both clusters are known as the HOX complex [159–167]. Hox complexes are now known to be present in all animals from cnidarians (jellyfish and corals) to humans. As mentioned earlier, the products of all the homeotic selector genes are similar in their DNA binding region. This region is 60 amino acid residues in length and is termed the homeodomain. The corresponding segment in the DNA is referred to as the homeobox, from which the abbreviation HOX is derived [168–170].

As “master and commander,” the homeotic genes preside over sets of genes that must be activated or repressed in a coordinated, time-sensitive manner to generate complex structures at the right time and in the right place [169, 171–173]. This is wonderfully shown by the coordination of segmentation itself in the development of the fly embryo. These elegant studies, which led to the 1995 Nobel Prize in Medicine, vindicated the approaches of experimental embryology and established many of the principles that apply to the inductive interactions taking place in the development of the retina (for excellent reviews see [158, 174–177]).

One class of Drosophila segmentation genes, the paired-axial homeobox ( pax) genes, is particularly relevant to the normal and pathological development of the human retina. Of the nine pax genes characterized to date, only four have been shown to cause abnormal development of the ocular structures: Waardenburg’s syndrome ( pax3) [178–183], Aniridia ( pax6) [184, 185], Peter’s anomaly (pax6) [186–188], and renal coloboma syndrome ( pax2) [189–193]. The corresponding spontaneous mouse mutants are Undulated (pax1) [183, 194], Splotch (pax3) [181, 183, 195–197], Small eye ( pax6) [183, 185, 196, 198–202]. Recently, analysis of spontaneous and transgenic mouse mutants has revealed that vertebrate Pax genes are key regulators during organogenesis of the kidney, eye, ear, nose, limb muscles, vertebral column, and brain [183, 188, 194, 203–209]. Like their Drosophila counterparts, vertebrate pax genes are involved in pattern formation during embryogenesis, possibly by determining the time and place of organ initiation or morphogenesis.

The role of pax6 in development has stimulated new thinking as well as controversy regarding the evolution of eyes. In his Origin of the Species, Darwin admitted that a structure as complex as the eye being formed by natural selection was difficult to accept. To address this problem, he proposed that eyes evolved from an imperfect eye prototype from which more advanced visual organs would have arisen gradually through natural

selection. Indeed, numerous intermediates between the most primitive eye and that of vertebrates have been described. Although these observations do at first appear to support Darwin’s model, the profound diversity of structure and function among different eyes in the animal kingdom appears to go against a monophyletic origin (here we define an eye as a light-sensitive organ capable of forming an image). To reconcile the remarkable diversity with a mechanism based on natural selection, it has been proposed that eyes arose independently at least 40 separate times [18, 210–212]. This view has been challenged by the observation that expression of the mouse pax6 gene in drosophila can induce the formation of ectopic functional eyes at various locations, including the legs, wings, and antennae [213–215]. These observations suggest to some biologists that pax6 is the master control gene for eye morphogenesis [20, 21, 213, 216]. This has led to the proposal that the various types of eyes in the animal kingdom evolved from a single prototype [23, 213, 216–220], a notion supported by the high degree of conservation of pax6 between flatworms to humans [197, 217, 221–224]. However, neither of the above models for the evolution of the eye appears to adequately embrace the profound diversity of eyes in the animal kingdom and the central role of pax6 in eye determination. Possibly resolving the controversy is the concept of intercalary evolution. This idea, which has recently been applied to the problem of eye evolution, provides an interesting model that could account for the monophyletic origin of the eyes and explain their fantastic diversity (the reader is encouraged to see [21, 22, 216]).

1.7  More than Meets the Optic Vesicle6

The molecular processes touched upon above provide some of the mechanisms for a set of reciprocal inductive interactions that culminate in the formation of the retina. In this choreography, whose participants are the surface ectoderm, neuroepithelium, and the neural crest-derived mesoderm, timing is everything. The performance has intrigued biologists for over a century [225].

The process begins with the formation of optic vesicles protruding from the neural tube bilaterally. As shown in the scanning electron photomicrographs of

6Margaret Saha et al. (1989) [56]

Fig. 1.9  Immerging model of the molecular basis of embryonic lens induction. Ogino et al. (2008) provide evidence that integration of signals from homeodomain protein Otx2 (pink) and SuH, a nuclear signal transducer of Notch signaling lead to lens expression in the presumptive lens ectoderm. These inputs converge­ to activate the major enhancer of Lens1, a gene essential for lens formation (green). (a, b) The head region, by expressing Otx2 is competent to respond to the Notch signaling. (c) According to the model, Otx2 and SuH although binding to the Lens1 enhancer, remain in a quiescent state until receiving

input through Notch signaling. Delta2 a Notch ligand, which is expressed in the adjacent­ optic vesicle, activates Notch signaling by releasing the intracellular domain of Notch (blue rectangle). This domain separates from Notch and moves to the nucleus. The Notch intracellular domain complexes with SuH promoting the activation of Lens1 transcription. The localized Delta / Notch signaling therefore defines lens formation at focal regions of the competent (Otx2 expressing) ectoderm. These unpublished diagrams of the author are based on the work of Ogino et al. Development 135, 249–58 (2008) [338]

Fig. 1.10, the optic vesicle extends to eventually make contact with the surface ectoderm. The surface ectoderm is induced to form the lens. As the lens expands, the optic vesicle involutes forming the optic cup. The lip of the cup forms the iris and ciliary body; remainder gives rise to the retina. This embryology explains the two-layer architecture of the adult eye. The lip of the cup, now the iris epithelium covering the posterior surface of the iris, consists of two pigmented epithelial layers continuous as the pupil margin. Proximally, the layers become the two layers of the ciliary body. The inner (vitreal) layer loses its pigment and is continuous with the neural retina, while the outer layer remains pigmented and is continuous with the retinal pigmented epithelium (RPE). Figure 1.11 shows the juncture between the developing ciliary body and retina.

How the different regions of the optic vesicle become specified to form the above structures is largely unknown although inductive interactions appear to be central to the process. We now appreciate that the surface ectoderm of the head region becomes biased to form the lens through a series of early inductive interactions. For this reason, this surface ectoderm is referred to as the “prelens ectoderm” [50, 56, 57, 226]. At the point of contact between the two tissues, the cells of the prelens ectoderm thicken and palisade to form the “lens placode” (see Fig. 1.10d). At the reciprocal point of

contact in the optic vesicle, the cells at the tip of the vesicle also begin to palisade to form the retinal disc. Interestingly, this contact is essential to the specification of the neural retina [227]. The prelens ectoderm is a source of fibroblastic growth factor (FGF) [228–230]. Furthermore, FGF-mediated signaling can substitute for the prelens ectoderm in specifying the neural retina [231, 232]. The tightly apposed prelens ectoderm and optic vesicle then invaginate to eventually form the early lens and optic cup, respectively (Fig 1.10).

A fundamentally important question is what are the mechanisms that bring about the involution of the optic vesicle to form the optic cup? Although reciprocal interactions are key to specification of the lens and neural retina, recent studies suggest that formation of the optic cup from the optic vesicle does not require assistance from the lens, as was previously thought [233]. This is not to say that the initiation of optic vesicle invagination does not require association of optic vesicle tissue with prelens ectoderm during a discrete temporal period. In an elegant set of experiments utilizing the developing chick embryo, Hyer et al. (2004) surgically removed lens tissue at various stages of development, from the prelens ectoderm stages to the invagination of the lens placode and optic cup [233]. Their findings are summarized in Fig. 1.12. Removal of the prelens ectoderm resulted in persistent optic

Fig. 1.10  Scanning electron photomicrographs of the developing mouse retina. (a) Fronto-lateral surface view of the head region at gestational age 8.5 days (corresponds to ~25 days in humans). Note that the left optic vesicle has created a lateral budge on the embryo. (b) Cut away of through the center of the bulging optic vesicle shown in (a). Note that the optic vesicle impinges upon the surface ectoderm (far left surface). (c) Cut away at gestational age day 9 (~28 days in humans). Note the formation of the optic stalk/optic vesicle. (d) Cut through thick-

ening lens placode and the adjacent portion of the optic vesicle, which is beginning to invaginate. Thickening of the ectoderm and beginning of the invagination of the optic vesicle (GA day 10; ~29 days in humans). (e) The invaginating lens placode forms the lens vesicle, which here is about to pinch off from the surface ectoderm. The invagination of the optic vesicle has formed the bilayered optic cup. Note that at this stage, the optic cup remains connected to the forebrain via the optic stalk. Images contributed by Dr. Kathleen K. Sulik

vesicles. That is, the lens did not form and the vesicles failed to invaginate into cups. Interestingly, these vesicles did show neural retinal differentiation, despite the failure to invaginate. If the surgery is conducted at a later stage, that is after the formation of the lens placode, again the lens failed to, but the vesicle did go on to invaginate normally. These results indicate two important points. First, the optic vesicle neuroepithelium requires a temporally specific association with prelens

ectoderm to undergo neurogenesis. Second, the optic cup can form in the absence of lens. In summary, the prelens ectoderm induces the optic vesicle to form an optic cup (Fig. 1.12). Although the molecular signaling pathways involved are not known, the role of retinoic acid signaling in this process may be a fruitful avenue of future research [235, 236].

As the optic vesicle invaginates, the outer layer differentiates into the RPE, while the inner layer becomes

Fig. 1.11  Photomicrograph of the rat retina at postnatal day 2. In rodents, differentiation of the retina occurs primarily postnatally. The lower panel is a higher magnification of the region designated by the white asterisk showing details of the RPE/ neural retinal interface. Arrowheads, retinal pigment epithelium (RPE); arrow in top panel, inner layer of the cilliary body; arrow in bottom panel, mitotic figure in neural retina. Note that ganglion cell layer has already differentiated (black asterisk, upper panel). Unpublished images from the author’s laboratory

the neural retina. Species differences exist in the overall structure of the optic vesicle. For example in teleosts, instead of a hollow vesicle, the retina arises from flat wing-like protrusions [237]. The RPE is a highly specialized­ epithelium that is a multifunctional and indispensable component of the vertebrate eye. Although a great deal of attention has been paid to its transdifferentiation capabilities and its functions in neural retina development, little is known about the molecular mechanisms that specify the RPE itself. Recently, advances in our understanding of the genetic network that controls the progressive specification of the eye anlagen in

vertebrates have provided some of the initial cues to the mechanisms responsible for RPE patterning. The emerging picture is that there are specific transcription factors, including otx2, mitf, and pax6, and a few signaling cascades that accomplish the onset of RPE specification in vertebrates (reviewed in [238]).

The cells that compose the early optic vesicle are indistinguishable from each other as they all express transcription factors otx2, pax6, rx1, and six3 [239]. As a result, any region of the optic vesicle is at first competent to give rise to neural retina, optic stalk, or RPE. As development proceeds, this capability is regionally restricted by signaling molecules that coordinate expression of a limited number of specific transcriptional regulatory pathways. In particular, fibroblast growth factor (FGF) suppresses RPE specification. The FGF signal may arise from the lens placode [230, 240, 241]. FGF signaling is transduced through tyrosinekinase type FGF receptors, which can activate a wide variety of signal transduction cascades. In contrast, other signals promote RPE differentiation. These signals, such as the tumor-derived growth factor (TGFb) family member activin A, appear to come from the surrounding extraocular mesenchyme [242–244]. Of central importance is the microphthalmia-associated transcription factor (mitf), which encodes a transcription factor of the basic helix-loop-helix and leucine zipper family. Mitf has a conserved and fundamental function in the development of melanin-producing cells and is activated through the receptor tyrosine kinase (RTK) pathway (reviewed in [245, 246]). Interestingly, the microphthalmia mouse was found to be deaf and had a white patch of fur, features resembling the human syndrome Waardenburg syndrome, type II. Using a candidate gene approach, the mouse mitf gene was instrumental in isolating its human homolog, which led to the identification of Waardenburg mutations in a DNA binding protein encoded by the human MITF locus [247–249]. Cells of the optic vesicle destined to become neural retina express the transcription factor chx10. One of the functions of chx10 is to inhibit mitf expression [250]. Interestingly, human microphthalmia is associated with mutations in CHX10 [251]. Finally, the orthodenticle-related transcription factors (otx), which are homeodomain-containing transcription factors with an important role in anterior head formation, are also initially expressed throughout the entire optic vesicle. However, their expression becomes restricted to the presumptive RPE during optic cup formation. Finally, the paired-box transcription factors

Fig. 1.12  How does the optic vesicle form? The emerging picture is that the prelens ectoderm, and not the lens, provides a critical signal to induce involution of the optic vesicle to form the optic cup. This diagram summarizes recent elegant microdissection studies of Hyer et al. [233]. The normal sequence of steps during the normal formation of the optic cup from the optic vesicle are shown: contact between the optic vesicle tissue and prelens ectoderm occurs from stage 11 to 13, after which the corresponding retinal placode and lens placode form at stage 13, and finally, the

lens vesicle begins to invaginate concomitantly with the distal tip of the stage 14 optic vesicle. Amazingly, if the prelens ectoderm is removed at stage 12 or earlier, the optic vesicle will not invaginate. However, if the prelens ectoderm is removed slightly later, at stage 13, the optic vesicle is capable if invaginating, forming a cup without a lens, leading to the conclusion that early communication between ectoderm and optic neuroepithelium provides the fundamental information for early optic cup formation. Reproduced from Hyer et al. [233] with permission of Academic Press

(pax) appear to make an important contribution. Pax2 is a negative regulator, while pax6 is a positive regulator of RPE specification. In summary, the optic vesicle is a dynamic structure. As it invaginates to form the optic cup, it segregates itself regionally into domains that will form the inner layer of the cup including the neural retina and domains that will form the RPE. This specification of different regions of the vesicle is accomplished through the interplay of a limited number of regulatory pathways. Disruption of these pathways is responsible for Waardenburg syndrome and some forms of microphthalmia.

The formation of the optic cup brings the inner and outer layers in close apposition. In fact, at the level of the iris and ciliary body, the two epithelial layers are physically attached by junctional complexes. At the level of the retina, the neural retina and RPE remain separated by a unique extracellular matrix that fills the subretinal compartment or space. Thus, the double layered retina is an innovation bringing into physical proximity the photoreceptors, RPE, and Müller cells [252–254]. Each of these cells borders the subretinal

compartment, which is filled with an interesting extracellular material termed the IPM [255, 256]. The IPM is a complex structure consisting of interphotoreceptor retinoid-binding protein (IRBP), growth factors [257, 258], metalloproteases [259], hyaluronan and hyaluronan binding proteoglycans [260], and sulfated glycosaminoglycans [261]. The IPM appears to mediate many of the critical interactions among the photoreceptors, RPE, and Müller cells, including retina/RPE adhesion, outer segment phagocytosis, outer segment structural stability, and nutrient exchange.

The possibility that the IPM has a significant role in development deserves further attention. Indeed, the development of the vertebrate retina depends on retinaRPE interactions [262, 263]. The presence, as mentioned above, of growth-promoting substances in the IPM is consistent with this notion. Furthermore, the expression pattern of some IPM components suggests a role in the development. Interestingly, IRBP, which is thought to function in the adult retina to transport retinoids in the vitamin A cycle, accumulates in the subretinal space before the retinoid cycle is operational

[264–268]. The gene for IRBP is expressed early during rodent retinal development and is up-regulated before that of opsin [266, 269–273] (Fig. 1.13). Perhaps IRBP participates in retinal development by facilitating the transport of retinoids or nutrients between the RPE and developing retina. Targeted disruption of IRBP results in early photoreceptor degeneration in transgenic mice [274]. In the Xenopus embryo, IRBP is first expressed

by photoreceptors in the central retina, and a central-to- peripheral gradient of IRBP appears to be established by diffusion of IRBP through the subretinal space [275]. Such a gradient could allow IRBP to transport retinoids and fatty acids from the RPE to the developing peripheral retina. The potential role of the IPM and its components in modulating interactions between the RPE and neural retina is a promising area for future research.

Fig. 1.13  Comparison of the emergence of the interphotoreceptor matrix (IPM) with photoreceptor differentiation. The graph summarizes quantitative densitometric analysis of autoradiograms at multiple exposure times. The integrated density of opsin and IRBP mRNA bands relative to actin mRNA is plotted against postnatal age. Densitometric measurements were taken from films exposed from three different lengths of time as detailed in [271]. The corresponding drawings depicts the postnatal development. Shortly before birth, junctional complexes between the RPE and neuroblastic epithelium are released. At P0 (day of birth), the neural retina is separated from the RPE by a thin extracellular matrix (stippled, area not drawn to scale). By P5, the matrix has greatly enlarged as inner segments protrude into the subretinal space. At P10, primitive outer segments are present over most of the inner segments. By P20, the matrix has accumulated to accommodate the expanded volume of the subretinal space. Arrowhead, external limiting membrane. Reproduced from Gonzalez-Fernandez et al (1993) with permission of Academic Press [271]

Integrated density relative to actin

Opsin

IRBP

1.8  Retinal Histogenesis:

A Controlled Explosion

In a short period of time, the retina generates in an almost explosive fashion more than the sufficient number of cells to comprise the mass of the retina. Simultaneously, in a highly coordinated manner, the retina achieves a wide array of different cell types located at the right position, in the right ratio, and connected to the right neurons [276, 277]. Accomplishing all of these goals is truly an amazing feat! Although we are far from having a complete picture of how this is orchestrated, some of the underlying mechanisms are beginning to emerge. The following paragraphs are meant to provide only an overview of what is one of the most exciting arenas of developmental neurobiology.

Once again, the retina shines as a model system providing one of the most elegant models not only into the development of its own complexity, but also into that of the central nervous system in general. A fundamental question in developmental neurobiology is how are cell fates in the CNS determined? A significant achievement of recent years has been a

better understanding of the relative contribution of nature vs. nurture to the process of cell fate determination. The emerging picture is that the different cell types are not produced from predefined lineages [278, 279]. Postmitotic retinal neurons and Müller glia are produced from a pool of cycling progenitors. As development proceeds, specific cell types leave the cell cycle in an orderly fashion. However, various cellular approaches using different species are converging on a model for cell-fate determination. This model combines the role of extrinsic as well as intrinsic regulators in controlling the cell-fate choices. Central to the model is the concept that the progenitor cell passes through intrinsically determined competence states [52, 280]. While passing through these states, the progenitors are capable of giving rise to a limited subset of cell types under the influence of specific extrinsic signals (reviewed in [281]) (Fig. 1.14).

The first component of this model is the temporal pattern of the appearance of the various retinal cell types. Cell birth dating performed over two decades ago provided important key early observations into the process [282, 283]. These impressive studies showed that the

various retinal cells are generated in a defined sequence. The order in which cells are born can be defined as the day on which they undergo their last S phase. This can be monitored by using [3H]thymidine labeling and autoradiography. In humans, as well as many other species, ganglion cells are the first to be born. This is illustrated in Fig 1.14, which shows a photomicrograph of the developing rat retina at postnatal day 2. At this stage, the only retinal neurons that have clearly differentiated are the ganglion cells, which have formed a clearly defined layer within the inner layer of the primitive retina. It is also important to point out that at these early stages of retinal development, the RPE is clearly present­ (Fig. 1.11). Figure 1.15, 16 show immunohistochemical staining­ for cellular retinaldehyde-binding protein (CRALBP). In the developing retina, CRALBP is present within the RPE (Fig. 1.14b). In the adult rat retina, CRALBP is also expressed by the Müller cell glia within the retina (Fig. 1.15). However, at postnatal day 2, the Müller cells have not differentiated, and CRALBP is noted only in the RPE (Fig. 1.15). Thus the primitive neural retina, which is sandwiched between formed RPE and ganglion cell layers, is positioned to be under the inductive influence of soluble morphogens elaborated by the RPE on one side and the ganglion cells on the other.

Following the retinal ganglion cells, the next cells to differentiate in the developing vertebrate retina are the cone photoreceptors, horizontal cells, and amacrine cells. Following these cells is a second wave consisting of the rod photoreceptors, bipolar cells, and Müller cells. The relative appearance of these cells during the histogenesis of the retina is summarized in Fig. 1.14. The final result is the highly organized vertebrate retina. A photomicrograph of the normal human retina shown in Fig. 1.17.

The second component of the current model for the histogenesis of the neural retina is the linear relationships among the retinal cells. Elegant lineage analysis studies in various species using either intracellular injection of tracers or retroviruses appear to converge on a similar mechanism. The ­retinal progenitor cells themselves appear to be multipotent. In fact, even in the last mitotic division of the progenitor cell, diverse cell types can still be produced [278, 279, 284–287]. Thus, the specific cell types that comprise the neural retina can be born at the same time. These observations suggest that specific extrinsic cues are important to direct the final cell fate. In vitro culture, systems have proved valuable information to identify the factors and the competence of the progenitors [288–290].

Fig. 1.15  Continuity of the developing peripheral retina with the ciliary body and iris epithelia. The rodent retina is a particularly useful experiment system as much of the development of the retina occurs during the 2 weeks following birth. Albino strains such as the one used here circumvent the problem melanin masking immunoperoxidase or immunofluorescence staining. The photomicrographs shown in this figure correspond to the peripheral rat retina at the second postnatal day of life. (a) H&E stained section showing that the neural retina is still largely undifferentiated at this age (white asterisk) except for the ganglion cells, which are clearly present (black asterisk). The ganglion cells have a larger nucleus compared to the undifferentiated retinoblasts and are restricted to their own layer in the inner retina. The RPE is continuous with the outer layer of the ciliary body (white arrows). At this location, the optic cup consists of the inner (black arrows) and outer epithelial layers of the ciliary body and iris. Both epithelial layers consist of a single row of cells. The two rows extend to the iris to line its posterior surface (far left pair of white and black arrows). The photomicrograph also shows the primitive angle and corneal endothelial layer (arrow heads). (b) Section similar to that in (a) except that due to processing artifact, there is compression of the angle with displacement of the lens (L) against the ciliary body. The section was probed with an antibody against cellular retinaldehyde-binding protein (CRALBP) and counterstained with toluidine blue, which has little affinity for the cytoplasm. The immunospecific staining for CRALBP labels the cytoplasm of the RPE and the outer ciliary body epithelial layer (arrows; brown reaction product). Sections treated with nonimmune serum showed no staining (data not shown). This staining for CRALBP in the ciliary body, although useful to show the continuity of the epithelial layers, is transient in the ciliary body disappearing in the adult retina (see also references [26, 330–333]). The antibody used for this immunohistochemical study was provided by Dr. Jack Saari. The images represent unpublished data from the author’s laboratory