that are highly homologous to the Drosophila NK-2 gene—Nkx2.2, TTF1, and Dlx—are expressed in the developing retina and in a complex pattern in other regions of the developing central nervous system (CNS) (25,26). Isl1, which contains both a LIM domain and a homeodomain, is expressed in endocrine organs, in the brain and spinal cord, and in the retina in subsets of cells in the inner nuclear and ganglion cell layers (27). Chx10, a homeodomain gene, is expressed in retinal neuroblasts but not in the developing ganglion cell layer; in the adult retina it is confined to the inner nuclear layer (28). Two transcription factors that do not contain homeodomains have been characterized in the retina. NRL, a member of the basic region leucine zipper family, is expressed only in the retina, where it is present in most or all neurons (29). Mash-1, a member of the basic region helix–loop–helix family, is expressed in many regions of the developing CNS; in the developing retina, it is present in neuroblasts and is absent from the ganglion cell layer (30). A number of more ubiquitous transcription factors have also been found in the retina but are unlikely to play a role in distinguishing cell types.
With respect to the generation and differentiation of retinal ganglion cells, four POU domain transcription factors are likely to be important, based on their expression in subsets of