Cytoskeletal and Cell Adhesion Defects in Wounded and Pax6 ^+/− Corneal Epithelia

Abstract

Purpose.: PAX6 heterozygosity (PAX6 ^+/−) causes aniridia and aniridia-related keratopathy (ARK) in humans, but the pathway from gene dosage deficiency to clinical disease has not been fully characterized. Recently, the authors suggested a model of a chronic wound state exacerbated by oxidative stress, showed the barrier function of Pax6 ^+/− corneas is compromised and suggested Pax6 ^+/− corneas show the molecular signature of a perpetual wound-healing state.

Methods.: Pax6 ^+/− mice were used as a model for Pax6-related corneal diseases and in vivo wound-healing assays. Immunohistochemistry and electron microscopy analyses were performed on mutant and wounded corneas.

Results.: This work reports defects in keratin, desmoplakin, and actin-based cytoskeletal structures in Pax6 ^+/− cells. During wild-type corneal reepithelialization, cell fissures and desquamation, intracellular vesicles, intercellular gaps, and filopodialike structures were apparent, similar to the phenotypes seen in “unwounded” Pax6 ^+/− corneal epithelia. Pax6 ^+/− cells and wounded wild-type cells showed changed patterns of desmoplakin and actin localization. Protein oxidation and ERK1/2 and p38 MAPK phosphorylation were barely detected in the basal cells of intact wild-type corneal epithelia, but they were found in basal wild-type cells near the wound edge and throughout Pax6 ^+/− corneal epithelia.

Conclusions.: These data show that cell junctions and cytoskeleton organization are dynamically remodeled in vivo by wounding and in Pax6 ^+/− corneas. This apparent wound-healing phenotype contributes to the clinical aspects of ARK.

The human cornea becomes opaque and vascularized in aniridia-related keratopathy (aniridic keratopathy [ARK]) resulting from heterozygous mutations in the gene encoding the transcription factor PAX6. ^1,2 PAX6 expression was also found to be reduced or absent in diseases such as Stevens-Johnson syndrome, chemical burn, and recurrent pterygium, ³ which, like ARK, are characterized by corneal opacity and thin, fragile, disrupted corneal epithelia. ^4,5 PAX6 functions in a dosage-sensitive manner such that both deficiency ^2,6,7 and overexpression ^8,9 affect development of the eye. The mouse Pax6 protein sequence is identical to that of humans, ¹ and Pax6 ^+/− mice have phenotypes similar ^10–13 to those of aniridic patients. ^14–16 Although aniridia in humans is treated clinically as a stem cell deficiency, in mice only a subtle deficit was found in the ability of Pax6 ^+/− cells to occupy the limbal niche and to produce corneal epithelium. ¹⁷ In addition, higher proliferative and normal apoptotic rates were observed in Pax6 ^+/− corneal epithelia, ^13,18 suggesting that the cornea should be able to maintain itself. Corneal epithelial abnormalities in glycoprotein trafficking, ¹⁹ wound healing, ^20–22 cell adhesion, ¹⁸ and migration ¹⁷ were reported in Pax6 ^+/− mice. Pax6 function may be directly involved in regulating corneal epithelial repair ^23,24 and cell migration, ^25,26 but the link between Pax6 dosage and clinical corneal disease has not been fully explained. 

The Pax6 ^+/− epithelium is fragile, and we recently found microlesions in Pax6 ^+/− mouse corneal epithelia and showed that oxidative stress exacerbates the dosage effect by hindering Pax6 nuclear transport and wound healing. ²⁷ Chronically increased levels of intracellular calcium and activated MAPK pathways in Pax6 ^+/− corneas suggested that they recapitulated a semipermanent wound-healing state, even when not wounded. Reduced levels of keratin (K) 12 and K5 were found in Pax6 ^+/− adult mouse corneal proteome. ²⁷ Diminished K12 expression in Pax6 ^+/− mice occurred during postnatal corneal development. ¹³ Keratins constitute the intermediate filaments of the epithelial cytoskeleton, ^28–30 and keratin mutations are associated with skin fragility and disorders. ^31–33 To test our hypothesis that the fragility, lesions, and oxidative stress observed in the Pax6 ^+/− corneal epithelia lead to a chronic wounded state in vivo, we compared the morphologic, ultrastructural, cytoskeletal, and molecular signatures of unwounded wild-type corneal epithelia with both Pax6 ^+/− corneal epithelia and wounded wild-type corneas. 

Methods

Mice

The Pax6^Sey-Neu phenotypic null was maintained on a CBA/Ca background. Pax6 ^+/Sey-Neu mice (Pax6 ^+/−) were mated to wild-type mice. Wild-type and Pax6 ^+/− littermates were compared. All experiments were performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

In Vivo Corneal Epithelial Wound Assay

Wild-type mice, 8 to 30 weeks old, were anesthetized under veterinary advice. Central circular (1.5-mm diameter) corneal epithelial wounds were made, and the epithelial sheet within the wound was removed. Anesthesia was immediately reversed with atipamezole hydrochloride (0.014 mg/10 g; Antisedan; Pfizer Animal Health, Exton, PA) administered subcutaneously to facilitate normal blinking and tear production. Twenty-one hours after wounding, eyes were enucleated and fixed in 4% paraformaldehyde (PFA) in PBS for 4 hours at 4°C. 

Reagents

Reagents were obtained as follows: rhodamine-labeled phalloidin (R415) from Invitrogen (Paisley, UK); anti–cytokeratin-12 (sc-17,101) from Santa Cruz Biotechnology (Santa Cruz, CA); anti–type II basic keratins (Ab6401, MS-342-P1) from Neomarkers and Abcam (Cambridge, UK); desmoplakin (AHP320) from Serotec (Oxford, UK); FITC-labeled α-tubulin (F2168) from Sigma (Poole, UK); p-ERK1/2 (9106) and p-p38 MAPK (9211) from Cell Signaling Technology (Beverly, MA). A protein oxidation kit (S7150, OxyBlot kit) was from Millipore (Billerica, MA). 

Cell Cultures

Each enucleated mouse eye was digested in 0.1 mL SHEM ³⁴ containing 15 mg/mL dispase II and 100 mM sorbitol at 4°C for 18 hours or at room temperature for 8 hours. The corneal epithelial sheet was peeled off and digested in 0.25% trypsin/1 mM EDTA, assisted by gentle pipetting. Dissociated cells were plated onto plastic coverslips (Thermanox; Thermo Fischer Scientific, Rochester, NY) and were cultured in SHEM at 37°C, 5% CO₂. 

Staining Protocol on Mouse Flat-Mount Whole Corneas

Corneas were dissected from PFA-fixed eyes and rinsed with Tris-buffered saline, 0.1% Tween-20, pH 8.5 (TBS/T). Corneal epithelia were permeated by rocking incubation for 24 to 48 hours at 4°C with TBS/T supplemented with 0.4% Triton X-100 and 0.4% NP-40; if F-actin was not to be stained, 5% methanol was included to facilitate the permeation of corneal epithelia. For protein oxidation assay, corneas were incubated with 1 × DNPH solution on a roller for 15 minutes at room temperature; the DNPH-carbonyl reaction was terminated by the addition of neutralization solution according to the manufacturer's instruction (S7150, OxyBlot kit; Millipore). Corneas were then washed 3 × 5 minutes in TBS/T, transferred to 0.2 to 0.5 mL primary antibody solution (TBS/T, 2% BSA), rotated for 24 hours at 4°C, and rinsed 3 × 5 minutes in TBS/T. Except for the one-step staining on F-actin and α-tubulin, other samples were incubated with 0.2 to 0.5 mL secondary antibodies diluted 1:800 to 1:1000 in TBS/T and 2% BSA overnight at 4°C. Corneas were washed 3 × 5 minutes in TBS/T and flat-mounted with a glass coverslip. Confocal images were taken under a laser scanning microscope (LSM 510; Carl Zeiss, Jena, Germany). 

Scanning Electron Microscopy

PFA-fixed eyes were dehydrated in graded ethanol and then subjected to critical point dehydration with CO₂ (030 Critical Point Drier; Bal-Tec, Bannockburn, IL). Samples were sputter-coated with gold (K550; Emitech K550, Ashford, UK) and viewed under a scanning electron microscope (JEOL35CF; Olympus, Tokyo, Japan) at 10 kV. 

Transmission Electron Microscopy

Freshly enucleated eyes were fixed in 1 to 5 mL 2.5% glutaraldehyde in PBS at 4°C for 2 days. A small slice of cornea was processed with a tissue processor (EMTP; Leica, Brisbane, Australia) to infiltrate in Spurr epoxy resin. Ultrathin 80-nm sections were prepared on an ultramicrotome (UC6; Leica). The sections were collected onto 200-mesh fine bar copper grids and stained using uranyl acetate and lead citrate. Sections were examined with a transmission electron microscope (CM10; Philips, Eindhoven, The Netherlands) and imaged with a charge-coupled device camera (Bioscan; Gatan, Pleasanton, CA). 

Fractionation of Protein Samples

Proteins from four wild-type and four Pax6 ^+/− corneas were separated into cytoskeletal and cytosolic fractions with a cell compartment kit (Qproteome; Qiagen, Valencia, CA) according to manufacturer's instructions. After standard polyacrylamide gel electrophoresis using 5 μg protein in each lane, samples were subjected to Western blot analysis and probed for α-tubulin (mouse monoclonal; Sigma, Poole, UK; 1:1000) and β-actin (mouse; Sigma; 1:1000) or cytokeratin-12 (goat polyclonal; Santa Cruz Biotechnology, Santa Cruz, CA; sc-17101, 1:1000) with secondary detection using donkey anti-mouse secondary antibody 680 nm (red) or donkey anti-goat 488 (green; Li-Cor Biosciences, Lincoln, NE). 

Results and Discussion

Scanning electron microscopy showed that the epithelial surface of a wild-type cornea was smooth and intact (n = 9; Fig. 1A). In contrast, all Pax6 ^+/− corneas examined (n = 10) had lesions or clusters of cells desquamating in the peripheral region, and 4 of 10 samples had cracks between cells in the central cornea (Fig. 1B). Twenty-one hours after reepithelialization in vivo, desquamation of surface cells near the wound edge was found in all wild-type cornea samples (n = 9; Fig. 1C), and small holes or fissures were observed in areas close to the epithelial migration edge in 4 of 9 samples (Fig. 1D). Increased epithelial desquamation was previously observed in rabbit corneas in the first 5.5 ± 3 hours of injury, during which time no overt epithelial migration was found. ³⁵  

Figure 1.

 

Scanning electron micrographs of the corneal surface. (A) Intact epithelial surface in adult wild-type mouse corneas. (B) Epithelial lesions in unwounded Pax6 ^+/− cornea. (C, D) Cell fissures and desquamation at or close to the wound edge in wild-type corneas. The wound edge is marked by a broken line (C). Scale bars: 100 μm (B); 10 μm (A, C, D).

Ultrastructural Analysis of Corneal Epithelia

Transmission electron microscopy was performed to investigate the ultrastructure in adult wild-type and Pax6 ^+/− corneal epithelia (n = 3 eyes each) and wild-type corneal epithelia 21 hours after wounding in vivo (n = 5 eyes). 

Wild-type mouse corneal epithelia were stratified as described previously, with basal cuboidal cells and five to seven layers of wing and squamous suprabasal cells; cells were tightly stacked, and the cellular contents were dense (Figs. 2A, 2B). Wild-type epithelial cells demonstrated smooth cell-cell borders with interdigitated filopodialike protrusions. Cell-cell borders were tightly sealed by alternating adherens junctions and desmosomes similar to the cell-cell adhesions seen in skin epidermal cells. ³⁶ Pax6 ^+/− adult mouse corneal epithelia had fewer cell layers (Fig. 2C). Intercellular gaps were found at the Pax6 ^+/− cell borders, as reported previously, ¹⁸ associated with large interdigitated filopodialike processes in the gaps (Figs. 2C, 2D). Large vesicles were obvious in Pax6 ^+/− cells. The extent of cell-cell apposition was variable in Pax6 ^+/− corneal epithelia (Fig. 2D): some cells had tightly sealed cell borders as in wild-type, but where intercellular gaps occurred, the surrounding plasma membranes were intact, and the intracellular adherens junctions were at least partly preserved judging by the electron-dense alignments at the membrane. The observed intracellular gaps were not artifacts of fixation because they were also observed in unfixed corneas processed by high-pressure freezing (Supplementary Fig. S1). Strikingly, 21 hours after wounding, intercellular gaps associated with prominent interdigitated filopodialike processes also occurred in wild-type corneal epithelia (Figs. 2E, 2F). Adherens junctions surrounding the gaps were preserved, and filopodialike processes protruded and maintained cell-cell adhesion. Large vesicles were also visible in the cytoplasm (Fig. 2F). Hence, Pax6 ^+/− corneal epithelia and wounded wild-type epithelia shared a phenotype of intercellular gaps, consistent with a partial loss of cell adhesion resulting from injury or cell migration. 

Figure 2.

 

Ultrastructural similarities between Pax6 ^+/− and healing wild-type corneal epithelia. (A) Corneal epithelium in wild-type adult mice comprised a basal layer of cuboidal cells (b) and five to seven layers of increasingly squamous cells (s). (B) 25,000 × magnification showing smooth cell boundaries sealed by alternating arrays of adherens junctions (aj) and desmosomes (ds) in wild-type corneal epithelia. (C) Pax6 ^+/− adult mouse corneal epithelia were also stratified, but the shapes of the basal cells (b) and two to four layers of squamous cells (s) were abnormal, with numerous gaps and filopodialike protrusions at the intercellular space. (D) 25,000 × magnification showing that Pax6 ^+/− cells had adherens junctions (aj) and desmosomes (ds) similar to those of wild-type cells. Prominent filopodialike processes appeared to interdigitate and retain cell-cell contact. (E) After healing for 21 hours in vivo, gaps and cellular protrusions developed in wild-type corneal epithelia. (F) 34,000 × magnification. Gaps associated with separated but apparently intact plasma membrane and adherens junctions (aj), filopodialike processes, and cytoplasmic vesicles in the healing wild-type corneal epithelia all resemble Pax6 ^+/− samples. Scale bars: 2 μm (A, C, E); 0.5 μm (B, D, F).

Cytoskeletal Defects in Pax6 ^+/− Corneal Epithelia

Actin-based microfilaments, keratin-based intermediate filaments, and tubulin-based microtubule structures were investigated by flat-mount staining of freshly fixed corneas. These cytoskeletal elements function to coordinate cell adhesion, migration, and tissue morphogenesis. ^37–39 Little is known about the architecture of the corneal epithelial cytoskeleton in vivo because the composition and organization of the cytoskeleton change during differentiation, disease, and in vitro culture. ^29,40–42  

Corneal epithelial cells were cultured from wild-type and Pax6 ^+/− adult mice according to a widely adopted protocol. ³⁴ Rhodamine-phalloidin staining revealed similar actin architecture in wild-type (n = 12 cultures; Fig. 3A) and Pax6 ^+/− cells (n = 12 cultures; Fig. 3B) in vitro. Actin proteins formed into bundles of “stress fibers” across and around the cell body. 

Figure 3.

 

Actin-based cytoskeleton in corneal epithelial cells. (A) Corneal epithelial cells of wild-type and (B) Pax6 ^+/− mice in vitro stained with rhodamine-phalloidin. (C) In vivo, F-actin was enriched at the borders of wild-type basal cells and did not form overt structures in the cytoplasm. (D) F-actin also aligned to the junctions of Pax6 ^+/− basal cells but was more diffuse and extended throughout the cytoplasm (arrowheads). (E) After healing in vivo for 21 hours, wild-type basal cells developed F-actin structures (arrowheads) similar to those of Pax6 ^+/− cells. (F) F-actin formed a meshwork in the cell body and aligned to the cell borders in wild-type superficial cells. (G) Lack of F-actin meshwork, numerous vesicles in the cell body, and branched alignments at the cell borders (arrowheads) in Pax6 ^+/− squamous cells. (H) After healing in vivo for 21 hours, some wild-type cells at the migrating edge form filopodia and lamellipodia (open arrowhead), and some cells form stress fibers. (I) F-actin stress fibers and branched alignments (arrowheads) at the junctions of some squamous cells of a healing wild-type corneal epithelium. Scale bars: 10 μm (A, B); 5 μm (C–I).

In vivo, actin architectures were distinct from those in culture, and basal cells showed different patterns of staining from apical cells (n = 12 wild-type eyes and n = 8 Pax6 ^+/− corneas; Figs. 3C–I). F-actin was concentrated at the borders of wild-type basal cells and did not form networks or stress fibers in the cytoplasm (Fig. 3C). The F-actin scaffold also aligned to the borders of basal Pax6 ^+/− cells but often appeared to split or expand to the cytoplasm (Fig. 3D, arrowheads). Separated cell borders and vesiclelike cytoplasmic structures were also revealed in basal wild-type cells 21 hours after wounding by actin labeling (n = 14 eyes; Fig. 3E). In wild-type squamous cells, actin assembled into short, winding fibers that joined into a fine, dense meshwork in the cell body and aligned to the cell-cell borders (Fig. 3F). F-actin also aligned to the Pax6 ^+/− cell-cell borders, but the fibers often split into a chainlike form (Fig. 3G, arrowheads). Notably, the F-actin meshwork was diminished or replaced by vesicles in the Pax6 ^+/− cell bodies (Fig. 3G). Lamellipodia and filopodia were visible in some corneal epithelial cells at the leading edge of wounded wild-type epithelia (Fig. 3H, open arrowhead). Stress fibers were seen in some cells (Figs. 3H, 3I), but most of the cells retained the actin meshwork structure (Fig. 3H). Intercellular gaps marked by split F-actin signals at the borders of reepithelializing superficial cells were also noticed (Fig. 3I, arrowheads), but they were less prominent than in Pax6 ^+/− corneas, whose superficial layer actin structure was generally more disrupted than in wounded wild-type epithelia. F-actin binds on adherens junctions ^36,43,44 ; therefore, the split chain form may confirm the presence of intercellular gaps found in Pax6 ^+/− and regenerating wild-type corneal epithelia under transmission electron microscopy (Figs. 2C–F). Previous analysis of actin localization in Pax6 ^+/− corneal epithelia ¹⁸ did not report the defects observed in this study, likely because of the differences in protocol-tissue sections previously and flat-mount staining. 

Keratin and Desmoplakin Defects in Pax6 ^+/− Corneal Epithelia

Keratins constitute the major intermediate filament in the corneal epithelium. ^29,45 K12 is a corneal epithelium-specific type 1 acidic keratin, apparently directly regulated by Pax6, ^13,46 and K12 knockout mice have fragile corneal epithelia. ⁴⁷ Desmoplakin is a desmosomal protein directly binding to epidermal type II keratins, ⁴⁸ and it functions in desmosome assembly, maintenance of cytoskeletal architecture, and reinforcement of membrane attachments. ³⁶ We studied corneal epithelial keratin architecture using antibodies against K12, type II basic keratins, and desmoplakin. 

In wild-type corneal epithelial cultures (n = 6), type II basic keratin filaments assembled into a dense network in the cell body (Fig. 4A), K12 constituted a meshwork structure (Fig. 4B), and strong desmoplakin signals were seen both at the cell-cell borders and in the cytoplasm (Fig. 4C). 

Figure 4.

 

Keratin and desmoplakin structures in cultured cells and basal corneal epithelial cells. In primary wild-type corneal epithelial cells cultured in SHEM, ³⁴ type II basic keratins (Type II K) formed a dense filamentous network (A), keratin 12 (K12) formed a nonfilamentous, meshworklike structure (B), and desmoplakin (Dpk) localized to the cell borders and the cell body (C) (n = 6 cultures). In vivo, staining of type II keratins (D) and K12 (E) in the basal cells of unwounded wild-type corneal epithelia did not reveal any overt structure, but (F) desmoplakin was concentrated at the cell borders (n = 15 eyes). (G) Pax6 ^+/− basal cells were positive for type II keratins, but (H) K12 proteins were barely detected, and (I) desmoplakin did not aggregate at the cell junctions (n = 16 eyes). After healing for 21 hours in vivo, type II keratins (J) and K12 (K) patterns did not change overtly in wild-type basal cells, but desmoplakin no longer concentrated at the cell junctions (L; n = 8 eyes). Scale bars, 5 μm.

In vivo, wild-type cells in the basal epithelial layer (n = 15 eyes) were positive for type II keratins (Fig. 4D) and K12 (Fig. 4E), but no structure could be distinguished. Desmoplakin aligned to the cell junctions, with weaker signals in the cytoplasm (Fig. 4F). Type II keratin signals in Pax6 ^+/− basal cells (n = 16 eyes) were similar to those of wild-type cells (Fig. 4G). K12 staining was heterogeneous between samples, as described previously ¹¹ but in the most extreme cases were undetectable (Fig. 4H). Desmoplakin was distributed in the mutant cytoplasm but not at the cell junctions (Fig. 4I). In wild-type mice 21 hours after wounding (n = 8 eyes), K12 staining was, in contrast to Pax6 ^+/−, similar to that in unwounded wild-type mice, but, as in Pax6 ^+/− epithelia, desmoplakin was no longer aggregated at the cell junctions (Figs. 4J–L). 

Superficial cells also showed a similar pattern of partial disruption of desmoplakin localization in Pax6 ^+/− corneas and wild-type corneas 21 hours after wounding (Supplementary Fig. S2). Some cell borders were strongly labeled, but other borders had apparently no desmoplakin associated with enlarged intercellular gaps (Supplementary Figs. S2B, S2C). 

It is likely that type II keratins normally interact with desmoplakin to integrate the K12 framework with desmosomes and regulate cell adhesion and epithelial morphogenesis by mechanisms similar to those in the skin. ^36,37,48 In a separate study, junctional proteins desmoglein, β-catenin, and γ-catenin were found at lower levels in Pax6 ^+/− corneas. ¹⁸ Desmoplakin, desmoglein, and γ-catenin (or plakoglobin) are desmosomal proteins; γ-catenin binds to desmoglein and desmoplakin, whereas desmoplakin and γ-catenin can bind to type II epidermal keratins. ⁴⁹ Desmosome assembly has been suggested to be a passive adhesion process after the interdigitated embedding of filopodia extended from neighboring cells. ⁵⁰ Therefore, failure of desmoplakin alignment to the cell junctions in Pax6 ^+/− cells and WT cells during wound healing may result passively from loss of cell-cell contact during epithelial migration. Similar to F-actin, we found that keratin architecture in vivo is profoundly different from that in primary cultures. 

α-Tubulin Dynamics in Mouse Corneal Epithelia

Staining of α-tubulin revealed mitotic bundles in wild-type (n = 12 eyes; Fig. 5A) and Pax6 ^+/− basal cells (n = 8 eyes; Fig. 5B). α-Tubulin aggregated cytoplasmically and at cell borders in wild-type basal cells (Fig. 5A) but was enriched in the cytoplasm, and away from the cell borders, of Pax6 ^+/− basal cells (Fig. 5B) and wild-type cells near a wound edge (n = 14 eyes; Fig. 5C). Some microtubules with no obvious orientation were seen in wild-type and Pax6 ^+/− squamous cells (Figs. 5D, 5E). Superficial wild-type cells dynamically assembled a dense microtubule network at the wound edge (Fig. 5F) that contrasted to unwounded wild-type and Pax6 ^+/− epithelia. 

Figure 5.

 

Organization of α-tubulin microtubules in corneal epithelia. (A) Staining of α-tubulin revealed mitotic spindles and α-tubulin protein enrichment near the cell borders in basal wild-type cells (n = 12 eyes); low levels of α-tubulin proteins and no overt structure were found in the cytoplasm. (B) Pax6 ^+/− basal cells (n = 8 eyes) had apparently normal mitotic spindles, but the cytoplasmic content of α-tubulin structures appeared to increase, (C) which was also observed in the basal wild-type cells after healing in vivo for 21 hours (n = 14 eyes). Superficial cells of (D) wild-type and (E) Pax6 ^+/− mouse corneal epithelia had microtubules showing no obvious patterns or organization. (F) After healing in vivo for 21 hours, the F-actin meshwork was apparently unchanged and stress fibers occasionally organized in wild-type cells. (G) In contrast microtubules appeared to undergo a dynamic change, assembling into microtubule networks to different extents in different cells. Arrowheads: cells that had apparently lower contents of microtubules. The wound edge is marked by a broken line (C, F, G). Scale bars, 5 μm.

The mild microtubule abnormality observed in Pax6 ^+/− may be related to desmoplakin function. Recently, desmoplakin was also found to regulate microtubule organization in epidermal cells. ⁵¹  

Protein samples from wild-type and Pax6 ^+/− corneas were separated into cytosolic, and cytoskeletal fractions and probed for α-tubulin, β-actin, and cytokeratin-12 by Western blot analysis. K12 was distributed evenly between soluble and insoluble fractions, but the cytosolic pool of α-tubulin was undetectable, and that of β-actin was minor. There was no evidence of increased solubility of cytoskeletal proteins in Pax6 ^+/− (Supplementary Fig. S3). 

Oxidative Stress and MAPK Signaling during Corneal Reepithelialization

Basal cells in normal mouse corneal epithelia are less differentiated ⁴⁶ and more proliferative than superficial cells. ¹³ Barrier function, mediated by tight junctions of the suprabasal cells, may protect basal cell functions from noxious stimuli. For example, wild-type superficial corneal epithelial cells exhibit levels of oxidative stress in which cellular proteins are oxidized and ERK1/2 and p38 MAPK are activated. ²⁷ However, basal epithelial cells in unwounded wild-type mouse corneas have no detectable amounts of active ERK1/2 (p-ERK), p38 MAPK (p-p38; Fig. 6A), or oxidized proteins (Fig. 6B; n = 23 eyes), suggesting the basal cell functions are little affected by oxidative stress. In contrast, the barrier function of Pax6 ^+/− corneal epithelia is compromised, ²⁷ and it was found that this was associated with high levels of p-ERK, p-p38 MAPK, and protein oxidation in the basal cells (Figs. 6C–E; n = 23 eyes). After healing in vivo for 21 hours, wild-type basal cells near the wound edge became positive for p-ERK, p-p38 MAPK, and protein oxidation (Figs. 6F–H; n = 3 eyes), suggesting that, similar to Pax6 ^+/− basal cells, wild-type basal cells are susceptible to oxidative stress once the epithelia are wounded. ⁵²  

Figure 6.

 

Oxidative stress, ERK1/2, and p38 MAPK signaling in the basal cells of corneal epithelia. In unwounded wild-type corneal epithelia (n = 23 eyes), (A) phosphorylated ERK1/2 (p-ERK) and p38 MAPK (p-p38) and (B) protein oxidation were not detectable in basal wild-type cells. In contrast, Pax6 ^+/− basal cells were positive for (C) p-ERK, (D) p-p38, and (E) protein oxidation activities. After healing in vivo for 21 hours (n = 3 eyes), in some regions near the wound edge, (F) p-ERK, (G) p-p38, and (H) oxidized proteins were detected in basal wild-type cells. (I) Negative control with secondary antibodies shows only unspecific (red) antibody binding in the stroma but not in the epithelial cells. The position of the wound edge is marked by a broken line (F, G, I) or is indicated by an open arrow (H). Scale bars, 5 μm.

These findings strongly support the hypothesis that oxidative stress may modulate cellular functions both in unwounded Pax6 ^+/− corneas and in wild-type corneas during the healing process. In our previous report, ²⁷ we described the compromised barrier function and higher levels of oxidative stress and stress-mediated calcium, ERK1/2, and p38 MAPK signaling in Pax6 ^+/− mouse corneal epithelia. In addition, oxidative stress slowed down epithelial wound healing in cultured whole eyes. The higher calcium levels in Pax6 ^+/− cells may mediate the formation of filopodialike protrusions. ⁵³ Transient activation of MAPK may protect the cells from oxidative stress, ^54,55 but prolonged activation of MAPK may lead to cell necrosis. ⁵⁶ Cell death and premature desquamation may explain why Pax6 ^+/− corneal epithelia experience higher proliferation and normal apoptosis but have fewer cells. ^13,18,22  

Conclusions

In this study, we developed a new whole-mount corneal immunostaining procedure and report for the first time the in vivo organization of actin, keratin, and α-tubulin architecture in mouse corneal epithelia. Specific defects of actin, K12, and desmoplakin localization in Pax6 ^+/− corneas are consistent with failure of the normal formation of cell junctions and their interaction with the cytoskeleton. These defects may underlie the intercellular gaps observed in Pax6 ^+/− corneal epithelia and, hence, the oxidative stress in basal layers. The cellular changes observed in Pax6 ^+/− corneas are similar at ultrastructural, immunohistochemical, and biochemical levels to those observed in wild-type cells during wound healing. The direct mechanistic link between increased oxidative stress and cytoskeletal changes in Pax6 ^+/− corneas, however, has not yet been demonstrated. Wound-healing and stress conditions may be important to treatments of ARK and other ocular surface disorders characterized by reduced Pax6 function, ³ particularly after the transplantation of limbal or corneal tissue or cultured epithelial cells, which have been the major methods to reepithelialize the corneas of patients. ^5,57–60  

Supplementary Materials

Supplementary Figure S1 - (PDF) 

Supplementary Figure S2 - (PDF) 

Supplementary Figure S3 - (PDF) 

Footnotes

 Supported by Biotechnology and Biological Sciences Research Council Grant BB/E 015840/1.

Footnotes

 Disclosure: J. Ou, None; C. Lowes, None; J.M. Collinson, None