Cultivation of Corneal Epithelial Cells on Intact and Denuded Human Amniotic Membrane

Abstract

purpose. Surgery to reconstruct the ocular surface is greatly facilitated by the use of amniotic membrane, either as a biologic drape or, more recently, as a substrate for the transplantation of cultivated corneal epithelial cells. This study was designed to compare the usefulness of intact and denuded human amniotic membranes as a substrate for corneal epithelial cell culture.

methods. Small (3-mm-diameter) biopsy specimens of superficial cornea including epithelium were excised from the central and limbal regions in rabbits. They were cultured on human amniotic membrane with or without amniotic epithelial cells and examined by light, scanning electron, and transmission electron microscopy.

results. Cellular outgrowth from the central explants (n = 10) after 14 days in culture measured 1.82 ± 2.62 mm² on intact amniotic membrane and 131.83 ± 28.31 mm² on denuded amniotic membrane. In contrast, outgrowths from the limbal explants (n = 10) at the same time measured 4.58 ± 4.56 and 505.39 ± 134.20 mm² on intact and denuded amniotic membranes, respectively. The leading edges of the outgrowths on intact amniotic membrane were much less uniform than those on denuded amniotic membrane, and, in the former, corneal epithelial cells appeared to migrate over the top of amniotic epithelial cells. Limbal cells cultivated on denuded amniotic membrane formed a nicely stratified layer that adhered well to the underlying amniotic membrane.

conclusions. Denuded amniotic membrane appears to be an excellent substrate for the cultivation of corneal epithelial cells, with a view to transplantation.

The corneal epithelium is a nonkeratinized, mucosal epithelial multilayer that covers the front surface of the cornea and is essential for proper vision. In healthy corneas the epithelium is in a constant state of turnover: Superficial epithelial cells are shed, whereas cells derived from stem cells in the limbus migrate in and replenish the epithelial cell population. ¹ In a severely diseased or injured cornea, however, in which the limbal and central epithelia are both absent, it is the neighboring conjunctival epithelial cells that invariably end up covering the corneal surface. ^{2 3} Because corneal and conjunctival epithelia are formed by two phenotypically distinct cell lines, ^{4 5 6 7 8 9} reepithelialization of the corneal surface by conjunctival cells is invariably accompanied by chronic inflammation, persistent epithelial defects, stromal scarring, and neovascularization. 

An increasingly popular surgical procedure for ocular surface reconstruction in individuals with severe thermal or chemical burns or serious ocular surface disorders, such as Stevens–Johnson syndrome, ocular cicatricial pemphigoid, and recurrent pterygium, ^{10 11 12 13 14 15 16} involves the use of preserved human amniotic membrane as a biologic drape to dress the bare stroma after the removal of abnormal conjunctival tissue. This approach is based on the rabbit model reported by Kim and Tseng, ¹⁷ in which the amniotic membrane is thought to inhibit conjunctival overgrowth and provide a good substrate for normal epithelial migration. The results of ocular surface reconstruction with amniotic membrane are generally good. Moreover, we, along with others, ^{10 13 15} find that in ocular surface disorders with stem cell deficiencies the use of limbal transplantation and keratoepithelioplasty in conjunction with amniotic membrane transplantation is often highly successful. Thus, it seems evident that the combination of amniotic/corneal epithelial cell transplantation is a potentially powerful one, and, recently, we demonstrated in rabbits the feasibility of cultivating corneal epithelial cells on amniotic membrane and transplanting them onto injured eyes with epithelial stem cell deficiencies. ¹⁸ The present study was designed to discover whether central or limbal corneal epithelial cells grown in vitro populate amniotic membrane more readily, and whether intact or denuded amniotic membrane promotes better epithelial colonization. 

Methods

Preparation of Amniotic Membrane

In accordance with the tenets of the Declaration of Helsinki and with proper informed consent, human amniotic membranes were obtained at the time of Cesarean section. Under sterile conditions the membranes were washed with sterile phosphate-buffered saline (PBS) containing antibiotics (5 ml of 0.3% ofloxacin) and stored at −80°C in Dulbecco’s modified Eagle’s medium (GIBCO BRL, Rockville, MD) and glycerol (Nacalai Tesqu, Kyoto, Japan) at the ratio of 1:1 (vol/vol). Immediately before use, the amniotic membrane was thawed, washed three times with sterile PBS, and cut into pieces approximately 2.5 × 2.5 cm in size. Some of the membranes were then deprived of their amniotic epithelial cells by incubation with 0.02% EDTA (Waco Pure Chemical Industries, Osaka, Japan) at 37°C for 2 hours to loosen the cellular adhesion, followed by gentle scraping with a cell scraper (Nalge Nunc International, Naperville, IL). Preliminary experiments on hematoxylin–stained, ethanol–fixed tissues confirmed that this protocol effectively removed epithelial cells from the amniotic membrane. Tissues were then washed twice more with sterile PBS. 

Primary Cultures of Corneal Epithelial Cells

Corneal epithelial cells were cocultured with inactivated 3T3 fibroblasts using established techniques, ¹⁹ with the following modifications: Inactivation was achieved by mitomycin C treatment, and the level of the culture medium was lowered after 14 days in culture to expose the upper surface of the cells to air (air-lifting). As a first step, confluent 3T3 fibroblasts were incubated with 4 μg/ml of mitomycin C for 2 hours at 37°C under 5% CO₂. These were then trypsinized and plated onto plastic dishes with a density of 2 × 10⁴ cells/cm². Intact and denuded amniotic membranes (measuring approximately 2.5 × 2.5 cm) were spread, epithelial basement membrane side up, on the bottom of polycarbonate membrane culture inserts (Iwaki Glass, Chiba, Japan), and these inserts were placed in dishes containing treated 3T3 fibroblasts. Biopsy specimens of anterior rabbit cornea, 3 mm in diameter and 100-μm-thick, were excised under anesthesia from the central and limbal areas of five rabbits housed and treated in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and according to an experimental procedure approved by the committee for Animal Research at Kyoto Prefectural University of Medicine. Four explants were obtained from the center of each cornea, and eight explants from the limbus. Each explant was placed directly, epithelial side up, on a portion of amniotic membrane (either intact or denuded) spread on a polycarbonate culture insert. The amniotic membrane containing the corneal explants was separated from the 3T3 fibroblasts by the polycarbonate membrane of the culture insert. The culture medium used was Dulbecco’s modified Eagle’s medium and Ham’s F-12 media (1:1 mixture) containing fetal bovine serum (10%), insulin (5 mg/ml), cholera toxin (0.1 nM), epidermal growth factor (10 ng/ml), and penicillin–streptomycin (50 IU/ml). Cultures were incubated at 37°C in a 5%CO₂–95% air incubator for up to 28 days with the medium changed every 2 days. Explants were left in the culture dish for the duration of the incubation. 

Measurement of Cellular Outgrowth

After 5 days of incubation, the extent of cellular outgrowth from the explant was measured on 20 separate cultures. Of these 20, 5 were central cornea on intact amniotic membrane, 5 were central cornea on denuded amniotic membrane, 5 were limbal cornea on intact amniotic membrane, and 5 were limbal cornea on denuded amniotic membrane. Within each group, cultures were selected at random. This process was repeated (using different cultures) after 7 days of cultivation and after 14 days of cultivation. Cultures were prepared as previously described ²⁰ to measure the area covered by the cultivated cells (the area of epithelial outgrowth). In brief, cultures were fixed in absolute ethanol (15 minutes), stained for 5 minutes with hematoxylin (Merck, Darmstadt, Germany), washed, and dried on glass slides. Photographs of the cultures were taken, and the area covered by cultivated cells was measured using a computerized image processing package (NIH image). 

Light and Electron Microscopy

Cultures of central and limbal corneal epithelial cells on intact and denuded amniotic membranes were examined by light, scanning electron, and transmission electron microscopy. On days 5, 7, 14, 21, and 28, two cultures on intact amniotic membrane and two cultures on denuded amniotic membrane were fixed in 2.5% glutaraldehyde in 0.1 M PBS, along with portions of intact and denuded amniotic membranes on which no cells had been grown. Specimens were then washed three times for 15 minutes in PBS, post-fixed for 2 hours in 2% aqueous osmium tetroxide, and washed three more times in PBS. After dehydration through a graded ethanol series, specimens were transferred to hexamethyldisilazane (Agar Scientific, London, UK) and allowed to air-dry. When dry, they were mounted on aluminum stubs and sputter-coated with gold before being examined on a JEOL 840a scanning electron microscope (Japanese Electron Optical Limited, Tokyo, Japan). 

After 28 days in culture, intact and denuded amniotic membranes containing cultivated central and limbal epithelial cells were prepared for light and transmission electron microscopy. In brief, tissues were fixed in 2.5% glutaraldehyde in 0.1 M PBS, post-fixed in 2% aqueous osmium tetroxide, dehydrated through a graded ethanol series, and embedded in Agar 100 epoxy resin (Agar Scientific). Semi-thin (1μ m) sections for light microscopy were collected on glass slides and stained for 30 seconds with toludine blue; ultrathin (70 nm) sections were collected on copper grids and stained for 40 minutes each with uranyl acetate and phosphotungstic acid. 

Results

An inspection of the cultures clearly shows that epithelial outgrowth from corneal (central and limbal) explants onto amniotic membrane was considerably quicker when the amniotic epithelial cells had been removed from the membrane before cultivation was begun (Fig. 1) . Also, epithelial cells migrating onto amniotic membrane from limbal explants colonize the membrane more rapidly than cells from central explants (Fig. 2) . Our data reveal that the area of amniotic membrane covered by outwardly migrating epithelial cells differs considerably when cells are grown on intact versus denuded amniotic membranes (Fig. 2) . Also, the areas covered by cells taken from central versus limbal regions of the donor corneas are markedly different (Fig. 2) . After 5 days in culture, epithelial cells from the central corneal explants (n = 5) had covered 0.35 ± 0.06 mm² of the intact amniotic membrane compared with 60.04 ± 13.01 mm² of the denuded amniotic membrane, whereas cells from the limbal explants (n = 5) had covered 0.60 ± 0.93 mm² of the intact amniotic membrane versus 59.80 ± 8.80 mm² of the denuded amniotic membrane. This pattern continued, and after 14 days in culture epithelial cells from the central corneal explants (n = 5) had covered 1.82 ± 2.62 mm² of the intact amniotic membrane as opposed to 131.83 ± 28.31 mm² of the denuded amniotic membrane, whereas cells from the limbal explants (n = 5) had covered 4.58 ± 4.56 mm² of the intact amniotic membrane versus 505.39 ± 134.20 mm² of the denuded amniotic membrane. 

As well as differences in the rates of cellular outgrowth on intact versus denuded amniotic membranes, we also noted differences in the nature of the cells at leading edges of the outgrowths. This is exemplified in low-magnification light micrographs of wholemounts, viewed from above (Figs. 3A and 3B ), which indicate that the leading edge of the outgrowth from a limbal explant on denuded amniotic membrane is nicely uniform over fairly large stretches (Fig. 3B) , whereas outgrowth on an intact amniotic membrane is considerably more irregular (and, as mentioned, a lot closer to the explant; Fig. 3A ). Higher magnification images (Figs. 3C and 3D) also demonstrate the irregularity of the leading edge on intact amniotic membrane and show corneal cells that appear to migrate over the top of the underlying amniotic epithelial cells (Fig. 3C) . 

An examination of the surface of the intact amniotic membrane by scanning electron microscopy reveals it to be completely covered by a layer of amniotic epithelial cells, 5 to 15 μm in size, with numerous microvilli and blebs on their apical surfaces (Fig. 4) . This is in line with previous studies. ²¹ The surface of the denuded amniotic membrane, on the other hand, contained no amniotic epithelial cells and consisted instead of exposed extracellular matrix (Fig. 4) . Scanning electron microscopy of the in vitro cultures indicated that cells were indeed migrating on both intact and denuded amniotic membranes and that they were becoming more numerous and more connected with time (Fig. 5) . Moreover, it appeared from the micrographs of cultures on intact amniotic membrane that corneal epithelial cells were migrating over the top of the existing amniotic epithelial cells (Fig. 5) , an impression that was supported by the flatmounts (Fig. 3) . After 28 days in culture both intact and denuded amniotic membranes contained regions covered by confluent, polygonal epithelial cells, 10 to 20 μm in size (Fig. 5) . 

A light microscopic examination of limbal epithelial cultures grown on intact amniotic membrane for 28 days revealed that the epithelial cell layer averaged some four to six cells thick (Fig. 6) , although at the leading edge, cells were more numerous and vacuolated (data not shown). Often, the most superficial cells appeared to be in the process of desquamation. Transmission electron microscopy of the same specimens revealed that corneal epithelial cells had migrated over the top of the preexisting amniotic epithelial cells and formed a confluent multilayer (Fig. 6) . The columnar amniotic epithelial cells could be readily distinguished from the overlying corneal epithelial cells of limbal origin because of their characteristic morphology, ^{21 22} specifically the presence in the former of abundant organelles including cisternal endoplasmic reticulum and Golgi apparatus. Desmosomal contacts were evident between amniotic epithelial cells, between corneal epithelial cells, and, interestingly, between corneal epithelial and underlying amniotic epithelial cells (Fig. 6) . 

Limbal explants grown on denuded amniotic membrane for 28 days exhibited an epithelial multilayer approximately six cells thick with a stratified appearance similar to that of normal corneal epithelium in which the basal cells are clearly more columnar than those above (Fig. 7) . The epithelium appeared to be nicely adhered to the underlying amniotic membrane, and, again, some superficial cells appeared to be in the process of shedding. The basal epithelial cells grown on denuded amniotic membrane (Fig. 7) took on a more columnar appearance than the deepest corneal epithelial cells grown on intact amniotic membrane (i.e., those lying immediately on top of the amniotic epithelial cells; Fig. 6 ). Desmosomal contacts were evident between limbal corneal epithelial cells cultured for 28 days on denuded amniotic membrane (Fig. 7) . Furthermore, the basal cells seemed to attach fairly well to the underlying extracellular matrix, although a well-defined basement membrane of the type seen in cornea had not reformed (Fig. 7) . 

Discussion

Human amniotic membrane is a widely used surgical material, ²³ and in recent years there has been a renewed interest in its use for ocular surface reconstruction. ²² It is a readily available tissue with many advantages for clinical use, not the least of which is its role in immune tolerance and expression of all classic HLA class I molecules together with HLA-G. ²⁴ Amniotic membrane acts so as to inhibit conjunctival overgrowth and provide a good substrate for normal epithelial migration. ¹⁷ Its basement membrane contains collagens ²⁵ as well as several adhesive glycoproteins found in corneal and conjunctival epithelial basement membranes, ²⁶ and amniotic epithelial cells themselves contain numerous growth factors thought to enhance corneal wound healing. ²⁷  

A recent development in the treatment of severe ocular surface disorders with stem cell deficiencies is the use of amniotic membrane on which corneal epithelial cells have been cultivated in vitro. ^{18 28} The benefit of transplanting cultivated corneal epithelial cells derived from the limbus onto diseased or injured eyes that have no corneal epithelial cells of their own has been illustrated recently in a small case series by Pellegrini and associates, ²⁹ who used petrolatum gauze or a soft contact lens as a carrier material for the cultivated cells. The transplantation of cultivated corneal epithelial cells in this manner is especially useful because potential immune complications can be avoided by the use of autologous grafts. ^{29 30} The feasibility of successfully grafting corneal epithelial cells grown on amniotic membrane onto bare corneal stroma was established in our initial study in rabbits. ¹⁸ We believe that the success of these animal surgeries, and the good early outcomes from our initial series of patients treated in this way (at the time of writing, we have successfully reconstructed the corneal surface in 13 eyes with severe limbal stem cell deficiencies by transplanting cultured human allo-corneal epithelial cells on denuded amniotic membrane), is due, in part, to the fact that the amniotic membrane may decrease the antigenicity of the corneal epithelial cells cultivated on it, a possibility that we are currently investigating. 

The present study was designed to assess the relative merits of cultivating corneal epithelial cells on intact versus denuded amniotic membrane and to investigate how central and limbal corneal explants migrate, adhere, and stratify on amniotic membrane. The results clearly indicate that denuded amniotic membrane promotes better corneal epithelial cell colonization than intact amniotic membrane does and that corneal cells from the limbal epithelium colonize denuded amniotic membrane more readily than epithelial cells from the central cornea. 

Amniotic membrane, especially denuded amniotic membrane, has proven to be more amenable for the cultivation of corneal epithelial cells than other substrates we have tried, ^{31 32 33} including denuded corneal stroma and type IV collagen sheets. Furthermore, the finding that, in vitro, corneal cells grow on denuded amniotic membrane better than they do on intact amniotic membrane is in line with our clinical experience, which shows that after combined amniotic membrane/corneal epithelial transplantation, cell migration from the epithelial grafts is slower on cellular amniotic membrane than on denuded corneal stroma. As mentioned, intact amniotic membrane is colonized by corneal cells much less quickly than denuded amniotic membrane. Moreover, wholemounts of the outwardly migrating limbal epithelial cells on denuded amniotic membrane reveal that they have a smooth, uniform leading edge. The leading edge of the limbal cells grown on intact amniotic membrane, on the other hand, is highly irregular, a finding that is perhaps reflective of the fact that corneal cells are migrating over the apical surface of the amniotic epithelial cells (Figs. 3 5 and 6) , a substrate that is likely to be less conducive to the migration of corneal epithelial cells than the exposed extracellular matrix with its collagens and complement of adhesive glycoproteins. ^{25 26} Our impression that corneal epithelial cells grown on intact amniotic membrane migrate over the top of amniotic epithelial cells is lent further backing by our recent immunohistochemical findings (unpublished data) of cornea-specific keratin 3 and keratin 12 in human epithelial cells cultivated in this manner. 

The better suitability of denuded amniotic membrane for the support of corneal epithelial cell growth is strengthened by morphologic observations of the epithelial multilayer grown on denuded as opposed to intact amniotic membrane. We find that the limbal epithelial cells that migrate slowly over the amniotic epithelial cells do not take on the appearance of normal corneal epithelium, whereas those grown on the exposed amniotic extracellular matrix do. Specifically, the basal cells on bare amniotic membrane are nicely columnar, and the more superficial cells seem fairly well differentiated into wing cells and surface cells (Fig. 7) . In contrast, the deepest limbal cells on the intact amniotic membrane (i.e., those on top of the amniotic epithelial cells) do not adopt such a columnar appearance (Fig. 6) , and the corneal epithelial cell layer, in general, doesn’t appear as well stratified as that grown on denuded amniotic membrane (Fig. 7) . In cultures of limbal cells grown for 28 days on intact and denuded amniotic membranes, cell–cell contact via desmosomes was noted between corneal cells, between amniotic epithelial cells, and, it appears, between both cell types (Figs. 6 and 7) . 

In the early stages of the in vitro cultivation, we found that corneal explants adhered well to the underlying denuded amniotic membrane. On the other hand, a proportion of the explants grown on intact amniotic membrane became dislodged during routine handling. After 28 days in culture on denuded amniotic membrane, basal cells derived from limbal explants seemed to attach fairly well to the underlying extracellular matrix, and some evidence of hemidesmosome-type structures was found, although a well-defined basement membrane of the type present in cornea had not reformed (Fig. 7) . 

When we began this line of research, we tried to culture the corneal epithelium without a 3T3 fibroblast feeder layer but were unable to obtain a well stratified/differentiated cellular multilayer. However, after changing to a coculture system with 3T3 fibroblasts based on the keratinocyte culture system established by Rheinwald and Green, ¹⁹ we found that we were able to create a cellular multilayer that better resembled corneal epithelium. Moreover, we now find that when limbal explants from humans are used, cells express cornea-specific keratin 3 and keratin 12. It is clear, then, that the presence of 3T3 fibroblasts is beneficial for the cultivation of corneal epithelial cells on denuded amniotic membrane, exactly why has not been established. It is also noteworthy that the limbal cells grown on denuded amniotic membrane in the present set of experiments form a more well-differentiated and better stratified multilayer than those in our previous set of experiments. ¹⁸ What are the reasons for this? Perhaps because in the present study we limited the damage to the membrane by removing the amniotic epithelium with EDTA. Previously, ¹⁸ we had used Dispase II to remove amniotic epithelial cells using a modified technique described by Gipson and Steven. ³⁴ Dispase II is a bacterial neutral protease that acts at the level of hemidesmosome basement membrane attachments ³⁴ and is an enzyme that can be used to obtain whole sheets of viable corneal epithelium. However, the possibility exists that it might damage the basement membrane of the amniotic membrane. EDTA, on the other hand, is a solution that influences cell–cell contacts regulated by calcium concentration, ³⁵ and as such might represent a potentially less disruptive method of removing amniotic epithelial cells from their underlying matrix. Additional possible reasons for the improved cultivation by the present technique include the facts that we left the explant in place for the entire culture period (previously it was removed after 5 days in culture ¹⁸ ) and exposed the apical surface of the cultivated cells to air (air-lifting) after 14 days in culture. 

We conclude that amniotic membrane with its amniotic epithelial cells still presents a valuable surgical material when spread on bare sclera to inhibit conjunctival overgrowth; however, amniotic membrane with epithelial cells removed appears to represent the more amenable substrate for the cultivation in vitro of corneal epithelial cells with a view to transplantation. 

 Supported by grants from the Japanese Ministry of Health and Welfare and the Japanese Ministry of Education (10470365), Tokyo; the Kyoto Foundation for the Promotion of Medical Science; the Intramural Research Fund of the Kyoto Prefectural University of Medicine; and the Royal Society, London.

 Submitted for publication January 7, 2000; revised March 14, 2000; accepted March 22, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

Cellular outgrowths from 3-mm-diamter limbal corneal explants (arrow) after 5 days in culture. On the intact amniotic membrane (A), only very limited outgrowth of corneal cells is evident (arrowheads). A much larger region of the denuded amniotic membrane (B) is covered by corneal epithelial cells (arrowheads). Magnification, ×5.

Figure 2.

 

The area of cellular outgrowth from central and limbal corneal explants on intact and denuded amniotic membranes (mean ± SD) measured from five separate cultures in each case.

Figure 3.

 

Wholemount of cultivated limbal corneal epithelium on amniotic membrane. After 7 days in culture, the leading edge of the migrating corneal epithelium on intact amniotic membrane (A) is irregular, whereas the leading edge on denuded amniotic membrane (B) is smooth and uniform. At higher magnification we can see the darkly stained corneal epithelial cells apparently migrating over the top of underlying, lightly stained amniotic epithelial cells (C). Corneal cells on denuded amniotic membrane migrate on a homogeneous exposed matrix (D). Magnification, (A, B) ×16; (C, D) ×50.

Figure 4.

 

Scanning electron micrographs of preserved amniotic membrane before culture. Numerous polygonal amniotic epithelial cells are evident on the intact membrane (A). Cells have been successfully removed from the denuded amniotic membrane (B). At higher magnifications, numerous microvillae and blebs (arrow) are evident on the apical surface of the amniotic epithelial cells (C), whereas the surface of the denuded amniotic membrane consists of exposed extracellular matrix (D). Magnification, (A, B) ×1250; (C, D)× 6600.

Figure 5.

 

Scanning electron micrographs of intact amniotic membrane after 7 (A), 21 (C), and 28 (E) days in culture, and on denuded amniotic membrane after 7 (B), 21 (D), and 28 (F) days in culture. With time, cell coverage spreads and more cell–cell contacts are made. On intact amniotic membrane (A, C) corneal epithelial cells appear to migrate over the surface of the amniotic epithelial cells below them. At 28 days (E, F) a confluent population of polygonal cells is present on both intact and denuded membranes (although it covers a lot more area in the case of the denuded membrane; see Fig. 2 ). Magnification, ×1200.

Figure 6.

 

Limbal cells grown on intact amniotic membrane for 28 days. (A) The epithelial cell layer is approximately 4 cells thick. (B) Corneal epithelial cells have migrated over the top of amniotic epithelial cells that can be readily identified by their characteristic morphology (specifically, the presence in the former of abundant organelles including cisternal endoplasmic reticulum and Golgi apparatus). (C) Desmosomal contacts are evident between the corneal epithelial cells and the amniotic epithelial cells and (D) between the corneal epithelial cells. Magnification, (A) ×370; (B)× 1,600; (C) ×16,000; (D) ×40,000.

Figure 7.

 

Limbal cells grown on denuded amniotic membrane for 28 days. (A) The epithelial cell layer is approximately 4 to 6 cells thick. It adheres well to the amniotic membrane and appears fairly well stratified/differentiated. (B) Basal corneal cells are nicely columnar; those above them resemble wing cells. (C) Basal corneal epithelial cells appear to adhere fairly well to the loose underlying extracellular matrix, although a complete basement membrane attachment zone is not present. (D) Desmosomal contacts are evident between the corneal epithelial cells. Magnification, (A) ×300; (B) ×1,600; (C) ×9,800; (D) ×23,000.