Following lens vesicle formation, which occurs between 4 to 5 weeks of gestation (27 to 36 days), surface ectodermal cells cover the defect left by lens vesicle invagination and become the primitive corneal epithelium composed initially of two cell layers. Similar to the avian cornea, the primitive epithelium of the primate cornea appears to be responsible for forming a prominent primary acellular corneal stroma, also known as the Bowman’s layer, which becomes detectable around 20 weeks of gestation. It should be noted, however, that the primate cornea’s ectodermal phase (primary acellular stroma) is de-emphasized compared to the mesodermal phase (secondary cellular stroma) and is less well organized in ultrastructure compared to that of the avian cornea.

Sometime between when the eyelids fuse together (8 weeks gestation) and by the time the eyelids open (26 weeks gestation), the epithelium stratifies to four to five cell layers thick. The basal epithelial cells, which secrete an underlying basement membrane called the basal lamina, are cuboidal to columnar. Above this layer are interdigitating wing cells and, most superficially, a layer of flattened squamous epithelial cells. Early in gestation, adhesion complexes (hemidesmosomes, anchoring fibrils, and anchoring plaques) on the basal surface of the epithelium are absent. Rudimentary adhesion complexes only become detectable by 19 weeks of gestation. With further development in utero, the number of hemidesmosomes increases, anchoring fibril penetration into the Bowman’s layer increases, and Bowman’s layer thickens.

A first wave of neural crest-derived mesodermal cells begins to extend beneath the corneal epithelial cells from the limbus around 5 weeks of gestation (33 days); these cells form the primitive endothelium. The primitive endothelium is initially composed of two cell layers, which by 8 weeks of gestation become a monolayer that starts to produce Descemet’s membrane. The epithelium and endothelium remain closely opposed until 7 weeks of gestation (49 days) when a second wave of mesoderm begins to grow centrally from the limbus between the epithelium and endothelium. This tissue forms the secondary cellular corneal stroma, and extracellular collagen is produced within a few days. During the third month of gestation, Descemet’s membrane is secreted by the endothelial cells and can be recognized in histologic sections. By the seventh month of gestation, the cornea resembles that of the adult in most structural characteristics other than size.

At birth in the full-term infant, the horizontal diameter of the cornea is only around 9.8 mm (surface area 102 mm²), or approximately 75% to 80% of the size of the adult human cornea. (Note that the posterior segment is less than 50% of adult size at birth.) Additionally, at birth, the cross-sectional thickness of the epithelium averages 50 μm, the Bowman’s layer averages 12 μm, the central cellular corneal stroma averages 450 μm, Descemet’s membrane averages 4 μm, and the endothelium averages 5-μm thick.

During infancy, the cornea continues to grow over the first years of life, reaching adult size at 2 years old with a horizontal diameter of 11.7 mm (surface area 138 mm²) and changes very little in size, shape, and optical properties thereafter. The only significant structure in the cornea that thickens with age is the Descemet’s membrane, which gradually increases an additional 6 to 11 μm from birth to senescence.

The average index of refraction of the cornea and tear film taken as a whole is about 1.376. The refractive power for the anterior surface of the cornea may be computed by the following formula:

where D equals diopters of optical power, n is the index of refraction of air (1.000), n′ is the index of refraction of the whole cornea, and r is the radius of curvature of the anterior corneal surface in meters.

The posterior surface of the cornea is bathed with aqueous humor. By applying the same formula, we arrive at the following value for the optical power of the posterior corneal surface, where n′ is the refractive index of aqueous humor (1.336):

Therefore, the total optical power of the cornea is 48.2 – 6.2, or approximately 42.0 D, which is about two-thirds of the 60.0 D total optical power of the human eye. Because the cornea is thinner in the center than in the periphery, it should act as a minus lens but functions as a plus lens because the aqueous humor neutralizes most of the minus optical power on the posterior corneal surface. If we compute the power of the posterior corneal surface in air, we would find the following:

D=1.000-1.375/0.0065 =-96.0

By subtracting the plus power of the anterior surface, the resulting net power for the cornea in air becomes -96.0 D + 42.0 D = -54.0 D, a powerful minus lens.

From the previous calculations, it is obvious that the most important refracting surface is that of the anterior cornea. However, if a large air bubble is placed in the anterior chamber so that it contacts the corneal endothelium or if the anterior surface of the cornea is submerged in water—thereby destroying the corneal-air interface—tremendous changes in the refractive power of the eye occur. When the eye is open underwater, the optical imagery is extremely blurred; the index of refraction of water is quite similar to that of the cornea and most of the optical power of the anterior corneal surface is lost. If the tear film–air interface is maintained by the use of a mask or goggles, then underwater mask vision is as sharp and clear as normal terrestrial vision.

A contact lens placed on the corneal surface is an efficient way of correcting optical errors in the dioptic system of the eye. This is particularly true for errors that are caused by abnormalities of corneal curvature. The contact lens placed upon the corneal surface has the same index of refraction as the cornea and becomes covered immediately with the normal corneal tear film. Therefore, the contact lens becomes, in effect, a new part of the cornea.

Several surgical procedures have also been developed to permanently alter the curvature of the cornea, thereby reducing nonpathologic refractive errors. The procedures most commonly performed today include laser-assisted in situ keratomileusis (LASIK), photorefractive keratectomy (PRK), and astigmatic keratotomy (AK). LASIK and PRK are surgical techniques that use an excimer laser (193 nm) to ablate the anterior corneal stroma with submicron accuracy and create a new refractive surface. LASIK is different from PRK in that it involves a mechanical or laser created lamellar keratectomy step before the excimer ablation step, which is applied to the underlying residual mid-stromal bed. PRK ablation directly occurs on the surface of the cornea (Bowman’s layer and anterior stroma). Both procedures have been used on myopic and hyperopic eyes with up to a mild degree of astigmatism and produce similar visual outcomes. AK is an incisional keratotomy technique that cuts up to 90% depth arcs in the peripheral cornea using a scalpel to reduce astigmatism by flattening the cornea in the visual axis over the area of the cut. AK is usually performed in association with cataract surgery.

When viewed anteriorly in the living eye, the adult human cornea is somewhat elliptical; the largest measurement is typically in the horizontal meridian (mean 11.7 mm) and the smallest is in the vertical meridian (mean 10.6 mm). The elliptical configuration is brought about by an anterior extension of the opaque scleral structure superiorly and inferiorly. When viewed from the posterior surface of the dissected eye, the cornea is circular, with an average horizontal and vertical measurement of 11.7 mm. The average radius of curvature of the anterior corneal surface is 7.8 mm, which is significantly less than the 11.5 mm average radius of curvature of the sclera. This results in a small, 1.5- to 2-mm wide transition zone that forms an external and internal surface groove, or scleral sulci, where the steeper cornea meets that of the flatter sclera. These sulcus typically are not as obvious to see clinically because they are filled in by overlying episclera and conjuctiva or trabecular meshwork.

The tissue in this transition zone is known as the limbus and is important because it is used as surgical landmark for various anterior segment surgeries, contains corneal epithelial stem cells, contains the conventional outflow pathway for the aqueous humor, is the inciting site of pathology in a few immunologic or cancerous diseases, and possibly contains stem-like corneal endothelial cells (Fig. 2). The limbus is a definite anatomic landmark, or geographical reference, used for planning surgical entry into the anterior segment because it appears clinically as a blue transition zone. Therefore, an incision placed anterior to blue zone is anatomically in the peripheral cornea, safely inside the trabecular meshwork.

The cornea is thinner in the center, measuring on average 520 μm and increases in thickness to approximately 700 μm as it reaches the limbus.^1,2,3,4 The central 4 mm of the cornea overlying the pupil contains the optical center of central vision. It typically has a regular surface and structure, and commonly is near exact in spherical configuration. The more peripheral portions of the cornea are often slightly irregular and torical in configuration rather than spherical, giving the whole cornea a hyperbolic shape. If the central cornea is not exactly spherical, the condition of astigmatism (stigma meaning a point) usually results. With astigmatism, an optical image without a single point focus forms within the eye resulting in two line foci.

The average radius of curvature for the posterior corneal surface is shorter than the anterior corneal surface at 6.5 mm. In adult humans, the conjunctival surface area has been measured at 17.65 cm², and the corneal surface area measures 1.38 cm², giving a c/c ratio of 12.8.⁵ Because topical lipophilic drug delivery to the anterior chamber occurs primarily through the cornea with the conjunctiva supplying a minor secondary delivery route, surface area ratios of the cornea and conjunctiva are important because a large conjunctiva-to-corneas (c/c) surface area ratio results in less drug delivery to the anterior chamber. Therefore, surface area ratios need to be considered when comparing drug delivery studies.

The anterior surface of the human cornea is covered by a transparent, nonkeratinized, stratified (five- to seven-cell layer) squamous epithelium uniformly around 50 μm in thickness that is continuous with the epithelium of the limbus and conjunctiva (Figs. 1, 2, and 3). The basal corneal epithelial cells actively secrete extracellular material (type IV collagen, laminin, heparin, and small amounts of fibronectin and fibrin) that forms an underlying 75-nm thick basement membrane called the basal lamina. On electron microscopy, the morphology of basal lamina appears to be composed of two distinct layers: a 25-nm thick lamina lucida and a 50 nm thick lamina densa (Fig. 3).

The cytoplasm of a corneal epithelial cell primarily contains cytoskeletal intermediate filaments and has sparse cytoplasmic organelles (i.e., mitochondria, endoplasmic reticulum, and golgi apparatus). The predominant cytoplasmic filament is keratin, whereas actin and microtubules are two other major types found in corneal epithelial cells. The epithelial cells are held together to one another by numerous anchoring junctions called desmosomes, whereas the basal surface of the epithelium adheres to the basal lamina and underlying Bowman’s layer through aa adhesion complex composed of hemidesmosomes, anchoring fibrils (type VII collagen), and anchoring plaques (Fig. 4). The function of the corneal epithelium is twofold: (a) to form a barrier from the environment to the corneal stroma of the cornea, and (b) to form a smooth refractive surface on the cornea through interaction with the tear film.

The epithelial cells differentiate from the basal layer to form two to three cell layers of wing cells and finally to form two to three cell layers of squamous cells (Fig. 3). The squamous cells form a barrier junction because they are surrounded by a continuous encircling band of zonula occludens tight junctions, which serve as a semipermeable, high-resistance (12–16 kΩ cm²) membrane^6,7 by closing off the intercellular space. This barrier prevents the movement of fluid from the tears into the stroma and also protects the cornea and intraocular structures from infectious pathogens. The apical surface of the corneal epithelium is specialized to maintain the tear film as microplicae and microvilli on the surface of the most superficial epithelial cells is covered with a glycocalyx and membrane-spanning mucins (MUC 1 and possibly MUC 4); altogether these structures and substances form the 1.0 μm thick mucinous layer of the tear film (Fig. 5).^8,9,10

The tear film typically measures 7 to 10 μm in thickness and contains three layers (mucinous, aqueous, and lipid layers). Recently, the aqueous layer of the tear film has been found to contain gel-forming mucins (MUC 5AC and 2) along with lysosyme, immunoglobulin A, transferring, defensin, and trefoil factor (Fig. 5, inset).¹⁰ Overall, the tear film functions in maintaining a healthy ocular surface by preventing ocular infections and forming a smooth optical surface required for clear vision. Deficiencies in the components that compose any of these layers potentially can cause ocular surface disease.

The corneal epithelium is in a state of constant healing as squamous cell are continuously

shed into the tear film (Fig. 6). It is estimated that all the cell layers of the corneal epithelium completely turn over every 7 to 10 days. The epithelial surface is maintained by basal epithelial cells, which can undergo mitosis resulting in two daughter cells that are found anterior to the basal cell layer, thus forming two wing cells and eventually squamous cells.^11,12,13 This delicate balance of shedding followed by proliferation is critical in maintaining a smooth and uniform epithelial surface around 50 μm in thickness. The shedding step is primarily induced by the friction that occurs from involuntary eyelid blinking, which happens on average every 7 seconds.

The signal for basal epithelial cell proliferation probably comes via the gap junctions, especially in the more basal layers, with connexin 43 being one of the common gap junction proteins present.^7,14 Interestingly, gap junction communication between corneal epithelial cells is promoted by low concentrations of bicarbonate, which is abundant in the aqueous humor and tear film.¹⁵ The barrier function of the epithelium itself is due to zonula occludens tight junctions between the most superficial epithelial junction, which are made up of the tight junction proteins ZO-1, occludin, and claudin-1 as well as other claudin sub-types.¹⁶ Under normal conditions, this barrier is extremely tight, preventing the movement of ions between the superficial squamous cells. In some instances (e.g., trauma, contact lens wear, infection), however, this barrier can be compromised. Activation of protein kinase C (PKC) by the mitogen-activated protein kinase (MAP kinase) signaling pathway, which in itself is activated by external stimuli such as growth factors and tyrosine kinases, appears to cause disruption of the epithelial barrier and perhaps is an initiating factor for basal cells to proliferate.¹⁷

In addition to basal epithelial cell mitosis, the corneal epithelium is maintained by migration of new basal epithelial cells into the cornea from the limbus. The cells migrate centripetally at about 120 μm/week and originate from a subpopulation of limbal epithelial cells.^18,19,20 Therefore, it appears that the corneal epithelium is maintained by a balance among the processes of limbal cell proliferation, basal corneal epithelial cell migration and proliferation, and shedding of superficial squamous corneal epithelial cells.^21,22 This concept has been coined the X, Y, Z hypothesis by Thoft [X (basal corneal epithelial cell proliferation) + Y (limbal cell proliferation and basal corneal epithelial cell centripetal migration) = Z (epithelial cell loss from the surface)].²²

When this equilibrium is disturbed, corneal epithelial cell wound healing typically begins. Interestingly, after injury, these processes typically return back to equilibrium with a certain degree of adaptability. For instance, after epithelial and stromal injury, if a stromal deficit occurs (e.g., corneal ulcers) then the epithelium can still maintain a smooth anterior surface either by developing elongated hypertrophic basal epithelial cells and/or by developing epithelial hyperplasia (> six cell layers).²³

As first described in the early 1960s by Maumenee, and expanded upon in the early 1970s by the work of Davanger and Evensen and in the early 1980s by Shapiro, Friend and Thoft, the limbal epithelium contains cells that have the ability to resurface the cornea following loss of the corneal epithelium.^24,25,26 This concept was applied clinically by Thoft in 1977, who described the use of autologous limbal epithelial transplants to treat unilateral chemical injury.²⁷ Subsequently, further study has determined that a portion of limbal basal epithelial cells are actually undifferentiated corneal epithelial cells, also known as corneal epithelial stem cells.^28,29 These stem cells have the characteristics of being extremely long-lived, with very slow cell cycle times and a high proliferative potential. They anatomically are found within a well-defined, protective microenvironment called the palisades of Vogt (Fig. 7) and are controlled via delicate regulatory mechanisms.^30,31,32 The progeny of stem cells are referred to as transient amplifying (TA) cells, which are cells that have more rapid cell cycles, yet have far less replicative potential than their parent stem cells. During terminal differentiation from the basalepithelial cell layer, the daughter cells migrate anteriorly in the epithelium becoming wing cells and, lastly, squamous cells.

When the mammalian cornea is mounted in an Ussing chamber, it generates a transepithelial potential of 25 to 40 mV. This relatively high voltage is consistent with the low ionic conductance of the apical epithelial cell membranes and the high resistance of the tight junctions of the paracellular pathway, which normally is in the range of 12 to 16 kΩ cm².⁶ Almost 50% of the short-circuit current across the corneal epithelium is carried by chloride ions moving through apical membrane channels into the tears. This current is due to ionic gradients set up by epithelial transport of Na⁺ and Cl^– ions. Na⁺ ions are pumped from the epithelial cells toward the stroma by the Na⁺/K⁺ adenosine triphosphatase (ATPase) in the lateral membranes of the cells, which, as in other cells, maintains an inward Na⁺ gradient. A Na⁺-chloride co-transporter, also located in the lateral membrane, facilitates the influx of Na⁺ down its electrochemical gradient, carrying along chloride ions. The chloride ions then diffuse through channels in the apical membranes. This chloride secretion is blocked when the Na⁺/K⁺ ATPase is inhibited by ouabain, demonstrating the coupling of chloride secretion to Na⁺ transport.^33,34,35,36

In addition to the transport mechanisms described, the corneal epithelial cells also contain a Na⁺/H⁺ exchanger and a lactate-H⁺ co-transporter. These transport mechanisms serve to regulate intracellular pH by extrusion of lactate and H⁺ ions.³⁷ The above discussion raises questions concerning the overall role of these transport mechanisms and second-messenger systems in corneal epithelial function. It has been demonstrated in vitro that ion transport can osmotically move water from the stroma to tears⁶; however, in vivo epithelial ion transport probably has a limited role, if any, in corneal deturgescence as compared to the endothelium.

Finally, it has been shown in both animal and human specimens that the corneal epithelium, although devoid of melanocytes, does contain immune cells. The basal epithelial cell layer of the peripheral cornea, along with the limbus and the conjunctiva, appears to have a subpopulation of cells intermixed that are bone-marrow–derived immune surveillance cells with high constitutive expression of major histocompatibility complex (MHC) type II antigen and co-stimulatory molecules.³⁸ This type of immune cell has previously been termed a Langerhans cell,³⁹ which functions as a “professional” antigen-presenting cell with an extraordinary capacity to initiate T-cell lymphocyte-dependent responses.⁴⁰ The functional steps of this cell type include the uptake and processing of antigens, the migration out of the cornea to lymph nodes where they stimulate naïve T-lymphocyte immune responses by presenting antigens and overexpressing co-stimulatory molecules. Recently, the central corneal epithelium also was found to have a similar subpopulation of immune cells that are apparently “immature” Langerhans cell types because their constitutive expression of MHC type II antigen and costimulatory molecules is low.⁴¹ Interestingly, these “immature” Langerhans cells under certain circumstances (e.g., inflammation or trauma) may develop the requisite signals for T-cell priming.⁴¹

The corneal stroma accounts for 90% of the corneal thickness. It is predominantly composed of water (78% hydrated or 3.5 g H₂O/g dry weight), which is stabilized by an organized structural network of insoluble and soluble cellular and extracellular proteins (Fig. 1).⁴² The dry weight of the adult human corneal stroma is made up of collagen (68%), keratocyte constituents (10%), proteoglycans (9%), and salts, glycoproteins, or other substances.⁴³

Collagen is a water-insoluble structural protein organized into an inextensible scaffold that forms the basic structural framework of connective tissues. In the human cornea, stromal collagens are poorly-ordered in the Bowman’s layer (i.e., the primary acellular stroma that is presumably ectodermally-derived), whereas they form a highly-ordered lattice-like structure in the secondary cellular stroma (neural crest-derived).⁴⁴ Overall, the corneal collagens are functionally important in establishing tissue transparency and in resisting tensile forces because collagen fibrils and filaments hold the cornea together (i.e., cohesive strength), ultimately defining the size of the tissue.

Keratocyte components make up the second major component of the cornea’s dry weight. Interspersed between collagen lamellae throughout the secondary cellular stroma, keratocytes form a closed, highly-organized syncytium. Keratocytes are neural crest-derived cells that enter the cornea during the second mesenchymal wave of corneal embryologic development. They function as modified fibroblasts during neonatal life, forming the extracellular matrix of the secondary cellular stroma. Subsequently, they typically remain in the cellular stroma throughout life as modified fibrocytes where they maintain the extracellular matrix of the corneal stroma, usually in an inconspicuous, or relatively transparent, fashion.

Proteoglycans, which are water-soluble glycoproteins, are the third major component of the cornea’s dry weight. Each proteoglycan molecule consists of a protein core with a covalently attached anionic polysaccharide side chain called a glycosaminoglycan (GAG). The proteoglycans are three-dimensionally ordered in the corneal stroma because the protein core of the molecule is noncovalently attached to collagen fibrils evenly throughout the tissue, whereas the GAG sidechain extends into the interfibrillar space where it acts as a pressure-exerting polyelectrolyte gel.^45,46,47 After Hedbys showed that the cornea collapses to around 20% of its volume if the proteoglycans in the tissue are precipitated with cetylpyridinium,⁴⁸ it has become quite apparent that the primary function of proteoglycans is to provide tissue volume, maintain spatial order of collagen fibrils, and resist compressive forces. Water, collagens, proteoglycans, and keratocytes work together to maintain and establish a transparent cornea, while also creating a tough and resistant structure that keeps ocular integrity intact and maintains a stable shape.

Interestingly, although the cornea primarily absorbs light between 200 to 295 nm (ultraviolet)⁴⁹ and transmits variable amounts of light from around 300 nm in wavelength (ultraviolet) to 2,500 nm in wavelength (infrared)⁵⁰, peak transmission (95% to 99% of the incident light) occurs along wavelengths where light is visible to humans (visible spectrum of light = 400 to 700 nm). The remaining portion that is not directly transmitted (1% to 5%) is scattered in all directions by the cornea in a wavelength dependent fashion or absorbed by the tissue (Fig. 8).⁵¹ Clinical slit-lamp examination and in vivo confocal microscopy suggests that most of the scattered light (i.e., reflected and scattered light) comes from cellular components in the cornea (endothelial cells > epithelial cells > nerve cells > keratocytes) as opposed to collagen fibrils (Figs. 1B and 8, bottom).^52,53

Collagen molecules measure 1.5 nm in width by 300 nm in length and are composed of a triple helix of three alpha chains (Fig. 9, top). There currently are 19 known collagen types as determined by the combination of alpha-chains that form the collagen molecule. The most common types usually aggregate into structural, banded (repeating pattern of polar amino acids seen on transmission electron microscopy [TEM]) fibrils by ordering themselves into a quarter-staggered parallel arrangement that is further stabilized in position by covalent intermolecular cross-links (Fig. 9, middle).

Collagen fibrils are generally heterotypic (composed of two or more types of collagen molecules) and reach certain diameters based on their composition of collagen types. The fibrils in the corneal stroma are a co-polymerization of collagen types I, III, and V molecules that form uniform 22-nm diameter fibrils in the acellular Bowman’s layer and uniform 25-nm diameter fibrils in the cellular corneal stroma with only slight variability (Figs. 9, bottom and 10).^54,55 Although the refractive index of collagen fibrils (1.47) is different from that of the extrafibrillar matrix (1.35), the highly uniform size (25 ± 2 SD nm) and interfibrillar spaces (40 ± 5 SD nm) along with the predominantly parallel directionality of these fibrils results in a highly-ordered lattice-like array of fibrils (i.e., not a true crystalline lattice, but one with short-range order) that allows transparency of the cornea due to destructive interference effects (Fig. 10B, C).⁵⁶

Collagen type VI is the third most common type of collagen in the corneal stroma, but is unique in that it is only able to aggregate into repeating tetramers of type VI molecules. Thus, it forms only 10 to 15 nm diameter, beaded (20 × 30 nm diameter ovals with a periodicity of 100 nm), nonbanded filaments (Fig. 11A). Functionally, it acts as a bridging filament that binds corneal lamellae together where they cross each other (Fig. 11B, C). Along with fibril-associated collagens with interrupted triple helices (FACIT collagens, type XII and XIV collagen molecules), it also bridges intralamellar fibrils together (Fig. 11D).^57,58 Overall, this three-dimensional, supramolecular scaffold created by human corneal stromal collagens results in a one-dimensional ordered (∼22 nm diameter collagen fibrils; ∼ 40 nm interfibrillar spaces × random directionality) 12-μm thick acellular Bowman’s layer (Figs. 12A and 13A) and a three-dimensionally-ordered (∼25 nm diameter collagen fibrils; ∼ 40 interfibrillar spaces × parallel directionality) series of successive stacks of lamellae in the cellular stroma measuring in the center of cornea around 450 μm in thickness.^59,60,61

Although difficult to count exactly, the central cornea has been found to have approximately 300 corneal lamellae, whereas the peripheral cornea has approximately 500.⁶¹ Although these lamellae are generally described as running parallel to the corneal surface in orthogonal directions to one another, regional difference exist in their size, directionality, and amount of interweaving. The anterior third of the cellular corneal stroma has thinner (0.2- to 1.2-μm thick), narrower (0.5- to 30-μm wide), and mostly obliquely oriented lamellae with extensive vertical and horizontal interweaving (Figs. 12B and 13B), whereas the posterior two-thirds has thicker (1- to 2.5-μm thick), wider (100- to 200-μm wide), and mostly parallel oriented lamellae with only slight horizontal interweaving (Figs. 12C and 13C).⁶⁰

Additionally, only lamellae in the posterior two-thirds of the cellular cornea stroma run in the classic orthogonal directions, whereas all of the lamellae of the anterior third and a minority of lamellae in the posterior two-thirds run in nonorthogonal directions. Because the lamellae in the anterior-most layers of cellular stroma branch so extensively, it is not surprising that they also interweave with (i.e., are attached to) the Bowman’s layer in a polygonal fashion, creating a mosaic appearance that can be seen on the anterior corneal surface under certain circumstances (Fig. 12A, inset). Finally, all corneal lamellae appear to stretch across the cornea from limbus to limbus where they turn and form an annulus approximately 1.5- to 2.0-mm wide around the cornea. This maintains the curvature of the cornea, while blending with limbal collagen fibrils.^61,62,63

Corneal proteoglycans previously were referred to as extrafibrillar amorphous ground substance because their water soluble state made them difficult to fully delineate with light and electron microscopy (Fig. 14A). It was not until an electron-dense, cationic dye called cupromeronic blue and a critical-electrolyte-concentration of 0.1 M MgCl₂ were used in combination to stain specifically for the sulfate-ester groups on corneal stromal proteoglycans that the shape, size, arrangement, and location of this material was observed with light and electron microscopy (Fig. 14B).⁶⁴ Since that time, it has become quite apparent that corneal proteoglycans are not amorphous, but rather are tadpole-shaped structural molecules composed of a globular core protein (10 to 15 nm in diameter) with a covalently attached GAG sidechain or tail (7 nm wide × 45 to 70 nm in length). They are arranged in the cornea stroma orthogonal to collagen fibrils with a constant separation around 65 nm between each other along the fibrils. Their core proteins noncovalently bind to collagen fibrils in specific gap zones along the peripheral portions of the collagen fibril (core proteins with dermatan sulfate sidechains bind to d and e gap zones and those with keratan sulfate side chains bind to a and c gap zones).⁶⁵ GAGs are highly negatively-charged, stiff polymers that extend into the interfibrillar space and form antiparallel duplexes with adjacent GAG sidechains, thereby crosslinking different collagen fibrils together by forming dumbbell-like structures (Fig. 14C).

Because the genes that produce the core proteins have now been cloned, four types of corneal proteoglycan core proteins have been identified: decorin, lumican, keratocan, and mimecan.⁶⁶ Subsequently, decorin was found to contain a single dermatan sulfate GAG sidechain, whereas lumican and mimecan had single keratan sulfate GAG sidechains and keratocan had three keratan sulfate GAG sidechains. Thus, there are four known types of proteoglycan core proteins and only two types of GAGs: keratan sulfate (60%) and dermatan sulfate (40%), found in the human cornea stroma. The GAGs are polymers of repeating disaccharide units of galactose and N-acetylglucosamine or iduronic acid and N-acetylgalactosamine, respectively (Fig. 14D).⁶⁵ Because GAG sidechains are posttranslationally added to the core protein portions in the golgi apparatus, there seems to be some flexibility in how long or how sulfated they can become depending of the type and functional needs of the connective tissue producing them. The human cornea is unique in that the GAGs are fibril-associated, small in length (keratan sulfate ≅45 nm and dermatan sulfate ≅70 nm), and with a higher proportion are oversulfated.

Interestingly, a comparative study of corneas from 12 mammalian species suggests that dermatan sulfate is the preferred proteoglycan in oxygen-rich environments such as thin corneas (e.g., mice) or in the anterior portion of thicker corneas (e.g., humans or rabbits), whereas keratin sulfate is a functional substitute produced through an alternate metabolic pathway in thicker corneas, especially in the posterior portion where oxygen levels may be dramatically reduced (Fig. 15, inset).^48,66 Functionally, this duality is quite useful because dermatan sulfate appears to be more efficient at holding water as it absorbs less total amounts of water, but most in a tightly-bound state (i.e., in a nonfreezable state) than does keratin sulfate.^47,67 These explanations are consistent with the fact that dermatan sulfate is highest in amount in the anterior portion of the corneal stroma in humans (region of highest oxygen tension and most affected by evaporation), whereas keratin sulfate is highest in amount in the posterior portion of the corneal stroma (region with the lowest oxygen tension, least affected by evaporation, and area where the need for loosely-bound water is required for transport across the endothelium through the endothelial cell metabolic pump) (Fig. 15).

The corneal stroma is maintained by a closed, highly-organized syncytium of keratocytes communicating with each other through gap junctions present on their long dendritic processes.⁶⁸ Keratocytes occupy 10% to 40% of the stromal volume (decreases from 40% in infancy to 10% in adulthood) that on two-dimensional, cross-sectional views appear as flattened (cell body = 20 μm in length × 1 μm in height), quiescent (i.e., scant intracytoplasmic organelles) cells lying between corneal lamella (Fig. 16A, B). In actuality, keratocytes are three-dimensional, stellate-shaped cells composed of a cell body (≅1 × 15 × 20 μm) with numerous dendritic-processes that extend up to 50 μm in length from the cell body. Two-dimensional tangential sections of the normal cornea suggest that these cells are more highly metabolically active in the resting state than initially presumed; in tangential sections (cell body view = 15 μm width × 20 μm), an abundance of cytoplasmic organelles is commonly seen (Fig. 16C, D).⁶⁹

Tangential sections also have shown more clearly that the anterior stromal keratocytes contain twice the number of mitochondria than the posterior two-thirds of the stroma, which correlates with the oxygen tension levels in the cornea. It also has demonstrated that a higher density or volume of cells reside in the anterior stroma than in the mid or posterior stroma (Fig. 17). Moreover, these views make it easy to appreciate that, in all levels, the keratocytes are highly spatially-ordered as they turn in a clockwise direction like a corkscrew.

In vivo confocal microscopy of normal human corneas has shown keratocyte densities average around 20,000 keratocytes/mm³ (or an area of 328 cells/mm²). The focal zone of increased cell density directly under the epithelial surface averages 35,000 keratocytes/mm³ in the anterior-most 9-μm layer and gradually tapers to 20,000 keratocytes/mm³ over the initial 60 μm in depth (Fig. 17C).⁷⁰ Confocal microscopy also concurs with previous studies as it shows that stromal cellular density decreases with age at approximately 0.45% per year of life. Interestingly, studies using immunohistochemistry on animal corneas^71,72 and ultrastructural studies on human corneas^69,72 suggest that not all the cells in the corneal stroma are keratocytes (i.e., fibrocytes), but some are one of three types of bone marrow-derived immune cells (“professional” dendritic cells, “nonprofessional” dendritic cells, and histiocytes) (Fig. 18). Thus, although the majority of cells in the stroma are keratocytes, which maintain the corneal stroma by synthesizing and secreting collagen type I, III, V, VI, XII, XIV, keratan sulfate, dermatan sulfate, and matrix metalloproteinases, a minority of cells in the stroma are also “professional” stromal dendritic cells, “nonprofessional” stromal dendritic cells, or stromal histiocytes. They appear to play a pivotal role in the induction of immune tolerance versus immune initiation into cell-mediated immune pathways. Additionally, the stromal histiocytes play a role in innate immunity as phagocytic effectors cells.

Two ion channels similar to those found in excitable cells have been characterized in keratocytes isolated from corneal stroma.⁷³ These channels include a delayed rectifer K⁺ channel and a tetrodotoxin-sensitive Na⁺ channel. The physiologic function of these channels has not been determined, but they may function in maintaining the ionic balance of the cells in their inactive state. In addition, because these channels allow the keratocytes to elicit action potentials when injected with a clamping current, it is possible they may act as intercellular signalers via the gap junctions between cells. This could be a route for information exchange between cells in different regions of the cornea.

The endothelium of the infant cornea is composed of a single layer of approximately 500,000 neural crest-derived cells, each measuring around 5 μm in thickness by 20 μm in diameter and cover a surface area of 250 μm².^58,85 The cells lie on the posterior surface of the cornea and form an irregular polygonal mosaic. The tangential appearance of each corneal endothelial cell is uniquely irregular, usually uniform in size to one another, and typically six-sided (i.e., hexagons). They abut one another in an interdigitating fashion with 20 nm wide intercellular space between each other. The intercellular space is known to contain discontinuous apical tight junctions (macula occludens) and lateral gap junctions, thereby forming an incomplete barrier to diffusion of small molecules. As corneal endothelial cells have numerous cytoplasmic organelles, particularly mitochondrial organelles, they have been inferred to have the second highest aerobic metabolic rate of cells in the eye next to retinal photoreceptors.⁵⁸

At birth, the central endothelial cell density of the cornea is around 5,000 cells/mm².⁸⁵ Because the corneal endothelium has very limited regenerative capabilities, there is a well-documented decline in central endothelial cell density with age that typical involves two phases: a rapid and a slow component.^85,86 Due to corneal growth and age-related, or developmental, selective cell death, during the fast component, the central endothelial cell density decreases exponentially to about 3,500 cells/mm² by age 5 years and 3,000 cells/mm² by age 20 years.⁸⁶ Thereafter, a slow component occurs where central endothelial cell density decreases to a linear steady rate of 0.6% per year, resulting in cell counts around 2,500 cell/mm² in senescence (Fig. 20).⁸⁵.⁸⁶ Because endothelium maintains its continuity by migration and expansion of surviving cells, it is not surprising that the percentage of hexagonal cells decreases (pleomorphism) and the coefficient of variation of cell area increases (polymegathism) with age.⁸⁶

It is important when reviewing this information to realize that these are average central corneal endothelial cell measurements from predominantly Caucasian U.S. populations, as racial and geographic differences can exist. Also, understand that this data applies only to central corneal endothelial measurements because recent work has shown that higher cell densities can typically be found in more peripheral aspects of the cornea (Fig. 21).⁸⁷ Therefore, based on these studies, it appears that central corneal endothelial cell numbers decrease on average about 50% from birth to death in normal subjects. Because corneal decompensation typically does not occur until central values reach around 500 cells/mm², there appears to be plenty of cellular reserve potential remaining after an average human life span.⁸⁶ Estimates suggest that healthy, normal human corneal endothelium could maintain corneal clarity up to a minimum of 215 years of life, if humans lived that long.⁸⁵