visual impairment and many cases of childhood blindness are caused by defects in the fovea and “central” visual loss [37–47]. While the development of the human fovea has been previously reviewed, there has yet to be an overview specifically aimed at incorporating data from retinal development in nonprimate species with data available from primates. In this chapter, four key steps necessary for foveal development are defined and discussed: (1) the specification of foveal location at the center of gaze; (2) the generation of the rod-free zone; (3) the progressive increase in cone packing density and initial pit formation; and finally (4) the lengthy centrifugal displacement of the inner retinal cells (cells move away from the foveal center), forming the adult foveal pit and increase in cone density overlying the fovea. These steps must be executed in the correct sequence or foveal development will be stalled, resulting in foveal hypoplasia (discussed at the end of the chapter).

FOVEAL DEVELOPMENT

Foveal development in primates occurs over a protracted period, beginning before birth and then extending far into postnatal life. While the region of the retina in which the fovea will become established is the first to differentiate in early fetal life, establishment of adult-like characteristics of the fovea occurs a considerable time after birth in the final stages of retinal development. Therefore, the mechanisms that control the development of the foveal region must be tightly regulated over a long temporal sequence. The proper execution of each step is dependent on the success of the previous step. Any change in the normal progression of development will affect foveal structure and function (e.g., the degree of visual acuity).

Specification of Foveal Location

In the adult eye, the fovea is located 4.9mm from the optic nerve head with little variability between eyes and individuals (Fig. 1A) [48]. These measures indicate that the placement of the fovea during development is tightly controlled and likely established early during retinal development. One possible scenario for placing the fovea would be that the foveal region is developing in a particular molecular environment that instructs progenitor cells to generate the unique properties of the macula.

What are the signals that set up this environment? In the case of the fovea, first the location must be established. Using a chicken model, it has been shown that the specification of spatial location in the retina is determined as early as optic vesicle formation (in humans, this is approximately fetal week 3.4) [49–52]. Dividing cells located toward the distal tip of the optic vesicle has been shown by fate mapping lineage analysis techniques to give rise to central retinal regions, and cells in the anterior and posterior optic vesicle will give rise to the nasal and temporal retina, respectively [50, 52, 53]. These experiments did not take into account specific retinal location or photoreceptor topography. Further refinement of this regional map will allow the exact mapping of the future center of gaze.

What controls the formation of the various axes and the central region of the retina? The address given to cells in different regions of the optic vesicle and optic cup is thought to be mediated by the expression of regionally specific genes, which segregate the developing eye into functionally distinct domains along the anterior posterior (AP) and dorsal ventral

Fig. 2. Examples of compartmentalized gene expression in the developing eye. A The DV axis of the eye and early neural retina is subdivided into several domains based on the expression of specific genes. The dorsal retina is delineated by Tbx5, ephrin B2, ephrin B1, BMP 4, and RALDH. The ventral retina expresses Vax2, Pax2, RALDH 3, and RALDH6. B The temporal retina expresses Foxd1 (brain factor 2), CBF3, and Eph A3. Foxg1 (brain factor 1), SOHo-1, GH6, and CBF-1 delineate the nasal retinal region. The molecules that are specific for this region are CYP26, BMP2, and FGF8. D dorsal, N nasal, T temporal, V ventral. Modified from Schulte D, Bumsted-O’Brien KM. 2008.

(DV) axes [50, 52, 54–57]. Work from many groups has led to the creation of a regionally distinct topographic map (Fig. 2). The dorsal retina is delineated by a number of transcription factors and signaling molecules, including Tbx5, ephrin B2, ephrin B1, BMP 4, and RALDH1 [56–66], while the ventral retina expresses Vax2, Pax2, RALDH 3, and RALDH6 [55, 56, 59, 65, 67, 68] (Fig. 2A). The temporal retina is delineated Foxd1 (brain factor 2), CBF3, and Eph A3 [69, 70]. Foxg1 (brain factor 1), SOHo-1, GH6, and CBF-1 are segregated to the nasal retinal region [55–71] (Fig. 2B). There is some overlap of the nasal and temporal expression gradients; however, this does not seem to be important in setting up the photoreceptor patterning (Fig. 2B, white zone). All of the nasal/temporal restricted genes tested so far are involved in retinal ganglion cell pathfinding as manipulating expression does not perturb photoreceptor topography [56, 57, 72].

The dorsal and ventral expression patterns do not overlap along the horizontal meridian where the center of gaze is located; instead, there is a middle ground where neither dorsal nor ventral genes are expressed (Fig. 2A). The molecules that are specific for this region are CYP26, BMP2, BMP7, and FGF8 [73–75]. The expression of CYP26 allows for an abrupt step in the diffusion of Retinoic acid (RA) levels in the central retina. Overexpression of CYP26 induces a loss of ephrin B2 expression, a dorsal-associated gene [76]. It has been shown previously that the disturbance of ephrin B2 leads to a disorganization of the normal topographically organized retinal ganglion cell projections [57]. The expression of BMP2 along the horizontal meridian and BMP7 in the chicken area centralis may indicate that it could play a role in locating the region that will form the area centralis

or fovea [73, 77]. Pax 6, an important regulatory gene in eye development, is expressed throughout the developing retina; however, the highest level of expression is in the central swath [78]. Overexpression of PAX6 in the developing eye extends the normally dorsally restricted domain of TBX5 and BMP4 into the ventral retina. PAX6 was shown to interact with VAX, leading to the modulation of PAX6 enhancer activity such that PAX6 function was inhibited in the ventral retina [78]. Collectively, these results lend strong support to the view that positional specification along both major retinal axes is already in place in the optic vesicle and early optic cup, a time before neurogenesis has been initiated. Thus, positional identities are assigned to the progenitor cells of the optic vesicle and optic cup long before the first postmitotic neurons differentiate in the retina.

Formation of a Rod-Free Zone

The next step in foveal development involves the formation of a rod-free zone. This region could be formed by either the generation of rods followed by selective cell death in the fovea or signaling (active or passive) to the progenitor cells to exclude the production of rods in the developing fovea. In this section, I argue that the latter mechanism is likely to be used in the formation of the human rod-free zone. Experiments in the chicken have provided evidence to support this argument. While chickens do not have a fovea, their retina contains a rod-free area centralis at the center of gaze that allows them to have an acuity of 7 cycles per degree [79, 80]. The chicken experiments also showed the importance of the regionally specific genes to the generation of the retinal axes and the specification of the rod-free zone. When Schulte and colleagues manipulated the ventrally expressed transcription factor VAX2, the rod-free zone was lost [56, 72]. Overexpression of the dorsal gene TBX5 causes local disturbances in the rod pattern but does not remove the rod-free zone [72]. This indicates that the boundaries set up by regional gene expression patterns are critical to localizing and specifying the rod-free zone of the central retinal region.

The data in the chicken suggest that rods are actively excluded from the developing fovea. In humans, by the time that the optic vesicle invaginates to form the optic cup (fetal week 4.5), the developing retina, which lies on the inner side of the cup, has a dramatic expansion in the number of dividing cells [81]. Cell birth starts at the site of the future fovea and then spreads from the retinal center toward the periphery in a wave of cell birth. There is no clear evidence regarding when ganglion cells are generated in the human. In a morphological study, Mann reported the appearance of the first axons in the optic nerve at fetal week 7 [28]. An estimation of human cell birth can be extrapolated from cell birth dating data obtained in the developing monkey. These data suggest that the first cells to exit the cell cycle in humans, the retinal ganglion cells, appear around fetal week 7 in the foveal region [81–84]. When correlated with morphology, the data indicate that retinal ganglion cells are born slightly before fetal week 7. The remaining retinal cell types are born in an orderly sequence beginning in the fovea and then spreading out into the periphery. Horizontal cells are born shortly after ganglion cells in the incipient fovea between fetal weeks 7 and 8, followed by cone photoreceptors. Next, amacrine cells and bipolar cells become postmitotic. The last cell types to be generated are the rod photoreceptors and Müller glia [84]. All cells are postmitotic in the fovea by fetal week 10. In the periphery, the last cells have been generated by fetal week 30 [85].

Fig. 3. Development of the rod-free zone. A Fetal week 11 human fovea with short cuboidal cones in the outer nuclear layer. Rods are not present in the fovea. The inner nuclear layer (INL) and ganglion cell layer (GCL) with the intervening synaptic layers are present in the incipient fovea. B Fetal week 11 edge of the fovea. Rods begin to be detected (arrows) on the edge of the fovea. C Rods on the edge of the fovea are labeled with an antibody to NR2E3 (arrows) and D Nrl (arrows) (modified from [89]). D The location of the rod-free zone (pale shading) in relation to the rod-dominant retina (dark shading). E The unknown signals (?) working on the progenitor cell (PC) to produce rods and a rod-free zone are indicated in the diagram to the right of the retina diagram. OD optic disk.

From the earliest point of identification, fetal week 11, rod photoreceptors are missing from the foveal center (Fig. 3A). Rods are first observed on the foveal edge (Fig. 3B–D, arrows). This lack of rods in the fovea in the human appears to be intricately linked to cell generation rather than resulting from the death of inappropriately generated rods. During the development of the fovea, it has been shown that there is little cell death in the human photoreceptor layer, indicating that it is unlikely that excess photoreceptors are generated then eliminated by cell death [86, 87]. In addition, molecules associated with rod differentiation, Nrl and NR2E3, are never detected in the foveal region, suggesting that rods are developmentally excluded from the fovea [88, 89] (Fig. 3C,D).