Integrin β1 Is Necessary for the Maintenance of Corneal Structural Integrity

Abstract

Purpose.: The precise role of a normal keratocyte in maintaining corneal structural integrity is unclear; it is generally considered to remain quiescent at the end of cell division. Given that integrins are essential for cell/extracellular matrix interactions, the authors tested the hypothesis that integrin expression by keratocytes is essential for corneal structure and function.

Methods.: Using a tamoxifen-dependent cre recombinase expressed under the control of a fibroblast-specific promoter/enhancer, the authors conditionally deleted the integrin β1 (Itgb1) gene in mouse keratocytes during the postnatal matrix maturation phase of the cornea. The effects of this deletion were monitored histologically and by macroscopic observation of the cornea.

Results.: The resultant cornea shows an initial thinning of the stroma, reduced space between collagen fibrils, loss of epithelial layers and subsequent edema, thickening of Descemet's membrane, and degenerative changes in the endothelial cell layer, with eventual scarring. These pathologic changes have some similarities to human corneal disease keratoconus. The phenotype did not develop when Itgb1 was deleted after complete corneal maturation.

Conclusions.: Loss of integrin β1 expression in keratocytes during the phase of stromal maturation results in corneal thinning and edema. Keratocyte-ECM interaction is essential for matrix maturation and thus in the maintenance of corneal structural integrity. This model has relevance in understanding corneal diseases such as keratoconus.

Understanding the mechanisms by which corneal structural integrity is maintained is extremely important because corneal diseases that cause loss of transparency are second only to cataracts as a cause of blindness affecting the general population. ¹ The cornea consists of an outer squamous epithelial layer, an inner endothelial layer, and the connective tissue layer stroma in between. The epithelial cells act as an immuno-barrier, protecting the posterior cornea, whereas the endothelial cells are vital in controlling the hydration of the cornea. The stroma is responsible for maintaining the structure of the cornea and constitutes nearly 85% of corneal thickness. ²  

Structurally, the stroma is composed of keratocytes (fibroblasts) that lie between lamellae of orthogonally arranged collagen fibrils surrounded by proteoglycans. ³ Several studies have investigated the role of keratocytes in the development of the cornea ^4,5 and in fibrotic scarring after surgery or injury. ⁶ However, the functional role of a normal keratocyte in the maintenance of structural integrity of the cornea is not well understood; it is considered to be mostly quiescent having left the cell cycle. ^7,8  

Connective tissue, in general, is essential for the maintenance of organ structure and function; defects in connective tissue function underlie disorders including chondrodysplasia and Marfan syndrome. ⁹ In addition, many studies have shown that cell-extracellular matrix (ECM) interactions are essential for the survival, migration, morphogenesis, differentiation, and transduction of mechanical signals; lack of such interactions is usually deleterious. ¹⁰ Because corneal connective tissue (stroma) is a structure under constant stress from intraocular pressure (IOP), we hypothesized that the adherence of keratocytes to ECM (to resist IOP) is essential to maintain corneal homeostasis and structural integrity. 

Cells interact with their ECM primarily through integrin receptors, which are heterodimeric proteins made up of alpha and beta subunits. Integrins mediate cell adhesion and mechanical stress, serving as a bidirectional conduit for signals originating from inside the cell and from outside the cell. ¹⁰ In the lens, for example, integrin β1 is known to have a role in the maintenance of the lens epithelial phenotype. ¹¹ Corneal keratocytes express collagen-binding integrins α2β1, α3β1, α5β1, and α6β1. ¹² Given that cells interact with their ECM primarily through integrin receptors, we conditionally deleted the integrin beta 1 (Itgb1) gene in fibroblasts (keratocytes) of the postnatal mouse, in vivo, to test our hypothesis. 

Consistent with our hypothesis, we found that deleting Itgb1 in keratocytes resulted in loss of structural integrity of the cornea, with progressive thinning of the stroma, loss of epithelial layers, and eventual edema and scarring. This pathology is observed only when Itgb1 is deleted during the postnatal maturation phase of the stroma, but not when Itgb1 is deleted after complete maturation of the cornea. Our study establishes new and valuable insights into the function of normal keratocytes in the maintenance of corneal homeostasis and structural integrity through their interaction with ECM. 

Materials and Methods

Conditional Knockout Mice

Mice with exon 3 of the Itgb1 gene flanked by loxP sites (Itgb1 ^f/f ) ¹³ were obtained from The Jackson Laboratory (Bar Harbor, ME) and were mated with mice expressing tamoxifen-dependent Cre recombinase under the control of the fibroblast-specific regulatory sequence from the pro alpha 2 (I) collagen gene [Col1a2-Cre(ER)T]. ¹⁴ Animal protocols were approved by the appropriate committee at the University of Western Ontario; the project adhered to the ARVO Statement for Use of Animals in Ophthalmic and Vision Research. To construct the expression vector for [Col1a2-Cre(ER)T], a 6.4-kb fragment incorporating 6 kb of the upstream 5′-flanking region of the Col1a2 gene (−19.5 kb to −13.5 kb upstream of the transcription start site was linked to an endogenous minimal promoter (−350 bp to +54 bp). Mice homozygous for Itgb1 ^f/f and hemizygous for Cre were administered tamoxifen (1 mg/mouse for 5 days) at 23 to 26 days of age to delete (or not) Itgb1 gene specifically in fibroblasts. Genetically identical littermates or littermates that did not have the Cre allele received corn oil (vehicle) or tamoxifen, respectively, and served as control mice (Itgb1 ^+/+). Corneal stroma was separated from epithelial and endothelial cells and was used to genotype mice for the presence of floxed Itgb1 allele by using primers 5′-CGGCTCAAAGCAGAGTGTCAGTC-3′ and 5′-CCACAACTTTCCCAGTTAGCTCTC-3′. The size of exon 3 after deletion was determined using primers 5′-TGAATATGGGCTTGGCAGTTA-3′ and 5′-CCACAACTTTCCCAGTTAGCTCTC-3′. ¹¹ The presence of the Cre sequence was determined using primers 5′-ATCCGAAAAGAAAACGTTGA-3′ and 5′-ATCCAGGTTACGGATATAGT-3′. ¹⁴  

β-Galactosidase Staining

ROSA26-lacZ reporter mice were obtained from The Jackson Laboratory and mated with mice expressing tamoxifen-dependent Cre recombinase under the control of fibroblast-specific regulatory sequence from Col1a2-Cre(ER)T. ¹⁴ Genotyping was conducted as described here and on The Jackson Laboratory Web site. Mice (23–26 days old) were injected with tamoxifen as described. For whole-mount β-galactosidase staining, eyes were fixed in 0.2% glutaraldehyde/1% formaldehyde/0.02% Nonidet P-40 in phosphate-buffered saline (PBS) overnight at 4°C. After washing with PBS for 3 × 10 minutes at room temperature, the tissues were incubated in X-gal staining solution [5 mM K3Fe(CN)6, 5 mM K4Fe(CN)6, 2 mM MgCl2, 1 mg/mL X-gal in PBS] overnight at room temperature on rocker. Tissues were kept in PBS for 2 days to allow for stain to develop. The staining was stopped by 4% paraformaldehyde, and the tissue was subsequently sectioned for histology. 

Corneal Keratocyte Culture

Corneal keratocytes were cultured using a previously published method ¹⁵ with some modifications. Mouse eyes were enucleated, washed, and placed in DMEM (Invitrogen, Carlsbad, CA) containing 20 mM HEPES buffer (Invitrogen), 15 mg/mL dispase (Invitrogen), 100 mM sorbitol (Sigma, St. Louis, MO) and 2% antibiotics/antimycotic solution (Invitrogen) at 4°C overnight. Corneal epithelium was then peeled off, and the cornea dissected excluding the limbal region. The endothelium was removed by gentle scraping. The resultant stroma then placed in DMEM containing 20 mM HEPES buffer, 1.25 mg/mL collagenase A (Roche Diagnostics, Indianapolis, IN), and 2% antibiotic/antimycotic at 37°C overnight. The keratocytes released from the stroma were then centrifuged at 800g for 5 minutes, and the pellet was plated in their culture medium as follows: DMEM with 20 mM HEPES buffer, 10% fetal bovine serum (Invitrogen), reagent (ITS Liquid Media Supplement; Sigma), and 1% antibiotic and anti-mycotic solution. 

Immunofluorescence Microscopy

Mouse eyes were enucleated and cornea dissected and fixed in 4% paraformaldehyde in PBS overnight at 4°C, then processed either for cryosectioning or paraffin-embedding. Immunofluorescence staining was performed on cryosections. For cryosectioning, the corneas were placed in 30% sucrose in PBS at 4°C overnight and embedded in a 1:2 solution of OCT and 20% sucrose, and sections were taken at 7 μm thickness. Paraffin-embedded corneas were sectioned at 5-μm thickness. Sections were blocked with 2% BSA and 0.1% Triton X-100 in PBS for 1 hour and incubated with primary antibody added in the blocking solution for 1.5 hours, washed, and incubated with appropriate fluorescence-tagged secondary antibody in blocking solution at 37°C for 45 minutes. Aqueous medium containing DAPI was used for mounting, and images were recorded using a microscope (Axio Imager M1; Zeiss, Oberkochen, Germany). Antibodies used were integrin β1 (R&D Systems, Minneapolis, MN), NFκB (Santa Cruz Biotechnology, Santa Cruz, CA), FSP-1 (Abcam, Cambridge, MA), cyclin D1, Fak, p-Fak, p-Akt, p-Jnk, and p-Erk (Cell Signaling, Beverly, MA). 

Transmission Electron Microscopy

Corneas were fixed overnight in 2% glutaraldehyde in 0.1 M sodium phosphate buffer (pH 7.2) at 4°C. Osmium tetroxide (1%) in 0.1 M sodium phosphate buffer (pH 7.2) was used for secondary fixation for 1 hour. Tissue was dehydrated through a graded ethanol series and infiltrated with propylene oxide/epoxy resin. Ultrathin sections from central cornea were collected on copper grids and stained with lead citrate and uranyl acetate. Sections were examined with a model 410 microscope (Philips, Eindhoven, The Netherlands). 

Results

Fibroblast-Specific Cre Deletes Itgb1 in Corneal Keratocytes

Whole body deletion of the Itgb1 gene causes embryonic lethality ^16,17 ; to study the impact of loss of Itgb1 in adults, the use of conditional knockout animals is required. ^13,18 By deleting the Itgb1 gene specifically in the fibroblasts of adult animals, we intended to decipher the function of keratocyte (fibroblast)-ECM interaction in the maintenance of corneal structural integrity. To delete genes specifically in fibroblasts, we used a tamoxifen-dependent Cre recombinase expressed under the control of the fibroblast-specific type I collagen promoter/enhancer [Col1a2-Cre(ER)T]. ^14,19,20 To confirm the specificity of the [col1a2-cre(ER)T] in the eye, 23- to 26-day-old mice harboring [Col1a2-Cre(ER)T] and ROSA26-lacZ (which expresses β-galactosidase where Cre is expressed) were generated. Administration of tamoxifen to ROSA26-lacZ-Cre mice resulted in the translocation of the Cre recombinase into the nucleus, enabling the expression of the lacZ reporter. When mice were examined 21 days after tamoxifen injection, β-galactosidase expression was limited to the fibroblasts (keratocytes); however, the occasional endothelial cell (<5%) stained positive for β-galactosidase (Fig. 1A). 

Figure 1.

 

Conditional deletion of Itgb1 in corneal keratocytes (fibroblasts). (A) Specificity of the type I collagen promoter/enhancer in cornea. Mice hemizygous for [Col1a2-Cre(ER)T] and for the ROSA26-lacZ reporter (which permits expression of β-galactosidase in cells expressing Cre recombinase) were generated. Mice (23–26 days after birth) were injected with tamoxifen. Corneas were stained for β-galactosidase 21 days after tamoxifen injection. Note that β-galactosidase expression is restricted to the keratocytes (however, the occasional endothelial cell, ∼2–3 cells/5-μm cross-section of cornea) stained positive for β-galactosidase; arrow). (B–E) Mice in which the Itgb1 gene was conditionally deleted in keratocytes were generated as described in this study and previously. ¹³ (B) Genotyping of mice. Deletion of Itgb1 gene was tested by PCR genotyping of DNA extracted from corneal stroma. Mice containing the Itgb1 gene (Itgb1 ^+/+; vehicle administration, lanes 1, 2) or in which the Itgb1 gene was conditionally deleted (Itgb1 ^{− / −}; tamoxifen administration, lanes 3, 4) in the fibroblasts (keratocytes) are shown. Corneal sections derived from Itgb1 ^+/+ (C) or Itgb1 ^−/− (D) mice were subjected to indirect immunofluorescence with anti–integrin β1 antibody (red). Sections were also stained with DAPI (blue) to detect cell nuclei. Mice 21 days after tamoxifen (Itgb1 ^−/−) or corn oil injection (Itgb1 ^+/+) are shown. Note that loss of integrin β1 expression is restricted to keratocytes. (E) Western blot analysis of keratocytes derived from Itgb1 ^+/+ or Itgb1 ^−/− mice was performed with anti–integrin β1 antibody. For all assays, cells from five mice were analyzed. A representative Western blot is shown.

Based on these results, to delete Itgb1 in keratocytes, we generated 23- to 26-day-old mice homozygous for a Itgb1 ^f/f allele ¹³ and hemizygous for a tamoxifen-dependent Cre recombinase expressed under the control of a fibroblast-specific promoter/enhancer. ¹⁴ Control mice (Itgb1 ^+/+) were maintained. PCR analysis of DNA obtained from corneal stroma showed that the Itgb1 gene was deleted only in animals in which tamoxifen-activated Cre enzyme was present (Fig. 1B; lanes 3, 4). When Itgb1 ^−/− mice were examined 21 days after tamoxifen injection, the expression of ITGB1 protein was undetectable in the stromal keratocytes compared with control mice (Figs. 1C, 1D). Consistent with our analysis of the Col1a2-Cre(ER)T ROSA26-lacZ mice showing specificity of the promoter/activity in fibroblasts, the expression of ITGB1 protein in the adjacent epithelial cell layers and in the endothelial layer of the cornea remained unaffected (Figs. 1C, 1D). An assessment by Western blot analysis of ITGB1 protein expression in cultured keratocytes derived from these mice confirmed that there was no detectable amount in animals in which the Itgb1 gene has been deleted in fibroblasts (Fig. 1E). The loss of Itgb1 and subsequent changes in cell morphology as well as deficiency in cell adherence, spreading, and contraction were supported by in vitro experiments with findings (data not shown) similar to our previously published results. ²⁰  

Deletion of Itgb1 Results in Loss of Corneal Structural Integrity

At 21 days after tamoxifen injection, the Itgb1 ^−/− corneas appeared normal and were similar to control Itgb1 ^+/+ corneas (Figs. 2A, 2B). However, by 42 to 45 days after tamoxifen injection, compared with Itgb1 ^+/+ corneas, Itgb1 ^−/− corneas displayed considerable thinning of most of the stroma and loss of epithelial layers (Figs. 2C, 2D). Examination of Itgb1 ^−/− corneas 35 days after tamoxifen by electron microscopy supported thinning of the stroma as collagen fibrils were arranged more closely to each other in the stromas of Itgb1 ^−/− corneas than in Itgb1 ^+/+ corneas (Supplementary Fig. S1A). However, the reduction in space was not caused by the loss of proteoglycans because there was no consistent and appreciable difference in Alcian blue (pH 1.0; sulfated acid proteoglycan stain) staining between Itgb1 ^+/+ and Itgb1 ^−/− corneas (Supplementary Figs. S1B). 

Figure 2.

 

Deletion of Itgb1 in corneal keratocytes results in alterations in corneal structure resembling a keratoconus phenotype. Hematoxylin and eosin staining of corneal sections. Corneal thickness (μm) was assessed from paraffin sections (n = 5) taken from similar regions of corneas (±75 μm); sections from frozen corneas with similar changes were not added to the analysis (unlike paraffin-embedded corneas, it was challenging to consistently obtain sections from similar regions of frozen corneas). Average thickness ± SD are shown (*P = 0.05, Student's t-test, unpaired). (A, B) No obvious change in morphology of the cornea was seen 21 days after tamoxifen injection in Itgb1 ^−/− animals compared with Itgb1 ^+/+ animals. (C, D) By 42 to 45 days after tamoxifen injection, there were areas of thinning (P = 0.01) in the Itgb1 ^−/− corneas compared with the Itgb1 ^+/+ corneas. (E, F, and insets) By 50 to 55 days after tamoxifen injection, Itgb1 ^−/− corneas exhibited patches of edematous changes with increased corneal thickness (P = 0.02), and the epithelial layer numbers were further reduced compared with Itgb1 ^+/+ corneas. Within edematous regions, there were degenerative changes in the endothelial layer with endothelial cells (F, arrow) peeling away from a thickened Descemet's membrane (F, arrowhead). (G, H) By 75 days after tamoxifen injection, epithelial layers are reduced to almost two layers in Itgb1 ^−/− corneas, with the beginnings of corneal epithelial erosion (H, arrows); neutrophils and endothelial cells around capillary lumen can also be seen (see Fig. 3B, higher magnification). (H, inset) At 75 days (Fig. 3B), an Itgb1 ^−/− cornea with a large number of keratocytes (fibroblasts) (fibroblast-specific protein-1 staining; frozen cornea section) was present in the area of scarring. (I and inset) Itgb1 ^−/− corneas also exhibited patches of extra layers of epithelial cells, causing distortions in the curvature of the cornea.

By 50 to 55 days after tamoxifen administration, each Itgb1 ^−/− cornea showed localized regions of edema, and the epithelial layers showed further reduction in number compared with Itgb1 ^+/+ corneas (Figs. 2E, 2F, and insets). Within edematous regions, keratocytes were enlarged, and the endothelial layer showed degenerative changes, with endothelial cells peeling away (arrow) from a thickened Descemet's membrane (Fig. 2F, arrowhead). 

By 75 days after tamoxifen injection, the epithelial cells in Itgb1 ^−/− corneas were mostly reduced to a single layer, with some (3 of 6 animals 75 days after tamoxifen injection) corneas showing signs of the beginning of epithelial erosion (Fig. 2H, arrowhead) or scarring, with a large number of fibroblasts (as deduced by FSP-1 staining) present in the area of erosion (Fig. 2H, inset). These corneas also showed evidence of external scarring (Fig. 3A, arrow) and haziness of the cornea, with one cornea showing the visible presence of neovascularization (Fig. 3A, arrowheads); the normal cornea has neither blood vessels nor macrophages. Some Itgb1 ^−/− corneas (3 of 6) exhibited patches of extra layers of epithelial cells, causing distortions in the curvature of the cornea (Fig. 2I and inset). Moreover, hematoxylin and eosin staining of these Itgb1 ^−/− corneas revealed the presence of thin spindle-shaped endothelial cells surrounding capillary lumen (Fig. 3B, arrows) and neutrophils (with their typical lobular nucleus; Fig. 3B, asterisk) in their stromas. The presence of neovascularization and neutrophils indicated inflammation but were likely to be secondary events resulting from the compromise of corneal structural integrity. 

Figure 3.

 

External morphology of Itgb1 ^−/− corneas. (A) Uneven epithelial surface due to loss of epithelial layers (asterisk) can be seen in this Itgb1 ^−/− cornea 75 days after tamoxifen injection, along with a region of scar tissue (arrow) and associated haziness of the cornea; neovascularization is also visible. (B) Magnification of image as in Figure 2H. Stromas of all Itgb1 ^−/− corneas that had severe loss of epithelial layers and edema showed the presence of neutrophils with their typical multilobular nucleus (asterisk). Also present were spindle-shaped endothelial cells surrounding capillary lumen (arrows).

Itgb1 ^−/− Corneas Have Reduced Cyclin D1 Expression in Basal Corneal Epithelial Cells

Basal corneal epithelial cells divide and stratify upward to replace the outer epithelial cell layers that are regularly sloughed. When Itgb1 ^−/− corneas at 42 days after tamoxifen were examined, we found that the basal epithelial cells had reduced cyclin D1 expression compared with controls (Fig. 4A), indicating that the loss of epithelial layers was due to deficiency in cell division; disruption in keratocyte-epithelial interactions ^21,22 is likely to be a contributing factor to this observation. However, in edematous patches of Itgb1 ^−/− corneas, basal epithelial cells stained intensely for cyclin D1 (data not shown). 

Figure 4.

 

Changes in protein expression in Itgb1 ^−/− corneas. (A) Reduced expression of cyclin D1 expression (arrowheads) in basal epithelial cells of Itgb1 ^−/− corneas indicated that the loss of epithelial layers occurred because of the lack of cell division. Shown are representative (n = 5) corneas 42 days after tamoxifen administration. However, (B) keratocytes remained quiescent even after the deletion of Itgb1. Itgb1 ^+/+ and Itgb1 ^−/− corneal keratocytes did not express FAK. (C) Five Itgb1 ^−/−corneas with edema were examined. NFκB (p65) expression was seen in keratocytes of Itgb1 ^−/− corneas only within areas of edema and severe loss of epithelial cells, indicating activation of inflammation. A representative Itgb1 ^−/− cornea 75 days after tamoxifen administration is shown. (C, inset) A magnified image of NFκB (p65)–positive keratocytes.

Keratocytes of Itgb1 ^−/− Corneas Show No Changes in Adhesive Signaling

To further understand the corneal pathology and the changes in keratocyte function after deletion of the Itgb1 ^−/− gene, we evaluated by immunostaining whether Itgb1 ^−/− corneas showed defects in activation of known downstream mediators of ITGB1 adhesive signaling. Surprisingly, both Itgb1 ^+/+ and Itgb1 ^−/− corneal keratocytes showed no expression of Fak (Fig. 4B); the marked downregulation of Fak has been reported in lens fibroblasts. ²³ Expression of many other signaling molecules, such as phospho-Fak, -Akt, -Erk, and -Jnk (Supplementary Fig. S2), were also not readily detectable in keratocytes of both Itgb1 ^+/+ and Itgb1 ^−/− corneas. These data are consistent with previous observations that keratocytes are quiescent after cell division, ^7,8 and they seem to remain quiescent even after the deletion of Itgb1. However, at later stages, some Itgb1 ^−/− corneas showed expression of NFκB (Fig. 4C) exclusively within the patches of edema that were also accompanied by severe loss of epithelial layers. Given that there was no expression of NFκB at early time points after Itgb1 deletion, its expression at later stages of the pathology are secondary, possibly due to the activation of inflammation-mediated pathways on severe compromise of corneal integrity. Thus mechanisms other than changes in adhesive signaling are likely to be responsible for the pathology of Itgb1 ^−/− corneas. 

Lack of Mature Stromal Matrix Organization in Itgb1 ^−/− Corneas

One of the major mechanisms implicated as a cause of keratoconus pathology is the lack of interlamellar cohesion. ²⁴ Thus, an alternative mechanism underlying the keratoconus-like phenotype of Itgb1 ^−/− corneas could be that the loss of integrin β1 results in a failure to properly organize corneal matrix during maturation of cornea. Cornea continues to mature after birth; keratocyte cell division in the stroma plateau around 21 days in rabbit cornea, but the stroma continues to mature, with much thicker lamellae at 30 days. ⁸ In mouse too, the corneal stroma reached adult thickness only around 30 days after birth. ²⁵ Our results described above were initiated on 23 to 26 day-old mice. To assess whether the keratoconus-like phenotype can be induced in mature cornea, we decided to delete Itgb1 gene in keratocytes 40days after birth of mice. We examined these Itgb1 ^−/− corneas at 75 days after tamoxifen administration and found no structural changes; indeed they were structurally similar to control corneas (Fig. 5). Thus, these results suggest that keratocyte-ECM interaction is necessary for the maturation of matrix organization during corneal development and thus is critical in maintaining the structural integrity of the cornea. 

Figure 5.

 

Integrin β1 is not required for maintenance of the adult corneal structure. Mouse corneal stroma reached adult thickness only around 30 days after birth. Itgb1 was deleted in keratocytes of mature cornea (40 days after birth of mice). Deletion of the Itgb1 gene in the mature cornea was tested by PCR genotyping of DNA extracted from corneal stroma. Mice containing the Itgb1 gene (Itgb1 ^+/+; vehicle administration, lanes 1, 2) or in which the Itgb1 gene was conditionally deleted (Itgb1 ^−/−; tamoxifen administration, lanes 3, 4) in the fibroblasts (keratocytes) are shown. When Itgb1 was deleted in the mature cornea, the cornea developed no structural changes and was similar to control corneas; shown are the Itgb1 ^−/− cornea 75 days after tamoxifen administration and the corresponding control cornea (n = 5). Note again that the expression of ITGB1 protein in the endothelium was not affected. These results indicate that keratocyte-ECM interaction through integrin β1 is necessary for the maturation of stromal matrix organization and thus the structural integrity of the cornea.

Discussion

By deleting the Itgb1 gene in keratocytes in mice 23 to 26 days after birth, we were able to reveal the importance of adhesion in maintaining the structural integrity of the cornea. Structural changes in Itgb1 ^−/− corneas include thinning of the stroma and reduction in the number of epithelial cell layers at the beginning. Edema develops later in the cornea, with associated changes in the Descemet's membrane and the endothelial layer. Patches of thickened epithelium, scarring, and stromal haze can also be seen. The pathology exhibited by Itgb1 ^−/− corneas (summarized in Table 1), such as the thinning of stroma and the loss of epithelial layers, are similar to those in human corneas affected by the degenerative disorder keratoconus ^26,27 ; however, features such as edema and the degenerative changes to the endothelial layer are rarely seen in keratoconus. The cause of keratoconus is unknown; our results suggest, however, that keratoconus could involve defective keratocyte-ECM interactions. 

Table 1.

 

Itgb1 ^−/− Corneal Pathology

Days after Tamoxifen Administration	Corneal Pathology
21	None
42–45	Thinning of stroma, loss of epithelial layers
50–55	Patches of edema, further loss of epithelial layers, thickening of Descemet's membrane, degenerative changes in the endothelium
75	Epithelial cells reduced almost to a single layer, epithelial erosion, scarring, neovascularization, presence of neutrophils

 All Itgb1 ^−/− corneas examined had pathologic conditions as described at time points tested, though to varying degrees.

The corneal stroma continues to mature until nearly 30 days after birth. ^8,25 Intriguingly, when we deleted Itgb1 after complete corneal maturation (i.e., 40 days after birth), the cornea did not exhibit any obvious alterations in pathology and appeared to be normal even after 75 days after tamoxifen administration. These observations suggest that integrin β1 is required for matrix maturation of stroma in the developing cornea. In this regard, defects in the stromal matrix organization have been observed in keratoconus corneas. ^28,29 Our study emphasizes the importance of integrin-mediated adhesion and stromal maturation in maintaining corneal structural integrity in adults. 

The thinning of Itgb1 ^−/− corneas is associated with reduction in the space between collagen fibrils of the stroma; however, we did not observe any loss of proteoglycans (as seen by Alcian blue stain). Indeed, thinning of the keratoconus cornea is thought to be due to increased sliding of collagen; tissue mass area is the same as in normal corneas except in advanced stages of disease. ^28,30 Meek et al. ²⁸ have hypothesized that in keratoconus, the physiological events that “lock” corneal and limbal lamellae during childhood are not activated, leading to a gradual development of the disease. Consistent with this hypothesis is the absence of pathology in Itgb1 ^−/− corneas when Itgb1 is deleted after corneal/stromal maturation. Interlamellar displacement and slippage in keratoconus stroma is thought to occur because of the loss of cohesion between collagen fibrils and noncollagenous matrix (proteoglycans) through an as yet unidentified mechanism, resulting in corneal thinning and ectasia. ^24,28,29,31  

In the mouse cornea, keratocytes and endothelial cells originate from a single wave of migration of neural crest cells. ³² This fact might explain why the type I collagen promoter/enhancer we used for our study is expressed in a very small subset of endothelial cells (∼2–3 cells/5-μm cross-section of cornea). However, immunofluorescence analysis using an anti–integrin β1 antibody clearly indicated that integrin β1 expression was lost in the keratocytes, but not in the endothelial layer. Moreover, deleting Itgb1 after complete corneal maturation (40 days after birth) caused no pathologic change in the cornea even at 75 days after tamoxifen administration. Because edema was seen only in the corneas of mice in which Itgb1 was deleted at 23 to 26 days after birth, the major cause of the edema observed in integrin β1–deficient corneas was likely to have been the inability of the immature stroma to resist normal intraocular pressure. In this regard, it has been shown that localized thinning of the stroma, as in LASIK surgery, is known to result in interface fluid accumulation ^33,34 and reduction in lamellar tension–induced acute expansion of peripheral stromal volume. ³⁵ Our results are consistent with previous observations that connective tissue cells through the integrin β1 system can actively modulate the physical properties of interstitial matrix and, thus, interstitial fluid homeostasis. ³⁶  

Our study has uncovered a hitherto unappreciated role for integrin-mediated keratocyte adhesion in the maintenance of corneal homeostasis and structural integrity. It once again highlights the quiescent character of keratocytes in a normal cornea. The results suggest defects in keratocyte-ECM adhesion may underlie diseases such as keratoconus and could serve as a model for developing a therapeutic treatment. 

Supplementary Materials

Figure sf01, TIF - Figure sf01, TIF 

Figure sf02, TIF - Figure sf02, TIF 

Footnotes

 Supported by a grant from the Canadian Institutes of Health Research (AL) and by a fellowship from the Canadian Scleroderma Research Group (SKP).

Footnotes

 Disclosure: S.K. Parapuram, None; K. Huh, None; S. Liu, None; A. Leask, None

The authors thank the staff at the Electron Microscopy Laboratory (London Health Sciences Center) for technical help with transmission electron microscopy and Stephen Sims and Jeffrey Dixon for helpful critical comments on the manuscript.