Expression of the α4 Integrin Subunit Gene Promoter Is Modulated by the Transcription Factor Pax-6 in Corneal Epithelial Cells

Abstract

purpose. Expression of several membrane-bound integrins is thought to be altered during corneal wound healing as a consequence of the massive secretion of fibronectin occurring during this process. Examination of the α4 integrin subunit gene promoter revealed the presence of three putative binding sites for the transcription factor Pax-6 expressed in the basal cells of the corneal epithelium during corneal wound healing. This study was undertaken to investigate whether the α4 integrin subunit is expressed in primary cultures of rabbit corneal epithelial cells (RCECs) and to test whether Pax-6 binds the α4 gene promoter and regulates its transcriptional activity.

methods. Both flow cytometry and immunocytochemical analyses, along with an antibody-directed receptor interference assay, were used to examine expression of the α4 subunit in RCECs. Expression of Pax6 was investigated by immunoblot analysis. Binding of PAX6 to the α4 gene promoter was tested in electrophoretic mobility shift assays (EMSAs). The regulatory influence exerted by Pax6 on the α4 promoter was studied by transfections in RCECs.

results. Expression of α4 was detected at both the mRNA and protein levels. Pax-6 was expressed in a cell-density–dependent manner in RCECs and altered the activity of the α4 promoter by interacting with multiple sites in both the promoter and 5′-flanking sequences. Pax-6 was also identified as the major protein component from the Bp5 complex, one of five protein complexes reported to bind the α4.1 element from the α4 basal promoter in vitro.

conclusions. These results provide evidence that the integrin subunit α4 and Pax-6 are coexpressed in RCECs and raise the possibility that Pax-6 directly regulates the expression of the α4 gene during corneal wound healing.

The corneal epithelium is a typical stratified squamous epithelium. Because of its position bordering the outside world, it is also subjected to abrasive pressure and occasional mechanical, chemical, and biological traumas. ^{1 2} Corneal wounds account for approximately 37% of all visual disabilities and 23% of medical consultations for ocular problems in North America. ³ Corneal epithelial cells adhere to the basement membrane by an integrin-mediated linkage of cytoplasmic actin filament to extracellular matrix (ECM) proteins such as fibronectin (FN), a recognized wound-healing agent required for the adhesion, migration, proliferation, and orientation of these cells. ^{4 5 6} Expression of this ECM component has been reported to be upregulated in inflammation and tissue injury. Indeed, corneal injury triggers the production of supranormal amounts of FN secreted by both the corneal basal cells and stromal keratocytes, ^{7 8 9 10 11} whose secretory activity increases during wound healing and then decreases on completion of the healing process. ^{7 12 13 14} In response to this increased secretion of FN, the basal epithelial cells flatten and migrate to cover the denuded area. ^{15 16} This process is mainly dependent on the migration and adhesion properties of the epithelial cells which, in turn, depend on the expression of membrane-bound integrins. ^{2 17}  

Integrins are a family of cell surface heterodimeric transmembrane glycoproteins composed of noncovalently associated α and β subunits. Membrane-bound integrins are involved in cell–cell and cell–ECM interactions, and many of them are known to transduce signals affecting the behavior of the cell. ^{2 6 17 18 19 20} In a recent survey of the human genome, 24 α and 9 β integrin subunits were identified ²¹ : 6 novel α- and 1 novel β-subunit relative to the previously recognized 18 α- and 8 β-subunits known to form 24 different heterodimers. However, the existence of these new integrin subunits remains to be firmly documented. Understanding the dependence of the ligand-integrin relationship in both the normal and the regenerating cornea is an important prerequisite for unraveling the mechanism behind corneal wound healing. ¹⁷ The massive increase of FN secretion that typically characterizes this process was postulated to be coordinated with the expression of its corresponding integrin receptors. ^{7 22} Surface expression of integrins has been suggested to increase during cell migration. ^{18 23} Indeed, expression of the FN-binding integrin α4β1 has been observed in epithelial cells from the cornea and has been postulated to play an important role in corneal wound repair. ^{24 25} Epithelial expression of α4β1 increases in accordance with levels of FN at the wounded surface and decreases as wound healing is completed. ²⁶  

Wound healing in the cornea, as in other tissues, is a complex process that can be modulated in different ways by the involvement of numerous mechanisms. ¹³ The expression of integrins depends on the action of various nuclear transcription factors. Of particular interest is the fact that the transcription factor Pax-6 has been found to be essential in vertebrate eye development, maintenance, and healing. ^{27 28 29 30} PAX genes are members of a family of developmental control genes that encode nuclear transcription factors involved in the regulation of early development and tissue maintenance in several adult organs. Nine paired box genes have been identified in the mouse (Pax1 to Pax9) and humans (PAX1 to PAX9). ^{27 29} Pax-6 is expressed in the developing eye, nose, pancreas, pituitary, and central nervous system. Pax-6 has several postdevelopmental roles, and its expression continues in the human lens, retina, corneal epithelium, ^{31 32} and elsewhere (e.g., brain and pancreas). Two major forms of the protein, Pax-6 and Pax-6(5a), have been reported and shown to result from an alternative splicing event. ³³ In humans, heterozygous Pax-6 mutations have been linked to aniridia, ³⁴ Peter’s anomaly, ³⁵ autosomal dominant keratitis, ³⁶ foveal hypoplasia, ³⁷ and the small eye phenotype (Sey) in mice. ³⁸  

As for both FN and the integrins, a significant upregulation of Pax-6 expression has been observed during the proliferation phase necessary to restore the stratified structure of the corneal epithelium during wound healing (Kays WT, IOVS 1997;38:ARVO Abstract 4406). The human cornea was recently shown to express equal levels of Pax-6 and Pax-6(5a); the Pax-6–Pax-6(5a) ratio was approximately five times less than that measured in lens epithelium but 20 times more than that measured in lens fibers. ³⁹ A detailed examination of the α4 gene basal promoter revealed the presence of several putative binding sites for Pax-6. In the present study, we provided evidence that the α4 integrin subunit gene is expressed in primary cultures of rabbit corneal epithelial cells (RCECs) and that its transcription is modulated through the recognition by Pax-6 of multiple target sites located along the promoter and 5′-flanking regulatory sequences of the α4 gene. 

Materials and Methods

All experiments described in this article were conducted in voluntary compliance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and all procedures were approved by the Laval University Animal Care and Use Committee. 

Cell Culture and Medium

RCECs were obtained from the central area of freshly dissected rabbit corneas, as described previously, ⁴⁰ and were grown to subconfluence (near 15% coverage of the plates), midconfluence (near 75% coverage), or full confluence (100% coverage for >48 hours) under 5% CO₂ in supplemented hormonal epithelial medium (SHEM) supplemented with 5% FBS and 20 μg/mL gentamicin. When indicated, human plasma FN (obtained as previously described) ⁴¹ was coated for 18 hours at 37°C on culture dishes at varying concentrations (1–10 μg per cm²). Coated Petri dishes were washed twice with PBS and blocked at 37°C with 2% BSA in PBS. Cells were then seeded and grown as just described. Inhibition of signal transduction induced by the MAP kinase pathway was performed by culturing midconfluent RCECs in the presence of 10 μM of the MEK/kinase inhibitor PD98059 (Sigma-Aldrich, Oakville, Ontario, Canada) for 48 hours before cells were harvested. 

Rabbit choroidal melanocytes were isolated and grown as primary cultures as follows. A circumferential incision was made below the ora-serrata and another around the optic nerve. The cornea, lens, iris, vitreous, and retina were removed. The remaining eyecup was cut in half and put in dissection saline solution (modified Hanks’ balanced salt solution [MHBS], containing 0.3 mg/mL albumin, 5 mg/mL polyvinylpyrrolidone [PVP], 10,000 U/mL penicillin, and 10 mg/mL streptomycin). Pieces of RPE-choroid-sclera were transferred to a 60-mm culture dish containing 5 mL dispase solution (MHBS containing 2.4% dispase; Roche Diagnostics, Laval, Québec, Canada), 10 mg/mL PVP, 2% chicken serum, and 88 mM CaCl₂ and incubated at 37°C for 1 hour. The choroid was then separated from the RPE at room temperature by gently spraying MHBS on the tissue. Choroidal melanocytes were obtained by further spraying dissection saline vigorously over the remaining RPE-free choroid. The melanocyte-containing saline was transferred to a centrifuge tube containing keratinocyte serum-free medium (SFM) supplemented with 5% bovine calf serum and various vitamins and proteins (200 mg/mL lipid rich bovine serum albumin [Albumax; Invitrogen-Gibco, Burlington, Ontario, Canada], 45 mg/mL ascorbic acid, 1 mg/mL carnitine, 500 mg/mL glucose, 112 mg/mL fructose, 5 mg/mL glutathione, 6 mg/mL hypoxanthine, 67 mg/mL oxalacetic acid, 0.15 mg/mL retinol acetate, 5 mg/mL taurine, 0.025 mg/mL d-α-tocopherol, 50 mg/mL transferrin, 0.3 mg/mL uridine, and 1% bovine retina extract (BRE), which was obtained as described elsewhere). ⁴² Sedimented melanocytes were washed twice, transferred to 100-mm tissue-culture dishes and maintained at 37°C under 5% CO₂ in SHEM. Culture medium was changed every 2 to 3 days. 

Flow Cytometry

RCECs, from both confluent and midconfluent primary cultures, and choroidal melanocytes, were harvested in PBS-EDTA 0.05% and washed with PBS containing 0.1% BSA and 0.05% sodium azide. Approximately 8 × 10⁵ cells were incubated on ice for 45 minutes with mAbs directed against either the α4 (ALC1/3 and HP1/2 [1:4 dilution], both provided by Francisco Sánchez-Madrid, Immunology Service, Universidad Autonoma de Madrid, Madrid, Spain) or the α5 (IIA₁ [1:50 dilution]) integrin subunit. A 1:50 dilution of a mouse monoclonal antibody (C-2-10) ⁴³ raised against the C-terminal end of the DNA binding domain of bovine poly(ADP-ribose)polymerase (PARP) was used as a negative control. After the cells were washed and labeled for 30 minutes on ice with a 1:50 dilution of a FITC-conjugated secondary antibody (Sigma-Aldrich, St-Louis, MO), cells were resuspended in 500 μL PBS and analyzed using a flow cytometer (Epics XL; Coulter Electronics, Miami, FL). 

Immunocytochemistry

RCECs (1 × 10⁵ cells at midconfluence; 5 × 10⁵ at postconfluence) were seeded into 500 μL complete SHEM in 24-well culture plates on microscope coverslips. The culture medium was changed 24 hours later and the cells were maintained in SHEM (1 mL) for 2 days. Cells on coverslips were treated with 0.3% H₂O₂ for 15 minutes to inhibit endogenous peroxidase activity. The background staining was blocked by using an avidin–biotin-blocking kit, according to the manufacturer’s protocol (Vector Laboratories, Burlingame, CA), as well as by incubating the cells for 30 minutes in 50 mM potassium PBS (KPBS) containing 5% goat serum and 0.2% Triton X-100. In the same buffer solution, the cells were then incubated for 2 hours at room temperature with an antibody directed against the α4 integrin subunit (1:50 dilution; Serotec, Raleigh, NC). A biotinylated goat anti-mouse antibody (1:400 dilution; Jackson ImmunoResearch, West Grove, PA) was then added. Incubation proceeded at room temperature for another 90 minutes, at which time antifade medium (Vectastain ABC; Vector Laboratories) was added for 1 hour. The staining was developed in nickel–diaminobenzidine solution (5% nickel ammonium sulfate, 100 mM sodium acetate, 0.5 mg/mL diaminobenzidine, 2 mg/mL β-d(+)-glucose, 0.4 mg/mL ammonium chloride, and 1 μL/mL glucose oxidase; Sigma-Aldrich) for 10 minutes. Each of the above steps was followed by four 5-minute rinses in KPBS. The coverslips were dehydrated and mounted onto slides with DPX (a mixture of distyrene, tricresyl phosphate, and xylene; Electron Microscopy Sciences, Fort Washington, PA). Control cells were treated as described, except the primary antibody was omitted. 

Plasmids and Oligonucleotides

The plasmids −1000α4-CAT, −400α4-CAT, −300α4-CAT, −200α4-CAT, −120α4-CAT, −76α4-CAT, and −41α4-CAT have been described previously. ^{44 45} They bear the chloramphenicol acetyl transferase (CAT) reporter gene from the plasmid pSKCAT fused to DNA fragments from the human α4 gene upstream regulatory sequence extending up to 5′ positions −1000, −400, −300, −200, −120, −76, and −41, respectively, but all sharing a common 3′ end located at position +15. The recombinant plasmid derivative of the parental vector −76α4-CAT that bears mutations in the α4.1 element (−76α4/4.1m) has been described before. ²⁷ The plasmid derivatives −120α4/4.1m, −400α4/4.1m, and −1000α4/4.1m that bear mutations in the α4.1 element were produced by PCR using a site-directed mutagenesis kit (QuickChange) from Stratagene (La Jolla, CA) according to the manufacturer’s instructions. The parental plasmids −120α4-CAT, −400α4-CAT, and −1000α4-CAT were used as templates. The double-stranded oligonucleotide bearing the sequence from the α4.1 regulatory element between positions –62 and –28 (5′-GATCTGTGGGGAGGAAGGAAGTGGGTATAGAAGGGTGCT-3′) or derivatives bearing mutations (in bold) into the 3′ most GTGGG element that has been used for site-directed mutagenesis (top-strand: 5′-GCAGAGGAAGTGTGGGGAGGAAGGAAAAAAATATAGAAGGGTGCTGAGATGGGG-3′; bottom strand: 5′-CCCACATCTCAGCACCCTTCTATATTTTTTTCCTTCCTCC-CCACACAACCTCTGC-3′), were chemically synthesized (Biosearch 8700 apparatus; Millipore, Bedford, MA). Double-stranded oligonucleotides bearing the DNA-binding site for the human HeLa CTF/NF-I in adenovirus type 2 (Ad2) (5′-GATCTTATTTTGGATTGAAGCCAATATGAG-3′), ⁴⁶ or high-affinity binding sites for Pax-6 (P6CON: 5′-GGATGCAATTTCACGCATGAGTGCCTCGAGGGATCCACGTCGA-3′) ⁴⁷ and Pax-6(5a) (5aCON: 5′-AATAAATCTGAACATGCTCAGTGAATGTTCATTGACTCTCGAGGTCA-3′) ³³ were also used as unlabeled competitors in electrophoretic mobility shift assays (EMSAs). Each recombinant plasmid was sequenced by chain-termination dideoxy-sequencing ⁴⁸ to confirm these mutations. 

Transient Transfection and CAT Assay

RCECs were plated at densities of 5 × 10⁴ (5%–10% confluence), 1 × 10⁵ (25%–30% confluence), 3 × 10⁵ (50%–60% confluence), 7 × 10⁵ (80%–85% confluence), and 1.5 × 10⁶ (100% confluence) cells per 35-mm well and transiently transfected using a polycationic detergent (Lipofectamine; Invitrogen-Gibco, Burlington, Ontario, Canada) as recommended by the manufacturer. When indicated, RCECs were incubated on ice for 20 minutes with increasing concentrations (25 ng to 1 μg) of an α4 integrin-specific monoclonal antibody (ALC1/3). Cells were then grown on tissue culture plates coated or not coated with FN (8 μg/mL) before they were transiently transfected 24 hours later as just described. Each transfected plate received 1 μg of the α4-CAT test plasmid, 1 μg of the human growth hormone (hGH)–encoding plasmid pXGH5, ⁴⁹ and 1 μg of either the Pax-6 expression plasmid pKW10 ⁵⁰ or an empty vector (pBluescript II KS; Stratagene). Levels of CAT activity for all transfected cells were determined as described, ⁵¹ normalized to the amount of hGH secreted into the culture medium, and assayed using a kit for quantitative measurement of hGH (Immunocorp, Montréal, Québec, Canada). The value presented for each individual test plasmid transfected corresponds to the mean of at least three separate transfections performed in triplicate. To be considered significant, each value had to be at least three times the value of the background level caused by the reaction buffer used (usually corresponding to 0.15% chloramphenicol conversion). To compare the groups, Student’s t-test was performed. Differences were considered to be statistically significant at P < 0.005. All data are expressed as the mean ± SD. 

Nuclear Extracts and Recombinant Pax-6

Crude nuclear extracts were prepared from both confluent and midconfluent RCECs and dialyzed against DNase I buffer (50 mM KCl, 4 mM MgCl₂, 20 mM K₃PO₄ [pH 7.4], 1 mM β-mercaptoethanol, 20% glycerol), as described. ⁵² Extracts were kept frozen in small aliquots at −80°C until use. Preparation of crude extracts from lens tissues reported to express Pax-6 (chicken embryonic lens, mouse αTN4-1 lens cells) has been previously described. ⁵³ The truncated human Pax-6 protein (designated GST-Pax-6(PD/HD)) bearing both the paired domain (PD) and the homeodomain (HD; amino acids 1-285) fused to glutathione-S-transferase (GST) was expressed in Escherichia coli and purified as described elsewhere. ⁵⁴  

SDS-Polyacrylamide Gel Fractionation of Nuclear Proteins

Crude nuclear proteins (75 μg) from both confluent and midconfluent RCECs were added to 2× loading buffer (62.5 mM Tris-HCl [pH 6.8], 2% SDS, 10% glycerol, 695 mM β-mercaptoethanol, 0.0013% bromophenol blue) and gel fractionated as described. ^{55 56} Approximately 10 μL from each eluted fraction was incubated with α4.1 labeled oligomer (3 × 10⁴ cpm) in the presence of 250 ng poly(dI-dC) · poly(dI-dC) in buffer D. The formation of DNA/protein complexes was analyzed by EMSA. 

Electrophoretic Mobility Shift Assays

EMSAs were performed with the radioactively labeled 35-bp α4.1 probe. Approximately 2 × 10⁴ cpm labeled DNA was incubated with crude nuclear proteins (1 μg) from either chicken embryonic lens or mouse αTN4-1 lens cells in the presence of 500 ng poly(dI-dC) · poly(dI-dC) in buffer D (5 mM HEPES [pH 7.9], and 10% glycerol [vol/vol], 25 mM KCl, 0.05 mM EDTA, 0.5 mM dithiothreitol, 0.125 mM PhMeSO₂F). Incubation proceeded at room temperature for 10 minutes at which time DNA-protein complexes were separated by gel electrophoresis through 6% native polyacrylamide gels run against Tris-glycine buffer, as described. ⁵² Gels were dried and autoradiographed at −80°C to reveal the position of the shifted DNA-protein complexes generated. EMSAs in the presence of antibodies were conducted by first incubating 1 μg crude nuclear proteins from the lens nuclear extracts in the presence of 250 ng poly(dI-dC) · poly(dI-dC), either alone or in the presence of 2 μL of an affinity-purified antiserum raised against human Pax-6 (afP6 antiserum), in buffer D. As a negative control, 2 μL of a nonimmune serum was also incubated with the lens nuclear proteins. Then, the α4.1-labeled probe was added and incubation was extended for another 15 minutes at room temperature. Samples were finally loaded onto polyacrylamide gels. 

SDS-PAGE and Western Blot

Crude nuclear proteins were obtained from RCECs grown at both confluence and midconfluence, as detailed earlier. Protein concentration was determined by Bradford procedure, which was further validated after Coomassie blue staining of SDS-polyacrylamide fractionated nuclear proteins. One volume of sample buffer (6 M urea, 63 mM Tris [pH 6.8], 10% [vol/vol] glycerol, 1% SDS, 0.00125% [wt/vol] bromophenol blue, and 300 mM β-mercaptoethanol) was added to 20 μg proteins before they were size fractionated on a 10% SDS-polyacrylamide minigel and transferred onto a nitrocellulose filter. As the positive control, chicken embryonic lens, and αTN4-1 lens cells expressing endogenous Pax-6 proteins were used. The blot was then washed once in TS buffer (150 mM NaCl and 10 mM Tris-HCl [pH 7.4]) and four times (5 minutes each at 22°C) in TSM buffer (TS buffer plus 5% [wt/vol] fat-free powdered milk and 0.1% Tween 20). Next, a 1:500 dilution of a Pax-6-specific (afP6 antiserum) was added to the membrane containing TSM buffer and incubation proceeded further for 4 hours at 22°C. The blot was then washed in TSM buffer and incubated an additional 1 hour at 22°C in a 1:1000 dilution of a peroxidase-conjugated goat anti-mouse immunoglobulin G (Jackson ImmunoResearch Laboratory). The membrane was successively washed in TSM buffers (four times, 5 minutes each) and TS (twice, 5 minutes each) before immunoreactive complexes were revealed using Western blot chemiluminescence reagents (Renaissance; NEN Dupont, Boston, MA) and autoradiographed. 

RT-PCR Analyses

Total RNA was isolated from midconfluent and 2-day and 5-day postconfluent RCECs (Tri reagent; Molecular Research Center, Inc., Cincinnati, OH), and reverse transcribed using a kit (Superscript II Transcriptase; Invitrogen-Gibco). Briefly, 10 μg total RNA was incubated in the presence of 10 mM dNTPs, 3 μg oligo dT primer, 15 mM DTT, and 1 μL RNAguard (Pharmacia-LKB Biotechnology, Pleasant Hill, CA), in first-strand buffer (Superscript II; Invitrogen-Gibco). Reverse transcriptase was added (2 μL of 200 U/μL), and incubation proceeded at 37°C for 90 minutes at which time newly synthesized first-strand cDNAs were column purified (QIAquick nucleotide removal kit; Qiagen, Santa Clarita, CA) and used for PCR amplification of both the α4 and the 18S ribosomal RNAs. The sequence of the template primers used for the amplification of the α4 transcript were 5′-ATGCTGCAAGATTTGGGAA-3′ (sense) and 5′-GCACCAATGCTACATCTAC-3′ (antisense). The oligonucleotide primers used for the amplification of the 18S ribosomal RNA were provided in an internal standards kit (Quantum RNA 18S Internal Standards kit; Ambion Inc., Austin, TX). Taq polymerase (Pharmacia-LKB) was selected for PCR amplification. Cycle parameters were the same for all primers used (denaturation 95°C, 30 seconds; annealing 56°C, 30 seconds; extension 72°C, 30 seconds) with an identical number of cycles (26, 28, 30, 32, 34, and 36 cycles) for both sets of primers. Both the α4 transcript and the 18S rRNA were coamplified from the same mixture of cDNAs. The PCR products were analyzed on a 1.5% agarose gel. The image was scanned (Visage 110S Bioimage analyzer; Millipore) to quantify differential gene expression at various cell density. 

Results

Expression of the α4 Integrin Subunit in RCECs

To clarify whether the FN-binding integrin subunit α4 is expressed in cultured corneal epithelial cells in vitro, flow cytometric analyses were conducted on both confluent and midconfluent RCECs. A recent study conducted by Kamata et al. ⁵⁷ has shown that most anti-human α4 or anti-mouse α4 mAbs recognize only human and mouse α4, respectively, even though both α4 subunits share at least an 85% identity. We therefore obtained several α4-antibodies, ⁵⁸ of which some were reported to recognize rabbit α4. Indeed, when used in the flow cytometry analysis, both the ALC1/3 (Fig. 1A) and HP1/2 (data not shown) antibodies revealed a low but clearly detectable expression of α4 in confluent, but not in midconfluent, RCECs. ALC1/3, however, yielded a higher relative fluorescent intensity and proved to be the most efficient of both antibodies. The ALC1/3 Ab also revealed expression of α4 on primary cultured rabbit choroidal melanocytes, which were used as a positive control. As a control for integrin expression in corneal epithelial cells, high levels of α5 expression were detected by flow cytometry in both confluent and midconfluent RCECs and in rabbit choroidal melanocytes (Fig. 1A , MEL). We then used a sensitive immunoperoxidase assay that involves the use of a biotinylated secondary antibody in conjunction with an avidin-conjugated peroxidase to confirm expression of α4 in RCECs. Expression of the α4 subunit was observed in postconfluent cells, although low levels were also expressed in midconfluent RCECs (Fig. 1B) . No such staining was observed when primary Ab was omitted (which has been used as a negative control; Fig. 1B , Ctl). Although both mid- and postconfluent RCECs had a patchy pattern of α4 expression, there was a clear tendency for this integrin subunit to accumulate at cell–cell contacts (Fig. 1B , arrows). 

Because the proliferative state of any particular cell type changes as a result of altered gene expression, we used semiquantitative RT-PCR analyses to investigate whether transcription of the α4 integrin subunit gene is under the influence of cell density in primary cultured RCECs (Fig. 1C) . PCR amplifications were performed on reverse-transcribed total RNA obtained from both midconfluent and postconfluent (2 and 5 days after confluence) RCECs. Coamplification of the 18S ribosomal RNA was performed for normalization purposes. The specific α4 PCR product, which appeared as a single DNA fragment of the expected size of 265 bp using RNA prepared from both midconfluent and 2-day postconfluent RCECs appeared visible on the gel after 30 cycles of amplification and remained linear for up to 34 cycles (Fig. 1C) . However, a significant reduction of product was observed in 2-day postconfluent cells. Moreover, nearly no signal corresponding to the α4 PCR product could be detected in 5-day postconfluent cells. Normalization of the 265-bp α4 band to that of the 489-bp product of the 18S rRNA in both midconfluent and postconfluent RCECs provided evidence that the amount of α4 transcript is, on average, 1.7 ± 0.1 and 4.9 ± 0.4 times lower in 2- and 5-day postconfluent RCECs than in midconfluent cells, respectively. We therefore conclude that the α4 integrin subunit is indeed expressed in both mid- and postconfluent RCECs and that transcription of the α4 gene is subject to a cell-density–dependent regulatory process. 

Effect of Cell Density on Transcriptional Activity Directed by the α4 Promoter

Because the results from semiquantitative RT-PCR analyses suggested the transcription of the α4 gene to be altered by cell density, we investigated whether the activity directed by the human α4 gene promoter is similarly affected on transient transfection of both mid- and postconfluent RCECs. Recombinant constructs bearing the CAT reporter gene under the control of various α4 gene promoter fragments were transfected into RCECs cultured at various cell densities. We had demonstrated the usefulness of using both mid- and postconfluent RCECs as a model to mimic wound healing. ⁵² As Figure 2B indicates, the basal promoter of the α4 gene −76α4-CAT construct, ⁴⁵ was sufficient to yield substantial CAT activity in midconfluent RCECs. The maximum α4 promoter activity was observed, however, when the 5′ end was extended up to position −120. Extending the 5′ end further up to position −1000 (in plasmid −1000α4-CAT) resulted in a near threefold reduction of α4 promoter activity compared with the plasmid −120α4-CAT, as previously reported. ⁴⁵ When RCECs were transiently transfected at confluence, the activity of the α4 promoter dramatically diminished by 9-, 90-, 37-, and 12-fold for the α4/CAT recombinant constructs −76α4-CAT, −120α4-CAT, −200α4-CAT, and −1000α4-CAT, respectively (Fig. 2B) . Such a dramatic reduction of CAT activities in postconfluent RCECs did not result from the reduced ability of such cells to be efficiently transfected relative to midconfluent cells. High amounts of hGH, used to normalize transfection data, were also secreted into the culture medium by postconfluent cells. On average, hGH was secreted into a concentration of approximately 450 ± 55 ng/mL for subconfluent cells and 115 ± 24 ng/mL for 48-hour postconfluent RCECs. The background caused by the culture medium corresponded to 0.01 ng/mL. 

Regulation of the α4 Integrin Subunit Gene Promoter Function by Fibronectin

As stated earlier, FN secretion is dramatically increased during corneal wound healing. We recently provided evidence that transcription of the α5 integrin subunit gene promoter increases in a dose-dependent manner when RCECs are grown in the presence of FN. ⁵² The ability of the α4 promoter to respond to various concentrations of FN was therefore investigated. RCECs were grown to midconfluence on tissue culture plates coated with increasing concentrations of FN. The recombinant plasmid −1000α4-CAT, which harbors the longest α4 promoter segment, was selected for this experiment. The α4 promoter activity increased in a dose-dependent manner as a consequence of growing RCECs on FN-coated culture plates (Fig. 3A) . The maximum FN responsiveness, measured as the ratio of CAT activity in RCECs cultured on FN over that measured in cells grown solely on plastic, was reached (5.9-fold increase) when RCECs were grown in the presence of 10 μg/cm² FN. 

To demonstrate further that FN responsiveness of the α4 promoter is mediated through the binding of FN to the α4β1 integrin, an antibody-directed receptor interference assay was conducted. In this experiment, RCECs were first incubated in the presence of increasing concentrations of the blocking anti-α4 Ab ALC1/3, then plated on FN-coated culture plates (8 μg/cm²) and grown to midconfluence. A threefold increase in α4 promoter function was observed when RCECs were cultured in the presence of FN (Fig. 3B) . However, exposing the cells to increasing concentrations of ALC1/3 Ab dramatically impaired the FN responsiveness of the α4 promoter. Indeed, as little as 200 ng was sufficient to prevent completely the FN-mediated activation of the α4 promoter (Fig. 3B) . The use of an unrelated monoclonal antibody (C-2-10) raised against bovine PARP, a nuclear protein absent from the cell surface, ⁴³ was totally ineffective in preventing the FN-mediated responsiveness of the α4 promoter (Fig. 3B , bottom), which therefore provides evidence for the specificity of the α4 promoter induction. 

Culturing murine Swiss 3T3 or rat REF52 fibroblasts on surfaces coated with either FN or with a synthetic peptide containing the RGD sequence has been shown to result in the activation of mitogen activated protein kinases (MAPKs), ⁵⁹ which are recruited to the ECM ligand/integrin-binding site. ⁶⁰ Recently, we demonstrated that binding of FN to its α5β1 integrin triggered hyperphosphorylation of the positive transcription factor Sp1 through activation of the MAPK pathway. ⁵² As transcription of the α5 integrin subunit gene is partly controlled by Sp1, culturing RCECs on FN-coated culture dishes resulted in increased α5 promoter activity in vitro. To determine whether the FN responsiveness directed by the α4 promoter is due to the activation of the ras-Erk signaling pathway, ⁶¹ RCECs grown to midconfluence on FN-coated culture dishes (8 μg/cm²) or not treated with FN were transfected with either the plasmid –1000α4-CAT or the α5-92 plasmid bearing the basal promoter from the human α5 gene upstream from the CAT reporter gene. ⁵² The experiment was also conducted with RCECs grown on untreated culture dishes. Cells were grown in the absence or presence of 10 μM of the MEK 1 kinase inhibitor PD98059. As expected, CAT activity directed by the transfected plasmids –1000α4-CAT and α5-92 was strongly increased (5.5- and 9.9-fold increases, respectively), when RCECs were cultured on FN-coated dishes (Fig. 3C) . The addition of 10 μM of the PD98059 inhibitor totally abolished the FN responsiveness directed by the α5 promoter, but had no influence at all on the FN responsiveness directed by the α4 promoter. We therefore conclude that the transcriptional activity directed by the α4 promoter is positively regulated in RCECs grown on FN-coated culture plates. However, the signal transduction pathway induced by the binding of FN to the α4β1 integrin that accounts for the increase in α4 promoter activity is clearly distinct from that induced by the binding of FN to the α5β1 integrin, as inhibition of the MAPK pathway did not suppress any positive regulatory influence of FN on the α4 promoter. 

Influence of the Transcription Factor Pax-6 on Activity Directed by the α4 Promoter

As Pax-6 expression increases during corneal wound healing in the basal cells of the corneal epithelium (Kays WT, IOVS 1997;38:ARVO Abstract S950), we investigated whether expression of the α4 subunit gene might be under the regulatory influence of Pax-6 during this process. A detailed examination of the α4 basal promoter revealed the presence of two putative target sites for both the PAI and RED subdomains of Pax-6. In addition, these two regions were also found to share homologies with the Pax-6 target site recently identified in the matrix metalloproteinase (MMP)-9 promoter ^{62 63} (Fig. 4) . These two preserved regions are contained within one short regulatory sequence, the α4.1 element, that was previously shown to be essential for proper transcription directed by the α4 basal promoter. ⁴⁶ The α4.1 element was also reported to bind nuclear proteins from RCECs that yield five different DNA-protein complexes in EMSA (Bp1 to Bp5). ⁵⁵  

A series of EMSAs were conducted to determine whether Pax-6 indeed interacts with the α4.1 element from the α4 promoter. Incubation of a labeled DNA probe corresponding to the α4.1 element with nuclear extracts obtained from mouse αTN4-1 cells or from whole chicken embryonic lens containing endogenous Pax-6 yielded formation of two specific DNA-protein complexes (Fig. 5A , complexes A and B). Addition of an affinity-purified antiserum (afP6 antiserum) raised against human Pax-6 to the embryonic chicken lens nuclear extract completely prevented formation of these complexes (Fig. 5B) . No alterations in the formation of complexes A and B were observed when a nonimmune serum (NIS) was added as a negative control. In addition, EMSAs were repeated using recombinant human Pax-6 protein (GST-Pax-6(PD/HD)) that contained both the paired domain (PD) and the homeodomain (HD) (amino acids 1-285) fused to glutathione-S-transferase (GST). ⁵⁴ Incubation of the α4.1-labeled probe with GST-Pax-6(PD/HD) revealed the formation of a DNA-protein complex (Fig. 5C , Pax-6). Formation of this complex was abolished in the presence of excess of unlabeled oligonucleotides containing high-affinity binding sites for Pax-6 (P6CON) ⁴⁷ and Pax-6(5a) (P5aCON). ³³ In contrast, an oligonucleotide corresponding to a binding site for the unrelated transcription factor NF1 could not compete this complex. We therefore conclude that Pax-6 indeed possesses the ability to interact with the α4.1 element from the α4 gene promoter in vitro. 

To determine whether Pax-6 exerts any regulatory influence on the activity directed by the α4 promoter, cotransfection experiments were conducted in RCECs. Cells were plated at various densities (1 × 10⁵ to 1.5 × 10⁶ cells per 30-mm culture well) and transfected with the α4 promoter-bearing plasmid −76α4-CAT or −1000α4-CAT, either alone or cotransfected with a Pax-6 expression plasmid. ⁵⁰ Pax-6 induced the activity of the α4 basal promoter in the plasmid −76α4-CAT by five- to ninefold. No statistically significant differences were observed between the cell densities (Fig. 6A , filled columns). A similar pattern of Pax-6–mediated activation of the α4 promoter was observed with the plasmid −1000α4-CAT (Fig. 6A , unfilled columns). 

To define more precisely the location of the α4 promoter sequences necessary for transcriptional activation by Pax-6, truncated plasmids derived from the α4 promoter were cotransfected into midconfluent RCECs (5 × 10⁵ cells per 30-mm culture well), either alone or with the Pax-6 expression plasmid. The activity directed by the −41α4-CAT plasmid, lacking the previously characterized α4.1 element, was not altered by Pax-6, as shown in Figure 6B . However, extension of the α4 promoter from position –41 up to position –76 in plasmid −76α4-CAT yielded a fourfold increase in CAT activity on cotransfection with the Pax-6 expression plasmid. Of note, the positive influence of Pax-6 on the –76 basal α4 promoter completely reverted to a fivefold repression, when the α4 promoter was extended to position –120. This negative influence of Pax-6 on the α4 promoter was maintained in plasmid –300α4-CAT. However, Pax-6 inducibility was recovered in plasmid –400α4-CAT, which extended up to α4 position –400, and was also maintained in plasmid –1000α4-CAT. 

A detailed examination of the α4.1 element revealed that the sequence GTGGG between positions –42 and –46 is also found in the well-characterized promoter of the gelB gene (Fig. 4) . Using site-directed mutagenesis, we changed the α4.1 GTGGG sequence to AAAAA ⁴⁵ on the background of the −76α4-CAT, −120α4-CAT, −200α4-CAT, and −1000α4-CAT plasmids. To determine whether mutations in the GTGGG element would prevent Pax-6 inducibility, both the wild-type and mutated plasmids were transfected in midconfluent RCECs, either alone or with the Pax-6 expression plasmid. A near fivefold increase in CAT activity was observed when the wild-type −76α4-CAT was transfected along with the Pax-6 expression plasmid. However, Pax-6 induction of the α4 promoter activity was abolished when the GTGGG element was mutated. In contrast, when the same element was mutated in the context of the −120α4-CAT plasmid (in −120α4/ 4.1m), the Pax-6–mediated repression (see Fig. 6B ) turned into a near threefold increase in the activity directed by the α4 promoter (Fig. 6C) . This observation raised the possibility for the presence of another Pax-6–binding site between positions –76 and –120. Mutation of the α4.1/Pax-6–binding site in the plasmid400α4/4.1m was not significantly different from that encoded by both the intact parental plasmid400α4-CAT and the mutated −120α4/4.1m plasmid, indicating that the detrimental influence of the mutations in the proximal α4.1 Pax-6–binding site are counteracted by the presence of the second –76/−120 Pax-6 target site. When the α4.1/Pax-6–binding site was mutated in the background of the parental plasmid −1000α4-CAT (to yield the mutant −1000α4/4.1m), a substantial increase (from 3.4 ± 0.7 to 8.8 ± 2.0) in the ability of Pax-6 to positively influence α4 promoter-mediated transcription was observed. These results suggest the presence of a third Pax-6 target site between positions –400 and –1000. They also suggest that binding of Pax-6 to the proximal α4.1/Pax-6–binding site influences function of the more distal Pax-6–responsive sites. 

Effect of Cell Density on Expression of Pax-6 in RCECs

Expression of Pax-6 has been shown to be altered during the proliferative phase necessary to restore the stratified structure of the wounded corneal epithelium (Kays WT, IOVS 1997;38:ARVO Abstract S950). As further evidence that Pax-6 is expressed in RCECs, Western blot analyses were conducted on nuclear extracts obtained from both confluent and midconfluent RCECs. Crude lens nuclear extracts were included as the positive control. As shown in Figure 7A , both the 46-kDa Pax-6 and 48-kDa Pax-6(5a) proteins, which migrated as a doublet (Pax-6 in the figure), could be detected in all lens-derived control extracts. The same Pax-6/Pax-6(5a) protein doublet was also detected in the extract from midconfluent (MC) RCECs. However, neither Pax-6 nor Pax-6(5a) was observed in the extract prepared from confluent RCECs when equal amounts of proteins were used (Fig. 7A) . The lack of both Pax-6 and Pax-6(5a) in confluent cells did not result from degradation of the proteins contained in the extract, as Coomassie-blue staining of SDS-gel fractionated proteins from both confluent and midconfluent cells yielded a similar pattern of nuclear proteins with a normal content in high molecular mass proteins (Fig. 7B) . We therefore conclude that both Pax-6 and Pax-6(5a) are expressed in nearly equal amounts in midconfluent, but not in confluent, RCECs. 

Pax-6 in the Bp5 Protein Complex Binding to the α4.1 Element

The α4.1 element from the α4 promoter was shown to form at least five distinct complexes involving nuclear proteins. These proteins are expressed in midconfluent RCECs, and their estimated molecular masses range from 39 to 91 kDa. The protein complexes were designated Bp1 to Bp5. ⁴⁵ The proteins yielding both the Bp4 and Bp5 complexes had molecular masses similar to those reported for both Pax-6 proteins. ³³ Because expression of Pax-6 was shown to decrease when RCECs reach a high cell density (see Fig. 7 ), expression of the Bp1 to Bp5 proteins was examined in nuclear extracts from RCECs cultured at various cell densities. Equal amounts of nuclear proteins from both confluent and midconfluent cells were first SDS-gel fractionated after a protein recovery procedure. ⁴⁵ Next, proteins from each fraction were tested for their ability to bind the α4.1-labeled probe in an EMSA. Examination of the fractions 3 to 7 was omitted, because their corresponding proteins were unlikely to be Pax-6 proteins. Nuclear proteins from fractions containing Bp2 to Bp5 activities were observed in midconfluent RCECs (Fig. 8A) . Of the Bp2 to Bp5 proteins, only formation of Bp4 and Bp5 was significantly reduced when RCECs reached postconfluence (compare the results between Figs. 8A and 8B ). Expression of Bp2 remained unaltered, whereas that of Bp3 was increased when cells reached postconfluence. Only the Bp5 protein, but not any of the other Bp proteins, including Bp4, had a pattern of electrophoretic mobility that matched that of the Pax-6/Pax-6(5a) proteins contained in the chicken embryonic lens extract in EMSA (Fig. 9A) . To verify whether Bp5 indeed corresponds to Pax-6/Pax-6(5a), EMSAs were conducted in the presence of the Pax-6 antiserum used in Figure 5B . Nuclear proteins from the fractions shown to support Bp5 binding in both confluent and midconfluent RCECs (fraction 15) were incubated with the α4.1-labeled probe in the presence or absence of 2 μL of the Pax-6 antiserum. The protein fraction 11 that supports Bp2 binding, which has a molecular mass clearly distinct from that of Pax-6 and Pax-6(5a) and whose formation was not significantly altered by cell density, was also used as a negative control. As shown on Figure 9B , addition of the Pax-6 antiserum did not alter the formation of the Bp2/α4.1 DNA-protein complex from either confluent or midconfluent RCECs. In contrast, formation of the Bp5 complex from midconfluent RCEC nuclear extracts was strongly impaired by the Pax-6 antiserum, whereas no such change was observed using the extract from postconfluent cells. This result is consistent with the lack of any Pax-6 protein detectable by Western blot in the extract from postconfluent RCECs. We conclude that the Bp5 complex formed with the α4.1 probe contains Pax-6 proteins present in midconfluent RCECs. 

Discussion

Injury to the corneal epithelium results in the massive secretion of FN very shortly after corneal tissue damage has occurred. Expression of specific integrins, notably integrin subunits α6, α9, and β4, ^{64 65 66 67} is affected by the wounding process. The present study was designed to investigate the variations that may exist in the level of expression of the FN-binding integrin subunit α4 at the promoter level in a cell culture model replicating some aspects of corneal wound healing. Both flow cytometry and immunoperoxidase analyses provided evidence that α4 is indeed expressed at low levels in both mid- and postconfluent RCECs. Because transcriptional activity directed by the α4 promoter responded to the presence of FN when coated on the culture plates, we used an antibody-directed receptor interference assay to provide additional evidence that the α4 integrin subunit is expressed in RCECs. FN responsiveness of the α4 promoter was abolished by saturating the receptor with the α4 integrin-specific antibody ALC1/3, establishing that RCECs indeed express this integrin subunit, although at a low level. These results were further validated by semiquantitative RT-PCR analyses, which also showed moderate levels of expression of the α4 transcript in actively proliferative RCECs but not in quiescent (5-day postconfluent) cells. The fact that little change is observed at 2 days after confluence is consistent with results we recently reported on RCECs using this promoter as a control. ⁶⁸ It also suggests that unlike the α5 integrin gene, transcription of the α4 gene is clearly not under the control of the transcription factors Sp1 and -3, both of which nearly disappear in 2-day postconfluent RCECs. Overexpression of Sp1 in the Sp1-deficient Drosophila Schneider cells results in a dramatic increase in α5 promoter activity, ⁶⁸ whereas it had no influence on such activity directed by the α4 promoter in similar experiments (data not shown). Of particular interest is the fact that binding of FN to α4β1 was shown to trigger regulatory signals activated through a transduction pathway distinct from the MAPK pathway, as the addition of the MEK kinase inhibitor PD98059 did not abolish the FN responsiveness of the α4 promoter. Consistent with our results, both the α4β1 and α5β1 integrins were recently shown to use distinct signaling pathways to control focal adhesion formation and cell migration in the A375-SM melanoma cell line. ⁶⁹ Although RCECs apparently express only low levels of α4β1, it is likely that the presence of even minute amounts of this integrin at the cell surface triggers a physiological response from the cell when exposed to its ligand FN, as is suggested by the experiments described herein. 

The transcriptional activity of the α4 integrin promoter has been shown to be strongly modulated by RCEC cell density. Indeed, α4 promoter activity was strongly repressed when RCECs reached confluence. As for the α4 integrin subunit, we found that the expression of the transcription factor Pax-6 was also similarly modulated by cell density in RCECs. The lack of Pax-6 expression that we observed in confluent, nonproliferative RCECs corresponds with the reduction in Pax-6 expression in unwounded rabbit corneal epithelium. ⁶² Activation of the proliferative and migrating properties of RCECs by plating them sparsely on the culture plate is frequently used as a model to mimic wound healing. ^{45 52 55} Under such conditions, we found that expression of both Pax-6/Pax-6(5a) was dramatically increased in proliferative cells at the protein level, which is also consistent with the increased Pax-6 activity reported to occur at the migrating edge of corneal wounds in vivo. ⁶² Equal amounts of both Pax-6 and Pax-6(5a) were detected by Western blot analyses in midconfluent RCECs. The human cornea also express equal ratios of Pax-6/Pax-6(5a). ³⁹  

Tissue repair and remodeling is a delicate process in which a transient change in the composition of cell adhesion molecules may be critical for cells to progress through wound healing. The cells have to respond to this process at the level of transcription by altering the pattern and activities of several transcription factors, such as Pax-6. These factors are in turn required for transient changes in the expression of cell adhesion molecules and their receptors. Pax-6 was shown to alter the expression of the human α4 gene promoter in cotransfection experiments conducted in RCECs grown at different cell densities by interacting with at least three distinct regulatory elements of the α4 integrin gene. Our data demonstrate direct binding of Pax-6 to the α4.1 proximal element of the α4 gene promoter. This regulatory element was shown to bind to several nuclear proteins from RCECs that yielded five distinct DNA-protein complexes designated Bp1 to Bp5. ⁵⁵ Pax-6 was recognized as a major protein component from the Bp5 complex for the following reasons: (1) Pax-6 possessed the ability to bind the α4.1 element in EMSA and exhibited the same electrophoretic mobility as Bp5 in native polyacrylamide gels; (2) the protein(s) yielding Bp5 had molecular masses that match those described for both Pax-6 and Pax-6(5a); (3) the Bp5 complex was substantially reduced when nuclear extracts were prepared from confluent RCECs, which is also consistent with the lack of any detectable Pax-6 in confluent RCECs; and (4) addition of the Pax-6–specific antiserum to the Bp5 complex significantly reduced its formation. It is likely that the proteins yielding the Bp1 to Bp5 complexes compete with each other for the availability of the α4.1 target site. A similar mode of regulation has been reported recently for the expression of the keratin K3 gene in the cornea. Indeed, the ratio between Sp1 and the transcription factor AP-2 has been found to be critical in transcriptional activation and repression of the K3 gene in RCECs. ⁷⁰  

Collectively, our data suggest the presence of al least three binding sites for Pax-6 in approximately 1 kb of 5′-flanking sequence of the α4 gene. The most proximal Pax-6–binding site is located 45 bp upstream from the α4 mRNA start site, the α4.1 element, whereas the two remaining sites are located somewhere between positions –76/−120 and –400/−1000. Indeed, a detailed examination of the DNA sequence between positions –76 and –120 revealed the presence of a putative Pax-6–binding site (−96… cTGcTCACGCATGcA… −82) related to the P6CON binding site. ⁴⁸ A similar candidate Pax-6–binding site (−953… AGGTTCACcaTgtATT… −936) is predicted in the –400/−1000 segment of the α4 promoter. Our study showed binding of Pax-6 proteins in vitro to the proximal Pax-6–binding site. Whether Pax-6 binds to the two more distal predicted sites remains to be determined. 

The results in a series of cotransfections showed that the individual Pax-6–binding sites functionally interacted with each other. A site-directed mutagenesis of the proximal binding sites resulted in the loss of Pax-6–mediated activation in the absence of the distal Pax-6-responsive sites. However, an extended promoter region including two presumptive Pax-6–binding sites responded to Pax-6 activation differently. The wild-type region extending to position –120 was repressed fivefold in the presence of Pax-6. In contrast, mutagenesis of the proximal Pax-6 binding site with the intact distal site resulted in threefold activation. This pattern was maintained with the promoter fragment extending to the position of –1000 and harboring another candidate Pax-6–binding site. Although the mechanism of these effects is not presently known, the possibility that Pax-6 can act as transcriptional activator and repressor is documented by both transcriptional studies ^{53 71 72 73} and studies of differential gene expression in mouse lens. ^{74 75} Two adjacent Pax-6–binding sites acting in tandem as an activator and a repressor have been shown in the chicken δ1-crystallin enhancer. ⁷¹  

To understand the regulatory functions of Pax6 in the transcriptional regulation of the α4-integrin promoter at the molecular level, identification of cis elements and their cognate transcription factors is needed, followed by studies on protein–protein interactions of Pax6 proteins. Pax-6 has been shown to interact physically with several transcription factors through direct protein–protein interactions, including homeodomain-containing transcription factors, ^{76 77 78} CBP/p300 cofactors, ⁷⁹ members of the tumor-suppressor retinoblastoma protein (pRB), ⁵⁴ and Mitf, a basic helix-loop-helix leucine zipper transcription factor essential for proper development of the retinal pigment epithelium. ⁸⁰ Both DNA-binding domains of Pax-6 were shown to interact with the b-HLH-LZ domain of Mitf, such interaction strongly reducing the transactivating properties of Pax-6. ⁸⁰ A near perfect Mitf target site (CATGCG instead of CATGTG) is located on the –76/−120 α4 promoter segment between positions –84 to –89, a position that also overlaps the putative Pax-6–binding site from –82 to −96. It is tempting to speculate that Mitf may interact with its –84/−89 α4 promoter segment and then prevent Pax-6 from interacting with either its α4.1 or –82/−96 target site or both, through direct protein–protein interactions. However, expression of Mitf in RCECs has yet to be demonstrated. Other transcription factors reported to regulate expression of the α4 gene, such as the transcriptional repressor ZEB, which binds to two distinct target sites along the α4 promoter between positions –300 and –400, ⁸¹ or members of the Ets family, ⁸² may all contribute to the switch in the regulatory function mediated by Pax-6 in RCECs. 

The present data add a novel gene into the growing list of genes directly regulated by Pax-6. Genes known to be regulated by Pax-6 in the corneal epithelial cells include gelatinase B and keratin K12. ^{62 63 83 84} The identification of the α4 integrin as a direct Pax-6 target gene in the cornea is consistent with the earliest hypothesis that one of the major functions of Pax-6 is to control cell adhesion. ⁸⁵ Our study also indicates that Pax-6, apart from its essential regulatory functions during the development and maintenance of ocular tissues, may also play an important role in tissue repair by its stress-mediated induction caused by the corneal wound. 

 Supported by Canadian Institutes of Health Research Grant 37946 (SLG), the Vision Health Network of Fonds de la Recherche en Santé du Québec (FRSQ), and National Eye Institute Grant EY12200 (AC). AC was supported by a Career Development Award from the Research to Prevent Blindness, Inc.; LV is the recipient of National Alliance for Research on Schizophrenia and Depression Young Investigator Award; and SLG is an FRSQ scholar.

 Submitted for publication August 20, 2003; revised January 14 and February 6, 2004; accepted February 13, 2004.

 Disclosure: K. Zaniolo, None; S. Leclerc, None; A. Cvekl, None; L. Vallières, None; R. Bazin, None; K. Larouche, None; S.L. Guérin, None.

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Flow cytometry and RT-PCR analysis of α4 in mid- and postconfluent RCECs. (A) Surface expression of the integrin subunit α4 in confluent and midconfluent RCECs was quantified by flow cytometry using the anti-α4 mAb ALC1/3 (α4) (thick lines). As a negative control, a monoclonal antibody directed against bovine PARP was used as the primary antibody (dashed lines). Primary cultures of rabbit choroidal melanocytes were used as a positive control for cell surface expression of α4. Expression of the FN integrin subunit α5 was monitored as a positive control in both confluent and midconfluent RCECs, as well as in choroidal melanocytes, by using the mAb IIA1 (anti-α5). In the histograms, relative fluorescence as a logarithmic scale is on the x-axis and the number of cells as a linear scale is on the y-axis. Data from one experiment of three similar experiments are presented. (B) Expression of the α4 integrin subunit was monitored through immunocytochemistry on both midconfluent (MC) and 2-day postconfluent (C) RCECs. Ctl: post-confluent RCECs treated as just described (C), except that the primary antibody was omitted. Arrows: predominant expression of α4 at cell–cell contacts. (C) Total RNA from midconfluent (MC) and both 2-day (C2d) and 5-day (C5d) postconfluent RCECs were reverse transcribed and PCR coamplified using the α4 and 18S ribosomal RNA-specific primers. The positions of both the amplified 265-bp α4 (α4) and 489-bp 18S fragments (18S) are indicated, along with that of the most relevant markers (left). Results are shown for PCR amplification cycles 28, 30, 32, and 34.

Figure 2.

 

Transient transfection analysis in RCECs. Plasmids harboring various lengths from the α4 promoter (−76α4-CAT, −120α4-CAT, −200α4-CAT, and −1000α4-CAT) were transfected into RCECs plated at various cell densities. (A) Schematic representation of each α4 promoter constructs. The positions of both the α4.1 and α4.2 elements previously identified ⁴⁶ are indicated (▪). (B) CAT activities from the transfection of the α4-CAT constructs into RCECs cultured at midconfluence (▪) or after confluence (□). Cells were harvested 48 hours after transfection, and CAT activities were determined and normalized to the level of hGH secreted into the culture medium. ( ) CAT activities at postconfluence that are significantly different from those at midconfluence (P <0.005; paired samples, t-test).

Figure 3.

 

FN responsiveness of the α4 promoter in RCECs grown in the presence of FN. (A) The plasmid −1000α4-CAT was transfected into midconfluent RCECs grown either on plastic or on culture dishes coated with increasing concentrations of FN (1, 2, 4, 6, or 10 μg/cm²). Cells were harvested and CAT activities determined. Results are expressed as the ratio of the CAT activity from cells grown in the presence of FN over that from cells grown on plastic. *CAT activities from transfected cells cultured on FN-coated dishes that are significantly different from those cultured on untreated dishes (P < 0.005; paired samples, t-test). (B) Cultures of RCECs were exposed to increasing concentrations (25–1000 ng) of the mAb ALC1/3 directed against the α4 integrin subunit before they were plated either on plastic (−FN) or on culture dishes coated with 8 μg/cm² FN (+FN), and transfected at midconfluence with the plasmid −1000α4-CAT. As a negative control, 1 μL (which corresponded to approximately 12 μg proteins) of a monoclonal antibody raised against bovine PARP (C-2-10) was added to the cells before their seeding on culture plates coated or not with FN (8 μg/cm²). Results are expressed as the ratio of the CAT activity from cells grown in the presence of FN over that from cells grown solely on plastic. ( ) CAT activities from transfected cells cultured in the presence of the α5 Ab that are significantly different from those cultured with no added α5 Ab (P < 0.005; paired samples, t-test). (C) The recombinant plasmids –1000α4-CAT and α5-92 were transfected into RCECs grown on plastic or FN-coated (+FN; 8 μg/cm²) culture dishes with either no or 10 μM of the MEK/kinase inhibitor PD98059. ( ) CAT activities from transfected RCECs cultured on FN-coated dishes in the presence of PD98059 that are significantly different from those cultured with no added inhibitor (P < 0.005; paired samples, t-test).

Figure 4.

 

Sequence similarities between the gelB Pax-6 target site and the α4.1 element from the α4 gene promoter. The DNA sequence from the target area of the MMP gelatinase B (gelB; MMP-9) gene promoter recently shown to bind Pax-6 is aligned with the α4.1 regulatory element identified in the basal promoter of the α4 gene promoter. ⁴⁶ The preserved nucleotide residues are in bold uppercase letters. The circled nucleotides correspond to those G residues whose methylation by DMS has been shown to interfere with the binding of a yet uncharacterized nuclear protein from RCECs, and the area protected by this unknown protein in DNase I footprinting is identified by the boxed residues. ⁴⁶ The position of two putative Pax-6 target sites is shown: one that is expected to interact with both the PAI and RED subdomains from Pax-6 (PAI/RED), whereas the other is expected to interact preferentially with the RED subdomain (RED).

Figure 5.

 

Binding of Pax-6 to the α4.1 element from the α4 promoter. (A) Crude nuclear extracts prepared from cells or tissues reported to express Pax-6 (chicken embryonic lens [lens], mouse αTN4-1 lens cells) were incubated with the α4.1-labeled probe in the presence of 1 μg poly(dI-dC) · poly(dI-dC) in buffer D. Formation of DNA-protein complexes was revealed through EMSA. The position of two distinct DNA-protein complexes (A and B) is shown. (B) Crude nuclear proteins from chicken lens (lens) were incubated with either no (C+) or 2 μL of the afP6 affinity-purified antiserum (Pax-6 Ab) raised against human Pax-6, or with 2 μL of nonimmune serum (NIS) as a negative control. The position of the DNA-protein complexes A and B is indicated along with that of the free probe (U). (C) The α4.1 probe was incubated with bacterially expressed GST-Pax-6(PD/HD) either alone (C) or in the presence of unlabeled competitor oligonucleotides (100- and 250-fold molar excesses) bearing target sites for either Pax-6 or Pax-6(5a) or the unrelated NF1 protein. P, labeled probe with no proteins added; U, unbound fraction of the labeled probe, C, control.

Figure 6.

 

Influence of Pax-6 overexpression on the activity directed by the α4 promoter in RCECs. (A) Both the −76α4-CAT (▪), and −1000α4-CAT plasmids (□) were transfected with either an empty vector or a Pax-6 expression plasmid in RCECs plated at various cell densities (1 × 10⁵ to 1.5 × 10⁶ cells per 35-mm well). Cells were harvested and the CAT activity measured. The ratio of the CAT activity measured with cotransfected Pax-6 over that without Pax-6 is shown. (B) CAT recombinant plasmids bearing various lengths (up to positions −1000, −400, −300, −120, −76, and −41 relative to the α4 mRNA start site) from the human α4 gene promoter were transfected with either the empty vector or the Pax-6 expression plasmid, in RCECs plated at an intermediate cell density (5 × 10⁵ cells per Petri). Cells were harvested and CAT activity measured and expressed as detailed in (A). ( ) CAT activities from RCECs transfected with the Pax-6 expression plasmid that are significantly different from those transfected without (P < 0.005; paired samples, t-test). (C) The wild-type plasmids −76α4-CAT, −120α4-CAT, −400α4-CAT, and −1000α4-CAT or their derivatives α4-76/4.1m, α4-120/4.1m, α4-400/4.1m, and α4-1000/4.1m that bear mutations in the α4.1 element, were cotransfected either with the empty vector or the Pax-6 expression plasmid in midconfluent RCECs. Cells were harvested 48 hours later and CAT activity determined. The results are expressed as the ratio of the CAT activity obtained with cotransfection of Pax-6 over that measured without Pax-6. ( ) CAT activities from RCECs transfected with the Pax-6 expression plasmid that are significantly different from those transfected without (P < 0.005; paired samples, t-test).

Figure 7.

 

Western blot analysis of Pax-6 expression in confluent and midconfluent RCECs. (A) Nuclear proteins were obtained from both confluent (C) and midconfluent (MC) RCECs and tested in Western blot analyses using the Pax-6 antiserum afP6. As positive controls, nuclear extracts from cells or tissues expressing endogenous Pax-6 (chicken embryonic lens [lens], mouse αTN4-1 lens cells) were also loaded. The molecular weight markers correspond to phosphorylase B (98.5 kDa), bovine serum albumin (66.7 kDa), ovalbumin (42.7 kDa), and carbonic anhydrase (28.6 kDa). The position of the Pax-6 doublet is shown. (B) Twenty micrograms nuclear proteins from both the confluent (C) and midconfluent (MC) RCECs that were used in (A) were loaded on a 12% SDS-polyacrylamide gel and stained with Coomassie blue.

Figure 8.

 

SDS-gel fractionation of crude nuclear proteins from confluent and midconfluent RCECs. Crude nuclear proteins isolated from both midconfluent (A) and confluent (B) RCECs were size-fractionated onto a 10% SDS polyacrylamide minigel, eluted, and renatured. A sample (10 μL) from each eluted fraction (labeled with numbers) was incubated with the α4.1-labeled probe and formation of DNA-protein complexes was tested in EMSAs using an 8% native polyacrylamide gel. The molecular weights indicated (bovine serum albumin: 66 kDa; ovalbumin: 45 kDa) correspond to molecular weight markers. The position of the free probe is indicated (U) as well as that of the previously reported Bp2 to Bp5 DNA-protein complexes. ⁵⁶ P, labeled probe with no proteins added.

Figure 9.

 

Immunologic detection of Pax-6 in complex Bp5 formed by nuclear proteins isolated from confluent and midconfluent RCECs. (A) The protein fractions from midconfluent RCECs forming the Bp4 and Bp5 complexes, along with proteins from the Pax-6–containing lens nuclear extract (lens), were incubated with the α4.1-labeled probe and formation of DNA-protein complexes was analyzed by EMSA. The position of the free probe is shown (U) along with that of Pax-6 from the chicken embryonic lens extract. (B) Nuclear proteins from the fractions that support formation of the Bp2 (used as a negative control) and Bp5 complexes in crude nuclear extracts from both confluent (C) and midconfluent (MC) RCECs were incubated in the absence (−) or presence (+) of 2 μL of the Pax-6 antiserum afP6. The position of both the Bp2 and Bp5 complexes is indicated, along with that of the free probe (U). P, labeled probe with no proteins added.

The authors thank Glenn D. Rosen (Department of Internal Medicine, Stanford University Medical School, Stanford, CA) for his generous gift of the α4 promoter-CAT constructs; Christian Salesse (Ophthalmology Research Unit, CHUL Research Center, Québec, Canada) for critically reviewing the manuscript; and Ales Cvekl Jr for help in revising and editing the manuscript.