P2X₇ knockout mice (P2X₇ ^−/−), strain B6.129P2-P2rx7^tm1Gab/J, and their wild-type (WT) counterparts, strain C57BL/6J, were obtained from Jackson Laboratories (Bar Harbor, ME) 5 weeks postnatally and were acclimated for 16 days. The antibody directed against P2X₇ was purchased from Lifespan Biosciences (Seattle, WA), whereas the antibody directed against P2Y₄ was from Chemicon (Temecula, CA). Antibodies to paxillin (pY¹¹⁸) and FAK (pY³⁹⁷) were from Biosource (Camarillo, CA), and the antibody against pannexin-1 was from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-fade, nuclear dye (To-Pro 3AM), Alexa Fluor-conjugated secondary antibodies, reagent (TRIzol), DNaseI, random hexamers, MMLV reverse transcriptase, and RNaseH were purchased from Invitrogen (Carlsbad, CA). Cuprolinic blue and other supplies for electron microscopy were from Electron Microscopy Sciences (Hatfield, PA). RNase inhibitor and gene expression assays (TaqMan) were from Applied Biosystems (Foster City, CA). Routine chemicals were obtained from Sigma (St. Louis, MO) or American Bioanalytical (Natick, MA). 

All experiments were carried out in accordance with the guidelines and the approval of The George Washington University Institutional Animal Use and Care Committee and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Manual debridement wounds were performed on corneas of 7-week-old mice, as described by Stepp and Zhu. ³² Briefly, mice were anesthetized with ketamine (5 mg/mL; Aveco Co., Inc., Fort Dodge, IA) and xylazine (20 mg/mL; Miles Inc., Shawnee Mission, KS). A topical anesthetic (proparacaine ophthalmic solution) was applied to the ocular surfaces until the blink sensation was absent. Corneas were scraped with a dull scalpel to remove the epithelium within a 1.5-mm central corneal area, which had been demarcated with a dull trephine. After wounding, eyes were treated with erythromycin ophthalmic ointment to minimize inflammation and to keep the ocular surface moist. Animals were humanely killed 16 hours after wounding. Corneas were immediately stained with Richardson vital stain and were photographed through a dissecting microscope. For morphologic analysis, eyes were enucleated and fixed in a 50:50 mixture of methanol and dimethyl sulfoxide for 2 hours at −20°C and were then stored in 100% methanol at −20°C. Images of 12 WT and 12 P2X₇ ^−/− eyes were analyzed using NIH ImageJ software. Each wound area was measured independently by three investigators (n = 12 eyes in each group). 

Enucleated, wounded eyes from each group were embedded in paraffin. Six-micrometer serial sections were cut and mounted on glass slides. Mounted sections were deparaffinized by two sequential 10-minute washes in xylene, followed by washes in 100%, 95%, and 70% ethanol. Sections were stained with Masson trichrome to examine collagen and stromal thickness. In addition, unwounded eyes from each group were embedded in epon/araldite, sectioned, and stained with toluidine blue. 

Immunohistochemistry was performed in a humidified chamber. Sections from three eyes in each group containing leading edges of the wound were blocked with 3% BSA in PBS, followed by incubation with appropriately diluted primary antibody in 1% BSA in PBS for 3 hours. Control sections were incubated with 1% BSA in PBS in place of primary antibody. Sections were washed and incubated with Alexa Fluor 488-conjugated secondary antibody for 1 hour Slides were washed three times in PBS and incubated for 15 minutes in nuclear dye (To-Pro 3AM; Invitrogen) diluted 1:1000 in PBS. Slides were washed, covered with anti-fade, and coverslipped. Sections were imaged and analyzed using a confocal microscope (LSM 510 200M; Zeiss, Thornwood, NY), as previously described. ⁷  

For routine electron microscopy, four unwounded corneas were processed for transmission electron microscopy (TEM), as described in Trinkaus-Randall and Gipson. ³³ Images were taken under an electron microscope (300 TM; Philips; Eindhoven, Netherlands). 

An electron-dense reagent, cuprolinic blue, which defines the sizes and directional orientations of proteoglycan (PG) chains, was used. In other studies, it has been used to demonstrate the association between sulfated PGs and matrix molecules such as collagen. ³⁴ Two eyes in each group were stained with 0.2% cuprolinic blue (in buffer containing 0.3 M MgCl₂) for 24 hours ³⁵ and were dehydrated and processed for TEM, as described previously. 

Corneas were removed from unwounded eyes, and epithelium was scraped off. Corneal stromas were homogenized in reagent (TRIzol; Invitrogen), and RNA was extracted according to manufacturer’s guidelines. Genomic DNA was removed by incubation with DNaseI in the presence of 1 U/μL RNase inhibitor. RNA was annealed with random hexamer primers, and first-strand cDNA synthesis was carried out with M-MLV reverse transcriptase. Negative controls were performed without the reverse transcriptase. The resultant cDNA was treated with RNaseH. Real-time PCR was performed (ABI 7300; Applied Biosystems). Gene expression assays (TaqMan; Invitrogen) used were Mm00489842_m1 for alpha3(v), Rn00801649_g1 for alpha(1)I, and eukaryotic 18S rRNA endogenous control. Cycling parameters were as follows: 50°C for 2 minutes, 95°C for 10 minutes, 45 cycles of 95°C for 15 seconds, and 60°C for 1 minute Results are presented as relative expression normalized to 18s rRNA and were calculated using the ΔΔCt method. 

For epithelial wounds, three independent investigators scored the wounds, the results were averaged and expressed as ± SEM, and Student’s t-test was performed. For stromal collagen, measurements of collagen fibril size and interfibrillar distance and size and number of proteoglycans were made using the Universal Desktop Ruler application. Results were averaged and expressed as ± SEM, and Student’s t-test was performed. 

Corneas from P2X₇ ^−/− and WT mice were examined, and the epithelium and stroma were compared. Mature corneas appeared grossly normal and had no apparent opacity. There were no differences in weight between the two groups and no detectable abnormalities in mobility or skin. This is the first report demonstrating a role for P2X₇ in corneal wound healing and stromal ultrastructure. 

Epithelial debridement wounds measuring 1.5 mm in diameter were made, and wound closure was evaluated after 16 hours in WT and P2X₇ ^−/− animals. Visual evaluation and photography were performed after topical administration of Richardson stain. Wounds were measured by three independent investigators. There was a trend toward delayed wound closure for the P2X₇ ^−/− mice (Table 1) . 

Cross-sections of the epithelial debridements of WT and P2X₇ ^−/− corneas were stained with Masson trichrome 16 hours after injury. Anterior sections of the cornea demonstrated that the region adjacent to the wound edge in the P2X₇ ^−/− was fragile, and the epithelium was routinely separated from the anterior stroma rather than along the basal lamina (Fig. 1A , arrow). Separation may be attributed to loose collagen fibrils in unwounded corneas. In addition, the stroma of the P2X₇ ^−/− was thicker than that of the WT, possibly because of the spaces in the mid and posterior cornea (Fig. 1B) . 

Immunohistochemical staining was performed, and the localization of proteins was visualized using confocal microscopy. Tissue sections were costained with nuclear dye (nuclear dye [To-Pro 3AM; Invitrogen]; see Materials and Methods). The absence of the receptor was verified in the P2X₇ ^−/− mice corneas using an antibody directed against the P2X₇ receptor. As predicted, P2X₇ was present in the cornea of the WT mice (Fig. 2) . To determine whether P2Y₄ receptors, shown to have a role in in vitro cell migration, were modulated by the P2X₇ receptor, we examined its localization in P2X₇ ^−/− and WT. The P2Y₄ receptor was detected along the apical epithelium in the WT, but in the P2X₇ ^−/−, P2Y₄ appeared diffuse throughout the apical and basal cells. Pannexin-1 was examined because it has been shown to interact with the P2X₇ receptor and form nonspecific pores. ^{20 21} In WT and P2X₇ ^−/−, pannexin-1 was localized to punctate regions between cells distal to the leading edge (defined as 105 μm back from the leading edge (Fig. 2 , inset). However, it was only detected at the leading edge in the WT. Use of appropriate rabbit or goat secondary antibodies alone showed negligible staining (Fig. 2) . 

Because wound healing was compromised in the P2X₇ ^−/− animals, we examined proteins (paxillin and FAK) known to be associated with cell migration. The localization of phosphorylated (p) paxillin differed between WT and P2X₇ ^−/−. Phosphorylated paxillin was diffuse throughout the epithelium in WT and P2X₇ ^−/− animals. However, intense staining was found distal to the leading edge along the epithelial stromal interface in the WT (Fig. 2 , inset). In addition, p-paxillin was more prominent in WT than P2X₇ ^−/− when quantified with the Zeiss image analysis program (data not shown). There were no detectable differences in phosphorylated focal adhesion kinase (p-FAK) between P2X₇ ^−/− and WT epithelium. This was consistent at the leading edge and distal from the wound. Sections stained only with secondary antibody were negligible (Fig. 2) . 

The epithelial basal lamina region was examined at the electron microscopic level, and hemidesmosomal structures and anterior collagen were examined. Filaments projected through the lamina lucida, and electron-dense regions of intermediate filaments splayed into the cytoplasm in WT and P2X₇ ^−/− corneas (Fig. 3) . Although anchoring fibrils are not prominent in mice, they were detected in WT and P2X₇ ^−/−. No morphologic change was observed in either the basal lamina or Descemet membrane of the P2X₇ ^−/− animals (data not shown). 

The ultrastructure of the unwounded, central stroma was examined and divided into anterior, middle, and posterior regions. Orthogonal and perpendicular lamellae in the P2X₇ ^−/− corneal stromas were thicker and fewer than in the WT. A typical P2X₇ ^−/− stroma revealed orthogonal and perpendicular lamellae with an average thickness of 1259 ± 100 nm, whereas WT averaged 770 ± 56 nm (Fig. 4) . Collagen in the P2X₇ ^−/− stromas lacked the orthogonal organization typical of the WT. For example, in the P2X₇ ^−/−, the collagen fibrils within an array were not aligned in a parallel pattern, and there was a paucity of collagen fibrils in cross-section (Figs. 4A 5B) . Instead of the prototypical parallel lamellae, the fibrils were present in a swirling configuration (Figs. 5A 5B) . Electron microscopic images indicated that the difference was greater in the mid and posterior parts of the cornea. 

Collagen fibril diameter and interfibrillar distance were determined for each region of the stroma. The average fibril diameter in P2X₇ ^−/− mice was smaller in all regions of the stroma than in WT mice. Although fibril diameter steadily increased from the anterior to the posterior stroma in WT, that of the P2X₇ ^−/− remained consistent over all regions (Fig. 6) . There was a significant difference between WT and P2X₇ ^−/− in average fibril diameter over all regions (P ≤ 0.0001; data not shown). The change in collagen fibril diameter was accompanied by a change in packing of lamellae. Distributions of the interfibrillar distances of various regions were examined (Fig. 7A) . There was a wide range in P2X₇ ^−/− and WT corneas, with the greatest differences in the posterior region. In all the regions examined, the interfibrillar distance was greater in P2X₇ ^−/− mice than in WT mice. The difference in interfibrillar distance was significant over all regions (Fig. 7B) . Although the mean interfibrillar distances in corneas of both mouse strains increased from anterior to posterior, the trend was more pronounced in P2X₇ ^−/− mice. 

Real-time PCR was performed on corneal stromas and a number of other tissues (tail, lung, aorta) to examine collagen expression levels. In the stromas, the mRNA expression of both collagen types, alpha1(I) and alpha3(v), was reduced by greater than 75% (Fig. 8) . Interestingly, though the expression of alpha(1)I was reduced in the P2X₇ ^−/− stroma, it was increased by more than twofold in the aorta (data not shown), indicating that the regulation was tissue specific. 

We examined whether a change in sulfated PG was associated with the change in collagen organization. Corneas were incubated in cuprolinic blue, which binds sulfated moieties, before processing for TEM. Three distinct populations of electron-dense proteoglycans were observed in WT and P2X₇ ^−/−. These included filaments regularly associated with collagen fibrils, elongated filaments that run along the fibrils, and filaments seen in cross-section (Fig. 9A) . This arrangement of PG filaments supported that previously described in WT mice. ³⁶ In the WT stroma, the electron-dense filaments were present along the collagen fibril, with a PG to fibril ratio of 9.06 ± 0.63 PG/μm. In contrast, in P2X₇ ^−/−, the ratio of sulfated PGs detected along collagen fibrils was 6.07 ± 0.60 PGs/μm (Fig. 9B) . 

The P2X₇ receptor is an ATP-gated ion channel thought to be associated with a number of proteins, including pannexin-1. It has a large cytoplasmic domain, and the structurally important regions have been studied through the examination of a number of mutant receptors. ^{16 18 19 37 38} Previous studies report that P2X₇ ^−/− mice have lower bone mineral content with reduced periosteal bone formation and hypothesize that the receptor is required for proper skeletal response to mechanical loading. ^{28 39} This is the first study that examines the role of P2X₇ in the cornea. We found that it mediates not only the organization and expression of stromal collagen but also the expression of proteins involved in wound repair. 

We demonstrated in P2X₇ ^−/− mice that there was a downward trend in the rate of epithelial wound repair. The change in rate may be associated with the morphology of the leading edge. In the WT, the epithelium directly behind the leading edge consists of two to three cell layers, whereas in the P2X₇ ^−/−, it remained a single cell layer for a longer distance. At the same time, the leading edge of the P2X₇ ^−/− was more fragile, and tearing was observed along the stroma. These morphologic differences may be attributed to changes in protein expression at the leading edge. 

Paxillin is involved in migration and is known to be phosphorylated in wounded and migratory cells. ^{40 41 42} Phosphorylated paxillin was diffusely localized in all layers of the epithelium in P2X₇ ^−/− and WT. However, in the WT, there was additional localization of p-paxillin in a margin at the epithelial-stromal interface that was not observed in the P2X₇ ^−/−. This agreed with previous in vitro studies in which the stimulation of corneal epithelial cells with nucleotides induced the phosphorylation of paxillin. ^{7 43} Our current results suggest that P2X₇ has a role in the nucleotide-induced phosphorylation of paxillin. 

Recently, investigators have shown that P2X₇ associates with pannexin-1 to form a large, nonspecific pore. ^{20 21} Although pannexin-1 can couple, and form an active pore, with P2X₇, it has other functions independent of P2X₇, such as the formation of hemichannels. We do not understand how the pannexin was regulated, but we detected its presence, which was altered compared with WT. Our results show that pannexin-1 was not localized at the leading edge of the wound in P2X₇ ^−/− mice but that it was detected 105 μm from the leading edge in distinct areas between the cells. In WT mice, it was found at the leading edge and throughout the epithelium. 

Loose collagen fibrils were detected under the epithelial wound edge of the P2X₇ ^−/− animals; therefore, we examined the epithelial-basal lamina interface and the stroma. Although we detected no difference in the hemidesmosomal region, we observed a change in the regulation and the organization of the stroma. The stromal matrix is composed primarily of collagen fibrils stacked in orderly lamellae surrounded by PGs. This organization of PGs and collagen fibrils is thought to be responsible for the optical and structural features of the stroma. ^{44 45 46} It is possible that the change in stromal organization in the P2X₇ ^−/− results from reduced collagen expression because we found that the expression of collagens alpha1(I) and alpha3(V) mRNA was reduced in P2X₇ ^−/− mice at 4 weeks of age (Fig. 8) . However, it is unknown whether the expression is decreased throughout development. Additionally, collagen fibrils in P2X₇ ^−/− stromas demonstrated an increased interfibrillar distance (Fig. 7)that would result in a reduction in the number of collagen fibrils within a given area. This reduced packing could result in the disorganization observed. Interestingly, an increased interfibrillar distance is detected at specific developmental time points in the lumican KO mouse, ⁴⁷ suggesting that P2X₇ ^−/− mice may develop at a different rate than their age-matched WT counterparts. In other studies, a thin stroma was observed in collagen V mutations where the stromal thickness was decreased over the entire cornea. This was accompanied by an overall decrease in fibril density. ⁴⁸ In the type V collagen mutations, the orthogonal arrays were maintained, but in the P2X₇ ^−/−, the orthogonal arrays of collagen were fewer and thicker than those in the WT stroma. In addition, there were irregularities throughout the stroma with collagen fibrils that appeared to swirl in the posterior cornea. The mechanisms underlying these changes are not understood, and studies to examine epithelial-mesenchymal interactions such as those described by Hay ⁴⁹ are under way. 

The structural basis for corneal transparency requires a homogeneous distribution of collagen fibrils, fibril packing, and arrangement of orderly orthogonal arrays. It was surprising, then, that the P2X₇ ^−/− mouse corneas were transparent. It is possible that the change in fibril size was offset by the increase in interfibrillar spacing and the decrease in sulfated PGs. For example, transparency is retained in Ehlers-Danlos syndrome, in mice with type V collagen mutations, ⁴⁸ and in mucopolysaccaridosis III, ⁵⁰ where there is a difference in diameter accompanied by a retention in packing. Some investigators have shown a lack of association between central corneal thickness and refraction, whereas others have detected differences in the central corneal thickness among myopic, hyperopic, and emmetropic eyes. ^{51 52}  

The irregularity in spacing of the collagen fibrils observed in the P2X₇ ^−/− mice may be indicative of changes in glycosaminoglycans (GAGs). ⁵³ When the stroma is compromised because of disease or injury, biochemical and structural changes occur. In general, after injury, collagen fibrils have larger diameters accompanied by irregular interfibrillar spacing, thought to be caused in part by a change in the expression of sulfated GAGs. ^{54 55 56} The size and number of PGs has been demonstrated in the literature to indicate phases of development or injury. ^{55 57} Further analysis of the stroma with cuprolinic blue staining revealed sulfated PGs, with fewer PG chains in P2X₇ ^−/− detected along collagen fibrils. In fact, Young ³⁶ has provided an important study on mice GAGs for the explicit purpose of comparing WT with KO. They report that GAGs found in mice were detected as filaments associated with the collagen fibrils or as long, thick filaments that run along and traverse the fibrils. Although P2X₇ ^−/− mice had fewer GAG chains than WT, both groups exhibited the overall structures described by Young. ³⁶ Future studies will determine the expression of specific PGs and examine how mice with irregular collagen and fewer PGs respond to and repair stromal injuries. 

These studies show for the first time that the P2X₇ receptor plays a critical role in the corneal epithelium and stroma. The results indicate that P2X₇ is necessary for timely healing of abrasion wounds and normal stromal collagen structure. Our current findings indicate that the P2X₇ receptor plays a critical role in the formation of a properly organized stroma. Because P2X₇ may contribute to the regulation of a number of proteins, future studies will examine whether the decrease in migration is related to changes in epithelial-mesenchymal interactions and whether the changes in stromal architecture depend on changes in collagen regulation, as demonstrated by the type V mutation. ⁴⁸  

 

The authors thank Haiyan Gong and Matthew Nugent for their discussions of the manuscript, and Rozanne Richman for technical expertise.