June 2004
Volume 45, Issue 6
Free
Cornea  |   June 2004
A Mouse Model for the Study of Recurrent Corneal Epithelial Erosions: α9β1 Integrin Implicated in Progression of the Disease
Author Affiliations
  • Sonali Pal-Ghosh
    From the Departments of Anatomy and Cell Biology and
  • Ahdeah Pajoohesh-Ganji
    From the Departments of Anatomy and Cell Biology and
    Biological Sciences, The George Washington University, Washington, DC; and the
  • Marcus Brown
    From the Departments of Anatomy and Cell Biology and
  • Mary Ann Stepp
    From the Departments of Anatomy and Cell Biology and
    Department of Ophthalmology, The George Washington University Medical Center, Washington, DC.
Investigative Ophthalmology & Visual Science June 2004, Vol.45, 1775-1788. doi:10.1167/iovs.03-1194
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to Subscribers Only
      Sign In or Create an Account ×
    • Get Citation

      Sonali Pal-Ghosh, Ahdeah Pajoohesh-Ganji, Marcus Brown, Mary Ann Stepp; A Mouse Model for the Study of Recurrent Corneal Epithelial Erosions: α9β1 Integrin Implicated in Progression of the Disease. Invest. Ophthalmol. Vis. Sci. 2004;45(6):1775-1788. doi: 10.1167/iovs.03-1194.

      Download citation file:


      © 2016 Association for Research in Vision and Ophthalmology.

      ×
  • Supplements
Abstract

purpose. To describe an in vivo mouse model for the study of recurrent corneal erosion syndrome (RCES) in mice and to characterize the changes in α9 integrin expression during wound healing.

methods. Corneal epithelial debridement wounds of two sizes (1.5 and 2.5 mm) were made on the ocular surface of BALB/c mice and were evaluated at various times after wounding. Corneas were processed either as whole mounts and stained with propidium iodide and an antibody against α9 integrin or for bromodeoxyuridine analyses of cell proliferation. A separate study involved analyses of corneal wound healing over time in individual mice with large and small debridement wounds. Mice were anesthetized once per week and their corneas stained with fluorescein to assess the quality of the corneal epithelium. After 6 weeks, mice were killed and eyes processed for study by immunofluorescence in either whole mounts or frozen sections.

results. Whole mount confocal microscopy showed open wounds on the ocular surface of mice at 1 and 2 weeks after large wounds were created, but not after small wounds. In addition, α9 integrin was upregulated during healing, and changes were observed in α9 integrin localization at the limbus with large wounds but not with small wounds. Although only 1 of 16 corneas with small wounds had erosions at 1 and 2 weeks, 11 of 16 with large wounds had erosions. However, by 6 weeks, 13 of 16 eyes showed signs of erosion whether wounds were small or large. With large wounds, RCES corneas frequently showed numerous goblet cells adjacent to a limbus lacking α9 integrin. Corneas from mice with documented RCES showed both retention of α9 integrin and tenascin-C expression at the anterior stromal–epithelial interface as well as impaired relocalization of α3β1 integrin to the basement membrane zone.

conclusions. These data show that spontaneous recurrent corneal erosions occur in a mouse model after manual creation of a single wound by debridement. Differences between the healing of small (1.5 mm) and large (2.5 mm) wounds were observed. Large wounds often resulted in the presence of goblet cells on the central cornea and a loss of α9 integrin at the limbus. Small wounds never showed differences in the localization of α9 integrin at the limbus, and no goblet cells were observed in the central cornea. More studies are needed to understand the causes of erosions in these mice.

Patients with recurrent corneal erosion syndrome (RCES) experience pain, discomfort, and a reduced quality of life. 1 2 The two major causes of RCES are trauma or various forms of epithelial basement membrane dystrophy, and the disease process can involve small erosions that heal within a few hours (microcytic) or large erosions that take days or weeks to heal (macrocytic). 3 4 Most erosions can be treated nonsurgically and generally heal well, but some persist and require intervention. Current surgical interventions include phototherapeutic keratectomy, 5 diamond burr superficial keratectomy, 6 and intrastromal puncture. 7 8 However, if the disease is caused by a basement membrane dystrophy, the patient is likely to experience chronic episodes of painful erosions. Trauma to the ocular surface is the most frequent ophthalmic complaint in emergency rooms, and some studies have estimated erosions to occur after trauma as frequently as 1 in every 150 cases. 9 The most common site for erosions is the inferior third of the cornea, and women are treated for RCES more often than men. 3  
The apparent cause of RCES is failure of the epithelial cells to regain tight adhesive contacts with the underlying corneal stroma after trauma. In epithelial basement membrane disease, it is the failure of the epithelial cells to maintain tight adhesion to a defective basement membrane. Cell adhesion molecules of the integrin family are in large part responsible for the adhesion of the corneal epithelium to the stroma. Integrins function as αβ heterodimers and are involved in mediating actin-based cell adhesion and migration and intermediate filament-based cell–substrate adhesion through hemidesmosomes (HDs). HDs are the adhesion complexes that mediate tight adhesion of skin and corneal epithelium to their underlying basement membranes. 10 11 They are composed—from inside to outside the cell—of keratin intermediate filaments, plectin, BP230, BP180, α6β4 integrin, laminin-5 anchoring filaments, and type VII collagen anchoring fibrils. 12 13 Plectin and BP230 form the intracellular plaque of the HD and mediate interaction with keratins. BP180 and α6β4 integrin are both integral membrane components of HDs, laminin-5 is a specialized component of the basement membrane, and type VII collagen fibrils penetrate deep into the corneal stroma to ensure firm adhesion of the epithelium to the underlying connective tissue. In the cornea the optical clarity essential to allow light to focus on the retina requires an optically smooth surface. This anatomic necessity results in an exposed, sheer surface vulnerable to debridement by friction and rubbing. During migration after an injury, HDs disassemble, and epithelial sheet migration allows open wounds to be covered quickly and efficiently. 11 14 Once migration is complete, the HDs must reassemble to restore tight epithelial adhesion. 
Efforts to determine the molecular origin of and to develop more effective treatments for RCES have been slowed by the lack of rodent animal models. Studies of spontaneous erosions have typically used various purebred dogs known to be susceptible to RCES. 15 16 Other studies have looked at chronic corneal erosion in animals by repetitive mechanical debridement of the cornea. 17 18 For a number of years, we have studied the role of cell adhesion molecules in regulating reepithelialization after mechanical debridement of rat and mouse corneas, 11 19 20 21 22 using a model in which a circular area of epithelial cells was removed with a dulled scalpel. The sizes of the wounds varied, depending on the study; either a 1.5- or a 2.5-mm diameter region of corneal epithelial cells was routinely removed from the central corneas of 8-week-old mice. The epithelial basement membrane was left intact at the time of injury and was not denatured by chemical or thermal means after wounding. 
Among the proteins whose localizations change after wounding is α9 integrin. 22 In the normal, unwounded adult mouse limbus α9 integrin localizes to the apical aspect of the basal cells where it is found both in the cytoplasm and in the cell membrane. 23 Studies of the developing mouse eye have shown that this limbal basal cell localization is a relatively late event in the development of the ocular surface. When evaluated at 1 week after birth, mice have a uniform distribution of α9 integrin over the entire ocular surface. By 2 weeks, this distribution pattern changes dramatically, and the central corneal epithelial cells express less α9 integrin and the limbal basal cells more. The adult pattern is established by between 6 and 8 weeks. 23  
Because the focus of our past work has been reepithelialization, our efforts have focused primarily on the study of healing over short time spans—generally, from hours to several days. However, to provide strain-specific mouse control corneas for studies of knockout mice, we began to assess corneas at time points from 1 to 12 weeks after wounding. During the course of those studies, we observed that, after a single, initial manual debridement wound to an 8-week-old BALB/c mouse cornea, corneal epithelial erosions occurred spontaneously, at times ranging from 1 to 8 weeks after wounding. The results that follow—from studies of wild-type healthy BALB/c mice—are an extension of initial observations after studies of genetically altered mice. 24  
In the current studies, we characterize, for the first time, an in vivo mouse model of erosions that spontaneously developed in response to a single, initial corneal debridement wound. We show that the RCES epithelium was characterized by both a higher cell proliferation rate than that of unwounded corneal epithelium and by altered localization of α9 integrin. Goblet cells were observed on the corneal surface with large wounds but not with small wounds. We then assessed the healing history of the individual eyes over time (1–6 weeks after wounding), using fluorescein to demonstrate that we were describing RCES rather than chronic open wounds. We were thus able to observe recurrent erosions in the BALB/c mouse after creating both small and large wounds. With the healing history documented for individual eyes, we demonstrated that after the large wounds were made, most RCES eyes had less α9 integrin at the limbus and numerous goblet cells present over the corneal surface. Finally, we show that, in RCES eyes, α3β1 integrin localization was disrupted, but α6β4 was localized primarily at the basal aspect of the basal cells. These data will improve our understanding of RCES in both animals and humans. Furthermore, having an experimental model to test hypotheses about causes and possible treatments for RCES should, in time, reduce the pain and improve the quality of life of people with this condition. 
Methods
Materials and Chemicals
Most of the chemicals used in these studies were obtained from Sigma-Aldrich (St. Louis, MO), including EGTA, bovine serum albumin (BSA), sodium dodecyl sulfate (SDS), and horse serum. Ketamine was purchased from Fort Dodge Veterinary Supplies (Fort Dodge, IA), xylazine from Vedco (St. Joseph, MO), and sodium pentobarbital from Veterinary Laboratories, Inc. (Lenexa, KS). Optimal cutting temperature (OCT) compound (Tissue Tek II) was obtained from Laboratory Tek (Napierville, IL), and proparacaine eye drops and erythromycin ophthalmic ointment were obtained from Bausch & Lomb (Tampa, FL). We purchased coverslips and paraformaldehyde from Fisher Scientific (Somerville, NJ), slides from Shandon (Pittsburgh, PA), and the pap pen and mounting medium (Fluormount) from Electron Microscopy Sciences (Washington, PA). Cell proliferation was studied using a 5-bromo-2′-deoxy-uridine (BrdU) labeling and detection kit (catalog no. 1 296 736; Roche Diagnostics, Indianapolis, IN). For detecting mouse primary antibodies on mouse tissue, another kit (MOM; FMK-2201; Vector Laboratories, Burlingame, CA) was used. For primary antibodies, we used a rat anti-mouse antibody against tenascin-C (Mtn-12, T-3413; Sigma-Aldrich), a rat anti-mouse β4 monoclonal antibody (346-11A, catalog no. 09491D; BD-Pharmingen, San Diego, CA), and a monoclonal anti-mouse antibody against mucin 5AC Ab-1 (45M1, catalog no. MS-145-P1; NeoMarkers, Fremont, CA). For α3 and α9, we used polyclonal antibodies we had previously characterized. 25 Propidium iodide (PI; P-1304), Alexa Fluor 488 anti-rabbit (A-11,006), anti-mouse (A-11,001), and Alexa Fluor 594 anti-rat (A-11,007) secondary antibodies were obtained from Molecular Probes (Eugene, OR). 
Manual Corneal Debridement
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 Animal Care and Use Committee. Eight to 10-week-old male and female BALB/c mice (22 to 25 g) were anesthetized with 250 μL of a 1:10 dilution of a 1:1 mixture of ketamine (100 mg/mL; Aveco Co., Inc., Fort Dodge, IA) and xylazine (20 mg/mL; Miles Inc., Shawnee Mission, KS). Once the animals were anesthetized, a topical anesthetic (proparacaine ophthalmic solution) was applied to their ocular surfaces until the blink sensation was lost, and their corneas were scraped with a dull scalpel to remove the corneal epithelium. For small wounds, a 1.5-mm central corneal area was demarcated with a trephine and removed manually by gentle scraping with a dulled blade. For large wounds, a 2.5-mm area of epithelial tissue was removed, with care taken to avoid limbal blood vessels. After they were wounded, all eyes were treated with erythromycin ophthalmic ointment to minimize inflammation and to keep the ocular surface moist while the mice were under anesthesia. Corneas with large wounds were allowed to heal in vivo for 1, 1.5, 2, and 3 days, and 1, 2, and 4 weeks and those with small wounds for 6, 12, 18, 24, 48, and 72 hours, and 1 and 2 weeks. At least three eyes were assessed for each time point for the data presented in Figures 1 2 3 4 5 . After mice were killed by lethal injection using sodium pentobarbital, eyes were enucleated and either frozen in OCT compound for immunofluorescence (IF) or fixed as whole mounts. 
Whole mounts
At the indicated time points, mice were killed and eyes were removed and fixed for 2 hours in a 4:1 dilution of prechilled 100% methanol and dimethyl sulfoxide at −20°C and then stored in 100% methanol. All eyes were transferred into phosphate-buffered saline (PBS) in a graded methanol series (70%, 50%, and 30% methanol and PBS, 30 minutes each). PBS (10×) was made as follows: 14.4 g Na2HPO4, 2.4 g KH2HPO4, 2 g KCl, 80 g NaCl, in a total volume of 1 liter of water (pH 7.4). All incubations were performed with gentle shaking and at room temperature, unless otherwise specified. After the eyes were washed in PBS twice, for 30 minutes each, they were incubated with blocking buffer for 2 hours. Blocking buffer was made as follows: to 100 mL 1× PBS, 1 g of BSA was added, the mixture was stirred for 30 minutes, 1 mL of horse serum was added, and the mixture was stirred for an additional minute. The tissues were then incubated overnight with primary antibody, at 4°C. The next day, the tissues were washed five times with PBS and 0.02% Tween 20 (PBST) for 1 hour each, blocked for 2 hours, and then incubated with goat anti-rabbit Alexa 488 secondary antibody overnight at 4°C. The next day, the eyes were washed three times with PBST for 1 hour each, followed by nuclear staining with PI for 5 minutes, and three washes with double-distilled water (Millipore, Bedford, MA) for 5 minutes each. For the goblet cell staining, the blocking buffer from the antibody detection kit (MOM; Vector Laboratories) was used. Then, under a dissecting microscope (Model SZ40; Olympus, Lake Success, Melville, NY), the retina, lens, and iris were discarded, and four incisions were made in each cornea. To achieve the best flattening, the corneas were placed epithelial-side-up on a black filter (25 mm, 0.45 μm, HABG02500; Millipore), mounting medium was then added, and they were coverslipped. The images were then captured with a confocal microscope. 
Immunofluorescence Microscopy on Tissue Sections
For sections, adult BALB/c mice were killed with a lethal injection, and eyes were frozen in OCT compound. Eight-micrometer cryostat sections were cut from the eyes and baked overnight at 37°C. The sections were then rehydrated in PBS for 15 minutes, blocked with blocking buffer, and incubated with primary antibodies (α3, α9, and β4 integrin and tenascin-C) for 1 hour in a moist chamber at room temperature. After they were washed with PBS, the sections were treated with blocking buffer for 15 minutes and were incubated with the appropriate Alexa secondary antibodies for 45 minutes in a moist chamber. They were then washed with blocking buffer for 15 minutes, covered in mounting medium, and coverslipped. The images were then captured with a confocal microscope. 
BrdU Cell Proliferation Analyses
Cell proliferation was assessed using the BrdU labeling and detection kit, as recommended by the manufacturer. All incubations were at room temperature unless specified. We assessed time points at 1.5 and 3 days, and 1, 2, and 3 weeks for large wounds, and 4 and 8 weeks for small wounds. Corneas were wounded as described earlier. Control (unwounded) eyes were obtained from mice that had not been subjected to wounding of either eye. At the given time points, the mice were killed and their eyes enucleated and placed immediately into a BrdU-containing labeling solution with complete minimum essential medium (MEM). 22 After 30 minutes in the labeling solution, the eyes were washed three times in MEM without BrdU and allowed to incubate for an additional 15 minutes in MEM. Whole eyes were embedded in OCT, and 8-μm sections cut and baked overnight at 37°C. Next, sections were rehydrated in PBS and fixed in ice-cold 70% methanol in 50 mM glycine buffer (pH 2) for 20 minutes at −20°C. After fixation, sections were incubated in PBS twice, for 2 minutes each, then with 2.5% trypsin in 0.1% CaCl2 in PBS, for 3 minutes. They were then briefly washed in PBS and transferred to a solution containing 4 M HCl, for 3 minutes. Sections were washed twice in PBS, for 5 minutes each, and then transferred to blocking buffer. Slides were processed for IF as recommended by the manufacturer (Roche Diagnostics). At least three corneas and two slides per time point were used—each slide containing four to six corneal cross sections. The number of labeled cells in at least three different sections from each slide were counted and the data expressed as the number of cells labeled per unit length of basal cell basal membrane, as measured with image-analysis software (ImagePro Plus ver. 4.1 software; Media Cybernetics, Silver Spring, MD). The data were then analyzed for significance on computer with the unpaired t-test (Instat 1.1 for Macintosh; GraphPad Software, San Diego, CA). 
Recurrent Erosion Study
To determine whether erosions were chronic or recurrent, we assessed healing over time in individual animals. We wounded eyes bilaterally, with the small wound in the left eye and the large wound in the right. The purpose of wounding both eyes in the same animal was to correlate the quality of healing of contralateral eyes. This experimental design was critical, because it allowed us to determine whether individual animals were poor healers or whether one size of wound healed more poorly than the other. After initial wounding, the mice were allowed to heal for 1, 2, 4, and 6 weeks. At each time point, the mice were placed under general anesthesia and their eyes stained with fluorescein and viewed using a dissecting microscope modified with blue-light illumination. Wounds were scored as open or closed based on fluorescein staining. Care was taken to make sure that the eyes remained moist after assessment and, before anesthesia wore off; each eye was treated with a single application of erythromycin ophthalmic ointment. At the last time point (6 weeks) wounds were assessed (using fluorescein) to be open or closed. Mice were killed by lethal injection, their eyes restained with a vital dye, and photographs taken using the dissecting microscope, to assess scaring and vascularization. After enucleation, eyes were carefully labeled and either processed for whole mounts or frozen for tissue sectioning. 
Confocal Microscopy
Confocal microscopy was performed at the Center for Microscopy and Image Analysis (CMIA) at the George Washington University Medical Center. A confocal laser scanning microscope (model MRC 1024; Bio-Rad, Hercules, CA) equipped with a krypton-argon laser and an inverted microscope (model IX-70; Olympus) was used to image the localization of Alexa 488 (488-nm laser line excitation; 522/35 emission filter) and PI (568 nm excitation; 605/32 emission filter). Optical sections (z = 0.5 μm) of confocal epifluorescence images were acquired sequentially at 1-μm intervals with a 20× objective lens (NA = 0.7) and/or a 10× objective lens (NA = 0.3) with image acquisition software (LaserSharp ver. 3.2; Bio-Rad). Typically, six to eight optical sections, each taken a 1-μm intervals, were merged and viewed en face. Image management software (Photoshop ver. 7.0; Adobe Systems, Mountain View, CA; with Bio-Rad plugins) was used both to convert images from pic (Bio-Rad) into tiff files and to create montages, and the images were presented. 
Results
Evidence of Erosion after Manual Debridement at 1 Week after Induction of Large Wounds
Because the corneal surface in the adult mouse is approximately 3 mm in diameter, creating the large wound involved removal of most of the corneal epithelium. Despite this, initial reepithelialization of these wounds was complete between 2 and 3 days (Fig. 1) . Although we have shown that α9 integrin expression is increased in cross sections taken from open wounds 2 days after large wounds, 22 this time point occurred late in the reepithelialization process. Most of the corneal epithelium had been forced to disassemble its HDs to migrate, and the entire epithelial sheet was loosely adherent. We have found that it is technically difficult to cut unfixed frozen sections at time points earlier than 2 days after large wounds are made 22 ; therefore, we did not know how soon after wounding the increase in α9 integrin occurred or how it was first manifested. It could have increased gradually over the entire corneal surface—first at the limbal region and later toward the leading edge—or it could have increased just at the leading edge. 
To study the spatial and temporal differences in α9 integrin localization after imposing large wounds, we used whole mounts of corneal tissues for immunofluorescence. In whole mounts of unwounded control cornea, α9 integrin was observed as a patchy pattern among a subpopulation of cells in the limbal region and was absent from the cells at the central cornea (Fig. 2A) . By sacrificing mice immediately after large wounds were made, we confirmed that these wounds left the α9 integrin–positive limbal basal cells intact at the limbus (Fig. 2B) . One day after imposition of large wounds, α9 integrin was upregulated in the limbal region, as well as in cells behind the leading edge (Fig. 3A) . Upregulation of α9 integrin at the limbus at 1 day was not uniform around the circumference of the cornea, and some regions of the limbus were more positive than others. The segment of the limbus shown in Figure 3A shows an intermediate level of α9 integrin staining. At sites where there were more α9-positive cells in the limbus, there were fewer cells expressing α9 integrin toward the leading edge. Two days after wounding (Fig. 3B) , more α9 integrin was present toward the leading edge but less localized in the limbus. These results indicate that, as a function of time after wounding, α9 integrin expression increased toward the leading edge and decreased at the limbal region (Fig. 3B) . All three wounds assessed at 1 and 2 days were open (Fig. 3 , asterisks). By 3 days, all large wounds assayed were closed, accompanied by decreased α9 integrin expression over the entire corneal surface (Fig. 3D) . In the limbal region, variability was seen in the number of α9 integrin–positive cells. Some regions had none (Fig. 3D) , whereas others had many α9 integrin–positive cells (Fig. 3C) , similar to that observed in unwounded corneas. 
The localization of α9 integrin was expected to return to the pattern observed in the unwounded control cornea after 3 days. However, that was not the case. At 1 week after making the large wounds, we found that open wounds had reappeared on the ocular surface, accompanied by increased expression of α9 integrin in the peripheral cornea and toward the leading edge (Fig. 4A 4B 4C) . We hypothesized that by 1 week after wounding, small patches of epithelial cells had eroded, and the cells at the periphery of the new wound site were forced once more to migrate to cover the newly exposed wound. All three large wounds studied showed signs of erosion at 1 week. By 2 weeks after wounding, we also observed α9 integrin on cells near the center of the cornea and around the corneal periphery (Figs. 4D 4E 4F) . While the specimen in Figure 4C showed an erosion at 1 week, the one in Figure 4F did not show an eroded area but rather a depression on the corneal surface, which was covered with α9 integrin-negative epithelial cells. In addition to observing the erosions, we saw fewer cells staining positively for α9 integrin in the limbus (compare Figs. 4A and 4D with Fig. 3 and with the unwounded limbus in Fig. 2 ). 
When we evaluated corneas 4 weeks after making the large wounds, all corneas assessed had intact closed epithelia (Figs. 5A 5B) ; however, the corneal epithelium did not appear to be normal. Although α9 integrin was no longer observed in the central cornea, the quality of the healed epithelium varied within and among eyes and the localization of α9 integrin at the limbus was reduced or absent. The images shown in Figures 5A and 5B are from different regions of the same eye. In Figure 5A , the number of α9 integrin–positive cells in the limbal region appeared to be almost back to normal levels (compare Fig. 5A to the limbal region shown in Fig. 2 ), and no α9-positive cells were seen in the central cornea. However, in a different region of the same eye (Fig. 5B) , no α9 integrin–positive cells were seen at the limbal region, and a diffuse low level of staining for α9 integrin was seen among clusters of epithelial cells in the central cornea. In addition, numerous gobletlike cells (some indicated by white arrows) were seen. To confirm that these gobletlike cells were in fact goblet cells, we stained similar corneas 4 weeks after large wounds with anti-mucin antibody MUC5AC (Fig. 5C) . We routinely observed more goblet cells on the corneal surface in regions without α9 integrin in the limbus. 
Evidence of Erosion after Manual Debridement in Small Wounds
Because, as shown in Figure 1 , initial reepithelialization was complete at 3 days, we were surprised to see open wounds after 3 days. It is possible that these results, suggesting recurrent epithelial erosions, could be a side effect of removing more than 50% of the corneal epithelial cells during the initial wounding process. To determine whether similar erosions could be observed after small wounds were made, involving removal of less than 40% of the corneal epithelium, we assessed corneas for α9 integrin localization and PI staining, using whole mounts at 6, 12, 18, 24, 48, and 72 hours and at 1 and 2 weeks after the wounds were made (Fig. 6) . No fewer than three eyes were assessed at each time point studied, and the data presented reflect typical results. There was no change in the number or the intensity of staining of α9 integrin–positive cells in the limbal region, at any time point studied (data not shown). From 6 to 12 hours after wounding (Figs. 6A 6B) , there was no expression of α9 integrin at or near the leading edge; between 12 and 24 hours, integrin localization became abundant in and around the margins of the wound as it closed (Figs. 6B 6C 6D) . At 48 hours after wounding, α9 integrin localization disappeared from the central cornea (Fig. 6E) ; however, at 72 hours (Fig. 6F) and at 1 week (Fig. 6G) , some α9 integrin–positive cells were present on the central cornea. The 2-week sample showed low but diffuse staining for α9 integrin on the corneal surface, but no signs of erosions or depressions (Fig. 6H) . These data suggest that erosions of the corneal surface can occur even after small wounding, but appear to be less frequent initially. The small depression seen in Figure 6F at 72 hours was similar to the larger depression seen in Figure 4F , 2 weeks after large wounding. These localized depression sites indicate thinning areas of the corneal stroma. 
Increased Corneal Epithelial Cell Proliferation Observed Several Weeks after Creation of Large and Small Wounds
If the corneal erosions occurred after debridement wounds initially closed, we hypothesized that the rate of corneal epithelial cell proliferation with both large and small wounds would be elevated above that of unwounded control corneas, at time points beyond 1 week. To assess this, we subjected no fewer than three 8-week-old mice per time point to large wounds and allowed the wounds to heal for 1.5 and 3 days and 1, 2, and 3 weeks (Fig. 7 , dark blue bars). As expected, during the initial reepithelialization phase of wound healing, at 1.5 days, cell proliferation, as measured by BrdU incorporation, was elevated and at its maximum compared with unwounded control corneas (Fig. 7 , light blue bars). Compared with 11 labeled cells per unit length of basement membrane in unwounded corneas, there were 23 labeled cells at 1.5 days after wounding. At 3 days, when migration was complete, cell proliferation decreased but remained elevated relative to unwounded controls with slightly under 15 labeled cells per unit length; however, this increase was not statistically significant. By 1 week, the increase had become statistically significant at slightly less than 16 labeled cells per unit length. At 2 and 3 weeks after wounding, proliferation remained significantly increased relative to the control, with 20 and 18 cells labeled per unit length, respectively. These data are consistent with the hypothesis that erosions occur after the primary migration event is complete. Cell proliferation remains elevated as tissues continue to struggle to achieve and maintain their normal epithelial thickness and cell count. 
To see whether similar results were observed after small wounds, we assessed cell proliferation at 4 and 8 weeks after wounding (Fig. 7 , hatched bars). No fewer than three 8-week-old mice were subjected to small wounds per time point studied. Corneas at 4 weeks after wounding had cell proliferation rates similar to unwounded controls, and corneas at 8 weeks after small wounds had corneal epithelial cell proliferation rates significantly higher than control corneas. We noticed the large error bars on the data for cell proliferation with small wounds and hypothesized that they reflect variability in the frequency of erosions after such wounds. Some corneas had cell proliferation rates at or below those observed in unwounded corneas, whereas others had values more than double those in the control. These data are consistent with the interpretation that only some of the corneas experienced erosions. 
Taken together, the data presented thus far indicate that erosion accompanied by increased cell proliferation occur frequently after creation of both small and large manual corneal epithelial debridement wounds. 
Similar Frequency of Recurrent Erosions Documented after Induction of Large and Small Wounds
Although the data presented thus far certainly suggest that recurrent erosions occurred after manual debridement wounding in the mice, they are not definitive, given that no individual animal with a corneal wound that first closed and then reopened was studied. It is possible that any one of the individual eyes with a corneal erosion and/or elevated corneal epithelial cell proliferation weeks after wounding had never completely closed after the initial wounding. 
To confirm that the erosions we documented were recurrent rather than chronic, we designed an experiment to allow us to assess wound closure over time in individual mice. We anesthetized mice at 1, 2, 4, and 6 weeks after imposing both large and small wounds, and assessed the wounds (using fluorescein eye drops) as either open or closed in individual eyes. After the final assessment at 6 weeks, the mice were killed. All assessments were made under the dissecting microscope, and care was taken to make these observations rapidly, so that a minimal amount of anesthesia could be used. The eyes were then treated with erythromycin ophthalmic ointment to keep them moist and to prevent ocular surface drying during anesthesia. One mouse (animal 14) died at 4 weeks due to anesthesia overdose, and another was killed at 4 weeks due to poor healing of the large wound (animal 5). Data are shown in Table 1
Of the large wounds, only 5 of 16 were closed at both 1 and 2 weeks. However, two of those wounds open at 2 weeks had been closed at 1 week. By 4 and 6 weeks after the large wounds were made, many more wounds were closed, with 12 of 16 wounds closed at 4 weeks (73%) and 12 of 14 eyes closed at 6 weeks (83%). Thus, after large wounds, recurrent erosions were confirmed in the majority of wounded eyes. Of 16 mice studied, only 3 had large wounds closed at every time point assessed. The remainder showed evidence of erosion at one or more time points. 
Of the small wounds, 15 of 16 were closed at both 1 and 2 weeks. Of note, the open wound at 1 week was closed at 2 weeks, and the one that was open at 2 weeks had been closed at 1 week. Rather than the number of open wounds decreasing (as occurred with the large wounds), after small wounds were created, the number of open wounds observed increased. At 4 weeks, 11 (69%) of 16 small wounds were closed and at 6 weeks, 8 (57%) of 14 were closed. Only 3 of 16 mice had small wounds closed at every time point assessed. The mice with small wounds that were closed at all four time points (mice 1, 2, and 3) were not the same mice with large wounds that were closed at all four time points (mice 11, 14, and 15). 
Despite the trend for the closure rate of large wounds to improve and that of the small wounds to worsen over time, the overall frequency of erosions after induction of small and large wounds was similar (13 of 16 eyes wounded). We took photographs of both eyes of each mouse at the time of death (6 weeks) and evaluated those images for evidence of scaring and neovascularization (data not shown). There were no differences in either the frequency of scaring or in neovascularization between eyes with large or small wounds. The majority of eyes (70% to 80%) had corneal scars, and 50 to 60% showed some evidence of neovascularization at 6 weeks. 
Poor Healing Observed in Some Corneas Despite Evidence of Wound Closure at All Time Points
Based on results of our studies of α9 integrin expression after wounding, using whole mounts (Figs. 2 3 4 5 6) , we predicted that the eyes with more frequently open wounds would show more evidence of goblet cells and a lack of α9 integrin in the limbal region. To address this question, 10 corneas from the animals described in Table 1 were used for whole mount IF to detect α9 integrin. Tissues were counterstained with PI. The animals that received the five large and five small wounds, the wound history data taken from Table 1 , and a summary of the results of the IF/PI staining are shown in Table 2
There was substantial variability in α9 staining and overall healing of large wounds. Typical staining of large (animals 16 and 6) and small (animal 15) wounds is shown in Figures 8A 8B 8C . In animal 16, whose large-wound history indicated that it was open at 1 and 4 weeks but closed at 2 and 6 weeks, α9 integrin was absent at the limbus, and there was extensive infiltration of the cornea with goblet cells (Fig. 8A) . For animal 6, whose large wound was open at weeks 1 and 2 but closed at weeks 4 and 6, IF/PI staining showed only a few goblet cells on the corneal surface and very few α9 integrin–positive cells at the limbus. However, numerous blood vessels were observed beneath the epithelial surface (Fig. 8B) . Despite the fact that it was closed at all four time points, the eye with the large wound in mouse 15 had numerous goblet cells and a reduced expression level of α9 integrin (data not shown but summarized in Table 2 ). For the small wound in animal 15, IF/PI staining showed that the cornea had numerous α9 integrin–positive cells at the corneal periphery, adjacent to a region of central depression in the cornea (Fig. 8C) . Whether or not there was α9 staining at the central cornea that had received a small wound, there was never any evidence of goblet cells on the corneal surface. Additional analyses of the small wounds at 6 weeks showed that there was always abundant α9 integrin at the limbus (data not shown but summarized in Table 2 ). 
Presence of Tenascin-C and Failed Relocalization of α3β1 Integrin in Corneas with Recurrent Erosions
Little is known about the causes of recurrent erosions. We hypothesized that, in this animal model, one cause could be the failure of the α6β4 integrin to reassemble into HDs at the basal aspect of basal cells after migration was complete. In previous studies, we showed, by transmission electron microscopy, that the disassembly of HDs occurs during migration after wounding, and that this disassembly is accompanied by relocalization of α6β4 from the basal aspect of the basal cells at the basement membrane zone (BMZ), to the lateral and apical aspects of the corneal basal and suprabasal cells. 21 26 Thus, if α6β4 fails to form stable HDs in the corneas with recurrent erosions, then localization of α6β4 would not be restricted exclusively to the BMZ. 
The whole mount procedure used for localization of α9 integrin is limited in its ability to show the relocalization of proteins within HDs, because the fixation method reduces the antigenicity of many antigens. Furthermore, it is easier to view the linear basement membrane in cross section than en face. Most of the tissues obtained from mice after the erosion experiment were processed for whole mounts. Six eyes of animals 7, 10, and 13 were frozen unfixed for IF. Typical results are shown in Figure 9 , after staining with antibodies against α9 integrin and tenascin-C—an α9 integrin ligand (Figs. 9A 9B 9C) —and with antibodies against α3 and β4 integrins (Figs. 9D 9E 9F) . The data show that in age-matched, unwounded control corneas, α3 integrin localized to lateral and apical membranes of the basal and suprabasal cells, as well as to the basal aspect of the basal cells, where it is thought to participate somehow in maintaining the organization of the basement membrane. β4 integrin was present exclusively at the basal aspect of the basal cells in unwounded corneas where it was an integral component of the HDs. After imposition of both large and small wounds, β4 integrin had relocalized to the BMZ—even in corneas known to have recurrent erosions—and there were only minor disruptions in linear BMZ staining. In contrast, α3 integrin did not relocalize as well to the BMZ in recurrently eroded corneas with both small and large wounds (Fig. 9E 9F) . This result could be best observed in the merged images of α3 and β4 integrin staining at sites where the red β4 staining was visible at the BMZ in the recurrently eroded eyes, but not in the unwounded eyes where there was colocalization of the two proteins, and intense green staining for α3 blocks the red β4 staining (compare Figs. 9E and 9F with Fig. 9D ). 
As expected, unwounded control corneas (Fig. 9A) did not show any evidence of expression or localization of α9 integrin or tenascin-C. The top panel in Figure 9A is a double-labeled limbal cross-section of a control cornea, where the two antibodies show the presence of the α9 integrin–positive cells in the limbus, with tenascin-C present both in front of the α9-positive epithelial cells and within the deep stroma at the limbus. However, after creation of both small and large wounds, tenascin-C expression in recurrently eroded corneas remained at the anterior aspect of the corneal stroma, and, although it correlated with the expression of α9 integrin at similar sites, there was no clear evidence for colocalization at these sites (compare Figs. 9B and 9C with Figs. 9E and 9F and note the lack of overlap, as indicated by yellow at the BMZ). 
Discussion
In the this study, we established for the first time a mouse model for the study of recurrent epithelial erosion. Although the initial erosions were induced by trauma (manual debridement), the recurrent erosions developed spontaneously by 1 week after imposition of large wounds and were seen at time points up to 6 weeks after creation of both small and large wounds. Recurrent erosions were frequent and were of both the microcytic and macrocytic types. Microcytic lesions in both large and small wounds could be documented in wounds that were assessed as closed at death, using a vital dye, but found to have small openings when viewed en face using whole mounts of the mouse cornea. For the small wound data presented in Figure 6 , we had assessed all three wounds to be closed at 24 hours when, in fact, all were slightly open. The fact that the healing of large wound in animal 11 was poor, even though the wound was documented to be closed when viewed by dissecting microscope and fluorescein eye drops at all four time points assessed, provides a reminder that, although wounds judged to be open were certainly open, wounds assessed as closed could have had lesions too small to be observed using the dissecting microscope. 
α9 Integrin Upregulation during Wound Repair
After both large and small wounds, our data showed that α9 integrin was upregulated within corneal epithelial cells migrating to cover the wound. α9 Integrin was most often observed just behind the leading edge of the epithelial sheet. Just before closure, the site of the most abundant localization of α9 integrin was where opposing epithelial sheets merged, suggesting a roll in cell–cell adhesion. 
α9 Integrin is one of the many partners of β1 integrin and is expressed in unwounded corneal limbus, 20 22 skin, airway epithelium, muscle, hepatocytes, 27 neutrophils, and intestine. 28 It mediates cell adhesion and migration through interaction with its ligands, such as tenascin-C, 29 vascular cell adhesion molecule (VCAM)-1, 30 the EIIIA segment of fibronectin (FN), 31 the polypeptide of von Willebrand factor (pp-vWF), blood coagulation factor XIII (FXIII), tissue transglutaminase (tTG), 32 L1-CAM, 33 and a disintegrin and metalloproteinase (ADAM) family members. 34 α9 integrin is known to share numerous functional properties with its closest integrin homologue, α4 integrin, because both integrins recognize the same repertoire of ligands. 34 35 Studies by Cai et al. 36 and Vitale et al. 37 showed, by in situ hybridization, that corneal epithelial cells synthesize alternatively spliced FN mRNAs that include EIIIA+ isoforms during corneal wound healing. In a previous study, we found no evidence for the production of α4 integrin by corneal epithelial cells before or during healing, 19 but the expression of α9 integrin we report here would provide a cellular receptor for EIIIA+ forms of FN. The timing of the changes in expression of the EIIIA+ isoforms of FN mRNAs in the cornea in response to injury 36 37 corresponds closely to the timing of the changes we observed in α9 integrin localization and suggest that α9 integrin might serve as a receptor for EIIIA+FN during healing. 
The Role of HDs in Maintaining Adhesion
The importance of HDs in maintaining epithelial adhesion has been demonstrated in studies of knockout mice. 38 39 40 41 Knocking out α6 or β4 integrin results in the death of a mouse at birth and completely detached skin. Knocking out the α3 integrin gene in mice results in a less severe blistering phenotype, but the mice still die at birth, probably of kidney failure. 42 43 It is clear from these gene-deletion experiments that HDs are essential for epithelial–stromal interaction. Whereas α3β1 is not an absolute component of HDs, it plays an essential role in basement membrane organization that impacts HD assembly. 39  
The wounds in these studies were created with a dulled scalpel that left the basement membrane intact and not denatured by temperature or chemical treatment. This feature of the model may be critical for the induction of spontaneous erosions after wounding. We have shown in a previous study 26 the partial disassembly as a function of time after wounding of the basement membrane after manual debridement in the mouse. The sheer nature of the wound and the partial disassembly of the lamina densa may both contribute to a situation in which reinsertion of deeply embedded anchoring fibrils are inhibited. 
Diabetic patients often have corneal abnormalities, including epithelial defects associated with recurrent erosions, and reduced corneal sensitivity. In experiments performed on animal and human diabetic corneas, reduced numbers of HDs and reduced insertion of anchoring fibrils have been reported. 44 45 46 47 Both factors have been hypothesized to contribute to the loose epithelial adhesion observed in the corneas of diabetic persons. In a detailed study of basement membrane, integrin, and related protein expression in healthy and diabetic human corneas, Saghizadeh et al. 48 found that diabetic corneas upregulated mRNAs for matrix metalloproteinase (MMP)-3 and -10. MMPs degrade extracellular matrix proteins during wound healing. They also play roles in remodeling the basement membrane when migration is complete. In addition to increased MMP-3 and -10 in diabetic patients, increased α3 and β1 integrin mRNA and protein were observed in extracts of total diabetic corneas, which included endothelial cells and fibroblasts as well as corneal epithelium. Surprisingly, the corneal epithelial cells alone showed no increase in α3 and β1 integrin. α3β1 has been shown to induce MMP expression. 49 It is possible that increased α3β1 integrin expression in the stroma by keratocytes in diabetic persons could lead to increased MMP expression, which in turn could alter the structure of the stroma and basement membrane making it difficult to maintain the assembly of HDs. The idea that elevation of MMP levels may play a role in the development of RCES is supported by the work of Garrana et al. 50 who showed elevated MMP levels in the epithelial tissues of patients with recurrent erosions. Whether the etiology of recurrent erosions in diabetic persons is similar to the etiology of erosions after trauma to normal eyes is unclear. 
Involvement of Both α6β4 and α3β1 Integrins in the Stability of HDs
We also found that, compared with α3β1 in unwounded control corneas, more α3β1 was localized away from the basal cell BMZ in corneas of RCES eyes. Although it is still not clear what role α3β1 integrin plays in the reassembly of HDs, data from other studies suggest that remodeling of laminin-5 in the matrix is required to make it competent for the assembly of α6β4 into HDs, and that this remodeling is regulated by α3β1 integrin. 51 52  
In quiescent cells, α3β1 is found both in the basolateral and basal membranes of the basal cells, whereas α6β4 is present virtually exclusively at the basal surface of the basal cells within HDs. The interaction of α3β1 and α6β4 with one another during migration has been shown to activate Rho-dependent cell adhesion. 53 As cells migrate, they secrete laminin-5 with an α3 chain that has a molecular weight of 190 kDa (LNα3-190). 51 LNα3-190 supports α3β1-mediated migration and upregulation of MMP-2 and -9 expression. 49 These MMPs cleave the LNα3-190 into the LNα3-160 form and play roles in mediating the cleavage of other matrix molecules as migration proceeds. LNα3-160 supports migration through α6β4, and as cells migrate over their underlying wound bed, a matrix containing LNα3-160 accumulates beneath them. 51  
We did not directly measure the number of HDs or the depth of insertion of the anchoring fibrils that develop in the RCES corneas in mice. However, we looked at the localization of the HD component α6β4 at the BMZ and found little difference between the unwounded corneas and those from RCES mice. Our data do not tell us whether HDs were present but suggest that the initial stages of their reassembly takes place even at sites where erosions occur. α6β4 is acted on during migration by various tyrosine kinases, including Fyn—phosphorylation of several residues on its cytoplasmic tail induces recruitment of Shc and Ras to the α6β4 integrin and leads to activation of both the ERK and JNK pathways. 54 Migration is regulated by actinomycin contraction, and it remains unclear whether α6β4 can directly bind to the actin filament network. 55 If α3β1 and α6β4 were complexed together during migration, α6β4 could function primarily by mediating those signaling events necessary to sustain migration and α3β1 could mediate actin cytoskeletal rearrangements. Once migration stopped, laminin-5 synthesis would be downregulated at the transcriptional level 53 . Most of the laminin-5 present beneath cells after migration is complete would be LNα3-160. In the absence of abundant de novo synthesis of LNα3-190, Rho-dependent activation of α3β1 and α6β4 would cease, and α3β1 would no longer support migration but would begin reorganizing the underlying matrix, making it competent for the nucleation of HDs. 14 52  
Tenascin-C Modulation of Cell Adhesion
In addition to accumulating laminin-5 in the underlying matrix, migrating keratinocytes also secrete molecules such as FN, fibrin, and tenascin-C. Tenascin-C is a large molecule that functions as a hexamer. It can bind to several different integrins, including α9β1, and to heparin and FN in the cell matrix. 56 57 58 Several studies have focused attention on the role that tenascin-C plays in interfering with adhesion, primarily to FN, during development and wound healing. 57 59 Studies of skin and corneal wound healing in tenascin-C knockout mice have shown that, in the absence of tenascin-C, less FN accumulates around cells at the wound site. 60 61 Our own studies of corneal debridement wounding in tenascin-C null mice failed to show any evidence of recurrent erosions—even when we looked as late as 8 weeks after creation of small wounds. 25 The data presented in this study on wild-type mice fit with the hypothesis that retention of tenascin-C in the anterior stroma could interfere with HD reassembly. The tenascin-C knockout mice were created on a genetic background very different from BALB/c mice. They were pigmented and on a GRS/A genetic background. 25 It will be important to confirm whether erosions develop after manual debridement of mice on different genetic backgrounds other than BALB/c. 
Integrin activation can induce Rho-mediated signaling; Rho-mediated signaling can also activate integrins. 53 This amplification of RhoGTP signaling through integrins is downregulated by mechanisms not well characterized. Tenascin-C has been shown to inhibit the activation of RhoGTP. 62 Thus, a model emerges, as proposed by Ridley 63 and Murphy-Ullrich, 64 in which cells respond to accumulated tenascin-C by inhibiting RhoGTP. Inhibition of RhoGTP stops the activation of α3β1 and α6β4 integrins and mediates the conversion of migrating cells from a migratory to a quiescent phenotype. 
Differences in the Healing of Small and Large Wounds
Important differences in the way the cornea responds to large and small wounds were observed in our studies. The limbal region in eyes after large wounds was left intact and the localization of α9 integrin–positive cells at this site resembles the limbus in unwounded eyes. Yet, goblet cells were observed on the ocular surface in most corneas by 4 weeks after large wounds. We did not observe goblet cells on the central cornea after small wounds. Further, depletion of α9 integrin–positive cells at the limbal region adjacent to sites containing numerous goblet cells on the cornea surface was seen in most of the large wounds but not in the small wounds. The presence of goblet cells on the corneal surface is sufficient evidence for many studies to conclude that a limbal stem cell deficiency (LSCD) condition exists. 
Our observations suggest that the LSCD resulting from large wounds may be caused by the depletion of α9 integrin–positive cells at the limbus. At both 1 and 2 weeks after large wounds were created, fewer α9 integrin–positive cells were observed at the limbus. α9 integrin was also absent at regions adjacent to sites containing numerous goblet cells at 4 and 6 weeks after imposition of large wounds. Conjunctival cells could also have displaced the α9 integrin–positive cells at the limbus. If stem cell proliferation could not keep up with the demand for additional corneal epithelial cells, then conjunctival cell proliferation and migration could contribute to healing. 
It has been known for several years that conjunctival epithelial cells can participate in the healing of wounds to the corneal surface when the limbus is removed surgically or damaged chemically or thermally. 65 66 67 Initial excitement about the hypothesis that conjunctival cells transdifferentiate into corneal epithelial cells turned out to be premature. Later experiments showed that conjunctival cells can resemble corneal cells morphologically in organ culture and yet not express cornea-specific keratins 68 and that healing from the conjunctiva in the absence of the limbus in vivo in mice does not represent transdifferentiation, since the cells on the corneal surface continue to express conjunctival-specific keratins. 69 The cause of erosions after large wounds occur is not clear; further studies are needed to clarify this question. 
The number of α9 integrin–positive cells in the limbal region remained unchanged with the small wounds and there were no goblet cells on the ocular surface. Therefore, the erosions that developed at later time points after the small wounds were created were not the result of an LSCD but were due to other causes. Finding those causes also necessitates additional experiments but our data implicate altered integrin-mediated cell signaling and/or adhesion involving α9β1-tenascin and/or α3β1-α6β4-laminin-5. 
Limitations of the Mouse Model for RCES
Although development of animal models for the study of human disease is important, it is also important to keep in mind that mouse and human corneas are very different. In addition to the lack of a Bowman’s layer, the vast differences in stromal thickness and proximity of mouse central corneal epithelial cells to their blood supply probably gave rise during evolution to differences in the biology of cells within mouse and human corneas. Because little is known about recurrent erosion in people, it is not clear how closely this mouse model mimics human RCES. In a study of patients with RCES and epithelial basement membrane dystrophy, using in vivo confocal microscopy, Rosenberg et al. 70 detected the presence of microfolds in the subbasal nerve plexus of the eyes of patients with RCES. These folds and depressions appear to be similar to the depressed sites we saw in several of the mouse RCES eyes. Much more study should be undertaken, but the availability of numerous transgenic and knockout mice now make it possible to ask questions about what causes recurrent erosion. 
In summary, we have described the characterization of an in vivo mouse model that spontaneously develops RCES in response to a single initial corneal debridement wound. We also show that RCES, after large (2.5 mm) wounding, is accompanied by a decrease in α9 integrin–positive cells at the limbal region and by the presence of goblet cells on the ocular surface. In addition, we demonstrate that tenascin-C accumulates at the epithelial stromal interface in RCES eyes and that α3β1 integrin mislocalization is observed after wounding. To improve our understanding of the etiology of RCES in animals and people, an experimental model is now available and can be used to test hypotheses about the causes and possible treatments for RCES. Our hope is to eliminate pain and suffering from corneal erosions through a better understanding of the cornea and how it responds to injury. 
 
Figure 1.
 
The initial reepithelialization of large wounds was complete 3 days after wounding. The number of eyes indicated were assessed for the timing of wound closure after the eyes were stained (obtained after sacrifice) with a vital dye at 1, 1.5, 2, and 3 days after wounding. Eyes were photographed under a dissecting microscope.
Figure 1.
 
The initial reepithelialization of large wounds was complete 3 days after wounding. The number of eyes indicated were assessed for the timing of wound closure after the eyes were stained (obtained after sacrifice) with a vital dye at 1, 1.5, 2, and 3 days after wounding. Eyes were photographed under a dissecting microscope.
Figure 2.
 
α9 Integrin localizes to the limbus in the unwounded mouse cornea and large wounds do not remove the α9 integrin–positive cells at the limbus. Whole mounts of unwounded corneas were flattened as indicated in the diagram (A); the site of the limbus is indicated by the inner circle. Tissues were stained to reveal the distribution of α9 integrin (green) within the epithelial cells of the ocular surface, which was also stained with the nuclear marker PI (red). Note that the central cornea was negative for α9 integrin, whereas the limbal region showed numerous cells with localization of α9 integrin. A cornea wounded using the technique for large wounds and killed immediately after wounding (0 hour) was also stained for α9 integrin and PI (B). Note that the wound did not remove any of the α9 integrin–positive cells at the limbus. Images are oriented (L, limbus; C, cornea) as indicated by the arrow on the left. Bar, 50 μm.
Figure 2.
 
α9 Integrin localizes to the limbus in the unwounded mouse cornea and large wounds do not remove the α9 integrin–positive cells at the limbus. Whole mounts of unwounded corneas were flattened as indicated in the diagram (A); the site of the limbus is indicated by the inner circle. Tissues were stained to reveal the distribution of α9 integrin (green) within the epithelial cells of the ocular surface, which was also stained with the nuclear marker PI (red). Note that the central cornea was negative for α9 integrin, whereas the limbal region showed numerous cells with localization of α9 integrin. A cornea wounded using the technique for large wounds and killed immediately after wounding (0 hour) was also stained for α9 integrin and PI (B). Note that the wound did not remove any of the α9 integrin–positive cells at the limbus. Images are oriented (L, limbus; C, cornea) as indicated by the arrow on the left. Bar, 50 μm.
Figure 3.
 
Changes in α9 integrin localization during reepithelialization after large wounding. The diagram above each image indicates the orientation of each montage below relative to the entire cornea; the innermost irregular circle in the schema above (A) and (B) indicates the remaining open wound. The montages in (A), (B), and (D) show typical regions of whole mount preparations of corneas obtained 1, 2, and 3 days after wounding, respectively. (C) The limbal region only, of a cornea 3 days after wounding, to highlight variability in staining (compare C with the limbal region in D). Images are oriented as indicated by the arrows, with the central cornea toward the bottom and the limbus toward the top. (A, B, Image Not Available ) Leading edge of migration. Note that the initial reepithelialization was complete by 3 days. Bars, 50 μm.
Figure 3.
 
Changes in α9 integrin localization during reepithelialization after large wounding. The diagram above each image indicates the orientation of each montage below relative to the entire cornea; the innermost irregular circle in the schema above (A) and (B) indicates the remaining open wound. The montages in (A), (B), and (D) show typical regions of whole mount preparations of corneas obtained 1, 2, and 3 days after wounding, respectively. (C) The limbal region only, of a cornea 3 days after wounding, to highlight variability in staining (compare C with the limbal region in D). Images are oriented as indicated by the arrows, with the central cornea toward the bottom and the limbus toward the top. (A, B, Image Not Available ) Leading edge of migration. Note that the initial reepithelialization was complete by 3 days. Bars, 50 μm.
Figure 4.
 
A secondary wave of cell migration was observed in most of the corneas evaluated 1 and 2 weeks after creation of large wounds. The diagram between (A) and (D) indicates the orientation of each image presented relative to the entire cornea, with L, P, and C indicating limbus, peripheral cornea, and central cornea, respectively. Whole mount images from the limbus to the central cornea are shown from typical corneas taken 1 (AC) and 2 (DF) weeks after wounding and stained to reveal the distribution of α9 integrin (green) within epithelial cells, which have been counterstained with the nuclear marker PI (red). (B, C, E, F, insets) Higher magnifications of the regions indicated by the asterisks. (B) The cut made to flatten the cornea. Note that α9 integrin was reduced at the limbal regions in (A) and (D) but was abundant on corneal epithelial cells 1 and 2 weeks after wounding. (C) Shows an erosion. (F) Shows a central depression covered by epithelial cells not expressing α9 integrin. α9 integrin–positive cells were present around the depression. α9 integrin was most abundant 50 to 100 μm away from the edge of the migrating epithelial sheet at 1 week. Bars: (BF) 100 μm; (A, D, insets) 50 μm.
Figure 4.
 
A secondary wave of cell migration was observed in most of the corneas evaluated 1 and 2 weeks after creation of large wounds. The diagram between (A) and (D) indicates the orientation of each image presented relative to the entire cornea, with L, P, and C indicating limbus, peripheral cornea, and central cornea, respectively. Whole mount images from the limbus to the central cornea are shown from typical corneas taken 1 (AC) and 2 (DF) weeks after wounding and stained to reveal the distribution of α9 integrin (green) within epithelial cells, which have been counterstained with the nuclear marker PI (red). (B, C, E, F, insets) Higher magnifications of the regions indicated by the asterisks. (B) The cut made to flatten the cornea. Note that α9 integrin was reduced at the limbal regions in (A) and (D) but was abundant on corneal epithelial cells 1 and 2 weeks after wounding. (C) Shows an erosion. (F) Shows a central depression covered by epithelial cells not expressing α9 integrin. α9 integrin–positive cells were present around the depression. α9 integrin was most abundant 50 to 100 μm away from the edge of the migrating epithelial sheet at 1 week. Bars: (BF) 100 μm; (A, D, insets) 50 μm.
Figure 5.
 
At 4 weeks, wounds were closed but variability was obvious in the quality of the epithelial surface. The schema between (A) and (B) indicates the orientation of each image presented relative to the entire cornea (L, limbus; C, cornea). The two montages shown are whole mounts from different quadrants of the same eye taken 4 weeks after a large wound was created. Data indicate that, although some limbal regions of the cornea reverted back to the α9 integrin expression profile observed before wounding (A), other regions (B) lacked α9 integrin in the limbus and showed evidence of gobletlike cells all the way to the center of the cornea—some of which are indicated by arrows. When present in the cornea, α9 integrin appeared diffuse and cytoplasmic rather than membranous. Similar variability was observed in all three corneas stained at 4 weeks. To confirm that the cells on the central cornea, which appeared to be goblet cells in (B), were goblet cells, corneas were stained with an antibody against the mucin MUC5AC (C). Data indicate numerous goblet cells on the ocular surface 4 weeks after wounding. (C, Image Not Available ) A goblet cell cluster shown at higher magnification in the inset. Images are oriented as indicated by the arrows on the left. Bars, 100 μm.
Figure 5.
 
At 4 weeks, wounds were closed but variability was obvious in the quality of the epithelial surface. The schema between (A) and (B) indicates the orientation of each image presented relative to the entire cornea (L, limbus; C, cornea). The two montages shown are whole mounts from different quadrants of the same eye taken 4 weeks after a large wound was created. Data indicate that, although some limbal regions of the cornea reverted back to the α9 integrin expression profile observed before wounding (A), other regions (B) lacked α9 integrin in the limbus and showed evidence of gobletlike cells all the way to the center of the cornea—some of which are indicated by arrows. When present in the cornea, α9 integrin appeared diffuse and cytoplasmic rather than membranous. Similar variability was observed in all three corneas stained at 4 weeks. To confirm that the cells on the central cornea, which appeared to be goblet cells in (B), were goblet cells, corneas were stained with an antibody against the mucin MUC5AC (C). Data indicate numerous goblet cells on the ocular surface 4 weeks after wounding. (C, Image Not Available ) A goblet cell cluster shown at higher magnification in the inset. Images are oriented as indicated by the arrows on the left. Bars, 100 μm.
Figure 6.
 
Erosions may also occur with small wounds. The images presented show the central cornea at 6 (A), 12 (B), 18 (C), 24 (D), 48 (E), and 72 (F) hours and 1 (G) and 2 (H) weeks after wounding. Tissues were processed to reveal the distribution of α9 integrin (green) within the epithelial cells, which have been counterstained with the nuclear marker PI (red). Note the increase in α9 integrin immediately before closure of the wounds during initial reepithelialization at 18 and 24 hours, and the immediate decrease once the wounds are closed at 48 hours. Note the central depression surrounded by α9 integrin–positive cells at 72 hours and the few cells staining positive at 1 week. These data suggest that at least some corneas do not permanently maintain a closed epithelial surface. Magnification varied, because each image was transformed to permit the presentation of the entire open wound area. Bars: (AD, G, H) 100 μm; (E, F) 50 μm.
Figure 6.
 
Erosions may also occur with small wounds. The images presented show the central cornea at 6 (A), 12 (B), 18 (C), 24 (D), 48 (E), and 72 (F) hours and 1 (G) and 2 (H) weeks after wounding. Tissues were processed to reveal the distribution of α9 integrin (green) within the epithelial cells, which have been counterstained with the nuclear marker PI (red). Note the increase in α9 integrin immediately before closure of the wounds during initial reepithelialization at 18 and 24 hours, and the immediate decrease once the wounds are closed at 48 hours. Note the central depression surrounded by α9 integrin–positive cells at 72 hours and the few cells staining positive at 1 week. These data suggest that at least some corneas do not permanently maintain a closed epithelial surface. Magnification varied, because each image was transformed to permit the presentation of the entire open wound area. Bars: (AD, G, H) 100 μm; (E, F) 50 μm.
Figure 7.
 
Corneal epithelial cell proliferation remained elevated for weeks after debridement to create both small and large wounds. BrdU labeling of unwounded control corneas and of corneas at 1.5 and 3 days and 1, 2, and 3 weeks after induction of large wounds and at 4 and 8 weeks after creation of small wounds. Light blue bar: control corneas; dark blue bars: large wounds; hatched bars: small wounds. *Significantly different from the control (P < 0.5).
Figure 7.
 
Corneal epithelial cell proliferation remained elevated for weeks after debridement to create both small and large wounds. BrdU labeling of unwounded control corneas and of corneas at 1.5 and 3 days and 1, 2, and 3 weeks after induction of large wounds and at 4 and 8 weeks after creation of small wounds. Light blue bar: control corneas; dark blue bars: large wounds; hatched bars: small wounds. *Significantly different from the control (P < 0.5).
Table 1.
 
Wound History After Debridement Wounding
Table 1.
 
Wound History After Debridement Wounding
  Image Not Available
Table 2.
 
Summary of Whole mount Data Obtained 6 Weeks after Wounding
Table 2.
 
Summary of Whole mount Data Obtained 6 Weeks after Wounding
  Image Not Available
Figure 8.
 
Whole mount staining for α9 integrin in recurrently eroded corneas show that large wounds heal less well than small wounds. The schema at the bottom of (B) indicates the orientation of each image presented in (A), (B), and (C) relative to the entire cornea. The montages shown are whole mounts taken 6 weeks after creation of large (A, B) and small (C) wounds. The numbers in the top right corners of each image are the mouse identification numbers followed by an L for large wounds or an S for small wound. The wound history of these eyes is indicated in Table 1 . Note that recurrent erosions with large wounds were frequently accompanied by decreased α9 integrin at the limbus, numerous goblet cells over the corneal surface, and vascularization. Recurent erosions with small wounds showed that, at the limbus, the number of α9 integrin–positive cells was similar to that in the unwounded control, and no evidence of goblet cells were observed on the ocular surface. Depressed regions of various sizes, which were surrounded by α9-positive cells, were frequently observed in RCES corneas with small wounds. When present in the central cornea, α9 integrin appeared diffuse and cytoplasmic rather than membranous. Images are oriented (L, limbus; P, peripheral cornea; C, central cornea) as indicated by the arrow on the left, with the central cornea toward the bottom. Bar, 100 μm.
Figure 8.
 
Whole mount staining for α9 integrin in recurrently eroded corneas show that large wounds heal less well than small wounds. The schema at the bottom of (B) indicates the orientation of each image presented in (A), (B), and (C) relative to the entire cornea. The montages shown are whole mounts taken 6 weeks after creation of large (A, B) and small (C) wounds. The numbers in the top right corners of each image are the mouse identification numbers followed by an L for large wounds or an S for small wound. The wound history of these eyes is indicated in Table 1 . Note that recurrent erosions with large wounds were frequently accompanied by decreased α9 integrin at the limbus, numerous goblet cells over the corneal surface, and vascularization. Recurent erosions with small wounds showed that, at the limbus, the number of α9 integrin–positive cells was similar to that in the unwounded control, and no evidence of goblet cells were observed on the ocular surface. Depressed regions of various sizes, which were surrounded by α9-positive cells, were frequently observed in RCES corneas with small wounds. When present in the central cornea, α9 integrin appeared diffuse and cytoplasmic rather than membranous. Images are oriented (L, limbus; P, peripheral cornea; C, central cornea) as indicated by the arrow on the left, with the central cornea toward the bottom. Bar, 100 μm.
Figure 9.
 
The distribution of tenascin-C, an α9 integrin ligand, and α3 integrin never returned to normal after induction of small and large wounds. The images shown are typical of those obtained from unfixed frozen corneas from a subset of the mice described in Table 1 , at 6 weeks after wounding. (AC) Localization of α9 integrin and tenascin-C in control corneas, 6 weeks after imposition of small and large wounds, respectively. (DF) Localization of α3 and β4 integrins in control corneas, 6 weeks after creation of small and large wounds, respectively. Double-labeled merged images are shown first, with single-labeled green and red images shown below. Schemata above (B) and (C) indicate the size of the initial wound relative to the mouse ocular surface. Animal identification numbers are indicated for each of the four wounded tissues studied. These numbers correspond to the animals presented in Table 1 . Control eyes were obtained from unwounded, age-matched mouse corneas. Bar, 50 μm.
Figure 9.
 
The distribution of tenascin-C, an α9 integrin ligand, and α3 integrin never returned to normal after induction of small and large wounds. The images shown are typical of those obtained from unfixed frozen corneas from a subset of the mice described in Table 1 , at 6 weeks after wounding. (AC) Localization of α9 integrin and tenascin-C in control corneas, 6 weeks after imposition of small and large wounds, respectively. (DF) Localization of α3 and β4 integrins in control corneas, 6 weeks after creation of small and large wounds, respectively. Double-labeled merged images are shown first, with single-labeled green and red images shown below. Schemata above (B) and (C) indicate the size of the initial wound relative to the mouse ocular surface. Animal identification numbers are indicated for each of the four wounded tissues studied. These numbers correspond to the animals presented in Table 1 . Control eyes were obtained from unwounded, age-matched mouse corneas. Bar, 50 μm.
The authors thank Robyn Rufner, the Director of the Center for Microscopy and Image Analysis at The George Washington University, for help with the confocal microscopy and Mary Rose and Pawandeep Aujla of the Children’s National Medical Center for advice on goblet cell staining. 
Brown N, Bron A. Recurrent erosion of the cornea. Br J Ophthalmol. 1976;60:84–96. [CrossRef] [PubMed]
Goldman JN, Dohlman CH, Kravitt BA. The basement membrane of the human cornea in recurrent epithelial erosion syndrome. Trans Am Acad Ophthalmol Otolaryngol. 1969;73:471–481. [PubMed]
Reidy JJ, Paulus MP, Gona S. Recurrent erosions of the cornea: epidemiology and treatment. Cornea. 2000;19:767–771. [CrossRef] [PubMed]
Heyworth P, Morlet N, Rayner S, Hykin P, Dart J. Natural history of recurrent erosion syndrome: a 4 year review of 117 patients. Br J Ophthalmol. 1998;82:26–28. [CrossRef] [PubMed]
Maini R, Loughnan MS. Phototherapeutic keratectomy re-treatment for recurrent corneal erosion syndrome. Br J Ophthalmol. 2002;86:270–272. [CrossRef] [PubMed]
Soong HK, Farjo Q, Meyer RF, Sugar A. Diamond burr superficial keratectomy for recurrent corneal erosions. Br J Ophthalmol. 2002;86:296–298. [CrossRef] [PubMed]
Hsu JK, Rubinfeld RS, Barry P, Jester JV. Anterior stromal puncture: immunohistochemical studies in human corneas. Arch Ophthalmol. 1993;111:1057–1063. [CrossRef] [PubMed]
McLean EN, MacRae SM, Rich LF. Recurrent erosion: treatment by anterior stromal puncture. Ophthalmology. 1986;93:784–788. [CrossRef] [PubMed]
Jackson H. Effect of eye-pads on healing of simple corneal abrasions. Br Med J. 1960;5200:713.
Gipson IK. Adhesive mechanisms of the corneal epithelium. Acta Ophthalmol Suppl. 1992.13–17.
Gipson IK, Spurr-Michaud S, Tisdale A, Elwell J, Stepp MA. Redistribution of the hemidesmosome components α6β4 integrin and bullous pemphigoid antigens during epithelial wound healing. Exp Cell Res. 1993;207:86–98. [CrossRef] [PubMed]
Jones JC, Asmuth J, Baker SE, Langhofer M, Roth SI, Hopkinson SB. Hemidesmosomes: extracellular matrix/intermediate filament connectors. Exp Cell Res. 1994;213:1–11. [CrossRef] [PubMed]
Koster J, Geerts D, Favre B, Borradori L, Sonnenberg A. Analysis of the interactions between BP180, BP230, plectin and the integrin α6β4 important for hemidesmosome assembly. J Cell Sci. 2003;116:387–399. [CrossRef] [PubMed]
Geuijen CA, Sonnenberg A. Dynamics of the α6β4 integrin in keratinocytes. Mol Biol Cell. 2002;13:3845–3858. [CrossRef] [PubMed]
Bentley E, Abrams GA, Covitz D, et al. Morphology and immunohistochemistry of spontaneous chronic corneal epithelial defects (SCCED) in dogs. Invest Ophthalmol Vis Sci. 2001;42:2262–2269. [PubMed]
Murphy CJ, Marfurt CF, McDermott A, et al. Spontaneous chronic corneal epithelial defects (SCCED) in dogs: clinical features, innervation, and effect of topical SP, with or without IGF-1. Invest Ophthalmol Vis Sci. 2001;42:2252–2261. [PubMed]
Kim WJ, Helena MC, Mohan RR, Wilson SE. Changes in corneal morphology associated with chronic epithelial injury. Invest Ophthalmol Vis Sci. 1999;40:35–42. [PubMed]
Bentley E, Campbell S, Woo HM, Murphy CJ. The effect of chronic corneal epithelial debridement on epithelial and stromal morphology in dogs. Invest Ophthalmol Vis Sci. 2002;43:2136–2142. [PubMed]
Stepp MA, Spurr-Michaud S, Gipson IK. Integrins in the wounded and unwounded stratified squamous epithelium of the cornea. Invest Ophthalmol Vis Sci. 1993;34:1829–1844. [PubMed]
Stepp MA, Zhu L, Sheppard D, Cranfill RL. Localized distribution of α9 integrin in the cornea and changes in expression during corneal epithelial cell differentiation. J Histochem Cytochem. 1995;43:353–362. [CrossRef] [PubMed]
Stepp MA, Zhu L, Cranfill R. Changes in β4 integrin expression and localization in vivo in response to corneal epithelial injury. Invest Ophthalmol Vis Sci. 1996;37:1593–1601. [PubMed]
Stepp MA, Zhu L. Upregulation of α9 integrin and tenascin during epithelial regeneration after debridement in the cornea. J Histochem Cytochem. 1997;45:189–201. [CrossRef] [PubMed]
Pajoohesh-Ganji A, Pal-Ghosh S, Stepp MA. Regional distribution of α9β1 integrin within the limbus of the mouse ocular surface. Dev Dyn. .In press.
Stepp MA, Gibson HE, Gala PH, et al. Defects in keratinocyte activation during wound healing in the syndecan-1-deficient mouse. J Cell Sci. 2002;115:4517–4531. [CrossRef] [PubMed]
Sta. Iglesia DD, Gala PH, Qiu T, Stepp MA. Integrin expression during epithelial migration and restratification in the tenascin-C-deficient mouse cornea. J Histochem Cytochem. 2000;48:363–376. [CrossRef] [PubMed]
Sta Iglesia DD, Stepp MA. Disruption of the basement membrane after corneal debridement. Invest Ophthalmol Vis Sci. 2000;41:1045–1053. [PubMed]
Palmer EL, Ruegg C, Ferrando R, Pytela R, Sheppard D. Sequence and tissue distribution of the integrin α9 subunit, a novel partner of β1 that is widely distributed in epithelia and muscle. J Cell Biol. 1993;123:1289–1297. [CrossRef] [PubMed]
Desloges N, Basora N, Perreault N, Bouatrouss Y, Sheppard D, Beaulieu JF. Regulated expression of the integrin α9β1 in the epithelium of the developing human gut and in intestinal cell lines: relation with cell proliferation. J Cell Biochem. 1998;71:536–545. [CrossRef] [PubMed]
Yokosaki Y, Matsuura N, Higashiyama S, et al. Identification of the ligand binding site for the integrin α9 β1 in the third fibronectin type III repeat of tenascin-C. J Biol Chem. 1998;273:11423–11428. [CrossRef] [PubMed]
Taooka Y, Chen J, Yednock T, Sheppard D. The integrin α9β1 mediates adhesion to activated endothelial cells and transendothelial neutrophil migration through interaction with vascular cell adhesion molecule-1. J Cell Biol. 1999;145:413–420. [CrossRef] [PubMed]
Liao YF, Gotwals PJ, Koteliansky VE, Sheppard D, Van De Water L. The EIIIA segment of fibronectin is a ligand for integrins α9β1 and α4β1 providing a novel mechanism for regulating cell adhesion by alternative splicing. J Biol Chem. 2002;277:14467–14474. [CrossRef] [PubMed]
Takahashi H, Isobe T, Horibe S, et al. Tissue transglutaminase, coagulation factor XIII, and the pro-polypeptide of von Willebrand factor are all ligands for the integrins α9β1 and α4β1. J Biol Chem. 2000;275:23589–23595. [CrossRef] [PubMed]
Silletti S, Mei F, Sheppard D, Montgomery AM. Plasmin-sensitive dibasic sequences in the third fibronectin-like domain of L1-cell adhesion molecule (CAM) facilitate homomultimerization and concomitant integrin recruitment. J Cell Biol. 2000;149:1485–1502. [CrossRef] [PubMed]
Eto K, Huet C, Tarui T, et al. Functional classification of ADAMs based on a conserved motif for binding to integrin α9β1: implications for sperm-egg binding and other cell interactions. J Biol Chem. 2002;277:17804–17810. [CrossRef] [PubMed]
Williams DA, Rios M, Stephens C, Patel VP. Fibronectin and VLA-4 in haematopoietic stem cell-microenvironment interactions. Nature. 1991;352:438–441. [CrossRef] [PubMed]
Cai X, Foster CS, Liu JJ, et al. Alternatively spliced fibronectin molecules in the wounded cornea: analysis by PCR. Invest Ophthalmol Vis Sci. 1993;34:3585–3592. [PubMed]
Vitale AT, Pedroza-Seres M, Arrunategui-Correa V, et al. Differential expression of alternatively spliced fibronectin in normal and wounded rat corneal stroma versus epithelium. Invest Ophthalmol Vis Sci. 1994;35:3664–3672. [PubMed]
Dowling J, Yu QC, Fuchs E. β4 integrin is required for hemidesmosome formation, cell adhesion and cell survival. J Cell Biol. 1996;134:559–572. [CrossRef] [PubMed]
DiPersio CM, Hodivala-Dilke KM, Jaenisch R, Kreidberg JA, Hynes RO. α3β1 Integrin is required for normal development of the epidermal basement membrane. J Cell Biol. 1997;137:729–742. [CrossRef] [PubMed]
Georges-Labouesse E, Messaddeq N, Yehia G, Cadalbert L, Dierich A, Le Meur M. Absence of integrin α6 leads to epidermolysis bullosa and neonatal death in mice. Nat Genet. 1996;13:370–373. [CrossRef] [PubMed]
van der Neut R, Krimpenfort P, Calafat J, Niessen CM, Sonnenberg A. Epithelial detachment due to absence of hemidesmosomes in integrin β4 null mice. Nat Genet. 1996;13:366–369. [CrossRef] [PubMed]
Fassler R, Meyer M. Consequences of lack of β1 integrin gene expression in mice. Genes Dev. 1995;9:1896–1908. [CrossRef] [PubMed]
Kreidberg JA, Donovan MJ, Goldstein SL, et al. α3β1 integrin has a crucial role in kidney and lung organogenesis. Development. 1996;122:3537–3547. [PubMed]
Azar DT, Spurr-Michaud SJ, Tisdale AS, Gipson IK. Altered epithelial-basement membrane interactions in diabetic corneas. Arch Ophthalmol. 1992;110:537–540. [CrossRef] [PubMed]
Azar DT, Spurr-Michaud SJ, Tisdale AS, Moore MB, Gipson IK. Reassembly of the corneal epithelial adhesion structures following human epikeratoplasty. Arch Ophthalmol. 1991;109:1279–1284. [CrossRef] [PubMed]
Azar DT, Spurr-Michaud SJ, Tisdale AS, Gipson IK. Decreased penetration of anchoring fibrils into the diabetic stroma: a morphometric analysis. Arch Ophthalmol. 1989;107:1520–1523. [CrossRef] [PubMed]
Azar DT, Gipson IK. Repair of the corneal epithelial adhesion structures following keratectomy wounds in diabetic rabbits. Acta Ophthalmol Suppl. 1989;192:72–79. [PubMed]
Saghizadeh M, Brown DJ, Castellon R, et al. Overexpression of matrix metalloproteinase-10 and matrix metalloproteinase-3 in human diabetic corneas: a possible mechanism of basement membrane and integrin alterations. Am J Pathol. 2001;158:723–734. [CrossRef] [PubMed]
DiPersio CM, Shao M, Di Costanzo L, Kreidberg JA, Hynes RO. Mouse keratinocytes immortalized with large T antigen acquire α3β1 integrin-dependent secretion of MMP-9/gelatinase B. J Cell Sci. 2000;113:2909–2921. [PubMed]
Garrana RM, Zieske JD, Assouline M, Gipson IK. Matrix metalloproteinases in epithelia from human recurrent corneal erosion. Invest Ophthalmol Vis Sci. 1999;40:1266–1270. [PubMed]
Goldfinger LE, Hopkinson SB, deHart GW, Collawn S, Couchman JR, Jones JC. The α3 laminin subunit, α6β4 and α3β1 integrin coordinately regulate wound healing in cultured epithelial cells and in the skin. J Cell Sci. 1999;112:2615–2629. [PubMed]
deHart GW, Healy KE, Jones JC. The role of α3β1 integrin in determining the supramolecular organization of laminin-5 in the extracellular matrix of keratinocytes. Exp Cell Res. 2003;283:67–79. [CrossRef] [PubMed]
Nguyen BP, Ren XD, Schwartz MA, Carter WG. Ligation of integrin α3β1 by laminin 5 at the wound edge activates Rho-dependent adhesion of leading keratinocytes on collagen. J Biol Chem. 2001;276:43860–43870. [CrossRef] [PubMed]
Mariotti A, Kedeshian PA, Dans M, et al. EGF-R signaling through Fyn kinase disrupts the function of integrin α6β4 at hemidesmosomes: role in epithelial cell migration and carcinoma invasion. J Cell Biol. 2001;155:447–458. [CrossRef] [PubMed]
Rabinovitz I, Gipson IK, Mercurio AM. Traction forces mediated by α6β4 integrin: implications for basement membrane organization and tumor invasion. Mol Biol Cell. 2001;12:4030–4043. [CrossRef] [PubMed]
Chiquet-Ehrismann R, Chiquet M. Tenascins: regulation and putative functions during pathological stress. J Pathol. 2003;200:488–499. [CrossRef] [PubMed]
Erickson HP, Bourdon MA. Tenascin: an extracellular matrix protein prominent in specialized embryonic tissues and tumors. Annu Rev Cell Biol. 1989;5:71–92. [CrossRef] [PubMed]
Mackie EJ, Tucker RP. The tenascin-C knockout revisited. J Cell Sci. 1999;112:3847–3853. [PubMed]
Mackie EJ, Halfter W, Liverani D. Induction of tenascin in healing wounds. J Cell Biol. 1988;107:2757–2767. [CrossRef] [PubMed]
Matsuda A, Yoshiki A, Tagawa Y, Matsuda H, Kusakabe M. Corneal wound healing in tenascin knockout mouse. Invest Ophthalmol Vis Sci. 1999;40:1071–1080. [PubMed]
Forsberg E, Hirsch E, Frohlich L, et al. Skin wounds and severed nerves heal normally in mice lacking tenascin-C. Proc Natl Acad Sci USA. 1996;93:6594–6599. [CrossRef] [PubMed]
Wenk MB, Midwood KS, Schwarzbauer JE. Tenascin-C suppresses Rho activation. J Cell Biol. 2000;150:913–920. [CrossRef] [PubMed]
Ridley A. Rho GTPases. Integrating integrin signaling. J Cell Biol. 2000;150:F107–F109. [CrossRef] [PubMed]
Murphy-Ullrich JE. The de-adhesive activity of matricellular proteins: is intermediate cell adhesion an adaptive state?. J Clin Invest. 2001;107:785–790. [CrossRef] [PubMed]
Shapiro MS, Friend J, Thoft RA. 1981. Corneal re-epithelialization from the conjunctiva. Invest Ophthalmol Vis Sci. 1981;21:135–142. [PubMed]
Kinoshita S, Friend J, Thoft RA. Biphasic cell proliferation in transdifferentiation of conjunctival to corneal epithelium in rabbits. Invest Ophthalmol Vis Sci. 1983;24:1008–1014. [PubMed]
Tseng SC, Hirst LW, Farazdaghi M, Green WR. Goblet cell density and vascularization during conjunctival transdifferentiation. Invest Ophthalmol Vis Sci. 1984;25:1168–1176. [PubMed]
Chen WY, Mui MM, Kao WW, Liu CY, Tseng SC. Conjunctival epithelial cells do not transdifferentiate in organotypic cultures: expression of K12 keratin is restricted to corneal epithelium. Curr Eye Res. 1994;13:765–778. [CrossRef] [PubMed]
Moyer PD, Kaufman AH, Zhang Z, Kao CW, Spaulding AG, Kao WW. Conjunctival epithelial cells can resurface denuded cornea, but do not transdifferentiate to express cornea-specific keratin 12 following removal of limbal epithelium in mouse. Differentiation. 1996;60:31–38. [CrossRef] [PubMed]
Rosenberg ME, Tervo TM, Petroll WM, Vesaluoma MH. In vivo confocal microscopy of patients with corneal recurrent erosion syndrome or epithelial basement membrane dystrophy. Ophthalmology. 2000;107:565–573. [CrossRef] [PubMed]
Figure 1.
 
The initial reepithelialization of large wounds was complete 3 days after wounding. The number of eyes indicated were assessed for the timing of wound closure after the eyes were stained (obtained after sacrifice) with a vital dye at 1, 1.5, 2, and 3 days after wounding. Eyes were photographed under a dissecting microscope.
Figure 1.
 
The initial reepithelialization of large wounds was complete 3 days after wounding. The number of eyes indicated were assessed for the timing of wound closure after the eyes were stained (obtained after sacrifice) with a vital dye at 1, 1.5, 2, and 3 days after wounding. Eyes were photographed under a dissecting microscope.
Figure 2.
 
α9 Integrin localizes to the limbus in the unwounded mouse cornea and large wounds do not remove the α9 integrin–positive cells at the limbus. Whole mounts of unwounded corneas were flattened as indicated in the diagram (A); the site of the limbus is indicated by the inner circle. Tissues were stained to reveal the distribution of α9 integrin (green) within the epithelial cells of the ocular surface, which was also stained with the nuclear marker PI (red). Note that the central cornea was negative for α9 integrin, whereas the limbal region showed numerous cells with localization of α9 integrin. A cornea wounded using the technique for large wounds and killed immediately after wounding (0 hour) was also stained for α9 integrin and PI (B). Note that the wound did not remove any of the α9 integrin–positive cells at the limbus. Images are oriented (L, limbus; C, cornea) as indicated by the arrow on the left. Bar, 50 μm.
Figure 2.
 
α9 Integrin localizes to the limbus in the unwounded mouse cornea and large wounds do not remove the α9 integrin–positive cells at the limbus. Whole mounts of unwounded corneas were flattened as indicated in the diagram (A); the site of the limbus is indicated by the inner circle. Tissues were stained to reveal the distribution of α9 integrin (green) within the epithelial cells of the ocular surface, which was also stained with the nuclear marker PI (red). Note that the central cornea was negative for α9 integrin, whereas the limbal region showed numerous cells with localization of α9 integrin. A cornea wounded using the technique for large wounds and killed immediately after wounding (0 hour) was also stained for α9 integrin and PI (B). Note that the wound did not remove any of the α9 integrin–positive cells at the limbus. Images are oriented (L, limbus; C, cornea) as indicated by the arrow on the left. Bar, 50 μm.
Figure 3.
 
Changes in α9 integrin localization during reepithelialization after large wounding. The diagram above each image indicates the orientation of each montage below relative to the entire cornea; the innermost irregular circle in the schema above (A) and (B) indicates the remaining open wound. The montages in (A), (B), and (D) show typical regions of whole mount preparations of corneas obtained 1, 2, and 3 days after wounding, respectively. (C) The limbal region only, of a cornea 3 days after wounding, to highlight variability in staining (compare C with the limbal region in D). Images are oriented as indicated by the arrows, with the central cornea toward the bottom and the limbus toward the top. (A, B, Image Not Available ) Leading edge of migration. Note that the initial reepithelialization was complete by 3 days. Bars, 50 μm.
Figure 3.
 
Changes in α9 integrin localization during reepithelialization after large wounding. The diagram above each image indicates the orientation of each montage below relative to the entire cornea; the innermost irregular circle in the schema above (A) and (B) indicates the remaining open wound. The montages in (A), (B), and (D) show typical regions of whole mount preparations of corneas obtained 1, 2, and 3 days after wounding, respectively. (C) The limbal region only, of a cornea 3 days after wounding, to highlight variability in staining (compare C with the limbal region in D). Images are oriented as indicated by the arrows, with the central cornea toward the bottom and the limbus toward the top. (A, B, Image Not Available ) Leading edge of migration. Note that the initial reepithelialization was complete by 3 days. Bars, 50 μm.
Figure 4.
 
A secondary wave of cell migration was observed in most of the corneas evaluated 1 and 2 weeks after creation of large wounds. The diagram between (A) and (D) indicates the orientation of each image presented relative to the entire cornea, with L, P, and C indicating limbus, peripheral cornea, and central cornea, respectively. Whole mount images from the limbus to the central cornea are shown from typical corneas taken 1 (AC) and 2 (DF) weeks after wounding and stained to reveal the distribution of α9 integrin (green) within epithelial cells, which have been counterstained with the nuclear marker PI (red). (B, C, E, F, insets) Higher magnifications of the regions indicated by the asterisks. (B) The cut made to flatten the cornea. Note that α9 integrin was reduced at the limbal regions in (A) and (D) but was abundant on corneal epithelial cells 1 and 2 weeks after wounding. (C) Shows an erosion. (F) Shows a central depression covered by epithelial cells not expressing α9 integrin. α9 integrin–positive cells were present around the depression. α9 integrin was most abundant 50 to 100 μm away from the edge of the migrating epithelial sheet at 1 week. Bars: (BF) 100 μm; (A, D, insets) 50 μm.
Figure 4.
 
A secondary wave of cell migration was observed in most of the corneas evaluated 1 and 2 weeks after creation of large wounds. The diagram between (A) and (D) indicates the orientation of each image presented relative to the entire cornea, with L, P, and C indicating limbus, peripheral cornea, and central cornea, respectively. Whole mount images from the limbus to the central cornea are shown from typical corneas taken 1 (AC) and 2 (DF) weeks after wounding and stained to reveal the distribution of α9 integrin (green) within epithelial cells, which have been counterstained with the nuclear marker PI (red). (B, C, E, F, insets) Higher magnifications of the regions indicated by the asterisks. (B) The cut made to flatten the cornea. Note that α9 integrin was reduced at the limbal regions in (A) and (D) but was abundant on corneal epithelial cells 1 and 2 weeks after wounding. (C) Shows an erosion. (F) Shows a central depression covered by epithelial cells not expressing α9 integrin. α9 integrin–positive cells were present around the depression. α9 integrin was most abundant 50 to 100 μm away from the edge of the migrating epithelial sheet at 1 week. Bars: (BF) 100 μm; (A, D, insets) 50 μm.
Figure 5.
 
At 4 weeks, wounds were closed but variability was obvious in the quality of the epithelial surface. The schema between (A) and (B) indicates the orientation of each image presented relative to the entire cornea (L, limbus; C, cornea). The two montages shown are whole mounts from different quadrants of the same eye taken 4 weeks after a large wound was created. Data indicate that, although some limbal regions of the cornea reverted back to the α9 integrin expression profile observed before wounding (A), other regions (B) lacked α9 integrin in the limbus and showed evidence of gobletlike cells all the way to the center of the cornea—some of which are indicated by arrows. When present in the cornea, α9 integrin appeared diffuse and cytoplasmic rather than membranous. Similar variability was observed in all three corneas stained at 4 weeks. To confirm that the cells on the central cornea, which appeared to be goblet cells in (B), were goblet cells, corneas were stained with an antibody against the mucin MUC5AC (C). Data indicate numerous goblet cells on the ocular surface 4 weeks after wounding. (C, Image Not Available ) A goblet cell cluster shown at higher magnification in the inset. Images are oriented as indicated by the arrows on the left. Bars, 100 μm.
Figure 5.
 
At 4 weeks, wounds were closed but variability was obvious in the quality of the epithelial surface. The schema between (A) and (B) indicates the orientation of each image presented relative to the entire cornea (L, limbus; C, cornea). The two montages shown are whole mounts from different quadrants of the same eye taken 4 weeks after a large wound was created. Data indicate that, although some limbal regions of the cornea reverted back to the α9 integrin expression profile observed before wounding (A), other regions (B) lacked α9 integrin in the limbus and showed evidence of gobletlike cells all the way to the center of the cornea—some of which are indicated by arrows. When present in the cornea, α9 integrin appeared diffuse and cytoplasmic rather than membranous. Similar variability was observed in all three corneas stained at 4 weeks. To confirm that the cells on the central cornea, which appeared to be goblet cells in (B), were goblet cells, corneas were stained with an antibody against the mucin MUC5AC (C). Data indicate numerous goblet cells on the ocular surface 4 weeks after wounding. (C, Image Not Available ) A goblet cell cluster shown at higher magnification in the inset. Images are oriented as indicated by the arrows on the left. Bars, 100 μm.
Figure 6.
 
Erosions may also occur with small wounds. The images presented show the central cornea at 6 (A), 12 (B), 18 (C), 24 (D), 48 (E), and 72 (F) hours and 1 (G) and 2 (H) weeks after wounding. Tissues were processed to reveal the distribution of α9 integrin (green) within the epithelial cells, which have been counterstained with the nuclear marker PI (red). Note the increase in α9 integrin immediately before closure of the wounds during initial reepithelialization at 18 and 24 hours, and the immediate decrease once the wounds are closed at 48 hours. Note the central depression surrounded by α9 integrin–positive cells at 72 hours and the few cells staining positive at 1 week. These data suggest that at least some corneas do not permanently maintain a closed epithelial surface. Magnification varied, because each image was transformed to permit the presentation of the entire open wound area. Bars: (AD, G, H) 100 μm; (E, F) 50 μm.
Figure 6.
 
Erosions may also occur with small wounds. The images presented show the central cornea at 6 (A), 12 (B), 18 (C), 24 (D), 48 (E), and 72 (F) hours and 1 (G) and 2 (H) weeks after wounding. Tissues were processed to reveal the distribution of α9 integrin (green) within the epithelial cells, which have been counterstained with the nuclear marker PI (red). Note the increase in α9 integrin immediately before closure of the wounds during initial reepithelialization at 18 and 24 hours, and the immediate decrease once the wounds are closed at 48 hours. Note the central depression surrounded by α9 integrin–positive cells at 72 hours and the few cells staining positive at 1 week. These data suggest that at least some corneas do not permanently maintain a closed epithelial surface. Magnification varied, because each image was transformed to permit the presentation of the entire open wound area. Bars: (AD, G, H) 100 μm; (E, F) 50 μm.
Figure 7.
 
Corneal epithelial cell proliferation remained elevated for weeks after debridement to create both small and large wounds. BrdU labeling of unwounded control corneas and of corneas at 1.5 and 3 days and 1, 2, and 3 weeks after induction of large wounds and at 4 and 8 weeks after creation of small wounds. Light blue bar: control corneas; dark blue bars: large wounds; hatched bars: small wounds. *Significantly different from the control (P < 0.5).
Figure 7.
 
Corneal epithelial cell proliferation remained elevated for weeks after debridement to create both small and large wounds. BrdU labeling of unwounded control corneas and of corneas at 1.5 and 3 days and 1, 2, and 3 weeks after induction of large wounds and at 4 and 8 weeks after creation of small wounds. Light blue bar: control corneas; dark blue bars: large wounds; hatched bars: small wounds. *Significantly different from the control (P < 0.5).
Figure 8.
 
Whole mount staining for α9 integrin in recurrently eroded corneas show that large wounds heal less well than small wounds. The schema at the bottom of (B) indicates the orientation of each image presented in (A), (B), and (C) relative to the entire cornea. The montages shown are whole mounts taken 6 weeks after creation of large (A, B) and small (C) wounds. The numbers in the top right corners of each image are the mouse identification numbers followed by an L for large wounds or an S for small wound. The wound history of these eyes is indicated in Table 1 . Note that recurrent erosions with large wounds were frequently accompanied by decreased α9 integrin at the limbus, numerous goblet cells over the corneal surface, and vascularization. Recurent erosions with small wounds showed that, at the limbus, the number of α9 integrin–positive cells was similar to that in the unwounded control, and no evidence of goblet cells were observed on the ocular surface. Depressed regions of various sizes, which were surrounded by α9-positive cells, were frequently observed in RCES corneas with small wounds. When present in the central cornea, α9 integrin appeared diffuse and cytoplasmic rather than membranous. Images are oriented (L, limbus; P, peripheral cornea; C, central cornea) as indicated by the arrow on the left, with the central cornea toward the bottom. Bar, 100 μm.
Figure 8.
 
Whole mount staining for α9 integrin in recurrently eroded corneas show that large wounds heal less well than small wounds. The schema at the bottom of (B) indicates the orientation of each image presented in (A), (B), and (C) relative to the entire cornea. The montages shown are whole mounts taken 6 weeks after creation of large (A, B) and small (C) wounds. The numbers in the top right corners of each image are the mouse identification numbers followed by an L for large wounds or an S for small wound. The wound history of these eyes is indicated in Table 1 . Note that recurrent erosions with large wounds were frequently accompanied by decreased α9 integrin at the limbus, numerous goblet cells over the corneal surface, and vascularization. Recurent erosions with small wounds showed that, at the limbus, the number of α9 integrin–positive cells was similar to that in the unwounded control, and no evidence of goblet cells were observed on the ocular surface. Depressed regions of various sizes, which were surrounded by α9-positive cells, were frequently observed in RCES corneas with small wounds. When present in the central cornea, α9 integrin appeared diffuse and cytoplasmic rather than membranous. Images are oriented (L, limbus; P, peripheral cornea; C, central cornea) as indicated by the arrow on the left, with the central cornea toward the bottom. Bar, 100 μm.
Figure 9.
 
The distribution of tenascin-C, an α9 integrin ligand, and α3 integrin never returned to normal after induction of small and large wounds. The images shown are typical of those obtained from unfixed frozen corneas from a subset of the mice described in Table 1 , at 6 weeks after wounding. (AC) Localization of α9 integrin and tenascin-C in control corneas, 6 weeks after imposition of small and large wounds, respectively. (DF) Localization of α3 and β4 integrins in control corneas, 6 weeks after creation of small and large wounds, respectively. Double-labeled merged images are shown first, with single-labeled green and red images shown below. Schemata above (B) and (C) indicate the size of the initial wound relative to the mouse ocular surface. Animal identification numbers are indicated for each of the four wounded tissues studied. These numbers correspond to the animals presented in Table 1 . Control eyes were obtained from unwounded, age-matched mouse corneas. Bar, 50 μm.
Figure 9.
 
The distribution of tenascin-C, an α9 integrin ligand, and α3 integrin never returned to normal after induction of small and large wounds. The images shown are typical of those obtained from unfixed frozen corneas from a subset of the mice described in Table 1 , at 6 weeks after wounding. (AC) Localization of α9 integrin and tenascin-C in control corneas, 6 weeks after imposition of small and large wounds, respectively. (DF) Localization of α3 and β4 integrins in control corneas, 6 weeks after creation of small and large wounds, respectively. Double-labeled merged images are shown first, with single-labeled green and red images shown below. Schemata above (B) and (C) indicate the size of the initial wound relative to the mouse ocular surface. Animal identification numbers are indicated for each of the four wounded tissues studied. These numbers correspond to the animals presented in Table 1 . Control eyes were obtained from unwounded, age-matched mouse corneas. Bar, 50 μm.
Table 1.
 
Wound History After Debridement Wounding
Table 1.
 
Wound History After Debridement Wounding
  Image Not Available
Table 2.
 
Summary of Whole mount Data Obtained 6 Weeks after Wounding
Table 2.
 
Summary of Whole mount Data Obtained 6 Weeks after Wounding
  Image Not Available
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×