FIG. 1. Order of birth of retinal cells in the mouse retina. A pulse of [3H]thymidine was administered to animals each day during development of the retina. Mature retinae were then processed for autoradiography to reveal the labeled cells. Cells that were in S phase during the pulse would incorporate the label. Those that continued to divide would dilute the label, and those that underwent their last S phase would retain the highest levels of label. By analyzing mature retinae for the presence of heavily labeled cells, the day of birth of each cell type is revealed. The percentage of cells born on a given day that are each type is shown on the ordinate. The data shown are for mouse retina. [Modified and reproduced with permission from ref. 11 (copyright Wiley, New York).]

types in disparate species (12). It appears that there are some conserved aspects of the order of birth. For example, ganglion cells are the first born in many species. However, cones, horizontal cells, and amacrine cells can be born at about the same time, although none before the first ganglion cells. Overlap in the birth of different cell types, and extreme differences in the numbers of different cell types, preclude simple models in which there is a set order of recruitment of different cells into the different cell fates, as in the development of the Drosophila retina (13,14).

FIG. 2. Lineage analysis of rat retinal cells. P0 rat retinae (Left and Center) or an E14 mouse retina (Right) were infected in vivo with replication-incompetent retroviruses encoding either β-galactosidase (Left and Right) or human placental alkaline phosphatase (Center). At maturity, the retinae were processed histochemically to reveal the presence of infected cells, and cross-sections were made on a cryostat. Clones of infected cells are arranged radially as a result of siblings migrating radially from the ventricular zone to their final location in the indicated layers. Cells were identified on the basis of their morphology and location within the retina. Retinae are shown with the photoreceptor outer segment layer at the top of the photograph, r, rod; b, bipolar cell; m, muller glial cell; g, ganglion cell; a, amacrine cell. [Left, reproduced with permission from ref.15 (copyright Macmillan Magazines); Right, reproduced with permission from ref.16 (copyright Cell Press.]

Lineal relationships among retinal cells have been defined. Several groups have performed lineage analysis of retinae of various species using either intracellular injection of tracers or retroviruses ( Fig. 2). These studies have yielded similar results with respect to clonal composition (1519). In all species, retinal progenitors appear to be multipotent. Infection or injection of mitotic retinal progenitors can produce clones with one to six cell types. Clones can also vary a great deal in terms of their size. In the rodent, clones composed of from 1 to 234 cells have been observed from infection at embryonic day 14 (E14).

The multipotency of retinal progenitors appears to extend to the last cell division. Clones of only two cells can consist of two different cell types. For example, in the rodent retina, cells as distinctive as rod photoreceptors and muller glia (the only nonneuronal cell type generated by retinal progenitors) can be the members of a two-cell clone (15). Even in the prenatal period of mouse development, two-cell clones can arise and consist of two different cell types (16). The only apparent exception to this is that of rod photoreceptors in mice and rats. Rods account for 70% of the cells in the rodent retina, and there are many multicellular clones (up to 33 cells in one clone) that are exclusively rods (15,16). This makes possible the hypothesis that there is a committed, mitotic progenitor that makes only rods. Lineage analysis is a technique that cannot address this issue, and other studies, described below, were undertaken to directly address it.

The observations that distinctive retinal cell types can be born at the same time and that retinal progenitor cells are multipotent favor the role of extrinsic cues in directing cell fates. However, as mentioned above, intrinsic properties of progenitor cells must contribute to choice of cell fate as well in that cells must be competent to respond to extrinsic cues to produce the appropriate cell types. To begin to define the factors making up the environment and the competence of cells to respond to these factors, we and others have undertaken studies of cell fate determination using in vitro culture systems (2025). One of the major advantages of the retina for such studies is that it is fairly autonomous in its development. While many areas of the CNS are intimately intertwined with other areas of the CNS during development, the retina is not. It is separated from the rest of the CNS, connected only by the



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement