Aberrant expression of a β-catenin gain-of-function mutant induces hyperplastic transformation in the mouse cornea

In mammalian cells, β-catenin is a highly conserved ubiquitous protein that shows at least two functions. It is a component of a adherent complex of cadherin that mediates cell-cell adhesion (Kemler, 1993; Orsulic et al., 1999). It can also act in conjunction with the transcription factor TCF/LEF to regulate specific gene expression upon the activation of Wnts-dependent signaling cascades (He et al., 1998; Tetsu and McCormick, 1999; Lin et al., 2000; Nelson and Nusse, 2004), resulting in dramatic alterations of many cellular events, such as proliferation, differentiation, migration. In the absence of Wnt ligands, cytoplasmic β-catenin is phosphorylated by glycogen synthase kinase 3β (GSK-3β), becomes labile and is degraded by the ubiquitin-proteasome system. By contrast, when Wnt ligand binds to its receptors Frizzled and Lrp5 or Lrp6, the assembly of proteins responsible for this phosphorylation state is abolished. Consequently, the unphosphorylated β-catenins accumulate in the cytoplasm and, subsequently, translocate into the nucleus to perform a variety of functions. In addition, β-catenin nuclear translocation can also be induced by inactivation of GSK-3β through wide variety of Wnt-independent pathways, such as fibroblast growth factors (FGFs) (Eblaghie et al., 2004; Wahl et al., 2007), epidermal growth factor (EGF) (Musgrove et al., 2004), insulin-like growth factor (IGF) (Richard-Parpaillon et al., 2002; Zhu et al., 2008), nuclear factor-κB (NF-κB) (Katoh and Katoh, 2007), phosphatase and tensin homolog (PTEN) (Zhao et al., 2005), the FP(B) prostanoid receptor (Fujino and Regan, 2001), integrin-linked kinase (Novak et al., 1998), nuclear hormone receptors (Mulholland et al., 2002; Saez et al., 2004; Mulholland et al., 2005; Schweizer et al., 2008) and oxidative stress (Mulholland et al., 2005; Almeida et al., 2007). It is noteworthy that aberrant activation of β-catenin is a crucial step in the pathogenesis of a wide variety of human cancers (Polakis, 2000).

Human ocular surface squamous neoplasia (OSSN) is the most common ocular surface pre-cancerous and cancerous lesion. It is also known by names such as conjunctival intraepithelial neoplasia, corneal intraepithelial neoplasia (CIN), or both together (CCIN) (Lee and Hirst, 1995). Clinically, OSSN manifests in different grades ranging from simple dysplasia to squamous cell carcinoma. Because of the high incidence of OSSN in the limbal area – where the corneal epithelial stem cells reside – the limbal transition zone theory (or stem cell theory) has been proposed for the development of CIN (Lee and Hirst, 1995). It has been suggested that the slow cycling limbal stem cells become hyper-proliferative by several factors, e.g. carcinogens, irradiation as well as the phorbolester and tumor promoter 12-O-tetradecanoylphorbol 13-acetate (TPA) (Tseng, 1989). Alterations in cell cycling within the limbal area can cause abnormal maturation of the conjunctival and corneal epithelium, and lead to the formation of CIN. In addition, human papilloma virus 16 (HPV16) and/or long-term UVB exposure are known to be the major risk factors associated with human OSSN (Scott et al., 2002; Kiire and Dhillon, 2006) but the actual molecular pathway(s) linking these two known clinical findings remain unknown. We have recently developed a Krt12^rtTA/rtTA/tet-O-FGF-7 double-transgenic mouse model in which overexpression of FGF-7 following induction by doxycycline (Dox) results in a tumor phenotype resembling OSSN (Chikama et al., 2008). In this tumor model, we found β-catenin nuclear translocalized in hyperplastic corneal epithelial cells upon Dox induction. These results implicated that β-catenin activation and nuclear translocation might be involved in the pathogenesis of human OSSN. In the present study, we conditionally expressed a Ctnnb1 gain-of-function mutant in the differentiated corneal epithelial cells in vivo to test whether β-catenin activation per se can induce their hyperplastic or neoplastic transformation in vivo.

The gene-targeted Ctnnb1^floxE3 mouse harbors two loxP sites flanking exon 3 of the Ctnnb1 gene. Exon 3 encodes the crucial Ser/Thr residues for priming by casein kinase1 (CK1) (Ser45) and phosphorylation by GSK-3β (Ser33, Ser37 and Thr 41). Therefore, deletion of exon 3 results in the β-catenin mutant ΔE3β-catenin, which is resistant to phosphorylation and the subsequent degradation by proteosome, by mimicing β-catenin activation (Harada et al., 1999). We found that expression of ΔE3β-catenin in the differentiated corneal epithelial cells resulted in the hyperplastic transformation of corneal epithelial cells. Interestingly, nuclear β-catenin was also found in all the human OSSN samples examined. Thus, our findings might consolidate the relationship between β-catenin activation and the pathogenesis of OSSN.

Corneal epithelial homeostasis is essential to maintain a normal visual function. To investigate whether the persistent activation of β-catenin has any effect in the differentiated corneal epithelium, we employed Dox-inducible and Cre-mediated expression systems to drive the expression of Ctnnb1 gain-of-function mutant (Harada et al., 1999) in differentiated – cytokeratin 12 (K12)-positive – corneal epithelial cells in vivo. Krt12^rtTA/Wt/tetO-Cre/Ctnnb1^floxE3/Wt triple-transgenic mice were generated and treated with or without Dox from embryonic day (E) E14.5 to postnatal day (P) P21 given that the earliest time point of endogenous K12 expression is E14.5 (Liu et al., 1993). As expected, un-induced mice developed eyes that seemed normal (Fig. 1A). By contrast, the Dox-induced group exhibited severe corneal opacity with numerous epithelial nodules and engorged neovascularization (Fig. 1B). Western blotting analysis showed that these triple-transgenic mice expressed 94 kD wild-type β-catenin under normal circumstances (Fig. 1C, lane 1), but an extra 66-kD band representing ΔE3β-catenin mutant was detected in the Dox-treated corneal lysates (Fig. 1D, lane 2).

We then examined whether the ΔE3β-catenin executed its transcriptional activity in collaboration with Lef/Tcf using a TOP-Gal reporter mouse line (DasGupta and Fuchs, 1999). The X-Gal-stained mouse eyeballs revealed that little or no β-galactosidase activity was shown in un-induced Krt12^rtTA/Wt/tetO-Cre/Ctnnb1^floxE3/Wt/TOP-Gal quadruple-transgenic mice (Fig. 2A,B), but many abnormal nodules with strong X-Gal-positive reactions (dark blue) appeared in the corneal surface of the Dox-induced quadruple-transgenic mice (Fig. 2C,D). These data demonstrate that the canonical Wnt/β-catenin signaling pathway is either inactive or suppressed in the normal cornea and limbus during fetal and neonatal stages. Also, the expression of ΔE3β-catenin mutant potentially triggers Tcf/Lef-dependent transcriptional activity, leading to the corneal nodule formation in differentiating and differentiated corneal epithelium.

To confirm that these epithelial nodules have Cre activity, we employed the global double-fluorescent Cre reporter mouse strain ROSA^mTmG, which expresses membrane-tagged red fluorescence before and membrane-tagged green fluorescence after Cre-mediated recombination in a wide variety of cell types (Muzumdar et al., 2007). Corneas from un-induced K12^rtTA/Wt/tetO-Cre/Ctnnb1^floxE3/Wt/ROSA^mTmG/Wt or K12^rtTA/Wt/tetO-Cre/Ctnnb1^Wt/Wt/ROSA^mTmG/Wt quadruple-transgenic mice, exhibited whole-body red fluorescence (data not shown). However, in both Ctnnb1 wild-type and Ctnnb1^floxE3/Wt mutant mice administrated Dox, red and green fluorescence patterns were mutually excluded by individual corneal epithelial cell (Fig. 2E-H). Moreover, as expected, no abnormality was found in the Ctnnb1 wild-type cornea (Fig. 2E,F) but various sizes of abnormal corneal nodules showing green fluorescence appeared in the Ctnnb1^floxE3cornea (Fig. 2G,H).

To investigate whether the pathological progression resulted from the expression of ΔE3β-catenin, Krt12^rtTA/Wt/tetO-Cre/Ctnnb1^floxE3/Wt triple-transgenic mice were administered Dox through IP-injection of the pregnant dam and examined at different developmental stages from E14.5 to P21. Morphological and immunohistochemical studies revealed that neither significant alterations nor nuclear β-catenin was detected during E14.5-E16.5 (data not shown). However, at E17.5 the corneal epithelium began to show epithelial protrusions in mice expressing ΔE3β-catenin (Fig. 3D), whereas age-matched Dox-induced single-transgenic (not shown) or un-induced triple-transgenic embryos exhibited normal corneal epithelium with typical two-cell layers (Fig. 3A). Immunofluorescence staining revealed that β-catenin expression was restricted to the cell membrane of the corneal epithelium (Fig. 3B,B′); however, Dox-treated Ctnnb1^floxE3/Wt mutants exhibited many epithelial cells with nuclear β-catenin (Fig. 3E,E′). In a consecutive section, we found that the expression of K12, normally expressed throughout the entire corneal epithelium (Fig. 3C,C′), was downregulated upon expression of ΔE3β-catenin (Fig. 3F,F′). Mice expressing ΔE3β-catenin until P2 displayed more-prominent hyperplastic nodules (compare Fig. 3G with J) than those found at E17.5 and, as before, cells within the nodules showed nuclear β-catenin staining (Fig. 3K,K′) and lost K12 expression (Fig. 3L,L′).

Consistent with the formation of hyperplastic epithelial nodules, expression of ΔE3β-catenin resulted in an eightfold and threefold increase in BrdU incorporation within the corneal epithelium and stroma, respectively (Fig. 4A-C). In addition, ΔE3β-catenin expression resulted in the upregulation of vascular endothelium growth factor-A (Vegf-A), its receptor Vegfr1 (Flt1), and angiopoietin (Angpt)-1 and Angpt-2 mRNA levels (Fig. 4D) – yet no obvious blood vessels were noticed in the cornea at this stage (Fig. 3J). These data suggest that expression of ΔE3β-catenin is responsible for the increase of corneal epithelial cell proliferation and the production of pro-angiogenic factors.

Corneal epithelial homeostasis was completely compromised, with profound angiogenesis and stromal invasion, in the Dox-treated Krt12^rtTA/Wt/tetO-Cre/Ctnnb1^floxE3/Wt mice at P21 (Fig. 5E-H). It prompted us to investigate whether the proteins involved in epithelial invasion rather than the maintenance of corneal epithelial integrity are altered by ΔE3β-catenin. Immunofluorescence staining demonstrated a reduction in E-cadherin expression levels, particularly in the epithelial nodules (Fig. 5L,L′) where anti-β-catenin staining showed strongly in the nuclei (Fig. 5K,K′). In addition, the major components of the basement membrane laminin-β1 and laminin-α1 (data not shown) were expressed in the epithelium during corneal morphogenesis and restricted to the basement membrane at P21 (Fig. 6A-C), but were drastically downregulated from E17.5 to P2 (Fig. 6D,E) and completely diminished at P21 (Fig. 6F) in mice expressing ΔE3β-catenin. Laminin has been known to be a proteolytic substrate for metalloproteases (MMPs) (Siu and Cheng, 2004), among which MMP-7 (also known as MMP7, matrilysin) is a direct downstream gene target of activated β-catenin (Gustavson et al., 2004). Indeed, MMP-7 was not detected in a normal corneal epithelium (Fig. 6G,G′), but aberrantly expressed in the ΔE3β-catenin mutant corneal epithelium (Fig. 6H,H′). These data implicate that ΔE3β-catenin directly downregulates E-cadherin gene expression and simultaneously upregulate expression of MMP-7 leading to the degradation of laminins and the subsequent disruption of the basement membrane.

To investigate the molecular and cellular mechanism by which the expression of ΔE3β-catenin had strong impact on proliferation and differentiation that might lead to disruption of corneal epithelial homeostasis, immunofluorescent staining revealed that both protein expression of PCNA and p63 was increased in ΔE3β-catenin expressing corneal epithelium (Fig. 7A-D). However, expression of ΔE3β-catenin resulted in the loss of Pax-6 (Fig. 7F) and K12 (Fig. 8B) expression in the hyperplastic epithelial nodules, indicating that the corneal differentiation phenotype was somehow lost. Moreover, the expression pattern of keratin 4 (K4; Fig. 8C,D) and Muc5A/C (data not shown) was not changed by the expression of ΔE3β-catenin. Likewise, no positive staining of keratin 10 (K10), vimentin or α-smooth muscle actin (α-SMA) was found in the epithelial nodules (data not shown), suggesting that ΔE3β-catenin expression did not induce transdifferentiation of corneal epithelium into conjunctival or epidermal epithelia or mesenchymal cells. Interestingly, these nodules were strongly positive for keratin 14 (K14; Fig. 8F), however, the expression of keratin 15 (K15), which is normally detected in the conjunctival and limbal epithelium (Yoshida et al., 2006), was aberrantly extended into the central corneal epithelium upon expression of ΔE3β-catenin (Fig. 8G,H). Collectively, these data implicate that expression of the ΔE3β-catenin mutant enhance cell proliferation and may induce ‘dedifferentiation’ of the corneal epithelium towards limbal epithelium, but not ‘transdifferentiation’ into conjunctival or epidermal epithelium or hair follicle (Pearton et al., 2005).

To investigate the effect of ΔE3β-catenin expression in the adult corneal epithelium, 6-week-old Krt12^rtTA/Wt/tetO-Cre/Ctnnb1^floxE3/Wt mice were administered Dox by a one-time IP-injection followed by Dox chow ad libitum for 2 weeks. Littermates of the same genotype were fed with normal chow and served as a control. Immunofluorescence staining showed that nuclear β-catenin was detected 3 days after Dox induction (Fig. 9A,B). Interestingly, as compared with un-induced corneal epithelium, which was well-organized into four to six layers (Fig. 9A,A′,A″), aberrant expression of ΔE3β-catenin caused mis-arrangement of the cells with many epithelial nodules (Fig. 9B,B′,B″). More interestingly, all mice (n=4) that expressed ΔE3β-catenin developed corneal neoplasia at 14 days after Dox treatment with leukoplakic lesion and profound neovascularization, resembling to human OSSN (Fig. 9E-H).

The etiology and pathogenesis of human OSSN remain elusive. To investigate whether nuclear translocation of β-catenin has any association with human OSSN, we examined clinical specimens from the eye pathology laboratory and performed immunohistochemistry. Out of eight human OSSN specimens, all of them showed moderate or strong nuclear β-catenin staining, whereas a normal eye-bank cornea revealed β-catenin immunoreactivity only in the cell membrane (Fig. 10; supplementary material Table S1).

In this study, we have provided strong evidence that a β-catenin gain-of-function mutant can disrupt corneal epithelial homeostasis and induce hyperplastic transformation in Krt12^rtTA/Wt/tetO-Cre/Ctnnb1^floxEx3/Wt tri-transgenic mice regardless the initial time of Dox-treatment during or following embryonic development. We have demonstrated that forced expression of stabilized ΔE3β-catenin in the differentiated corneal epithelium can cause tumorigenesis. The phenotypes are similar but not identical to those observed previously in Dox-treated Krt12^rtTA/rtTA/tetO-FGF-7 mice, which also displayed hyperplastic nodules on the corneal surface (Chikama et al., 2008).

Repression of the Wnt/β-catenin signaling pathway has been implicated as an essential prerequisite for the differentiation of corneal epithelial progenitor cells into a non-keratinizing stratified epithelium during corneal morphogenesis (Mukhopadhyay et al., 2006; Gage et al., 2008). It has been previously shown that stabilized ΔE3β-catenin driven by the Le-Cre promoter suppresses surface ectoderm invagination for lens formation at ~E9.5 to ~E10.5. These data suggested that Wnt/β-catenin signaling is unfavorable during the early stages of ocular surface morphogenesis (Smith et al., 2005). The K12^rtTA/tetO-Cre driver mice allow us to assess the role of Wnt/β-catenin signaling in a corneal cell-type specific and Dox-inducible manner. In the present study, we demonstrated that persistent expression of ΔE3β-catenin driven by the endogenous Ctnnb1 enhancer-promoter had strong impact to the differentiated (Pax-6-positive and K12-positive) corneal epithelial cells. Therefore, the stringent control of Wnt/β-catenin signaling is crucial for the maintenance of homeostasis and integrity in differentiated corneal epithelium.

We have documented that endogenous K12 can be detected in the corneal epithelium by in-situ hybridization at E14.5 and by immunohistochemistry at E15.5 (Liu et al., 1993). Therefore, we expected the phenotypes caused by ΔE3β-catenin to be observed ~E14 to ~E15.5 if mice were exposured to Dox at the beginning of life. However, we have never detected nuclear β-catenin or noticed any significant morphological change before E17.5 in the Krt12^rtTA/Wt/tetO-Cre/Ctnnb1^floxE3/Wt mouse embryos induced with Dox since E0.5. This could be attributed to the delay resulting from a series of cellular events including rtTA activation by Dox, transcriptional activation of tetO-Cre, generation of the Ctnnb1^ΔE3 allele followed by its transcription and/or translation, and the accumulation of the ΔE3β-catenin protein. Indeed, we previously documented that it took 48 hours following induction to drive the expression of a Cre-reporter gene using the K12^rtTA/tetO-Cre driver mouse line (Chikama et al., 2005). Expression of ΔE3β-catenin had a tremendous impact on the corneal epithelium as early as at E17.5. Initially, only a few protrusions were displayed on the corneal epithelium at E17.5 but these quickly progressed into hyperplastic epithelial nodules at P2. It is noteworthy that expression of ΔE3β-catenin in the corneal epithelium also enhanced underneath stromal cell proliferation. This result strongly indicated that Wnt/β-catenin signaling elicited from the corneal epithelium triggered an epithelium-stromal mesenchyme interaction via a putative growth-promoting factor to enhance stromal cell proliferation. Although the identity of this growth promoting factor is unknown, FGF family members are probably involved because it has been documented that Wnt/β-catenin signaling upregulated FGFs including FGF-4, FGF-9, FGF-18 and FGF-20 in various cellular context (Kratochil et al., 2002; Hendrix et al., 2005; Shimokawa et al., 2003; Chamorro et al., 2006). Since these hyperplastic nodules appeared only in the epithelium before the opening of the eyelid and we did not observe F4/80-positive macrophage in any of these corneas (data not shown), this dramatic pathological progression is probably caused by the autonomous effect of ΔE3β-catenin. This argument is somewhat supported by our results in that the Ctnnb1 gene is required for the Dox-induced corneal epithelial hyperplasia and/or neoplasia formation in the Krt12^rtTA/rtTA/tetO-FGF-7 mice (Y.Z. and C.-Y.L., unpublished results)

Although we cannot exclude additional secondary or tertiary effects following ΔE3β-catenin expression that might involve subsequent pathological progression (Fig. 5E), it is possible that ΔE3β-catenin expression still causes directly and actively epithelial invasion into the corneal stroma, and neovascularization at P21. To explore the mechanism by which stabilized ΔE3β-catenin causes the aforementioned pathology in the cornea, we have tested the expression levels and/or patterns of some related genes such as E-cadherin (Cdh1), MMP-7 (Mmp7), and Vegf-A (Vegfa). For example, E-cadherin, known to be the Wnt/β-catenin target gene (Jamora et al., 2003; ten Berge et al., 2008), was drastically downregulated in the epithelial nodules (Fig. 5L,L′), suggesting that the adherent junction was compromised by the expression of ΔE3β-catenin. Since reduced E-cadherin gene expression has been shown to be associated with epithelial tumor formation and invasion (Birchmeier, 2005), it is probable that the formation of the hyperproliferative epithelial nodules is directly associated with the downregulation of E-cadherin by ΔE3β-catenin. However, both Vegf-A and MMP-7 genes, which contain Lef/Tcf binding sites (Zhang et al., 2001; Crawford et al., 1999; Brabletz et al., 1999), were upregulated in the corneas of Krt12^rtTA/Wt/tetO-Cre/Ctnnb1^floxE3/Wt triple-heterozygous mice. It is obvious that upregulation of Vegf-A and other proangiogenic factors, such as Vegfr1, Angpt-1 and Angpt-2, through direct or indirect effects of ΔE3β-catenin expression resulted in profound corneal neovascularization. Likewise, ectopic upregulation of MMP-7 probably caused proteolytic degradation of laminins (Siu and Cheng, 2004; Gustavson et al., 2004) and disruption of basement membrane; therefore, the overall corneal homeostasis was compromised.

Both β-catenin and Lef/Tcf function as factors to sustain tissue progenitor cells in the adult (Lowry et al., 2005). Our immunofluorescence staining results showed that BrdU uptake (Fig. 4) aswell as expression of PCNA and p63 increased, but Pax-6 expression decreased (Fig. 7). This was consistent with the previous report stating that activation of Wnt/β-catenin signaling enhanced Ki67 expression and cell division, and induced corneal epithelial dedifferentiation (Pearton et al., 2005). Moreover, the cell-type-specific pattern of epithelial keratin expression showed downregulated expression of K12 (the corneal epithelial marker) and ectopic expression of K15 (the limbal marker) in the central cornea, but no changes in the expression patterns of K4 (the conjunctival marker) or K10 (the epidermal marker). Our results argue that the expression of ΔE3β-catenin in K12-positive corneal epithelial cells do not trigger their transdifferentiation into conjunctival or epidermal epithelium but rather induced dedifferentiation towards limbal epithelial progenitor cells, and that their further differentiating into corneal epithelium could be dependent in the presence of other growth and/or differentiation factors in the environment, such as FGF-7. This is probably why K12 expression in the FGF-7 model remained (Chikama et al., 2008) but was absent in β-catenin gain-of-function mice. Nevertheless, it requires further study to demonstrate whether these ΔE3β-catenin-expressing nodules, indeed, maintain the characteristic features of stem and/or progenitor cells and can differentiate into corneal epithelial cells in vitro and/or in vivo.

Whereas lack of the Wnt antagonist Dkk2 resulted in the ectopic appearance of conjunctiva, epidermis and hair follicles in the corneal epithelium due to direct and/or indirect effect (Mukhopadhyay et al., 2006; Gage et al., 2008), we did not observe these phenotypes in our animal model, regardless of the initial time of Dox-treatment to induce expression of ΔE3β-catenin in K12-positive corneal epithelial cells. One of the possibilities is that Wnt/β-catenin activation in the Dkk2 null-mutant affects corneal epithelial stem and/or progenitor cells, whereas only the K12-positive cells (including differentiating – or transient amplifying cells – and differentiated corneal epithelial cells) are influenced in our mouse model. Because Dkk2 knockout was done much earlier and might have targeted the limbal stromal mesenchyme, one can appreciate why more plasticity into other cell lineages is possible. Moreover, it is possible that loss of Dkk2 not only activates Wnt signaling, but also changes other signaling pathway(s) that are necessary for fating cornea into epidermis. Furthermore, activation of β-catenin per se is not sufficient to change the differentiation of corneal epithelium into epidermal or hair follicle. Other molecules involved in signaling pathways, such as noggin or the bone morphogenetic protein (BMP) antagonist, might also be needed to switch a dedifferentiated corneal epithelium into epidermal epithelium (Pearton et al., 2005).

It is of particular interest that, panni that have been surgically removed from human OSSN patients also exhibited nuclear translocation of β-catenin (Fig. 10). These findings implicate that β-catenin activation might be a common mechanism in tumorigenesis resulting from corneal progenitor cells that undergo oncogenic transformation as a result of insults, such as infection with the HPV16 and/or long-term UVB exposure (Scott et al., 2002; Kiire and Dhillon, 2006). In the Ctnnb1 gain-of-function mouse model, the generation of Ctnnb1^ΔE3 allele is dependent upon Dox-inducible Cre activity, but the expression of ΔE3β-catenin is under transcriptional control by its own promoter and/or enhancers. In human cancers, exon 3 of the CTNNB1 gene is a mutational hot spot for the gain-of-function isoform (Moon et al., 2002). Since the CTNNB1 gene is transcriptionally active in the corneal epithelial cells, it is possible to activate the gain-of-function β-catenin isoform through the somatic mutation in corneal epithelium. Nevertheless, it requires further investigation to determine whether the exon 3 mutation of CTNNB1 is linked with the pathogenesis of human OSSN.

Altogether, our data, which rely on the Dox-inducible corneal specific K12^rtTA/tetO-Cre driver mice, suggest that the aberrant expression of a dominant stable ΔE3β-catenin mutant triggers a variety of cellular alterations including cell proliferation and differentiation by regulating the expression of related genes and, thus, disturbing corneal morphogenesis and homeostasis with the formation of nodules and neovascularization in the cornea. Tight regulation of β-catenin signaling to control gene expression is indispensable for cornea development and correct maintenance of its homeostasis, and may otherwise lead to tumorigenesis such as OSSN (Fig. 11). Thus, inhibition of β-catenin activation might be beneficial for treating this disease.

Human OSSN samples and eye bank eye samples were obtained in accordance with the Institutional Review Board at Chang Gung Memorial Hospital (protocol #97-1256B to L.K.Y.) and all participants gave informed written consent.

Experimental animals were housed under pathogen-free conditions in accordance with institutional guidelines. Animal care and use conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. All animal protocols were approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Cincinnati.

Genetic modified mouse lines such as Krt12^rtTA (Chikama et al., 2005), tetO-Cre (Perl et al., 2002), Ctnnb1^floxE3 (Harada et al., 1999), TOP-Gal (DasGupta and Fuchs, 1999) and Gt(ROSA)26Sor^{tm4(ACTB-tdTomato-EGFP)Luo}/J (here referred to as ROSA^mTmG) mice (Muzumdar et al., 2007) have been previously described. Compound transgenic mice were generated by natural breeding of individual mouse lines, and the genotypes were identified by polymerase chain reaction (PCR) using oligonucleotide primers specific for each transgene. In all experiments, detection time of the vaginal plug was defined as E0.5 of embryonic development. To activate tetO-Cre expression, mice were intraperitoneally (IP) injected once with Dox (Sigma, 80 μg/g body weight) at a concentration of 10 mg/ml (Utomo et al., 1999) and fed Dox-chow (1 g/kg chow, Bioserv Corporation). Control animals were fed regular chow.

The following antibodies were used: rabbit anti-K12 antibody (2 μg/ml) (Liu et al., 1994); rabbit anti-Pax-6 (1:100), rabbit anti-K15 (1:100), rabbit anti-K10 (1:200) (Covance Inc.); rabbit monoclonal anti-MMP-7 (1:100), mouse 4A4 anti-p63 (1:50); mouse anti-BrdU (1:100); mouse PC10 anti-PCNA (1:100) (Lab Vision, Corp.); mouse anti-β-catenin (1:50, BD Biosciences Pharmingen); rabbit anti-β-catenin (1:2000, Sigma); mouse 6B10 anti-K4 (1:100, Abcam, Inc.); Goat anti-laminin-α1 (1:100), anti-laminin-β1(1:100) (Santa Cruz Biotechnology, Inc.). Alexa-Fluor-555-conjugated donkey anti-goat-IgG or goat anti-rabbit IgG was used for immunofluorescent staining. For immunohistological analysis, excised eyes were fixed in 4% paraformaldehyde in PBS and paraffin embedded. Tissue sections (5 μm) were de-paraffinized and hydrated in a graded ethanol series. The primary antibody was incubated at the dilutions mentioned above in TBST/BSA (3%) at 4°C overnight. Non-specific antibody binding was removed by washing with TBST three times. Alexa-Fluor-555-conjugated secondary antibody was then incubated at room temperature for 1 hour. Sections for immunofluorescence analysis were mounted (SlowFade Light Antifade Kit; Molecular Probes, Inc.) in the presence of 4′,6′-diamidino-2-phenylindole (DAPI), and observed using a Nikon Eclipse E800 epifluorescent microscope and photographed using a SPOT RT digital camera system (Diagnostic Instruments, Inc.). For western blotting analysis, total cellular lysates from mouse corneas were prepared in lysis buffer [Tris-HCl, 50 mM, pH 7.4; NaCl, 250 mM; EDTA, 5 mM; NP-40, 0.1%; NaF, 25 mM; 1× protease inhibitor cocktail (Sigma P8340), and subjected to sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (10%). First antibody was incubated at dilutions as mentioned above in TBST/BSA (3%) at 4°C overnight. Alkaline phosphatase (AP)-conjugated secondary antibody was then incubated at room temperature for 1 hour. Positive signals were visualized by Western blue substrate for AP (Promega, Inc.).

Total RNA (10 μg) was isolated from corneal epithelium, annealed to random primer and reverse transcribed with avian reverse transcriptase (RT) kits from Promega Inc. (Madison, WI) according to manufacturer's instructions. Single-strand cDNA was subjected to PCR using primer pairs. Real-time RT-PCR was performed by using the CFX96 real-time system equipped on a C1000™ Thermal Cycler (Bio-Rad Laboratories Inc.). After the initial step at 95°C for 5 minutes, 40 cycles at 95°C 30 seconds, 62°C 30 seconds, 72°C 30 seconds. The cycle threshold values were used to calculate the normalized expression of mouse Vegf-A, Vegfr1 (Flt1), Angp-1, and Angpt-2 against β-actin using Q-Gene software. Primer sets and their PCR products for the quantitative RT-PCR are listed as follows: mVegf-A forward primer 5′-TTCCTACAGCACAGCAGATG-3′, mVegf-A reverse primer 5′-TTACACGTCTGCGGATCTTG-3′, PCR product 122 base pairs (bps); mFlt-1 forward primer 5′-GTCTTGCCTTACGCGCTGCT-3′, mFlt-1 reverse primer 5′-AGAGTCTGGCCTGCTTGCAT-3′, PCR product 121 bps; mAngpt-1 forward primer 5′-ATGCGCTCTCATGCTAACAG-3′, mAngpt-1 reverse primer 5′-CCGCAGTGTAGAACATTCCA-3′, PCR product 80 bps; mAngpt-2 forward primer 5′-ACGAGGCGCATTCGCTGTAT-3′, mAngpt-2 reverse primer 5′-CTGGTTGGCTGATGCTACTT-3′, PCR product 117 bps; mβ-actin forward primer 5′-GCTCTGGCTCCTAGCACCATGA-3′, mβ-actin reverse primer 5′-CCTGCTTGCTGATCCACATCTG-3′, PCR product 127 bps.

Embryos were stained as whole mounts overnight at room temperature in a solution of X-Gal (Sigma) at a final concentration of 0.4 mg/ml made from a 40 mg/ml stock in DMSO, with 4 mM K₃Fe(CN)₆, 4 mM K₄Fe(CN)₆·6H₂O, 2 mM MgCl₂ in PBS. After staining, tissues and embryos were rinsed with PBS and photographed as whole mounts. These samples were then embedded in paraffin as previously described. Paraffin sections were counterstained with hematoxylin, observed with a Nikon Eclipse E800 epifluorescent microscope and photographed with a SPOT RT digital camera system (Diagnostic Instruments, Inc.).

Adult (6-weeks old) quadruple-transgenic mice K12^rtTA/Wt/tetO-Cre/Ctnnb1^Wt/Wt/ROSA^mTmG/Wt or K12^rtTA/Wt/tetO Cre/Ctnnb1^floxE3/Wt/ROSA^mTmG/Wt administered Dox chow for 3 days. Mouse eyes were enucleated and corneas were dissected, counterstained with DAPI, mounted on slides. The Z-stack images were collected using the Zeiss LSM 510 META two-photon laser scanning confocal microscope (Carl Zeiss MicroImaging, Inc.). The fluorescent signal detection from tomato red was obtained by using the 543-nm laser line of the red helium-neon laser. Emitted light was detected by using a 565-615 nm filter. For the detection of EGFP, the 488-nm laser line of the argon laser was selected and emitted light was caught by a filter of 500-550 nm. DAPI signals were captured by using a model-locked MaiTai laser (Carl Zeiss). Three dimensional images were acquired by using the Axiovision AxioVs40 V 4.7.0.0 software (Carl Zeiss). Images were treated using Adobe Photoshop CS2 (Adobe Microsystems).

Two-tailed Student's t-test (Excel, Microsoft, Redmond, WA) was used in analyzing the number of BrdU-positive cells. All quantification data are presented as the mean ± s.d.

We thank the reviewers for their instructive comments for this manuscript. This work was supported by grants from NIH/NEI RO1 EY12486 (C.-Y.L.) and EY13755 (W.W-Y.K.), Research Prevent Blindness, Ohio Lions Foundation for Eye Research. Deposited in PMC for release after 12 months.