Early Events in the Establishment of the Eye

A single continuous eye field begins to take shape around the area of the prechordal plate during late gastrulation. Cells in the eye field express RAX (retina and anterior neural fold homeobox). Mutations in RAX are the basis for anophthalmia, a rare condition characterized by the absence of any ocular structures in humans. Other prominent markers are Pax-6 and Lhx-2, which are heavily involved in patterning the eye fields (Fig. 13.2). With the secretion of sonic hedgehog (shh) by the prechordal plate and the ventral midline of the diencephalon, Pax-6 expression in the midline is repressed, and the single eye field splits into two separate eye fields, located on either side of the diencephalon. Rax and another important transcription factor, Six-3, protect the ability of the forebrain to secrete shh by suppressing Wnt activity. If Wnt is not suppressed, the anterior region of the developing brain becomes posteriorized and is unable to secrete shh. The absence of Six-3 activity results in the loss of shh secretion and prevents the splitting of the eye fields and leads to the condition of holoprosencephaly and the formation of only one eye (see p. 309 and Fig. 8.18).

Development of the eye is first evident at about 22 days’ gestation, when the lateral walls of the diencephalon begin to bulge out as optic grooves (Fig. 13.3). Within a few days, the optic grooves enlarge to form optic vesicles, which terminate very close to the overlying lens placode in the surface ectoderm. As the optic vesicle expands, the pattern for the future neural retina and retinal pigment epithelium is laid down. Initially, the transcription factor Mitf is expressed throughout the optic vesicle, but the subsequent expression of Vsx-2 in the distal optic vesicle (future neural retina) confines Mitf to more proximal regions, which will become the retinal pigment epithelium. Apposition of the outer wall of the optic vesicle to the surface ectoderm is essential for the transmission of an important inductive message that stimulates the cells of the lens placode to thicken and begin forming the lens (see Fig. 13.2; Fig. 13.4).

The interaction between the optic vesicle and the overlying ectoderm was one of the first recognized inductive processes. It was initially characterized by deletion and transplantation experiments conducted on amphibian embryos. When the optic vesicles were removed early, the surface ectoderm differentiated into ordinary ectodermal cells instead of lens fibers. Conversely, when optic vesicles were combined with certain types of ectoderm other than eye, the ectoderm was stimulated to form lens fibers. Subsequent research on amphibian embryos has shown that series of preparatory inductions from neural plate and underlying mesoderm condition the ectoderm for its final induction into lens by the optic vesicle. In mammals, an important mechanism underlying the severe microphthalmia (tiny eyes) or anophthalmia (absence of eyes) seen in the small eye and fidget mutants is a disturbance in the apposition of optic vesicles and surface ectoderm that interferes with lens induction.

The paired box gene Pax6 plays a prominent role throughout early eye development and at certain later stages of development of the retina and lens. Pax-6 is initially expressed in the lens and the nasal placodes and much of the diencephalon. In Drosophila, Pax6 has been called a master gene for eye development; that is, it can turn on the cascade of genes that guide development of the eye. The power of Pax-6 is shown by the formation of ectopic eyes on antennae and legs in Drosophila when the gene is improperly expressed. In the absence of Pax-6 expression (eyeless mutant), eyes do not form. In the small eye mutation, the mammalian equivalent of eyeless, the early optic vesicle forms, but, as previously noted, eye development does not progress because the surface ectoderm is unable to respond to the inductive signal emitted by the optic vesicle.

The identification in humans of two genes (Eya [eyes absent] and Six [sine oculis]) that are activated by Pax-6 in Drosophila strongly suggests that despite major differences in the structure and development of the vertebrate and insect eye, the basic genetic apparatus has been conserved throughout phylogeny. In mice, Eya-1 and Eya-2 are expressed in the lens placodes and are required for placodal induction and early differentiation, but in the absence of Pax-6 function, they are not expressed, and eye development fails to proceed.

As the process of lens induction occurs, the surface ectoderm stimulates the outer face of the optic vesicle to flatten and ultimately to become concave. This results in the transformation of the optic vesicle to the optic cup (see Fig. 13.3F). The progression from optic vesicle to optic cup requires the expression of Lhx-2 and the action of retinoic acid. In their absence, eye development is arrested at the optic vesicle stage (see Fig. 13.2). Meanwhile, the induced lens ectoderm thickens and invaginates to form a lens vesicle, which detaches from the surface epithelium from which it originated (Fig. 13.5; see Fig. 13.3). Then the lens vesicle takes over and becomes the primary agent of a new inductive reaction by acting on the overlying surface ectoderm and causing it to begin corneal development (see Fig. 13.4).

Formation of the optic cup is an asymmetric process that occurs at the ventral margin of the optic vesicle, rather than at its center. This results in the formation of a gap called the choroid fissure, which is continuous with a groove in the optic stalk (Fig. 13.6). During much of early ocular development, the choroid fissure and optic groove form a channel through which the hyaloid artery passes into the posterior chamber of the eye. Differential expression of Pax genes determines which cells become optic cup (future retina), and which cells become optic stalk (future optic nerve). Through exposure to high concentrations of shh, the expression of Pax-6 is inhibited, and Pax-2 is induced in the optic stalk, whereas a lower concentration of shh more distally permits the expression of Pax-6 in the optic vesicle, thus paving the way toward formation of the retina.

The optic stalk initially represents a narrow neck that connects the optic cup to the diencephalon, but as development progresses, it is invaded by neuronal processes emanating from the ganglion cells of the retina. Pax-2–expressing cells in the optic stalk provide guidance cues to outgrowing retinal axons that pass through the optic nerve and optic chiasm and enter the contralateral optic tract. After the neuronal processes have made their way to the appropriate regions of the brain, the optic stalk is properly known as the optic nerve.

Later in development, the choroid fissure closes, and no trace of it is seen in the normal iris. Nonclosure of the choroid fissure results in the anomaly of coloboma (see Fig. 13.19B). In certain forms of coloboma, especially forms that are associated with anomalies of the kidneys, mutations of Pax-2 genes are seen. In Pax-2–mutant mice, retinal axons do not cross the midline through the optic chiasm, but rather remain in the ipsilateral optic tract.

Formation of the Lens

The lens is derived from cells in the generic preplacodal region, as discussed earlier. From the earliest stages, formation of the lens depends on genetic instructions provided by Pax-6. Pax-6 expression is required for the surface ectoderm to respond to inductive signals (FGF and BMP) from the underlying optic vesicle by activating and combining with another transcription factor, Sox-2. This leads to the thickening of the surface ectoderm to form the lens placode (see Fig. 13.3D). At the same time, migrating neural crest cells, which do not penetrate the space between optic vesicle and prospective lens, give off signals that inhibit cells in other areas of the preplacodal region from forming lens. Inhibition of lens-forming capacity is signaled by the downregulation of Pax-6 by these cells. Pax-6 expression continues as the lens placode invaginates to form the lens vesicle, which ultimately breaks off from the surface ectoderm. At this point, Pax-6 plays a new role in regulating the activity of the genes governing the formation of the lens crystallin proteins.

When it is breaking off from the surface ectoderm, the lens vesicle is roughly spherical and has a large central cavity (see Fig. 13.3E). At the end of the sixth week, the cells at the inner pole of the lens vesicle begin to elongate in an early step toward their transformation into the long, transparent cells called lens fibers (Fig. 13.7A). The influence of the transcriptional regulator Foxe-3, which operates downstream of Pax-6, facilitates the breaking off of the lens vesicle from the surface ectoderm and the transformation of posterior cells into lens fibers.

Differentiation of the lens is a very precise and well-orchestrated process involving several levels of organization. At the cellular level, relatively nonspecialized lens epithelial cells, under the influence of Sox-2, Pax-6, and other proteins paired with an oncogene called Maf, undergo a profound transformation into transparent, elongated cells that contain large quantities of specialized crystallin proteins. At the tissue level, the entire lens is responsive to signals from the retina and other structures of the eye so that its shape and overall organization are best adapted for the transmission of undistorted light rays from the corneal entrance to the light-receiving cells of the retina.

At the cellular level, cytodifferentiation of the lens consists of the transformation of mitotically active lens epithelial cells into elongated postmitotic lens fiber cells. Up to 90% of the soluble protein in these postmitotic cells consists of crystallin proteins. The mammalian lens contains three major crystallin proteins: α, β, and γ.

The formation of crystallin-containing lens fibers begins with the elongation of epithelial cells from the inner pole of the lens vesicle (see Fig. 13.3). These cells become the fibers of the lens nucleus (Fig. 13.8). The remaining lens fibers arise from the transformation of the cuboidal cells of the anterior lens epithelium. During embryonic life, mitotic activity is spread throughout the outer lens epithelial cells. Around the time of birth, mitotic activity ceases in the central region of this epithelium, thus leaving a germinative ring of mitotically active cells surrounding the central region. Daughter cells from the germinative region move into the equatorial region of cellular elongation, where they cease to divide and take on the cytological characteristics of RNA-producing cells and begin to form crystallin mRNAs. These cells soon elongate tremendously, fill up with crystallins, and transform into secondary lens fibers that form concentric layers around the primary fibers of the lens nucleus. The midline region where secondary lens fibers from opposite points on the equator join is recognized as the anterior and posterior lens suture (see Fig. 13.7D). With this arrangement, the lens fibers toward the periphery are successively younger. As long as the lens grows, new secondary fibers move in from the equator onto the outer cortex of the lens.

The crystallin proteins show a very characteristic pattern and sequence of appearance, with the α-crystallins appearing first in the morphologically undifferentiated epithelial cells. Synthesis of β-crystallins is seen when the lens fibers begin to elongate, whereas the expression of γ-crystallins is restricted to terminally differentiated lens fiber cells. Each of the crystallin protein families contains several members. They show different patterns of activation (some members of a family being coordinately activated) and different patterns of accumulation. These patterns facilitate the optical clearing of the lens to allow the efficient transmission of light.

Throughout much of its life, the lens is under the influence of the retina. After induction of the lens, secretions of the retina, of which FGF is a major component, accumulate in the vitreous humor behind the lens and stimulate the formation of lens fibers. A striking example of the continued influence of the retina on lens morphology is seen after a developing lens is rotated so that its outer pole faces the retina. Very rapidly, under the influence of retinal secretions, the low epithelial cells of the former outer pole begin to elongate and form an additional set of lens fibers (Fig. 13.9). A new lens epithelium forms on the corneal side of the rotated lens. Such structural adaptations are striking evidence of a mechanism that ensures correct alignment between the lens and the rest of the visual system throughout development.

Formation of the Cornea

Formation of the cornea is the result of the last of the series of major inductive events in eye formation (see Fig. 13.4), with the lens vesicle acting on the overlying surface ectoderm. This induction results in the transformation of a typical surface ectoderm, consisting of a basal layer of cuboidal cells and a superficial periderm, to a transparent, multilayered structure with a complex extracellular matrix and cellular contributions from several sources. In keeping with its multifaceted role at almost all stages of eye development, Pax-6 expression in the surface ectoderm is a requirement for corneal induction.

The inductive influence of the lens stimulates a change in the basal ectodermal cells. These cells increase in height, largely as a result of the elaboration of secretory organelles (e.g., Golgi apparatus) on the basal ends of the cells. As these changes are completed, the cells begin to secrete epithelially derived collagen types I, II, and IX to form the primary stroma of the cornea (Fig. 13.10).

Using the primary stroma as a basis for migration, neural crest cells around the lip of the optic cup migrate centrally between the primary stroma and the lens capsule. Although mesenchymal in morphology during their migration, these cells become transformed into a cuboidal epithelium called the corneal endothelium when their migration is completed. At this point, the early cornea consists of (1) an outer epithelium, (2) a still acellular primary stroma, and (3) an inner endothelium.

The migration of neural crest cells between the lens and overlying ectoderm is subject to tight developmental control. A positive stimulus for migration is the production of transforming growth factor-β (TGF-β) by the lens. Modulating this influence is the presence of semaphorin 3A on the lens. The periocular neural crest cells express neuropilin-1, which, when combined with semaphorin, inhibits migration. At a critical time in development, a subpopulation of these neural crest cells ceases to express neuropilin. These cells are then able to migrate between the lens and corneal epithelium. A similar mechanism allows the penetration of sensory nerve fibers into the cornea.

After the corneal endothelium has formed a continuous layer, its cells synthesize large amounts of hyaluronic acid and secrete it into the primary stroma. Because of its pronounced water-binding capacities, hyaluronic acid causes the primary stroma to swell greatly. This provides a proper substrate for the second wave of cellular migration into the developing cornea (Fig. 13.11). These cells, also of neural crest origin, are fibroblastic. They migrate and proliferate in the hyaluronate-rich spaces between layers of collagen in the primary corneal stroma. The migratory phase of cellular seeding of the primary corneal stroma ceases when these cells begin to produce large amounts of hyaluronidase, which breaks down much of the hyaluronic acid in the primary stroma. In other parts of the embryo (e.g., limb bud), there is also a close correlation between high amounts of hyaluronic acid and cellular migration and a cessation of migration with its removal. With the removal of hyaluronic acid, the cornea decreases in thickness. When the migratory fibroblasts have settled, the primary corneal stroma is considered to have been transformed into the secondary stroma.

The fibroblasts of the secondary stroma contribute to its organization by secreting coarse collagen fibers to the stromal matrix. Nevertheless, prominent layers of acellular matrix continue to be secreted by epithelial and endothelial cells of the cornea. These secretions provide the remaining layers that constitute the mature cornea. Listed from outside in, they are (1) the outer epithelium, (2) Bowman’s membrane, (3) the secondary stroma, (4) Descemet’s membrane, and (5) the corneal endothelium (see Fig. 13.10).

The final developmental changes in the cornea involve the formation of a transparent pathway free from optical distortion, through which light can enter the eye. A major change is a great increase in transparency, from about 40% to 100% transmission of light. This increase is accomplished by removing much of the water from the secondary stroma. The initial removal of water occurs with the degradation of much of the water-binding hyaluronic acid. The second phase of dehydration is mediated by thyroxine, which is secreted into the blood by the maturing thyroid gland. Thyroxine acts on the corneal endothelium by causing it to pump sodium from the secondary stroma into the anterior chamber of the eye. Water molecules follow the sodium ions, thus effectively completing the dehydration of the corneal stroma. The role of the thyroid gland in this process was shown in two ways. When relatively mature thyroid glands were transplanted onto the extraembryonic membranes of young chick embryos, thereby allowing thyroid hormone to gain access to the embryonic circulation via the blood vessels supplying the membrane (chorioallantoic membrane), premature dehydration of the cornea occurred. Conversely, the application of thyroid inhibitors retarded the clearing of the cornea.

The other late event in the cornea is a pronounced change in its radius of curvature in relation to that of the eyeball as a whole. This morphogenetic change, which involves several mechanical events, including intraocular fluid pressure, allows the cornea to work with the lens in bringing light rays into focus on the retina. If irregularities develop in the curvature of the cornea during its final morphogenesis, the individual develops astigmatism, which causes distortions in the visual image.

Retina and Other Derivatives of the Optic Cup

While the lens and cornea are taking shape, profound changes are also occurring in the optic cup (see Fig. 13.3). The inner layer of the optic cup thickens, and the epithelial cells begin a long process of differentiation into neurons and light receptor cells of the neural retina. The outer layer of the optic cup remains thin and ultimately becomes transformed into the retinal pigment epithelium (see Fig. 13.7). Cells of the retinal pigment epithelium do not differentiate into neurons during normal embryogenesis, but in postnatal life some of these cells maintain stem cell properties and can differentiate into multiple cell types. At the same time, the outer lips of the optic cup undergo a quite different transformation into the iris and ciliary body, which are involved in controlling the amount of light that enters the eye (iris) and the curvature of the lens (ciliary body).

Formation of the optic cup and the distinction between neural retina and retinal pigment epithelium depend on molecular events occurring early in eye formation. Stimulated by FGFs from the overlying surface ectoderm, a cooperative interaction between Pax-2 and Pax-6 subdivides the optic vesicle into distal (closest to the surface ectoderm) and proximal (closest to the optic stalk) regions. Under the influence of Pax-6, the distal part of the optic vesicle invaginates to form the inner wall of the optic cup and begins to express the pairedlike homeodomain transcription factor Vsx-2, which characterizes future neural retina. The helix-loop-helix transcription factor Mitf (microphthalmia-associated transcription factor) is initially expressed throughout the entire optic vesicle, but through the inductive effect of BMP from the surrounding neural crest mesenchyme and the action of Pax-2 and Pax-6, its expression becomes restricted to the proximal part of the optic vesicle, which is the future retinal pigment epithelium.

The neural retina is a multilayered structure; its embryonic development can be appreciated only after its adult organization is understood (Fig. 13.12). When seen in cross section under a microscope, the neural retina consists of alternating light-staining and dense-staining strips that correspond to layers rich in nuclei or cell processes. The direct sensory pathway in the neural retina is a chain of three neurons that traverse the thickness of the retina. The first element of the chain is the light receptor cell, either a rod or a cone. A light ray that enters the eye passes through the entire thickness of the neural retina until it impinges on the outer segment of a rod or cone cell (photoreceptor) in the extreme outer layer of the retina. The nucleus of the stimulated rod and cone cell is located in the outer nuclear layer. The photoreceptor cell sends a process toward the outer plexiform layer, where it synapses with a process from a bipolar cell located in the inner nuclear layer. The other process from the bipolar neuron leads into the internal plexiform layer and synapses with the third neuron in the chain, the ganglion cell. The bodies of the ganglion cells, which are located in the ganglion cell layer, send out long processes that course through the innermost nerve fiber layer toward their exit site from the eye, the optic nerve, through which they reach the brain.

If all light signals were processed only through the simple three-link series of neurons in the retina, visual acuity would be much less than it actually is. Many levels of integration have occurred by the time a visual pattern is stored in the visual cortex of the brain. The first is in the neural retina. At synaptic sites in the inner and the outer plexiform layers of the retina, other cells, such as horizontal and amacrine cells (see Fig. 13.12), are involved in the horizontal redistribution of the simple visual signal. This redistribution facilitates the integration of components of a visual pattern. Another prominent cell type in the retina is the Müller glial cell, which sends processes to almost all layers of the retina and seems to play a role similar to that of astrocytes in the central nervous system.

Neural Retina

From the original columnar epithelium of the inner sensory layer of the optic cup (see Fig. 13.7), the primordium of the neural retina takes on the form of a mitotically active, thickened pseudostratified columnar epithelium organized in a manner similar to that of the early neural tube. During the early stages of development of the retina, its polarity becomes fixed according to the same axial sequence as that seen in the limbs (see Chapter 10). The nasotemporal (anteroposterior) axis is fixed first; this is followed by fixation of the dorsoventral axis. Finally, radial polarity is established.

To process distinct visual signals, the retina must develop according to a well-defined pattern, and that pattern must be carried on to the brain so that distinct visual images are produced. Despite its cuplike shape, patterning in the retina is often described in a two-dimensional manner, with dorsoventral and nasotemporal gradients serving as the basis for retinal pattern formation and differentiation.

Dorsoventral patterning is initiated through the presence of BMP-4 dorsally and shh ventrally (Fig. 13.13A). The presence of shh ventrally stimulates the production of Otx-2 in the outer layer of the optic vesicle and differentiation of the retinal pigment epithelium. Within the inner layer of the optic vesicle, shh and a protein, ventroptin, both antagonists of BMP-4, stimulate the expression of the transcription factors Vax-2 and Pax-2 in the ventral retina. In the dorsal part of the future retina, BMP-4 signals the expression of Tbx-5, the transcription factor involved in the formation of the forelimb (see p. 193). Although many molecules are unequally distributed along the two axes of the retina, opposing gradients of specific ephrins and their receptors are heavily involved in the characterization of the retinal axes (see Fig. 13.13A).

As the number of cells in the early retina increases, the differentiation of cell types begins. There are two major gradients of differentiation in the retina. The first proceeds roughly vertically from the inner to the outer layers of the retina. The second moves horizontally from the center toward the periphery of the retina (Fig. 13.13B).

Evidence obtained on fish retinas suggests that the horizontal gradient of neurogenesis may be driven by an initial point source of shh in a manner similar to a better-understood process that occurs in Drosophila. Differentiation in the horizontal gradient begins with the appearance of ganglion cells and the early definition of the ganglion cell layer (Fig. 13.14). As the ganglion cells differentiate, the surrounding cells are prevented from premature differentiation by the activity of the Notch gene. A major function of Notch is to maintain populations of cells in the nondifferentiated state until the appropriate local cues for their further differentiation appear. With the later differentiation of the horizontal and amacrine cells, the inner and outer nuclear layers take shape. As the cells within the nuclear layers send out processes, the inner and outer plexiform layers become better defined. The bipolar neurons and cone cells differentiate last, thus completing the first gradient.

The horizontal gradient of differentiation of the neural retina is based on the outward spread of the first vertical gradient from the center to the periphery of the retina. The retina cannot grow from within. During the phase of growth in the human eye (or throughout life in the case of continuously growing animals such as fish), developmentally immature retinal precursor cells along the edge of the retina undergo mitosis as an ever-expanding concentric ring on the periphery of the retina. Just inside the ring of mitosis, cellular differentiation occurs in a manner corresponding roughly to that of the vertical gradient.

Cell lineage experiments involving the use of retroviral or other tracers (e.g., horseradish peroxidase) introduced into neuronal precursors in the early retina have revealed two significant cellular features of retinal differentiation. First, progeny of a single labeled cell are distributed in a remarkably straight radial pattern following the vertical axis of retinal differentiation. There seems to be little lateral mixing among columns of retinal cells (Fig. 13.15). The second cellular feature of retinal differentiation is that a single labeled precursor cell can give rise to more than one type of differentiated retinal cell.

A later stage in retinal differentiation is the growth of axons from the ganglion cells along the innermost layer of the retina toward and into the optic stalk. Cells near the exit point express netrin-1, which acts as an attractor for the outgrowing nerve processes. When the axons reach the optic stalk, they grow into it, following cues provided by Pax-2–expressing cells. As the axons reach the area of the optic chiasm, they undergo a segregation so that the axons that arise from the temporal half of the retina remain on the same side, and the axons from the nasal half of the retina decussate through the optic chiasm over to the other side. The choice of pathways involves precise local signaling, with netrin-1 again serving as a growth cone attractor and shh playing a new role as a repulsive signal to axonal outgrowth. During this phase of retinal axonal outgrowth, the precise retinal map is maintained in the organization of the optic nerve, and it is ultimately passed on to the visual centers of the brain.

Iris and Ciliary Body

Mediated by a still poorly defined influence of the lens, differentiation of the iris and ciliary body occurs at the lip of the optic cup, where the developing neural and pigmented retinal layers meet. Rather than being sensory in function, these structures are involved in modulating the amount and character of light that ultimately impinges on the retina. In addition, the ciliary body is the source of the aqueous humor that fills the anterior chamber of the eye. The iris partially encircles the outer part of the lens, and through contraction or relaxation, it controls the amount of light passing through the lens. The iris contains an inner unpigmented epithelial layer and an outer pigmented layer, which are continuous with the neural and pigmented layers of the retina (Fig. 13.16). The stroma of the iris, which is superficial to the outer pigmented layer of the iris, is of neural crest origin and secondarily migrates into the iris. Within the stroma of the iris lie the primordia of the sphincter pupillae and dilator pupillae muscles. These muscles are unusual because they are of neurectodermal origin; they arise from the anterior epithelial layer of the iris through the transdifferentiation of pigmented cells into smooth muscle.

Eye color results from levels and distribution of pigmentation in the iris. The bluish color of the iris in most newborns is caused by the intrinsic pigmentation of the outer pigmented layer of the iris. Pigment cells also appear in the iridial stroma in front of the pigmente epithelium. The greater the density of pigment cells is in this area, the browner the eye color will be. Definitive pigmentation of the eye gradually develops over the first 6 to 10 months of life.

Between the iris and the neural retina lies the ciliary body, a muscle-containing structure that is connected to the lens by radial sets of fibers called the suspensory ligament of the lens. By contractions of the ciliary musculature acting through the suspensory ligament, the ciliary body modulates the shape of the lens in focusing light rays on the retina. Factors leading to the formation of the ciliary body are uncertain, but it seems to be induced by the lens (acting through FGFs) and surrounding neural crest mesenchyme (acting through BMPs).

The ciliary body of the eye secretes aqueous humor into the posterior chamber of the eye. The fluid passes in front of the lens through the pupil into the anterior chamber, where it maintains outward pressure on the cornea. It is then resorbed through a trabecular meshwork and into the canal of Schlemm in the angle of the eye (see Fig. 13.16). This outflow area arises from the organization of neural crest cells into a trabecular meshwork of lamellae covered by flat endotheliumlike cells, which abut onto an enlarged venous sinus, designated the canal of Schlemm. Development of this network in humans is not completed until around the time of birth.

Vitreous Body and Hyaloid Artery System

During early development of the retina, a loose mesenchyme invades the cavity of the optic cup and forms a loose fibrillar mesh along with a gelatinous substance that fills the space between the neural retina and lens. This material is called the vitreous body.

During much of embryonic development, the vitreous body is supplied by the hyaloid artery and its branches (Fig. 13.17). The hyaloid artery enters the eyeball through the choroid fissure of the optic stalk (see Fig. 13.6), passes through the retina and vitreous body, and terminates in branches to the posterior wall of the lens. As development of the retinal vasculature progresses, the portions of the hyaloid artery (and its branches supplying the lens) in the vitreous body regress through apoptosis of their endothelial cells and leave a hyaloid canal. The more proximal part of the hyaloid arterial system persists as the central artery of the retina and its branches.

Choroid Coat and Sclera

Outside the optic cup lies a layer of mesenchymal cells, largely of neural crest origin. Reacting to an inductive influence from the pigment epithelium of the retina, these cells differentiate into structures that provide vascular and mechanical support for the eye. The innermost cells of this layer differentiate into a highly vascular tunic called the choroid coat (see Fig. 13.7C), and the outermost cells form a white, densely collagenous covering known as the sclera. The opaque sclera, which serves as a tough outer coating of the eye, is continuous with the cornea. The extraocular muscles, which are derived from cranial mesoderm and provide gross movements to the eyeball, attach to the sclera.

Eyelids and Lacrimal Glands

The eyelids first become apparent during the seventh week as folds of skin that grow over the cornea (Fig. 13.18A; see Fig. 13.7). When their formation has begun, the eyelids rapidly grow over the eye until they meet and fuse with one another by the end of the ninth week (Fig. 13.18B; see Fig. 18.4). Many growth factors are involved in the migration of the epidermal cells across the cornea. Mutants of these result in lack of eyelid fusion in the embryo and impaired epidermal wound healing postnatally. The temporary fusion involves only the epithelial layers of the eyelids, thus resulting in a persistent epithelial lamina between them. Before the eyelids reopen, eyelashes and the small glands that lie along the margins of the lids begin to differentiate from the common epithelial lamina. Although signs of loosening of the epithelial union of the lids can be seen in the sixth month, BMP-mediated reopening of the eyelids normally does not occur until well into the seventh month of pregnancy.

The space between the front of the eyeball and the eyelids is known as the conjunctival sac. Multiple epithelial buds grow from the lateral surface ectoderm at about the time when the eyelids fuse. These buds differentiate into the lacrimal glands, which produce a watery secretion that bathes the outer surface of the cornea when mature. This secretion ultimately passes into the nasal chamber by way of the nasolacrimal duct (see Chapter 14). The lacrimal glands are not fully mature at birth, and newborns typically do not produce tears when crying. The glands begin to function in lacrimation at about 6 weeks.

Despite the many types of malformations of the eye and visual system, the incidence of most individual types of defects is uncommon. Mutations of almost any of the genes mentioned in this section result in a recognizable malformation of the eye. Among these, anophthalmia, microphthalmia, congenital cataracts, and colobomas are common phenotypes. Examples of a few of the many ocular malformations are given in Clinical Correlation 13.1.

Clinical Correlation 13.1   Congenital Malformations of the Eye

Anophthalmos and Microphthalmos

Anophthalmos, the absence of an eye resulting from a mutation in RAX, is very rare and can normally be attributed to lack of formation of the optic vesicle. Because this structure acts as the inductive trigger for much of subsequent eye development, many local inductive interactions involved in the formation of eye structures fail to occur. Microphthalmos, which can range from an eyeball that is slightly smaller than normal to one that is almost vestigial, can be associated with a large number of genetic defects (e.g., aniridia) or various other causes, including intrauterine infections (Fig. 13.19A). Microphthalmos is one of the common components of the rubella syndrome.

Coloboma of the Iris

Nonclosure of the choroid fissure of the iris during the sixth or seventh week results in its persistence as a defect called coloboma iridis (Fig. 13.19B). The location of colobomas of the iris (typically at the 5 o’clock position in the right eye and the 7 o’clock position in the left eye) marks the position of the embryonic choroid fissure. Because of the enlarged pupillary space resulting from the colobomatous fissure, individuals with this condition are sometimes sensitive to bright light, given their inability to contract the pupil properly.

Congenital Cataract

Cataract is a condition characterized by opacity of the lens of the eye. Not so much a structural malformation as a dysplasia, congenital cataracts first came into prominence as one of the triad of defects resulting from exposure of the embryo to the rubella virus.

Cyclopia

Cyclopia (bold) or synophthalmia (bold) represent different degrees of or lack of splitting of the single eye field into two separate bilateral eye fields. Generally due to deficient midline shh signaling from the brain, cyclopia is a malformation secondary to formation of the midline of the brain and face. Cyclopia (see Fig. 8.18) is often accompanied by the presence of a fleshy proboscis dorsal to the eye.

Development of the Inner Ear

Development of the ear begins with preliminary inductions of the surface ectoderm, first by the notochord (chordamesoderm) and then by the paraxial mesoderm (Fig. 13.20). These inductions prepare the ectoderm for a third induction, in which FGF-3 signals from the rhombencephalon induce the adjacent surface ectoderm to express Pax-2. Then Wnt signals greater than a certain threshold stimulate Pax-2–positive cells to form both the otic and epibranchial placodes (Fig. 13.21). The cells exposed to a subthreshold concentration of Wnt are destined to become epidermal cells. Late in the fourth week, possibly under the influence of FGF-3 secreted by the adjacent rhombencephalon, the otic placode invaginates and then separates from the surface ectoderm to form the otic vesicle, or otocyst (see Fig. 13.21C). Even at its earliest stages, the development of the main components of the inner ear is under separate genetic control: Pax-2 and Sox-3 for the auditory portion (cochlea) and Nkx-5 for the vestibular portion (semicircular canals).

When the otic vesicle has formed, the next steps in its development are molecular patterning and subdivision of the otic vesicle into sensory (future hair cells) and neuronal components of the inner ear. Similar to the developing limb and eye, three cartesian axes (anteroposterior, dorsoventral, and mediolateral) become specified early in the development of the otocyst. Ventrality is specified by shh coming from ventral sources, and dorsality is specified by Wnts coming from the dorsal neural tube.

The otic vesicle soon begins to elongate, thus forming a dorsal vestibular region and a ventral cochlear region (Fig. 13.22A). The paired box gene, Pax2, is heavily involved in early development of the otic vesicle. Expression of Pax-2 in the ventral otic vesicle is important for the continued development of the endolymphatic duct and the cochlear apparatus, whereas dorsal expression of Dlx-5 and Gbx-2 is important for development of the vestibular system. Quite early, the endolymphatic duct arises as a short, fingerlike projection from the dorsomedial surface of the otocyst (Fig. 13.22B). FGF-3, secreted by rhombomeres 5 and 6, is necessary for normal development of the endolymphatic duct. At about 5 weeks, the appearance of two ridges in the vestibular portion of the otocyst foreshadows the formation of two of the semicircular ducts (Fig. 13.22C).

As the ridges expand laterally, their opposing epithelial walls approximate each other, to form a fusion plate. Programmed cell death in the central area of epithelial fusion or epithelial cell migration away from this area converts the flangelike structures to canals by setting up a zone of resorption (see Fig. 13.22C). The epithelial precursors of the semicircular canals express the homeobox transcription factor gene Nkx-5-1, which is important for the development of the dorsal vestibular portion of the inner ear. Other transcription factors play a role in the formation of individual semicircular canals. In the absence of Otx1, the lateral semicircular canal fails to form, and the homeobox transcription factor Dlx-5 must be expressed for the anterior and posterior canals to develop. The cochlear part of the otocyst begins to elongate in a spiral; it attains one complete revolution at 8 weeks and two revolutions by 10 weeks (see Fig. 13.22C through F). The last half turn of the cochlear spiral (a total of two and one-half turns) is not completed until 25 weeks.

The inner ear (membranous labyrinth) is encased in a capsule of skeletal tissue that begins as a condensation of mesenchyme around the developing otocyst at 6 weeks’ gestation. The process of encasement of the otocyst begins with an induction of the surrounding mesenchyme by the epithelium of the otocyst (see Fig. 13.20). This induction, involving BMP-4, stimulates the mesenchymal cells, mainly of mesodermal origin, to form a cartilaginous matrix (starting at about 8 weeks). The capsular cartilage serves as a template for the later formation of the true bony labyrinth. The conversion from the cartilaginous to the bony labyrinth occurs between 16 and 23 weeks’ gestation.

The sensory neurons that compose the eighth cranial nerve (specifically the statoacoustic ganglion) arise from cells that migrate out from a portion of the medial wall of the otocyst (see Fig. 13.21). How the axons originating from the statoacoustic ganglion make their way to precise locations within the inner ear is not well understood, but guidance by neural crest cells and channeling by Slit-Robo interactions are known to play prominent roles in directing growth cones to their proper targets. The cochlear part (spiral ganglion) of cranial nerve VIII fans out in close association with the sensory cells (collectively known as the organ of Corti) that develop within the cochlea. Neural crest cells invade the developing statoacoustic ganglion and ultimately form the satellite and supporting cells within it.

The sensory cells of the organ of Corti are also derived from the epithelium of the otocyst. They undergo a very complex pattern of differentiation (Fig. 13.23). The generation of sensory neuroblast precursors in the inner ear seems to use the Notch pathway to control the proportion of epithelial cells that differentiate into neuroblasts versus supporting cells in a manner similar to that described for the differentiation of ganglion cells in the early retina (see p. 280). As in other sensory systems, highly regulated developmental controls ensure precise matching between sensory cells designed to receive sound waves at different frequencies or gravitational information and the neurons that transmit the signals to the brain. Formation of the graded array of hair cells that respond to different frequencies of sound is to a great extent based on execution of the planar cell polarity pathway (see p. 87).

Development of the Middle Ear

Formation of the middle ear is intimately associated with developmental changes in the first and second pharyngeal arches (see Chapter 14). The middle ear cavity and the auditory tube arise from an expansion of the first pharyngeal pouch called the tubotympanic sulcus (Fig. 13.24). Such an origin ensures that the entire middle ear cavity and auditory tube are lined with an endodermally derived epithelium.

By the end of the second month of pregnancy, the blind end of the tubotympanic sulcus (pharyngeal pouch 1) approaches the innermost portion of the first pharyngeal cleft. Nonetheless, these two structures are still separated by a mass of mesenchyme. Later, the endodermal epithelium of the tubotympanic sulcus becomes more closely apposed to the ectoderm lining the first pharyngeal cleft, but they are always separated by a thin layer of mesoderm. This complex, containing tissue from all three germ layers, becomes the tympanic membrane (eardrum). During fetal life, a prominent ring-shaped, neural crest–derived bone, called the tympanic ring, supports the tympanic membrane. Experiments have shown that the tympanic ring is actively involved in morphogenesis of the tympanic membrane. Later, the tympanic ring becomes absorbed into the temporal bone.

Just dorsal to the end of the tubotympanic sulcus, a conspicuous condensation of mesenchyme of neural crest origin appears at 6 weeks and gradually takes on the form of the middle ear ossicles. These ossicles, which lie in a bed of very loose embryonic connective tissue, extend from the inner layer of the tympanic membrane to the oval window of the inner ear. Although the middle ear cavity is surrounded by the developing temporal bone, the future middle ear cavity remains filled with loose mesenchyme until late in pregnancy. During the eighth and ninth months, programmed cell death and other resorptive processes gradually clear the middle ear cavity and leave the auditory ossicles suspended within it. Even at the time of birth, remains of the middle ear connective tissue may dampen the free movement of the auditory ossicles. Free movement of the auditory ossicles is acquired within 2 months after birth. Coincident with the removal of the connective tissue of the middle ear cavity is the expansion of the endodermal epithelium of the tubotympanic sulcus, which ultimately lines the entire middle ear cavity.

The middle ear ossicles themselves have a dual origin. According to comparative anatomical evidence, the malleus and incus arise from neural crest–derived mesenchyme of the first pharyngeal arch, whereas the stapes originates from second-arch mesenchyme (Fig. 13.25).

Two middle ear muscles help modulate the transmission of auditory stimuli through the middle ear. The tensor tympani muscle, which is attached to the malleus, is derived from first-arch mesoderm and is correspondingly innervated by the trigeminal nerve (cranial nerve V). The stapedius muscle is associated with the stapes, is of second-arch origin, and is innervated by the facial nerve (cranial nerve VII), which supplies derivatives of that arch.

Development of the External Ear

The external ear (pinna) is derived from mesenchymal tissue of the first and second pharyngeal arches that flank the first (hyomandibular) pharyngeal cleft. During the second month, three nodular masses of mesenchyme (auricular hillocks) take shape along each side of the first pharyngeal cleft (Figs. 13.26 and 13.27). The auricular hillocks enlarge asymmetrically and ultimately coalesce to form a recognizable external ear. During its formation, the pinna shifts from the base of the neck to its normal adult location on the side of the head. Because of its intimate association with the pharyngeal arches and its complex origin, the external ear is a sensitive indicator of abnormal development in the pharyngeal region. Other anomalies of the first and second arches are often attended by misshapen or abnormally located external ears.

The external auditory meatus takes shape during the end of the second month by an inward expansion of the first pharyngeal cleft. Early in the third month, the ectodermal epithelium of the forming meatus proliferates and forms a solid mass of epithelial cells called the meatal plug (see Fig. 13.24C). Late during the fetal period (at 28 weeks), a channel within the meatal plug extends the existing external auditory meatus to the level of the tympanic membrane.

The external ear and external auditory meatus are very sensitive to drugs. Exposure to agents such as streptomycin, thalidomide, and salicylates during the first trimester can cause agenesis or atresia of both these structures. Congenital malformations of the ear are discussed in Clinical Correlation 13.2.

Clinical Correlation 13.2   Congenital Malformations of the Ear

The ear is subject to a wide variety of genetically based defects, ranging from those that affect the function of specific hair cells in the inner ear to those that produce gross malformations of the middle and external ear. Defects of the middle and external ear are often associated with genetic conditions that affect broader areas of cranial tissues or regions. Many malformations involving the first or second pharyngeal arches are also accompanied by malformations of the ear and reduced acuity of hearing.

Congenital Deafness

Many disturbances of development of the ear can lead to hearing impairments, which affect 1 in every 1000 newborns. Conditions such as rubella can lead to maldevelopment of the organ of Corti that causes inner ear deafness. Abnormalities of the middle ear ossicles or ligaments, which can be associated with anomalies of the first and second arches, can interfere with transmission, resulting in middle ear deafness. Agenesis or major atresias of the external ear can cause deafness by interfering with the primary collection of sound waves.

The surge of molecular studies on the development of the inner ear has identified in mice and humans a variety of mutants of the inner ear that could lead to deficits in hearing, equilibrium, or both. Mutants, such as those of Pax-3 causing variants of Waardenburg’s syndrome (see p. 265), can affect development at levels ranging from gross morphogenesis of the ear to specific cellular defects in the cochleosaccular complex.

Auricular Anomalies

Because of the multiple origins of its components, there is great variety in the normal form of the pinna. Variations include obvious malformations such as auricular appendages, or sinuses (Fig. 13.28). Many malformations of the external ears are not functionally important, but rather are associated with other developmental anomalies, such as malformations of the kidneys and pharyngeal arches. Exposure to excess retinoic acid or derivatives commonly results in anomalies of the external ear (see Fig. 8.15).