Our research on the growth and development of the retina is focused on both the molecular and cellular levels of analysis. Understanding retinal development requires learning how cells make choices since the final phenotype of a retinal cell is the product of signals it receives from its microenvironment combined with changes in its gene expression. Extracellular signals, cell-cell contact and intracellular factors act together to regulate the developmental trajectories of cell phenotypes in the vertebrate retina. For these interactions to occur, the cell must express the appropriate genes for each particular time in development. We use an African cichlid fish, H. burtoni for these analyses because the adult teleost retina offers unique advantages for such developmental analyses. Moreover, discoveries about the control of retinal development in fish are likely to apply to higher vertebrates because teleosts, like other vertebrate species, rely heavily on vision.
H. burtoni grow throughout life and their eyes enlarge correspondingly. Retinal area increases as a result of two distinct processes. First, new retinal cells of all types except rods are continually added concentrically from the ciliary marginal zone (CMZ) at the ora terminalis. Second, existing tissue is gradually stretched, accommodating massive intercalation of new rods, generated by cell divisions in the outer nuclear layer (ONL). As a consequence of these processes, the rod density is held constant, preserving the threshold of visual sensitivity and the density of all other cell types drops, although visual acuity is maintained because the image on the retina is larger in a large eye. Because the density of all cell types except rods decreases with increasing eye size, the ratio of rods to all other retinal cells increases dramatically during the lifetime of the animal. This means that the cell production in the CMZ must be continually "adjusted" during the lifetime of the animal. As we have shown, this regulation of cell production is achieved by a second germinal zone that generates rods exclusively just inside the CMZ. There is also a population of slowly dividing cells in the inner nuclear layer (INL) that are hypothesized to be the source of these ONL cells. Fig. 1 illustrates these proliferative populations in H. burtoni
Four features of teleost retinal growth make it a particularly advantageous system in which to study vertebrate retinal development.
1. The adult H. burtoni retina recapitulates spatially the temporal events that occur during retinal embryogenesis. At the most peripheral edge of the retina (the CMZ) are undifferentiated, dividing cells, corresponding to those found at the site of original cytogenesis in the earliest embryonic retina. As new cells are added at this edge, older cells are displaced more centrally (Fig. 1). The secondary zone of rod neurogenesis, located central to the CMZ, corresponds to the late phase of rod cell addition during embryonic retinal construction (Fernald & Shelton, 1985) when only rods are added (Hagedorn & Fernald, 1992). The character of the cells changes toward the center so that after ca. 250 Ám, there is a fully assembled, functional retina (Fernald, 1985). This exact spatial recapitulation of embryonic retinal neurogenesis can be exploited to investigate specific corresponding to a specific embryonic time by analysis of a particular retinal location in the adult animal.
2. The growing retina has two different populations of progenitors with distinctly different behaviors. The CMZ produces all cell types except for rods, while ONL progenitors produce rods exclusively. In the CMZ, cones containing three different photopigments are produced in a nearly crystalline array (Fernald, 1984) with an exactly specified spatial relationship among the cone photopigment types (Fernald, 1981). Rods, on the other hand, are inserted all across the retina (Fernald & Johns, 1980; Johns & Fernald, 1981; Fernald, 1989) with a naso-temporal preference (Kwan et al., 1996).
3. The control of cell division and possibly differentiation appears to be different in the CMZ as compared with that of rod progenitor cells. As the animal grows older and the density of all cell types except rods decreases, the marginal germinal zone produces cells at a lower density and is itself less densely populated with stem cells (Fernald, unpubl. obs.). In contrast, the density of progenitor cells in the secondary zone appears to remain constant, implying that these cells can replicate themselves in addition to producing progeny which differentiate into rods (Fernald & Scholes, 1985; Hagedorn & Fernald, 1992).
4. Rod progenitor cells appear to be motile both in the secondary germinal zone and in the central ONL (Mack & Fernald, 1995). This motility may allow them to respond to local cues, including rod density, and to divide or differentiate accordingly.
Our studies use a variety of cellular and molecular techniques to understand the orchestration of retinal growth and development.