Essential Role for Pbx1 in Corneal Morphogenesis

Abstract

Purpose.: The Pbx TALE (three-amino-acid loop extension) homeodomain proteins interact with class 1 Hox proteins, which are master regulators of cell fate decisions. This study was performed to elucidate the role of the Pbx1 TALE protein in the corneal epithelium of mice.

Methods.: Pbx1^f/f mice were crossed with mice containing Cre recombinase under the control of the K14 promoter. Subsequently, the eyes of these mice were dissected and prepared for histologic or molecular analysis.

Results.: Tissue-specific deletion of Pbx1 in the corneal epithelium of mice resulted in corneal dystrophy and clouding that was apparent in newborns and progressively worsened with age. Thickening of the cornea epithelium was accompanied by stromal infiltration with atypical basal cells, severe disorganization of stromal collagen matrix, and loss of corneal barrier function. High epithelial cell turnover was associated with perturbed expression of developmental regulators and aberrant differentiation, suggesting an important function for Pbx1 in determining corneal identity.

Conclusions.: These studies establish an essential role of the Pbx1 proto-oncogene in corneal morphogenesis.

The cornea covers the outer surface of the eye, and its refractive and transparent properties are essential for unimpaired vision. Loss of corneal integrity due to disease, injury, or surgical interventions can lead to loss of sight, which affects more than 1.5 million individuals annually, highlighting the importance of understanding corneal integrity. ¹  

The cornea is composed of three distinct layers. A single cell layer of endothelium separates the inner aqueous humor of the eye chamber from the corneal stroma, which constitutes up to 90% of the corneal thickness. The stroma contains interspersed neural crest–derived keratocytes that secrete the highly structured collagen-rich matrix essential for corneal transparency. An external, nonkeratinized, self-renewing epithelium of ectodermal origin acts as a barrier to the external environment. It consists of both basal and stratified squamous cells that are separated from the corneal stroma by a basement membrane. ² Stratification of the cornea occurs in early development, and postnatally the corneal epithelium exhibits a dynamic homeostasis, turning over approximately every 10 days. ^3,4 Defining the mechanisms that orchestrate differentiation and self-renewal of corneal epithelial cells will facilitate our understanding of corneal epithelial cell homeostasis. 

The TALE (three amino acid loop extension) class homeodomain transcription factor Pbx1 was initially described as a proto-oncogene (PBX1, pre-B cell leukemia transcription factor 1 ^5,6 ) and has subsequently been characterized as a global developmental regulator in mice, zebrafish, ^7,8 Caenorhabditis elegans, and Drosophila melanogaster. ^9,10 The Pbx TALE proteins display unique abilities to interact with class 1 Hox proteins, which are master regulators of cell fate, and orphan homeodomain proteins, which are necessary for development of various tissues and organs. ^11–16 Pbx interactions with these homeodomain transcription factors increase their DNA-binding affinity and specificity. Pbx1 also functionally interacts with the bHLH class of transcription factors, which regulate skeletal muscle development, differentiation, and regeneration, thus suggesting that Pbx1 serves an even broader and more important role as a general transcriptional cofactor integrating transduction signals in embryogenesis and organ development. ^17–19  

In this report, a conditional knockout mouse model was used to specifically inactivate Pbx1 in corneal epithelial cells. This method resulted in alteration of corneal cell fate, high epithelial cell turnover, and marked disruption of the corneal basement membrane with deficient epithelial barrier function. These studies demonstrate an essential role for Pbx1 in morphogenesis and maintenance of self-renewing adult tissues of the cornea. 

Materials and Methods

Mice

K14Cre transgenic mice (Jackson Laboratory, Bar Harbor, ME) and Pbx1^+/−-knockout mice ²⁰ were intercrossed to generate K14Cre⁺.Pbx1^+/− mice. Subsequent breeding with Pbx1^f/f mice ¹⁶ produced K14Cre⁺.Pbx1^−/f or K14Cre⁺.Pbx1^f/f mutants and their control littermates. Mice were genotyped by PCR as previously described ¹⁶ and maintained in the Stanford animal facility. Mice aged 17 to 21 days were used for the experiments, except for those on neonatal (P0) mice. All experiments were performed with the approval of and in accordance with Stanford's Administrative Panel on Laboratory Animal Care and the ARVO Statement for the Use of Animals in Ophthalmic and Visual Research. 

Anatomy and Histology

Eye macroscopic morphology was photographed (model D100; Nikon, Tokyo, Japan). Enucleated whole eyes were washed in phosphate-buffered saline (PBS) and fixed in 4% paraformaldehyde in PBS for 45 minutes. Subsequently, the eyes were washed in PBS and stored at 4°C in 30% sucrose-buffered PBS. They were embedded in antifade Yazulla mounting medium and sectioned at 14-μm thickness. 

Immunohistochemistry

The presence of Pbx1b in tissue sections was detected with an anti-Pbx1b antibody ²¹ using an ABC reagent (Vectastain), MOM, and DAB (Vector Laboratories, Burlingame, CA) according to the manufacturer's instructions. 

For immunochemistry, the sections were blocked for 1 hour at room temperature (RT) in PBS, 2% NGS, and 0.2% Triton-X and incubated overnight at 4°C with the primary antibodies. The sections were washed three times for 5 minutes each at RT in PBS, and then incubated with secondary antibodies (1:2000 dilution) for 1 hour at RT. The slides were counterstained with DAPI, mounted with glycerol/PBS (AS1; Citifluor Ltd., London, UK), and photographed (BX51; Olympus, Tokyo, Japan). Primary antibodies consisted of anti-Notch1 (1:500; 15580-100; Abcam, Cambridge, UK), anti-Ki67 (1:500; 49990-100; Abcam), anti-decorin (1:10; R&D Systems, Minneapolis, MN), anti-Pax6 (1:100; Developmental Studies Hybridoma Bank, Iowa City, IA), anti-K12 (1:500; Santa Cruz Biotechnology, Santa Cruz, CA), anti-K10 (1:500; Santa Cruz), and anti-CD45 (1:300; Ebiosciences, San Diego, CA). Secondary antibodies consisted of anti-mouse (Alexa 488 and 494 [Invitrogen-Molecular Probes, Eugene, OR], A21121 and A1105 [Serotec, Oxford, UK] and anti-rabbit (Alexa 488 and 494; A11008 and A11012) antibodies. 

Cornea Fragility Assay

Corneal fragility was assessed postmortem, as previously described. ²² One cornea was brushed in a circular fashion two times with a microsponge soaked in PBS; the contralateral cornea was not disturbed. Subsequently, 0.5 mL of 2% fluorescein in PBS (Sigma, St Louis, MO) was applied for 1 minute to each eye, which was then copiously washed with PBS. The eyes were examined immediately afterward and photographed (D100; Nikon) under a binocular microscope (1500SMZ; Nikon) with a GFP filter. 

Electron Microscopy

Samples were fixed in Karnovsky's fixative (2% glutaraldehyde and 4% paraformaldehyde in 0.1 M sodium cacodylate [pH 7.4]) for 1 hour at RT, then sectioned, postfixed in 1% osmium tetroxide for 1 hour at RT, washed three times with ultrafiltered water, and stained en bloc for 2 hours at RT or held at 4°C overnight. Samples were then dehydrated through a series of ethanol washes for 15 minutes each at 4°C beginning at 50%, then 70%, then 95%, at which point the samples were allowed to warm to RT, and the wash was changed to 100% ethanol and then to propylene oxide (PO) for 15 minutes. The samples were infiltrated with resin (EMbed-812; EMS cat. no. 14120) mixed 1:1 with PO for 2 hours followed by two parts resin and one part PO overnight. The samples were placed in resin for 2 to 4 hours, and then placed into molds with labels and fresh resin, oriented, and held at 65°C overnight. 

The sections (75–90 nm) were picked up on formvar/carbon-coated 75-mesh copper grids, stained for 20 seconds in 1:1 saturated ∼7.7% uracetate in acetone followed by staining in 0.2% lead citrate for 3 to 4 minutes, observed by transmission microscopy (1230 TEM; JEOL. Tokyo, Japan) at 80 kV, and photographed with a digital camera (Multiscan 791; Gatan, Inc., Pleasanton, CA). 

Quantitative Real-Time PCR

Eyes were enucleated postmortem, washed profusely in PBS, and enzymatically digested at 4°C for 8 hours in EMEM without calcium (Cambrex Biosciences, Cambridge, UK) containing 15 mg/mL dispase II (Sigma). The ocular surface epithelial sheet was loosened by gentle shaking in PBS, transferred to RNA extraction reagent (TRIzol; Invitrogen), and mechanically sheared. RNA was isolated and cDNA synthesized by using commercially prepared reagents according to the manufacturer's instructions (Invitrogen), then subjected to real-time PCR (Prism 7700 Detector; Applied Biosystems, Inc. [ABI], Foster City, CA) using probes for Pbx1, cMyc, cyclin D1, p21, p27, p53, Notch1, Pax6, and Decorin1 (TaqMan; ABI). Expression levels were standardized to those of β-actin, which served as an internal control. Data were analyzed (using SDS ver. 1.9.1 software; SDS, Cary, NC; and Excel, Microsoft, Redmond, WA). 

BrdU Labeling

Mice were injected with BrdU intraperitoneally (20 mg/mL in PBS per gram mouse weight) and killed 4 hours later. The eyes were removed, fixed in 4% paraformaldehyde in PBS at RT for 45 minutes, washed in PBS, and held in 30% sucrose-buffered PBS at 4°C before embedding. 

Apoptosis Assay

Terminal dUTP nick end-labeling (TUNEL) assays to detect DNA strand breaks in cell nuclei were performed on frozen sections (10 μm) using commercially prepared reagents (Roche, Indianapolis, IN) according to the manufacturer's instructions. The sections were counterstained with DAPI, and after three washes in PBS, were mounted (AF1; Citifluor Ltd) and viewed by fluorescence microscope (BX51; Olympus) equipped with appropriate filters. TUNEL-positive cells were counted in at least three corneas of wild-type (wt) and knockout (ko) mice (for each cornea, three slides were obtained from the central portion of the cornea). 

Results

Efficient Deletion of Pbx1 in the Mouse Cornea

A genetic approach was used to conditionally delete Pbx1 in the mouse corneal epithelium, which contains abundant Pbx1b protein (Fig. 1A), the major isoform of Pbx1 present in developing and adult tissues. Cre recombinase was expressed under control of the keratin 14 (K14) promoter, which is normally active in basal and occasionally in suprabasal cells of the corneal epithelium ²³ as confirmed by immunohistochemistry (IHC; Fig. 1B). Mice with a Pbx1 floxed allele (Pbx1^f/f) ¹⁶ were crossed with Pbx1^+/− mice ²⁰ containing the K14Cre transgene to generate K14Cre:Pbx1^f/− (mutant) and littermate K14Cre:Pbx1^f/+ (control) mice (Fig. 1C), which will henceforth be referred to as mutant and control, respectively. Efficient and specific deletion of Pbx1 occurred in epithelial tissues of the cornea and skin, including the epithelial conjunctiva but not nonepithelial tissues (Fig. 1D), and was accompanied by a 90% reduction (<100% possibly due to inadvertent contamination from cells not expressing K14Cre) in Pbx1 transcript levels (Fig. 1E) and absence of Pbx1 protein (Fig. 1F) in the corneal epithelium of mutant mice, thus validating our approach to targeting Pbx1 in this cell type. 

Figure 1.

 

Efficient deletion of Pbx1 in the corneal epithelium. (A) IHC demonstrated the presence of Pbx1b in the adult mouse cornea and (B) K14 in corneal epithelium from control (K14Cre⁺.Pbx1^f/+) and mutant (K14Cre⁺.Pbx1^f/−) mice. (C) The floxed Pbx1 allele (top) and its inactivation by tissue-specific deletion of the third Pbx1 exon (middle) mediated by expression of K14Cre. The Pbx1 constitutive ko allele is also shown (bottom). (D) PCR performed on genomic DNA demonstrates efficient tissue-specific deletion of Pbx1 in the cornea and skin, but not the liver. (E) Relative levels of Pbx1 RNA in corneal epithelium isolated from control and mutant mice as assessed by real-time PCR (±SEM, n = 6). (F) Representative IHC demonstrated the absence of Pbx1 in the corneal epithelium of mutant (K14Cre⁺.Pbx1^f/f) versus control mice, which showed specific nuclear staining of greater intensity than that of diffuse background. st, stroma; ep, epithelium.

Corneal Clouding Due to Pbx Deficiency

The mutant mice were viable and fertile and opened their eyes at P13.5 to P14.5 similar to the control animals. However, the mice displayed a conspicuous corneal clouding phenotype (Fig. 2A) that was apparent in the newborn mice and progressively worsened with age (data not shown). In contrast to the outer ocular surface of the control mice, the entire cornea of the mutant mice appeared opaque (Fig. 2B). The mutant mice exhibited normal development of skin epithelium and pelage; however, in aged mice (>6 months), hair loss and skin abrasions developed around areas of stress (whiskers); these observations will be described elsewhere. 

Figure 2.

 

Loss of Pbx1 severely perturbed corneal morphogenesis. (A, B) Representative macroscopic images of in situ (A) and enucleated (B) eyes reveal corneal cloudiness in mutant but not control mice (3 weeks old). (C–J) H&E-stained sections show the histologic features of corneas visualized by bright-field microscopy at low-power (C–F) and their respective high-power (G–J) magnifications. (K–N) Macroscopic images from a corneal fragility assay of control (K, L) and mutant (M, N) eyes after application of fluorescein to the corneas of anesthetized mice, with or without prior mechanical brushing. st: stroma, ep: epithelium.

The corneal clouding phenotype displayed 88% penetrance in K14Cre:Pbx1^f/− mice (10% in K14Cre:Pbx1^f/f mice raised the possibility that hemizygosity in stromal cells might compromise mesenchymal/epithelial interaction for maintenance of corneal transparency), but increased when Pbx1 conditional mice were crossed with either Pbx2^−/− or Pbx3^−/− mice to create mice with compound deficiencies of Pbx family proteins (Table 1), which are broadly and dynamically present in a large number of fetal and adult tissues. These data are most consistent with a quantitative model ²⁴ in which the overall dosage of Pbx proteins regulates corneal development but nevertheless reveals Pbx1 to be the main regulator of this process. 

Table 1.

 

Corneal Phenotypes Caused by Pbx Deficiencies

Genotype	Phenotype	Viability	Penetrance (%)
K14Cre Pbx1f/f	Corneal cloudiness	Normal	10%
K14Cre Pbx1−/f	Corneal cloudiness; skin problems in older mice	Normal	88%
K14Cre Pbx1f/f Pbx2+/−	Corneal cloudiness; skin problems in older mice	Normal	74%
K14Cre Pbx1−/f Pbx2+/−	Corneal cloudiness; skin problems in older mice	Normal	100%
K14Cre Pbx1f/f Pbx2−/−	Corneal cloudiness; sunken eyes	Decreased	100%
K14Cre Pbx1+/f Pbx2−/− Pbx3+/−	Corneal cloudiness; sunken eyes	Decreased	91%
K14Cre Pbx1f/f Pbx2+/− Pbx3+/−	Corneal cloudiness when eyes open	Decreased	94%
K14Cre Pbx1−/f Pbx2+/− Pbx3+/−	NA	Embryonic lethality	NA
Pbx2−/−	Normal	Normal	NA
Pbx3−/−	Normal	Neonatal lethality	NA

 Mice with the indicated Pbx genotypes were assessed for corneal phenotypes, viability, and phenotype penetrance (n > 20 for all genotypes). NA, not applicable.

Effect of Pbx1 Deficiency on the Corneal Epithelium, Stroma, and Basal Membrane

Histologic examination revealed that the epithelium was markedly thickened and contained disorganized dysplastic basal and squamous cells (Figs. 2C–J, Fig. 1F). There was also increased cellularity of the corneal stroma, disorganization of its collagen matrix, and an apparent discontinuous basal epithelial layer compared with the littermate controls, suggesting a defect in the barrier function of the cornea. There were no abnormalities in the number or morphology of keratocytes in the endothelium of the cornea, and no morphologic differences in the retina and lens, consistent with a lack of K14Cre-driven Pbx1 deletion in these tissues (Figs. 2C–J, and data not shown). 

High-resolution electron microscopy showed a striking disorganization of the epithelial and stromal layers within mutant corneas. The squamous cell layer was reduced in thickness (Fig. 3B), and its normal stratified architecture (Figs. 3A, 3D) was severely disorganized by the presence of more immature cells with less squamoid differentiation. Basal cells displayed atypical ultrastructural features and comprised proportionally more of the epithelial layer than in normal corneas (Figs. 3A, 3B). Large intercellular gaps disrupted the tight-adhesions characteristic of normal basal cells, and the basement membrane was breached by abnormal extensions of basal cells into the stroma (Figs. 3C, 3E). The highly organized structure of collagen fibers that constitutes the extracellular matrix of the stroma (Fig. 3G) and is essential for the shape, flexibility, and transparency of the cornea was markedly disrupted in Pbx1 mutants (Fig. 3F). 

Figure 3.

 

Ultrastructural abnormalities of Pbx1-deficient corneas. (A–G) Electron photomicrographs show ultrastructural features of corneal epithelium from control (A, D, G) and mutant (B, C, E, F) mice. (H) IHC demonstrated increased decorin (green) in the mutant corneal epithelium compared with the control cornea (DAPI, blue). (I) Relative levels of decorin RNA in corneal epithelium, as detected by quantitative real-time PCR (±SEM, n = 6; *P = 0.031) bc, basal cells; sc, stratified squamous cells; st, stroma; bm, bowman's membrane. Scale bars: (A, B, D) 10 μm; (C) 4 μm; (E) 2 μm; (G) 1 μm; (F) 0.2 μm.

Studies have suggested that the proteoglycan decorin accumulates in corneas with disruptions in stromal lamellae, Bowman's layer, Descemet's membrane, and stromas with increased cellularity. ²⁵ Thus, decorin levels were assessed to address a possible defect in basal cell secretion of extracellular proteins involved in the formation of basement membranes. Protein and RNA levels for decorin were significantly elevated in mutant corneal stroma (Figs. 3H, 3I) in contrast to that in control stroma. Decorin was also increased in Pbx1-null cultured mouse embryonic fibroblasts (MEFs), suggesting that decorin gene expression is directly or indirectly regulated by Pbx1 (data not shown). 

Proliferation and Apoptosis in Pbx1-Deficient Corneal Epithelium

Significant epithelial thickening suggested that the dynamic homeostasis of the cornea, which renews itself every 10 days, ^3,4 was perturbed. Therefore, proliferation status was assessed by IHC with an anti-Ki67 antibody that stains cycling cells. This assay revealed a large increase of basal cell proliferation in mutant corneas (Fig. 4A) with 14% of cells in cycle compared with 4% in control (Fig. 4B). Quantitative real-time PCR for transcripts of key cell cycle regulators showed increased levels (3.3-fold) for c-Myc, which is essential for driving cells into the cycle, ²⁶ and cyclin D1 (3.1-fold), which is necessary for cell cycle progression, in corneal epithelium of mutant mice (Fig. 4C). Conversely, RNA for p21, a cyclin dependent kinase inhibitor (CDKI), was decreased 2.9-fold, whereas p27, another CDKI, was not significantly changed (Fig 4C). These data collectively support a significant increase in cell cycling activity due to loss of Pbx1 in corneal epithelial cells. Furthermore, the rate of apoptosis, which is typically very low, ²⁷ was markedly increased (Fig. 4D) suggesting an abnormally high cell turnover in the mutant epithelium. 

Figure 4.

 

Increased cycling and apoptotic cells in Pbx1-deficient corneal epithelium. (A) IHC showed an increased number of cells containing Ki67 (green) in Pbx1-deficient versus control corneal epithelium (DAPI, blue). (B) Quantification of Ki67-positive cells in the corneal epithelium (±SEM, n = 4). (C) Relative levels of RNA for the indicated cell cycle genes in the corneal epithelium as detected by quantitative real-time PCR (±SEM, n = 5; *P = 0.063, **P = 0.041). (D) TUNEL staining showed an increased number of apoptotic cells (green) in mutant compared with control corneal epithelium (n = 5).

Defective Corneal Barrier Function Induced by Loss of Pbx1

The severe morphologic disorganization of the corneal epithelium and stroma raised the possibility of a barrier defect in Pbx1-mutant corneas, which was addressed in corneal fragility assays. ²² When applied to the cornea, fluorescein dye was unable to penetrate the epithelial barrier of control mice, either before or after slight mechanical brushing of the cornea (Figs. 2K, 2L). However, Pbx1-deficient corneas were readily penetrated by the dye even before gentle brushing (Fig. 2M), indicative of a barrier defect consistent with the observed morphologic abnormalities of the corneal epithelium. 

Disruption of the integrity of the corneal barrier can lead to an inflammatory response, which was assessed by staining for CD45, a marker for leukocytes and macrophages. Focal staining in the stroma and epithelium was detected in mutants (Fig. 5A), suggestive of an inflammatory infiltrate in mature mice, but was not observed in neonatal mice (Supplementary Fig. S1). In addition, the mutant stroma contained Bigh3, an extracellular keratoepithelin that participates in cell adhesion and differentiation as a downstream target of the TGF-β signaling pathway, possibly triggered by secondary inflammatory processes in mature mice (Fig. 5B). 

Figure 5.

 

Characterization of cells within corneas. (A) IHC demonstrated CD45 staining in Pbx1-deficient, but not wild-type, corneas. (B) The BIGH3-encoding protein keratoepithelin was present within deposits in mutant corneas. (C) Pbx1-deficient cornea lacked the specific differentiation marker K12. (D) The K10 marker for skin epithelium was absent in wild-type corneas and weakly present in the stroma of Pbx1-deficient corneas. (E) Alizarin staining within corneal deposits present in mutant mice provided evidence of calcification.

Effect of Loss of Pbx1 on the Expression of Developmental Regulators in the Corneal Epithelium

Stratification of the cornea begins at postnatal day 4, ²⁸ and corneal-type epithelial differentiation is marked by specific expression of keratin12 (K12). ²⁹ IHC showed that K12 was present in the outermost epithelial cell layer in the control corneas, but was absent in mutant corneas (Fig. 5C), indicating a defect in corneal epithelial cell differentiation. Loss of K12 in the cornea is also a feature of mice deficient in Notch1, a broad regulator of embryonic and postnatal development in multiple tissues including the cornea. ^23,30 After persistent inflammatory insult, Notch1 deficiency is associated with a change in their fate from cornea to skin epithelium, ²³ which expresses keratin 10 (K10); however, Pbx1-mutant corneas lacked K10 (Fig. 5D) as well as skin-associated appendages (Figs. 2C–J) indicating that corneal epithelium does not aberrantly differentiate into skin in the absence of Pbx1. Nevertheless, Pbx1 deficiency was associated with perturbations of Notch1 as evidenced by substantially elevated Notch1 protein and RNA levels in the mutant corneal epithelium in contrast to control corneas, which lacked Notch1 at this stage of development (Figs. 6A, 6C). Consistent with increased Notch1, mutant corneas also contained elevated levels of Pax6 (Figs. 6B, 6D), a transcription factor that is a downstream target of Notch1 signaling ³¹ and is implicated in the maintenance of corneal epithelium. ³²  

Figure 6.

 

Developmental regulators are abnormally expressed in Pbx1-deficient corneas. (A) IHC shows increased levels of Notch 1 (green) in Pbx1-deficient corneal epithelium (DAPI, blue). (B) IHC shows increased levels of Pax6 in the outer layer of Pbx1-deficient corneal epithelium. (C, D) Quantification of RNA levels for Notch1 (C) and Pax6 (D) RNA in corneal epithelium. (±SEM, n = 6 *P = 0.0017; **P = 0.095).

Corneal precursor cells have the capacity to differentiate along the adipocyte and osteogenic lineages in culture. ³³ The absence of detectable lipid deposits by oil red O staining in corneas from mutant mice suggested a lack of aberrant differentiation toward an adipocyte fate (data not shown). However, alizarin red staining revealed large regions of calcification ³⁴ in the stroma of mutant corneas (Fig. 5E), consistent with aberrant differentiation toward an osteoblast fate in the absence of Pbx1. 

Discussion

Our study provided genetic evidence that the Pbx1 transcription factor plays an important role in controlling the integrity of the cornea. Epithelia-specific deletion of Pbx1 by Cre recombinase expressed under control of the keratin14 promoter led to a progressive eye defect characterized by severe corneal opacity, thickening and degeneration of the corneal epithelium, disorganization of the stroma, and perturbation of corneal epithelial differentiation. 

Corneal Morphology and Pathology

Corneal thickening in Pbx1-mutant mice is most likely due to increased cellularity and proliferation within the stromal layer, partially caused by disruption of the basal membrane and infiltration of dysplastic basal and squamous cells into the stroma. No abnormalities were detected in the number and morphology of keratinocytes. Disruption of the basal barrier, suggested in some studies to be the mouse Bowman's barrier ^35,36 is known to induce proliferation. ³⁷ In Pbx1-mutant corneas, increased proliferation was also accompanied by elevated apoptosis. 

Inflammatory changes were also present in mature but not in neonatal mice consistent with loss of corneal barrier function. Although corneas usually lack resident lymphoreticular cells, they are capable of actively participating in an immune response. ³⁸ Other authors suggest the presence of mesenchymal-derived cells, acting as antigen-presenting cells that can serve as immune sentinels. In the cornea these include macrophages and dendritic cells, ³⁹ possible candidates in the inflammatory process marked by the presence of CD45-expressing cells in mutant corneas. 

Increased stromal deposition of keratoepithelin, a downstream target of TGF-β signaling, suggests activation of this pathway in Pbx1-mutant corneas. Corneal opacities are induced by transgenic expression of TGF-β, whereas mice with the mutation of TGF-β2 have thinner corneas. ⁴⁰ Activation of the TGF-β pathway may be secondary to inflammation or loss of Pbx1 in mutant animals may lead to de-repression of TGF-β signaling and subsequent corneal thickening. In support of the latter, alterations of the TGF-β pathway recently reported in Pbx1 deficient hematopoietic stem cells ¹⁶ suggests a functional link between these pathways. 

Corneal Cell Fate

Loss of K12 expression, a specific corneal marker, indicates aberrant differentiation of cells within Pbx1-deficient corneas. Cranial neural crest cells contribute to corneal development after their migration into the space between the superficial ectoderm and the lens epithelium, lining the inner side of the corneal epithelium. ⁴¹  

Notch signaling plays an important role in cranial neural crest formation and lateral inhibition within neural crest. ⁴² Experiments in chick-quail chimeras demonstrate that neural crest grafts derived from different regions may contribute to ocular and periocular structures. ⁴¹ Thus, the loss of corneal epithelial differentiation may be due to misspecification or failure in the processes related to the upregulation of Notch signaling in mutant corneas. Since mesenchyme-derived cells are present in the cornea, it is also possible that their differentiation occurs abnormally within keratinocytes and after elevated Notch signaling leads to the appearance of adipocytes and osteoblasts. The presence of focal calcified deposits in Pbx1-mutant corneas supports the presence of osteoblast lineage cells. 

Cross-talk between Notch1 and the TGF-β ligand BMP-2 promotes osteogenic differentiation. ⁴³ Thus, activation of the TGF-β pathway may not only participate in the inflammatory process of Pbx1^−/− corneas but also, together with Notch signaling, in the change in cell fate of the mesenchymal cells within the cornea. It remains to be clarified whether other molecules known to play a role in the molecular control of osteogenesis are also activated/induced by the absence of Pbx1. 

Pax6 is a key developmental regulator capable of inducing ectopic eye formation ^44–46 and is generally essential for morphogenesis of mesenchyme-derived tissues in the eye. High and continuous expression of Pax6 is necessary for the expression of several transcription factors, structural genes, and signaling molecules that are essential in the development of cells that derive from surface ectoderm and optic cup (e.g., lens and corneal epithelium) and subsequently for the migration of neural crest cells into the eye by inductive processes. ^47,48 In contrast, low and transient expression of Pax6 serves a cell autonomous role in the differentiation of trabecular meshwork and in formation of corneal endothelium and keratinocytes. ^47,49,50 Thus, upregulation of Pax6 expression in Pbx1 ko mice may drive differentiation of corneal cells to other cell types. 

In summary, our data indicate a role for Pbx1 in corneal morphogenesis. The absence of Pbx1 leads to corneal opacification, which reflects high epithelial cell proliferation and disorganization of the collagen matrix. Deficiency of Pbx1 is accompanied by upregulation of Notch1 and components of its signaling pathway. It remains to be demonstrated whether Pbx1 acts directly or indirectly to influence cell fate within the cornea. In addition, compromised barrier function is associated with an acquired inflammatory process in mature but not in neonatal mice, most likely driven by the TGF-β pathway. The potential roles of other pathways remain to be elucidated. 

Supplementary Materials

Supplementary Figure S1 - (PDF) 

Footnotes

 Supported by funds from National Cancer Institute Grant CA90735.

Footnotes

 Disclosure: M.J. Murphy, None; B.K. Polok, None; D.F. Schorderet, None; M.L. Cleary, None

The authors thank Maria Ambrus, Cita Nicolas, and Natacha Nanchen for technical assistance.