Disruption of the Basement Membrane after Corneal Débridement

Abstract

purpose. To determine whether the native basement membrane left behind after manual débridement wounding is retained throughout healing in the Balb/c mouse.

methods. Mouse corneas were subjected to either 1.5 mm (small) or limbus-to-limbus (large) epithelial débridement wounds and allowed to heal for times ranging from 12 hours to 3 days. For the larger wounds, care was taken to leave an approximately 0.5-mm zone of epithelial cells near the limbal border. Unwounded corneas served as control specimens. At each time point, confocal immunofluorescence microscopy was used to localize several proteins found in the basement membrane including laminin-5, entactin, and perlecan. In addition, ultrastructural studies were performed using transmission electron microscopy (TEM) to assess the basement membrane zone (BMZ) of the corneas at various times after injury.

results. The smaller (1.5-mm) wounds healed within 24 hours, and the larger wounds healed at approximately 48 hours. Both wound sizes healed with little scarring or neovascularization. At all time points after 1.5-mm wounding, immunofluorescence confocal microscopy and TEM showed that both basement membrane proteins and the lamina densa were retained at the BMZ throughout healing. For the larger wounds, at time points after 24 hours, confocal microscopy showed patches along the denuded corneal stroma where there was a partial or complete loss of basement membrane markers at the BMZ. TEM confirmed that the lamina densa was partly or completely absent along the anterior surface of the exposed cornea at time points of more than 24 hours after the larger wounds.

conclusions. The denuded epithelial basement membrane was shown to be partially disassembled in response to manual débridement wounds when re-epithelialization took more than 24 hours. Regulated disassembly of the epithelial basement membrane probably plays a role in the healing of large-diameter débridement wounds.

Studies of corneal wound healing after manual débridement were first conducted more than 30 years ago and have been shown by numerous groups to leave the lamina densa of the basement membrane intact and in its native state after wounding. ^{1 2 3 4 5} Based on previous studies, ^{6 7} we know that the mouse cornea re-epithelializes after manual débridement between 20 to 24 hours in the 8-week-old Balb/c mouse after wounds removing no more than 40% of the epithelial surface (1.5 mm; small wounds). Re-epithelialization of small wounds is accompanied by increased expression of α6β4 integrin, ⁸ a structural component of the hemidesmosomes and a signaling molecule known to regulate epithelial cell proliferation. ^{9 10 11} Extracellular ligands for α6β4 integrin are members of the laminin family of adhesive glycoproteins, primarily laminin-1 and -5. Laminin-5 is a component of the anchoring filaments of hemidesmosome adhesion complexes. It also has been shown recently to be secreted by migrating epithelial cells in the skin in response to blisters and deeper, more penetrating wounds ¹² and by corneal epithelial cells in response to manual keratectomy wounds. ¹³ Laminin-1 is also a basement membrane component, ¹⁴ and studies have shown that both laminin-1 and 5 interactions with α6β4 can mediate cell migration. ^{12 15 16}  

Increased expression of the mRNA and protein for another integrinα 9β1 accompanies re-epithelialization of larger corneal débridement wounds, but not smaller wounds. ⁷ Unlikeα 6β4, the functions of α9β1 in epithelial cells are unclear. To determine whether alterations in the nature of the substrate, the basement membrane zone (BMZ), are associated with the increased expression of either α6β4 or α9β1 observed in response to corneal débridement wounding, we performed both small and large débridement wounds and evaluated tissues by immunohistochemistry for the localization of several different basement membrane proteins at a variety of different time points after wounding. We also evaluated the morphology of the corneal epithelial basement membrane directly by transmission electron microscopy (TEM). 

Materials and Methods

Animal Model

All experiments described in this article were conducted in voluntary compliance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and all procedures were approved by the George Washington University (GWU) Animal Care and Use Committee. Eight- to 10-week-old male Balb/c mice were anesthetized and a topical anesthetic applied to the ocular surface and the corneas scraped with a dull scalpel to remove the corneal epithelium within a 1.5-mm central corneal area demarcated with a trephine (small wound) or from limbus to limbus, taking care to avoid limbal blood vessels (large wound). This procedure has been determined previously to leave the basement membrane intact and nondenatured. The corneas were allowed to heal in vivo for 12, 18, and 24 hours (small wounds) or for 1, 1.5, 2, 3, 6, or 10 days (large wound). After mice were killed by lethal injection, eyes were enucleated and corneas dissected and frozen in O.C.T. embedding medium (Tissue Tek II; Laboratory Tek, Napierville, IL) for fluorescence immunohistochemistry or fixed in half-strength Karnovsky fixative for TEM. ¹⁷  

Immunohistochemical Analysis

The frozen corneas used for immunofluorescence microscopy were processed as described previously. ⁷ In brief, the tissues were sectioned (10 μm) onto poly-l-lysine–coated slides and stained with primary antibodies to basement membrane markers: laminin-β1, laminin-γ1, entactin, perlecan (provided by Alexander Ljubimov, Cedars–Sinai Medical Center, Los Angeles, CA), and J18, a polyclonal sera raised against basement membrane that has been shown to recognize primarily laminin-5 but may also react against laminin-6 and -7 (provided by Jonathan C. Jones, Northwestern University Medical School, Chicago, IL). Tissues were counterstained with the appropriate fluorescently labeled secondary antibody. Sections were viewed either with a fluorescence microscope (model BX60; Olympus, Lake Success, NY) or with confocal microscopy (1024 program; Bio-Rad, Cambridge, MA). Control sections incubated without addition of primary antisera were included in each immunofluorescence procedure. For each time point, no fewer than three corneas from three different animals were used. 

Electron Microscopy Analysis

The harvested corneas were processed for TEM using a modified procedure based on methods described in Tisdale et al. ¹⁷ and Orenstein et al. ¹⁸ with further modifications suggested by Robyn Rufner (Director, GWU Center for Microscopy and Image Analysis). Tissues were fixed in half-strength Karnovsky’s fixative for 1 hour at room temperature and then transferred to 4°C for storage until use. Specimens were then postfixed in 2% osmium tetroxide, stained en bloc with 0.5% uranyl acetate, passed through an alcohol series, and embedded in mounting compound (Embed 812). Thin sections were then stained with 5% uranyl acetate and lead citrate and examined on an electron microscope (model 1200EX; JEOL, Peabody, MA) at ×20,000 magnification at 60 kV. For each time point, at least three corneas from three different animals were examined. 

Results

Linear BMZ in Unwounded Corneas

Corneas were stained with a monoclonal antibody to the β1 chain found in laminin-1 and -10 (Fig. 1A ), a polyclonal serum that recognizes primarily laminin-5 (Fig. 1B) , and a monoclonal antibody against perlecan (Fig. 1C) and viewed with immunofluorescence. All three antibodies recognized antigens that are present within basement membranes. The corneal epithelial BMZ appeared as a continuous distinct line between the basal surface of the epithelial basal cells and the anterior stroma. Although the lamininβ 1 chain and perlecan were present in both the epithelial BMZ and Descemet’s membrane (Figs. 1A 1C , arrow), laminin-5 was exclusively localized to the epithelial BMZ (Fig. 1B , arrow). Results of entactin staining were similar to those observed for laminin β1 and perlecan (not shown). TEM (Fig. 1D) revealed contiguous lamina densa and lamina lucida regions and the presence of numerous hemidesmosomes spaced regularly along the basement membrane. 

BMZ Became Discontinuous after Large Wounds

Laminin-5 was localized at and behind the leading edge and on the denuded BMZ at 12 (Figs. 2A 1B 1C ) or 18 (Figs. 2D 2E 2F) hours after small corneal débridement wounds. Note that the denuded BMZ still showed uniform staining for laminin-5, even at 18 hours. The only location where laminin-5 was not abundant at the BMZ was at the leading edge (Figs. 2B 2E) . These confocal microscopy studies reveal that the denuded BMZ retained the localization of laminin-5 throughout re-epithelialization when corneas were subjected to small wounds that closed within 24 hours. 

In large wounds we consistently observed discontinuities or an apparent absence of localization of basement membrane markers at the BMZ in regions that remained uncovered by epithelial cells at times of more than 24 hours after injury. Data for laminin-5 are shown in Figure 3 at 24 (Figs. 3A 3B 3C) and 36 (Figs. 3D 3E 3F 3G 3H) hours after the larger wounds. At 24 hours, there were no breaks in the localization of laminin-5 over the bare stroma (Fig. 3C) . At the leading edge and behind the leading edge (Fig. 3B) there was absent or reduced staining of laminin-5 overall, consistent with the data presented for the small wounds (Figs. 2B 2E) ; however, somewhat farther away from the wound edge (Fig. 3A) , laminin-5 was localized within the cytoplasm of the migrating sheet of epithelial cells. The migrating epithelial sheet was thinner after larger wounds and consisted of only one or two cell layers. 

At 36 hours after larger wounds, the epithelium had migrated to cover more than 80% of the corneal surface. The region of denuded stroma at the center of wounds at 36 hours began to show loss of laminin-5 localization at discrete sites along the bare stroma in front of the leading edge (Figs. 3E 3F) . Shown at slightly higher magnification in Figures 3G and 3H are examples typical of the localized loss of laminin-5 staining observed at 36 hours after wounding. At some sites we saw diffuse staining, whereas at others, there was no staining for laminin-5. 

The primary antigen recognized by the J18 sera used in Figures 2 and 3 was laminin-5, the major structural component of the extracellular anchoring filaments of hemidesmosomes. In Figure 4 , we show that the localization of entactin and perlecan, additional components of the basement membrane of the cornea, was also disrupted 36 hours after larger wounds. Figures 4A 4B 4C 4D show the localization of entactin at the leading edge and bare central stroma at 18 hours after small wounds (Figs. 4A 4B) and at 36 hours after larger wounds (Figs. 4C 4D) , and Figures 4E 4F 4G 4H show the localization of perlecan at the leading edge and bare central stroma at 18 hours after small wounds (Figs. 4E 4F) and at 36 hours after larger wounds (Figs. 4G 4H) . Thus, for laminin-5, entactin, and perlecan, disruptions of the linear staining pattern at the BMZ were observed at 36 hours but not at 18 hours. 

Taken together, the data presented suggest that over time after wounding the basement membrane is subject to disassembly. In addition, Figures 3C and 3E clearly show that entactin and perlecan were produced by the migrating epithelial cells. By 3 days after wounding the immunofluorescence staining profile for these basement membrane proteins at the epithelial cell–stromal interface was again continuous and discrete (data not shown). 

TEM Confirmed Loss of Basement Membrane

The immunofluorescence data suggest that the basement membrane may be lost when débridement wounds take longer than 24 hours to close. To determine whether this is the case, TEM studies were conducted. Presented are TEM data from small (Fig. 5) and large (Fig. 6) wounds. Data show the BMZ at a region just behind the leading edge but within the region of active cell movement, the basal surface of a basal cell migrating at the leading edge of the wound, and the anterior aspect of the bare stroma at 12 (Figs. 5A 5B 5C , respectively) and 18 (Figs. 5D 5E 5F , respectively) hours after the small wound and at 24 (Figs. 6A 6B 6C , respectively) and 36 (Figs. 6D 6E and 6F , respectively) hours after larger wounds. Higher magnifications of the bare stroma at 36 hours are presented in 6G and 6H. A continuous, discrete lamina densa was present behind the leading edge, beneath the leading edge, and at the denuded anterior stromal surface before re-epithelialization at 12 and 18 hours after small wounds (Fig. 5) . Fragments of basal cell basal membranes can be observed that were left behind after débridement wounding at 12 hours but not at 18 hours (Fig. 5C) . Although there are no mature hemidesmosomes apparent at either time point, a few small hemidesmosomes were found toward the limbus (not shown). Compared with 12 hours (Fig. 5C) , at 18 hours (Fig. 5F) , the exposed lamina densa appeared somewhat thicker but was still readily apparent. Underneath the cells comprising the leading edge and behind the leading edge at 12 (Fig. 5A 5B) and 18 hours (Figs. 5D 5F) an apparent lamina lucida and densa were visible, and epithelial cells appeared closely associated with the underlying stroma, despite the absence of mature hemidesmosomes. 

At 24 hours after the larger wounds, the lamina densa was still apparent at the denuded anterior stromal surface (Fig. 6C) but looked more disrupted and less discrete than for small wounds at either 12 or 18 hours (Figs. 5C 5F) . However, by 36 hours the lamina densa was highly disrupted or completely absent (Figs. 6F 6G 6H) . The longer stretches of exposed stroma shown in Figures 6G and 6H demonstrate that the disassembly of the BMZ was substantial but partial because remnants of the lamina densa of the basement membrane could still be observed at some locations at 36 hours. Underneath the cells comprising the leading edge and behind the leading edge at both 24 and 36 hours after wounding, adhesion of the epithelial cells to the stroma appeared to become progressively less tight, suggested by the increased spacing between the basal cell membrane and the apparent lamina densa observed just behind the leading edge at 36 hours (Fig. 6D) compared with earlier time points (Figs. 5A 5D 6A) . 

Discussion

Basal Lamina Disassembled During Healing

Our previous studies of corneal wound healing after débridement have focused on the leading edge and behind the leading edge and on the nature of the adhesion complexes formed during migration. ^{6 8 19 20} In this article, we focus primarily on the denuded stroma and report that there was no longer a continuous epithelial basement membrane at the anterior aspect of the mouse cornea when re-epithelialization took longer than 24 hours. We also show that abundant entactin and perlecan were produced by epithelial cells during migration after larger wounds. 

Although it has been known for some time that epithelial cells produce laminin-5, it has been reported by Ekblom et al. ²¹ that entactin and perlecan are synthesized primarily by mesenchymal cells in the skin and not by epithelial cells. In the cornea, a rapid loss of stromal fibroblasts at the anterior stroma has been well documented after débridement wounding. ^{22 23} Thus, no mesenchymal cells are available at the anterior stroma to replenish the entactin and nidogen at the BMZ during re-epithelialization. If corneal epithelial cells did not synthesize entactin and perlecan themselves, a delay would be observed before the reappearance of these proteins within the basement membrane region after the larger wounds. Not only was no delay observed, but after the larger wounds, both proteins appeared abundant within the cytoplasm of the cells making up the single cell-layer migrating sheet (Fig. 4) . Therefore, entactin, and perlecan appeared to be upregulated in migrating corneal epithelial cells at time points when the basement membrane was partially disassembled. 

Based on the data presented and on data found in the literature, we propose that regulated BMZ disassembly could affect re-epithelialization in one or more of the following ways: 

First, by exposing epithelial cells to underlying stromal extracellular matrix proteins such as collagen I and/or to new ligands generated by proteolysis, disassembly of the BMZ may induce new integrin expression or activation in migrating corneal epithelial cells. Studies evaluating the function of the α2β1 integrin have demonstrated integrin-mediated migration and collagenase I induction in epidermal keratinocytes exposed to native collagen I but not in cells adhering to collagenase-I digested collagen or to gelatin. ²⁴ The increase in α9β1 integrin expression we have observed previously after larger corneal débridement wounds ⁷ could be induced by the partial loss of the epithelial basement membrane. 

Second, disassembly of the BMZ may modify intracellular signaling pathways in migrating epithelial cells. Cytokines such as transforming growth factor-β and basic fibroblast growth factor are found in the BMZ after injury and are held there by binding to molecules in the matrix, including heparan sulfate proteoglycans. ^{25 26 27} The partial disassembly of the BMZ could release molecules involved in modulation of cell proliferation, cell differentiation, and/or apoptosis. ^{9 28}  

Third, it could promote the formation of a more stable adhesion complex. Studies to determine the most effective treatments for recurrent epithelial erosions have concluded that requiring the epithelial cells to resynthesize a new basement membrane is one common parameter in successful treatments of this condition. ²⁹ In addition, Azar et al. ^{30 31} have shown that one of the hallmarks of the diabetic cornea is its failure to reassemble the adhesion complex correctly after injury. Ljubimov et al. ³² have shown that human diabetic corneas have basement membrane abnormalities including reduced expression of laminin-1 and -10 and entactin, and that these losses correlate with reductions in integrin localization within the epithelial cells. Studies in both animal and human corneas show that the structure and composition of the epithelial basement membrane affects the adhesion of the cells sitting on it. 

Both TEM and immunofluorescence microscopy have been used to evaluate the BMZ in healing rabbit corneas. ⁵ Twenty-two hours after 8.5-mm wounds in the rabbit, the basement membrane was intact and laminin-1 staining was continuous; data from later time points were not presented. Larger débridement-type wounds were made involving removal of all the corneal epithelial cells, the limbal epithelium, and a 1-mm portion of the conjunctiva. These large wounds took 1 to 2 weeks to close with conjunctival cells migrating onto the central cornea. TEM data on such a wound at 36 hours were presented, and the lamina densa was present; again, later time points were not presented. Additional ultrastructural studies of the basal lamina after manual débridement show retention of the lamina densa immediately after wounding ^{2 4} or at times soon after wounding ¹ but do not specifically examine whether this structure is maintained throughout re-epithelialization over regions of denuded stroma. 

Complicating our studies and those of other groups ^{1 2 3 4 5 6 7} is that at longer times after injury, the area of the remaining denuded basal lamina becomes quite small as the wound edges begin to merge. The migrating cells also deposit the components of the basement membrane as they migrate, leaving at their basal surface a continuous, lamina densa–like structure. Thinner epithelial sheets at later time points are prone to breaking off during processing, making discernment of the leading edge difficult. Despite these complications, inconsistencies between our results and those of others could also be due to differences in species used and the types of wounds studied. 

Relationship to Epidermal Healing

The re-epithelialization of the skin blister is similar in many ways to the healing of corneal débridement wounds. ^{33 34} Kainulainen et al. ¹² recently evaluated the expression and localization of the integrins α3β1 andα 6β4 and their ligand laminin-5 at the leading edge in response to suction blisters. They showed immunohistochemically that laminin-5 was retained on the floor of the blister at 2 days but by 4 days, micrographs indicated a loss of laminin-5 on the blister floor in front of the leading edge of migrating epithelial cells. Thus, skin blister models support the current results on the cornea regarding the disassembly of the laminin-5 component of the basement membrane at longer times after injury. 

Proteolysis of the BMZ Likely

Polymorphonuclear neutrophils have been demonstrated to bind to the surface of the exposed basement membrane after débridement wounding ^{2 35} and would therefore be available to release proteases capable of degrading the basement membrane. The tear film may also play a role, in that it is known to contain proteases, especially after corneal wounding. ^{36 37} Although studies show that protease activity is lower in corneal epithelial cells after manual débridement wounds than after wounds penetrating the BMZ, corneal epithelial cells can produce gelatinase B and matrilysin, ^{38 39} metalloproteinases capable of degrading components of the basement membrane. Further, both these metalloproteinases are present at the leading edge of migration, where we demonstrated a loss of localization of laminin-5. Studies on cultured epithelial cells recently showed that proteolytic cleavage of laminin-5 alters α6β4 adhesion and migration. ^{40 41} Protease digestion of the basement membrane over time after injury would alter both its structure and function. 

Although proteases may be involved in mediating the loss of basement membrane proteins, given their availability from multiple sources, we cannot rule out simple mechanical unraveling of the lamina densa over time after débridement. Given sufficient time after injury and the absence of integrins and other molecules to permit their stabilization and organization, the basement membrane proteins within the densa may diffuse into the tear film as a result of mechanical friction caused by the blinking eyelid. 

Potential Clinical Relevance

Corneal abrasions affect more than 2 million people in the United States annually. ²⁹ Most are successfully treated and do not recur, but in a small percentage, recurrent epithelial erosion (REE) develops, a painful condition characterized by repeated episodes of epithelial cell loss, usually near the site of initial trauma. REE has also been observed in patients with some classes of the syndrome epidermolysis bullosa. ^{42 43 44 45} Treatment for REE begins with débridement of the corneal epithelium and patching of the eye. When the condition persists, patients may be treated with either anterior stromal puncture ⁴⁶ or a modification of excimer laser photorefractive keratectomy. ⁴⁷ Both approaches involve manual débridement of the involved epithelium and demand that new basement membrane be synthesized to be effective. Although the causes of REE in patients not known to have a blistering disease remain unclear, studies looking at the structure of the BMZ of corneas from patients with REE document diffuse and poorly assembled basement membranes as a common feature in all patients. ^{48 49} In addition, other conditions including bullous keratopathy ⁵⁰ and diabetes ³² are also associated with defective basement membrane assembly and corneal erosions. 

In this study, we showed that there was a loss of several of the molecules present within the basement membrane at the BMZ, as well as a loss of the ultrastructure of the lamina densa, when epithelial débridement wounds were large enough that they required times longer than 24 hours to close. Future studies to uncover the underlying mechanism of this loss are under way, focusing primarily on the role that proteases play and their cellular sources. A more complete understanding of how epithelial cell–matrix interactions stabilize the corneal epithelium and the basement membrane during re-epithelialization after injury will improve our ability to treat those with poor corneal epithelial healing. 

 Supported by National Institutes of Health Grant RO1 EY-O8512-9 (MAS).

 Submitted for publication May 14, 1999; revised September 9, 1999; accepted October 5, 1999.

 Commercial relationships policy: N.

 

Figure 1.

 

The localization of several extracellular matrix components of the epithelial basement membrane is continuous in unwounded corneas. (A) Localization of the laminin β1 chain, a component of laminin-1 and -10; (B) localization of the J18 antigen, which has been shown to be primarily laminin-5; and (C) localization of perlecan. All three basement membrane markers showed a linear, continuous staining pattern in the epithelial BMZ in the unwounded cornea. (A, B, and C, arrows) Location of Descemet’s membrane and the corneal endothelial cell layer. Note that J18–laminin-5 was not present at Descemet’s membrane, whereas the laminin β1 chain and perlecan were abundant that this site. (D) Hemidesmosomes (∗) and lamina densa at the BMZ in an unwounded cornea by TEM. Bar (A, B, and C), 100 μm; (D), 278 nm.

Figure 2.

 

The localization of J18–laminin-5 after small wounds appears continuous and discrete. Localization of J18–laminin-5 behind the leading edge (A, D), at the leading edge (B, E), and at the denuded center (C, F) of mouse corneas at 12 hours (A, B, and C) and 18 hours (D, E, and F) after 1.5-mm manual débridement wounds. (B, E, arrows) Cells at the tip of the leading edge. Although there was no J18–laminin-5 staining beneath several cells at the leading edge, there was no evidence of a loss of J18–laminin-5 toward the center of the denuded basement membrane. Bar, 80 μm.

Figure 3.

 

The localization of J18–laminin-5 at the BMZ in larger wounds became increasingly discontinuous before wound closure. Localization of J18–laminin-5 behind the leading edge (A, D), at the leading edge (B, E), and in at the denuded center (C, F) of mouse corneas at 24 hours (A, B, and C) and at 36 hours (D, E, and F) after limbus-to-limbus manual débridement wounds. (G, H) Discontinuous localization of laminin-5 at the center of the wound area 36 hours after injury, at slightly higher magnification. (B, E, arrows) Cells at the tip of the leading edge. Again, J18–laminin-5 staining was absent beneath cells at the leading edge. (∗) Areas of J18–laminin-5 loss observed more frequently at 36 hours of wounding than at 24 hours. Bar (A through F), 80 μm; (G, H), 50μ m.

Figure 4.

 

The localization of entactin and perlecan confirms that the BMZ in larger wounds becomes discontinuous before wound closure. (A through D) Localization of entactin; (E through H) localization of perlecan. Leading edge (A, C) and denuded center (B, D) of mouse corneas at 18 hours after 1.5-mm smaller wounds (A, B) and at 36 hours after larger wounds (C, D) stained for entactin. Leading edge (E, G) and denuded center (F, H) of mouse corneas at 18 hours after smaller wounds (E, F) and at 36 hours after larger wounds (G, H) stained for perlecan. Arrows: The tip of the leading edge. Asterisks: Areas without entactin and perlecan were observed at 36 hours after wounding but not at shorter times. Bar (A through F), 80 μm; (G, H), 50μ m.

Figure 5.

 

Ultrastructural TEM studies show that the lamina densa was retained after 1.5-mm wounds. TEM was used to identify the state of assembly of the basement membrane at various times after 1.5-mm wounds. Areas behind the leading edge (A, D), cell at the very tip of the leading edge (B, E), and area at the denuded center (C, F) of mouse corneas at 12 hours (A, B, and C) and 18 hours (D, E, and F) after 1.5-mm manual débridement wounds. Note the maintenance of the lamina densa, an electron-dense structure at the anterior aspect of the denuded stroma (C, F) at both 12 and 18 hours after injury. Bar, 290 nm.

Figure 6.

 

Ultrastructural TEM studies confirm that the BMZ after larger wounds becomes increasingly discontinuous before wound closure. Areas behind the leading edge (A, D), at the leading edge (B, E), and at the denuded center (C, F) of mouse corneas at 24 hours (A, B, and C) and at 36 hours (D, E, and F) after limbus-to-limbus manual débridement wounds. (G, H) Center of the wounded cornea at 36 hours after large wounds (similar to F) showing that the BMZ disassembly, which is substantial, did not affect the entire corneal surface equally but occurred in patches at intervals across the bare stroma. Bar, 290 nm.

The authors thank Ann Tisdale, Jan Orenstein, and Robyn Rufner for help with the TEM, the GWU Center for Biomedical Communications for help with photography, and Temmy Qiu for technical assistance.