Laminin Reduces Expression of the Human α6 Integrin Subunit Gene by Altering the Level of the Transcription Factors Sp1 and Sp3

Abstract

purpose. Damage to the corneal epithelium results in the massive secretion of fibronectin (FN) shortly after corneal injury, which is later replaced by the secretion of laminin (LM). Laminin is recognized by receptors (α6β1 and α6β4) that belong to the integrin family. The authors characterized the regulatory influence exerted by laminin on the Sp1/Sp3-mediated α6 gene promoter function.

methods. Recombinant plasmids bearing the CAT reporter gene, fused to various segments from the α6 promoter, were transfected into rabbit corneal epithelial cells (RCECs) grown on untreated or on LM-coated culture plates or into Sp1-deficient Drosophila Schneider cells. Expression and DNA binding of Sp1/Sp3 was monitored with the use of Western blot and electrophoretic mobility shift assays (EMSAs), respectively. DNA target sites for Sp1/Sp3 in the α6 gene promoter were identified in vitro by DNAse I footprinting. Binding of Sp1 and of a few other transcription factors was also examined in vivo by chromatin immunoprecipitation (ChIP) assays. Expression of the α6 mRNA in RCECs grown on LM-coated culture plates was also assessed by Northern blot and RT-PCR analyses.

results. Transfection experiments provided evidence that both Sp1 and Sp3 positively influence α6 promoter activity in Sp1/Sp3-deficient SL2 Schneider cells. Two GC-rich target sites for Sp1/Sp3 (a proximal and a distal site) were identified in the α6 promoter by DNAse I footprinting. Binding of Sp1/Sp3 was further validated in vivo by ChIP. Transfections conducted into RCECs grown on LM-coated culture plates resulted in repression of the activity directed by the α6 promoter. This LM-mediated negative regulatory influence also translated into a similar reduction in the expression of the endogenous α6 mRNA as revealed by both RT-PCR and Northern blot analyses. Most of all, results from EMSA and Western blot analyses suggested that the LM-mediated repression of the α6 promoter activity results in part from the proteolytic cleavage of Sp1/Sp3.

conclusions. Unlike FN, which functions as an activator of α5 gene transcription, LM repressed transcription, directed by the α6 gene promoter, by altering the nuclear levels of Sp1 and Sp3. Reappearance of LM in the basement membrane after repair of the corneal damage is therefore expected to substantially contribute to the final steps of this process by influencing the degree to which genes that encode protein products requested for cell adhesion and migration, such as integrin genes, are expressed during corneal wound healing.

The corneal epithelium is made up of five to seven layers of regularly arranged squamous epithelial cells. On corneal injury, the damaged cells should be replaced rapidly to maintain proper visual acuity, a process that is ensured by the centripetal migration of stem cell-derived transient amplifying cells from the corneal limbus. ^{1 2} The early events of the wounding process are chiefly characterized by the disassembling of the hemidesmosomes in the epithelial cells 50 to 70 μm from the wound edge ³ and the massive secretion of fibronectin (FN), which peaks as late as 3 to 12 hours after corneal damage and whose expression is transitory in the basal membrane as it starts disappearing 1 week later. ^{4 5} It is thought that the newly synthesized FN serves as a temporary matrix for the attachment and migration of the basal epithelial cells that border the injured area. ⁶ It is produced by subepithelial stromal cells in superficial wounds that alter the epithelium or by stromal cells and corneal epithelial cells in more important wounds that also involve the basal membrane and the stroma. ⁷ As FN staining progressively diminishes, the secretion of laminin (LM), a major component of the basal membrane (BM), increases to reach maximal expression 1 week after corneal damage. ⁴ LM-1 (recently reclassified as LM-111 ⁸ ), LM-5 (recently reclassified as LM-332 ⁸ ), and LM-10 (recently reclassified as LM-511 ⁸ ) are the major LM isoforms in the corneal BM. ^{9 10} The primary function of FN and LM is in cell-matrix attachment, but many additional biological activities, including the promotion of cell migration, wound repair, and mediated cell-signaling events, have been demonstrated. ^{9 11} The BM has many functions that help maintain the normal multilayer structure in the corneal epithelium. Interaction of its basic components, collagen and LM, with their corresponding transmembrane integrin receptors, control cell shape, gene expression, cell migration and proliferation, and programmed cell death. 

Integrins are widely expressed, glycosylated, heterodimeric transmembrane adhesion receptors made up of noncovalently bound α- and β-subunits that link the extracellular matrix (ECM) to the cell’s cytoskeleton. They mediate bidirectional transfer of information through diverse signal transduction pathways (for reviews, see Plow et al. ¹² and van der Flier and Sonnenberg ¹³ ). In a recent survey of the human genome, 24 α- and 9 β-integrin subunits have been identified, ¹⁴ which implies 6 novel α- and 1 novel β-subunits relative to the previously recognized 18 α- and 8 β-subunits reported to form 24 different heterodimers. ^{12 15 16 17} However, the existence of these new integrin subunits remains to be firmly established. The α6β1 and the α6β4 integrins recognize LM as their ligand. ^{18 19} The integrin α6β4 is a component of the hemidesmosomes (HDs) and is therefore located to the basal side of the basal epithelial cells. ^{20 21 22 23} HDs are specialized junctional complexes that contribute to the attachment of epithelial cells to the BM ²⁴ and are implicated in signal transduction through α6β4. ²⁵ The pattern of integrins expressed by the basal cells of the corneal epithelium is considerably altered as the leading edge progresses toward the wounded corneal surface. ^{4 20 26 27} A few studies reported dynamic changes in the distribution of the α6 subunit during corneal wound healing in vivo. ^{20 28} In response to wounding, HDs are disassembled, and the expression of α6 ²² and β4 ²⁷ increases during migration of the epithelial cells in vivo. Stepp et al. ²⁷ suggested that the increased β4 expression must play a role in cell-cell or cell-substrate adhesion mechanisms that are required during the corneal wounding or in the preparation of cells for the restratification process. 

Although FN deposition can be observed as early as 3 hours after damage to the corneal epithelium, no staining of LM occurs beneath the leading edge of migrating epithelial cells until 24 hours after the injury. ²⁹ Interestingly, FN has been shown to influence the transcriptional activity not only of the α5 subunit gene ³⁰ but also that of the LM binding α6 integrin subunit ²² in in vitro cultures of rabbit corneal epithelial cells (RCECs). These results suggest that while corneal epithelial cells start migrating over the newly synthesized temporary FN matrix, expression of the α5 and α6 integrin subunit genes is turned on, ensuring early expression of the α6 subunit well before LM accumulates beneath the leading edge. As a consequence, the signal transduction pathway activated on binding of the α6β1 and α6β4 integrins to their ligand LM might trigger regulatory signals distinct from those resulting from the binding of FN to the α5β1 integrin. Activation of the MAPK pathway through the binding of FN to its α5β1 integrin was recently reported to result in the transcriptional activation of the α5 gene through a modification in the phosphorylation state of Sp1. ³⁰ Both the promoter and the 5′-flanking region of the human α6 subunit gene have recently been cloned. ^{31 32 33} Although multiple target sites for Sp1 were identified along the α6 proximal promoter sequence by DNAse I footprinting, ³² only the most proximal Sp1 site (position −48 to −43) was assessed for its regulatory influence through site-directed mutagenesis. ³¹ In addition, all α6 promoter transfection experiments conducted to date were performed in cancer cell lines of various origins. Nontransformed primary cultured cells are much closer to their in vivo counterparts than are transformed cells or cell lines; consequently, they are attractive as a source of cellular material for gene expression studies ^{34 35 36} (also see Gaudreault et al. ³⁷ ). Considering that cancer cell lines and nontransformed primary cultured cells of similar tissue origin often behave differently, regulatory elements with negligible regulatory influence in transformed cells may be required for gene transcription in nontransformed cells. 

The purpose of this study was to investigate in more detail the role exerted by the positive transcription factors Sp1 and Sp3 on the transcriptional activity directed by the α6 promoter in primary cultured and established tissue-cultured cells and to evaluate whether LM may alter the regulatory influence mediated by these nuclear proteins. DNAse I footprinting analyses identified two specific target sites for Sp1 (a proximal and a distal site) along the α6 promoter sequence. Binding of Sp1 to the α6 promoter was also demonstrated in vivo through ChIP assays. Site-directed mutational analyses provided evidence that different cells use either Sp1 target site, proximal or distal, with varying efficiency. Most of all, the activity directed by the α6 promoter was found to be downregulated when RCECs were grown on LM-coated culture plates. The reduced α6 promoter activity was shown to rely, at least in part, on a reduced level of expression of Sp1 and Sp3 in corneal epithelial cells (RCECs and HCECs) when they are grown on LM. 

Materials and Methods

All experiments 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 Media

HCECs were isolated from the limbal area of normal eyes from a 44-year-old donor and were obtained through the Eye Bank from the CHUL Research Center in accordance with a procedure we previously described. ^{37 38} RCECs were obtained from the central area of freshly dissected rabbit corneas and were grown into SHEM medium (fetal bovine serum, 15% wt/vol), as recently described. ³⁹ Human skin keratinocytes (HSKs) were obtained from normal newborn foreskin specimens from a 17-day-old donor and were cultured with a feeder layer, as recently described. ⁴⁰ Human skin fibroblasts (HSFs) were isolated from skin biopsy specimens of healthy male donors and underwent primary culture in Dulbecco modified Eagle medium (DMEM, Gibco BRL, Burlington, ON, Canada) supplemented with 10% fetal bovine serum (FBS; Gibco BRL). Human epithelioid carcinoma HeLa (ATCC CCL 2) cells were grown in DMEM supplemented with 10% fetal bovine serum (FBS). Human colon adenocarcinoma Caco2/15, a derivative of the Caco2 cell line (ATCC HTB-37; Rockville, MD) characterized by Beaulieu and Quaroni, ⁴¹ was grown in DMEM high glucose, 10% FBS, HEPES 20 mM, and glutamine 10 mM. ARPE-19 cells were purchased from the ATCC (CRL-2302) (Manassas, VA) and grown in DMEM/F12 growth medium (Canadian Life Technologies, Burlington, ON, Canada) with 3 mM glutamine (Sigma, Oakville, ON, Canada) and supplemented with 10% fetal bovine serum (Hyclone, Logan, UT). SL2 Drosophila Schneider cells (ATCC CRL-1963) were cultured at 28°C without CO₂ in Schneider insect medium (Sigma) supplemented with 10% FBS and 20 μg/mL gentamicin. All cells were grown under 5% CO₂ at 37°C, and gentamicin was added to all media at a final concentration of 15 μg/mL. 

Flow Cytometric Analysis

RCECs, HeLa, HSFs, HSKs, and ARPE-19 cells were grown as described and removed from the culture dishes (Sarstedt, Montréal, QC, Canada) with PBS/citrate/EDTA solution and were washed twice with serum-free medium containing 1% BSA and fixed with 1% paraformaldehyde-PBS (1% PFA/PBS) for 20 minutes. Aliquots of fixed cells (1 × 10⁶) were washed twice and resuspended in PBS (pH 7.2) containing 1% BSA. One microgram of a primary antibody directed against the α6 integrin subunit (MA6; Chemicon, Temecula, CA) was added to cell suspensions for 60 minutes at room temperature and agitated on a rotator. After 1 hour, cells were washed twice, and a dilution of 1:100 of fluorescein isothiocyanate-conjugated (FITC) human anti-rat IgG suspended in PBS-1% BSA was added to cells and incubated for 30 minutes on a rotator at room temperature. Finally, cells were resuspended in 1% PFA/PBS and analyzed with a flow cytometer (Epics XL; Beckman Coulter, Miami, FL). 

Laminin and Fibronectin Coating

Skin keratinocytes (2.5 × 10⁴/cm²) known for their ability to secrete large amounts of LM-5 ⁴² and obtained from an adult skin donor (kindly provided by Lucie Germain, Laboratoire d’Organogénèse Expérimentale [LOEX], Hôpital du Saint-Sacrement du CHA, Québec, QC, Canada) were plated on an irradiated 3T3 feeder layer (2 × 10⁴ 3T3/cm²) on tissue culture plates. Keratinocytes were cultured in Dulbecco-Vogt modified Eagle medium (Life Technologies, Grand Island, NY), supplemented with 10% fetal calf serum (FCS; Hyclone-PDI Bioscience, Aurora, ON, Canada), 100 IU/mL penicillin, and 25 μg/mL gentamicin (Sigma, St. Louis, MO) until they reached 30%, 70%, or 100% cell density for various periods of time (1, 3, 5, and 10 days) and then were completely removed from the plates by incubation in 20 mM NH₄OH for 5 to 10 minutes. The plates on which LM was deposited were then washed with PBS and reseeded either with RCECs (2.5 × 10⁴/cm²) or HCECs (5 × 10⁴/cm²) grown to midconfluence before transfection. A similar approach was used with Caco2/15 cells, which are known for their ability to secrete LM-10 ⁴³ and LM-1. ⁴⁴ When indicated, coating of tissue-culture plates with commercially purified LM (LM-1, 0.5–8 μg/cm²; Sigma) before RCECs were added, as described. ³⁰ Human plasma FN (obtained as previously described ⁴⁵ ) was coated for 18 hours at 37°C on the culture dishes at 8 μg/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 described. 

Plasmids and Oligonucleotides

Recombinant plasmids α6–84, α6–141, α6–181, α6–352, α6–430, and α6–590—all of which bore the chloramphenicol acetyltransferase (CAT) reporter gene fused to DNA fragments from the human α6 gene upstream regulatory sequence extending to 5′ positions −84, −141, −181, −352, −430, and −590 but shared a common 3′ end located at position +76—were constructed. The PGEM-3Z f(+) AvaI/α6 plasmid that bears the human α6 gene promoter from position −930 to +76 relative to the α6 mRNA start site (generously provided by Schei Kitazawa, Kobe University School of Medicine, Kusunoki-Cho, Chuo-Ku, Kobe, Japan) was linearized at its unique KpnI site (5′ end) and was treated with the nuclease Bal31 before it was blunted. The 5′-digested α6 promoter fragments generated were then digested with HindIII (3′ end at α6 position +76) before they were ligated into the unique SmaI/HindIII sites of the plasmid PGL3 to produce the PGL3/α6–590 vector. The PGL3/α6–590 construct was digested with NheI and blunt-ended with Klenow (which then preserved the α6–590 5′ end), before it was second-digested with XbaI (which preserved the α6 +76 3′ end). The resultant α6 promoter-bearing DNA fragment was ligated upstream of the CAT reporter gene from the pCATBasic vector (Promega, Madison, WI), which was first digested with PstI (5′ end), blunt-ended with Klenow, and second digested with XbaI (3′ end) before its dephosphorylation with alkaline phosphatase (to yield pCATBasic/α6–590). Removal of the α6 promoter fragments from the common SphI site (located upstream from α6 promoter position −590 in the multiple cloning site of the parental plasmid pCATBasic) to the internal restriction sites SacII (at α6 position −84), KpnI (at −141), SacI (at −181), NsiI (at −352), and AccI (at position −430) yielded, after blunt-ending with Klenow and ligation, the plasmids α6–84, α6–141, α6–181, α6–352, and α6–430, respectively. 

The plasmid derivatives α6–84/mSp1p, α6–141/mSp1p, and α6–352/mSp1p (which bear mutations into the proximal Sp1 site), α6–352/mSp1d (which bear mutations into the distal Sp1 site), and α6–352/mSp1pd (which bear mutations into the proximal and the distal Sp1 sites) were all produced by PCR with a site-directed mutagenesis kit (QuickChange; Stratagene, La Jolla, CA) according to the manufacturer’s instructions. Parental plasmids α6–84, α6–141, and α6–352 were used as templates. Double-stranded oligonucleotides used to introduce mutations in the Sp1 sites (mutations are shown in bold) contained the following sequences: proximal Sp1 site (top strand, 5′-GGGGCGAGAGGGTGGGGAGGATCGCGGCCGGCGTCCTCGTC-3′; bottom strand, 5′-GACGAGGACGCCGGCCGCGATCCTCCCCACCCTCTCGCCCC-3′); distal Sp1 site (top strand, 5′-GTCCACAGAGATCGCCGCAGTGGGGCTGCTTCGCCG-3′; bottom strand, 5′-CGGCGAAGCAGCCCCACTGCGGCGATCTCTGTGGAC-3′). Sp1 and Sp3 expression plasmids pPacSp1 and pPacSp3, respectively, were obtained from Guntram Suske (Institute für Molecularbiology und Tumorforschung, Philipps Universität, Marburg, Germany). Double-stranded oligonucleotides used for competition assays in the EMSAs contained the following DNA sequences: proximal (5′-GAGGGTGGGGAGGGGCGGGGCCGGCGTCCTCGTCA-3′) and distal (5′-TTCTGTCCACAGAGGGCGGCGCAGTGGGGCTGCTT-3′); Sp1 sites identified in this study in the α6 gene promoter and in the alternative Sp1p_a site (5′-GGGCGCGCAAGGAGGGGCGAGAGGGTGGGGAGGGG-3′); the high-affinity binding sites for the transcription factors Sp1 (5′-GATCATATCTGCGGGGCGGGGCAGACACAG-3′) ⁴⁶ and NFI (5′-TTATTTTGGATTGAAGCCAATATGAG-3′) ⁴⁷ ; the consensus sequence for the transcription factor E2F1 (5′-CAGAGCCGTTTCGCGCGGTGCGGGCGGTGC-3′); and the AP-1 site identified in the promoter of the human α5 integrin subunit gene (5′-GATCCCCGCGTTGAGTCATTCGCCTC-3′). ⁴⁸ All oligonucleotides used in the present study were chemically synthesized (Biosearch 8700 apparatus; Millipore, Beford, MA). 

DNAse I Footprinting

DNAse I footprinting was performed in DNAse I buffer ⁴⁹ by incubating 3 × 10⁴ cpm labeled probe with increasing amounts (1–10 μL) of recombinant Sp1 protein (GST-Sp1–8xHis protein; kindly provided by Claude Labrie, Oncology and Molecular Endocrinology Research Center, CHUL). The recombinant GST-Sp1–8xHis protein was overexpressed in Escherichia coli BL21 CodonPlus R1L cells with the pGEX-Sp1–8xHis plasmid before purification, as described. ⁵⁰ The 217-bp HindIII-XbaI DNA fragment from the α6 promoter extending from positions −141 to +76 and the 430-bp fragment spanning α6 promoter sequences from −352 to +76 were used as labeled probes in these assays. DNAse I digestion and further analysis of the digestion products onto polyacrylamide sequencing gels were performed as described. ⁴⁹  

Chromatin Immunoprecipitation Assays

ChIP analyses were conducted as previously described. ⁵¹ Briefly, HCECs were grown on 150-mm tissue culture dishes to approximately 75% confluence in complete medium. Cells were then cross-linked with 1% formaldehyde for 10 minutes, harvested, and chromatin immunoprecipitated with 1 μg polyclonal antibodies directed against the Sp1 (sc-59; Santa Cruz Biotechnology, Inc., Santa Cruz, CA), Sp3 (sc-644; Santa Cruz Biotechnology), and NFI (SC-5567; Santa Cruz Biotechnology) and E2F1 (sc-193x; Santa Cruz Biotechnology) transcription factors. The resultant DNA was analyzed by PCR using a pair of primers (ITGA6-U,5′CTATCAAGGTGTGCGCTGTG-3′; ITGA6-L,5′CAGAATTGTGGTTGCCGAGTA-3′) spanning the entire ITGA6 gene-promoter area. As a negative control, each ChIP sample was also subjected to PCR using primers (p21-U [5′AATTCCTCTGAAAGCTGACTGCC3′] and p21-L [5′OAGGTTTACCTGGGGTCTTTA-GA-3′]) specific to a region located approximately 2 kbp upstream of the human p21 promoter. Standard deviation is provided and has been calculated from three separate experiments. 

Transient Transfections and CAT Assays

HCECs and RCECs were grown to midconfluence (near 80% coverage) into six-well (35-mm) tissue culture plates and transiently transfected using a polycationic detergent reagent (Lipofectamine; Gibco BRL) according to a procedure we previously described. ^{30 52} Each reagent (Lipofectamine)-transfected plate received 1 μg α6-CAT test plasmid and 0.5 μg human growth hormone (hGH)-encoding plasmid pXGH5. ⁵³ Primary cultured HSFs, established tissue culture HeLa cells, and Drosophila SL2 Schneider cells were transfected according to the calcium phosphate precipitation procedure ^{49 54} at a density of 5 × 10⁵ HSF, 4 × 10⁵ HeLa, or 1 × 10⁶ SL2 cells per 60-mm culture plate. Levels of CAT activity for all transfected cells were determined as described ⁵⁴ and were normalized to either β-galactosidase or the amount of hGH secreted into the culture media and were assayed using a kit for quantitative measurement of hGH (Immunocorp, Montréal, QC, Canada). The value presented for each individual test plasmid transfected corresponded 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 over the background level caused by the reaction buffer used (usually corresponding to 0.15% chloramphenicol conversion). Student’s t-test was performed for comparison of the groups. Differences were considered statistically significant at P < 0.05. All data are expressed as mean ± SD. 

Nuclear Extracts, Electrophoretic Mobility Shift Assays, and Supershift Experiments

Crude nuclear extracts were prepared from midconfluent RCECs (5 × 10⁴ cells/cm² cultured for 72 hours usually yields almost 70% coverage of the culture flasks) grown either on BSA- (2% BSA in PBS) or LM-coated culture dishes, as detailed previously. ⁵⁵ Electrophoretic mobility shift assays (EMSAs) were conducted as described ³⁰ by using the Sp1 oligomer as a probe and 5 μg nuclear proteins. When indicated, unlabeled oligomers bearing the sequence of various target sites for known transcription factors (SP1p, Sp1d, Sp1p_a, Sp1, NFI, AP-1) were added as unlabeled competitors (50-, 100-, 125- and 500-fold molar excesses) during the assay. DNA-protein complexes were then separated by gel electrophoresis through 6% native polyacrylamide gels run against Tris-glycine buffer. ⁵⁶ Supershift experiments in EMSA were conducted by incubation of 5 μg nuclear proteins from RCECs in the presence of no or 2 μL polyclonal antibodies raised against the transcription factors Sp1 and Sp3 (Santa Cruz Biotechnology, Inc.). 

SDS-PAGE and Western Blot

Approximately 30 μg protein was added to 1 vol sample buffer (6 M urea, 63 mM Tris [pH 6.8], 10% [vol/vol] glycerol, 1% SDS, 0.00125% [wt/vol] bromophenol blue, 300 mM β-mercaptoethanol) and was size-fractionated on a 10% SDS-polyacrylamide minigel before transfer to a nitrocellulose filter. The blot was then washed in TS and TSM buffers, as described. ³⁰ A 1:5000 dilution of a rabbit polyclonal antibody raised against Sp1 or Sp3 was added, and incubation proceeded for 2 hours at 22°C. The blot was then washed and incubated with a 1:1000 dilution of a peroxidase-conjugated goat anti-rabbit immunoglobulin G (Jackson ImmunoResearch Laboratories, Inc., Bar Harbor, ME), and immunoreactive complexes were revealed with the use of Western blot chemiluminescence reagents (Amersham Biosciences, Arlington Heights, IL), as recently described. ³⁰  

RT-PCR and Northern Blot Analyses

Total RNA was isolated from midconfluent RCECs grown on BSA or on LM-coated culture plates using a reagent (Tri Reagent; Molecular Research Center, Inc., Cincinnati, OH) and was reverse transcribed using a transcriptase kit (Superscript II; Gibco-BRL), as recently detailed. ⁵² The resultant newly synthesized first-strand cDNAs were column purified (QIAquick nucleotide removal kit; Qiagen, Santa Clarita, CA) and used for PCR amplification of the α6 integrin subunit transcript and the 18S ribosomal RNA. The DNA sequence of both the 5′ and the 3′ template primers for α6 (5′ primer, CAAGATGGCTACCCAGATAT-bp; 3′ primer, CTGAATCTGAGAGGGAACCA-bp; PCR product, 210 bp) was derived from the corresponding human α6 integrin gene (GenBank accession no., NM 000210). Oligonucleotide primers for the amplification of the 18S ribosomal RNA were provided (Quantum RNA 18S Internal Standards kit; Ambion Inc., Austin, TX). Taq polymerase (Pharmacia-LKB, Uppsala, Sweden) was selected for PCR amplification. Cycle parameters were the same for all primers used (denaturation 94°C, 30 seconds; annealing 58°C, 30 seconds; extension 72°C, 30 seconds) with an identical number of cycles (24, 26, 28, 30, 32, and 34 cycles) for both sets of primers. PCR-amplified DNAs were fractionated on a 2% agarose gel, and their positions were revealed by ethidium bromide staining. The gel photograph was scanned (Visage 110S Bioimage analyzer; Millipore) to quantify the alterations in the amount of the α6 and the 18S PCR-amplified fragments. 

Northern blot analyses were conducted on total RNA isolated from midconfluent RCECs grown on BSA- or LM-coated culture plates, as detailed. Total RNA was size-fractionated on a 1.2% formaldehyde-agarose gel and blotted onto a Hybond-N+ membrane (Amersham Canada, Oakville, ON, Canada), as described elsewhere. The membrane was then hybridized at 65°C in buffer (PerfectHybrid Plus Hybridization; Sigma). Blots were washed once at 25°C in 2× SSC-0.1% SDS (1× SSC is 150 mM NaCl; 15 mM sodium citrate) and twice at 50°C in 0.2× SSC-0.1% SDS. The labeled probe used for hybridization consisted of a 665-bp EcoRI–EcoRI restriction fragment derived from the α6 cDNA. Membranes were autoradiographed at −80°C for 18 hours before development. 

Results

DNAse I Footprinting of the Sp1 Binding Sites along the Promoter of the α6 Integrin Gene Subunit

To determine precisely where Sp1 binds along the α6 integrin subunit gene promoter, in vitro DNAse I footprinting was performed using a recombinant preparation of Sp1. The DNA sequence, located between positions +76 to −352 and constituting the entire α6 promoter, was examined on both strands for Sp1 binding. Two distinct target sites for Sp1 were identified along the α6 promoter sequence: a proximal site, located between positions −38 to −62 (Sp1p; Fig. 1Aand also depicted on Fig. 1C ), and a more distal site, located between positions −195 and −215 (Sp1d; Fig. 1Band also depicted on Fig. 1C ). No binding site for Sp1 except Sp1p and Sp1d could be identified on the α6 promoter under the experimental conditions used. 

Regulatory Influence Played by Both the Sp1p and the Sp1d Sites on α6 Promoter Activity

To assess the functional relevance of the proximal and distal Sp1 sites, transfection of recombinant constructs bearing the CAT reporter gene fused to DNA segments from the α6 promoter bearing the proximal or both the proximal and the distal Sp1 sites was conducted into primary cultured cells (HSFs, HSKs, RCECs) and established tissue cultured cells (HeLa, ARPE19). All these cells were shown to express the α6 integrin subunit to varying degrees in the following order: RCECs > HSKs > ARPE19/HeLa > HSFs, as revealed by FACS analysis (Fig. 2) . As shown on Figure 3 , maximal basal promoter function is ensured by the first 84 bp located immediately upstream of the α6 mRNA start site (in plasmid α6–84) in all types of transfected cells. Surprisingly, extending the 5′ end further up to position −352 to include the Sp1d site resulted in a dramatic reduction in CAT activity on transfection of the plasmid α6–352 into all types of cells (near 5-, 3-, 15-, 9-, and 50-fold reduction relative to the level directed by α6–84 in RCEC (Fig. 3B) , HSK (Fig. 3C) , HeLa (Fig. 3D) , ARPE19 (Fig. 3E) , and HSF (Fig. 3F)cells, respectively. Elongating the 5′ end up to position −430 or −590 did not result in CAT activities statistically different from those measured with the α6–352 construct when transfected in RCECs and in HeLa cells (data not shown), suggesting that most of the regulatory elements required for the transcription of the α6 gene are located downstream from position −352. In addition, comparison of CAT activity yielded by the transfection of the α6–84 construct clearly indicated that the strength of the α6 promoter is higher in RCECs than in any of the other cell types transfected (32,118 ± 4941, 17,301 ± 3818, 6584 ± 1353, 1219 ± 455, and 60 ± 16 in RCECs, HSKs, HeLa, ARPE-19 and HSFs cells, respectively; similar results were observed with the α6–352 construct). 

We then exploited site-directed mutagenesis to establish the individual contribution of Sp1p and Sp1d in the transcription directed by the α6 gene promoter. As shown on Figure 3 , mutating the Sp1p site in the basal α6 promoter from the parental plasmid α6–84 (to yield the α6–84/mSp1p mutated derivative) resulted in dramatic reductions in CAT activity on transfection of primary cultured cells and established tissue culture cells (24-, 29-, 27-, 14-, and 150-fold reduction in RCECs, HSKs, HeLa, ARPE19, and HSFs, respectively). However, mutating the Sp1p site in the context of the longer construct α6–352 (which bears Sp1p and Sp1d) had a statistically significant influence on the α6 promoter activity only in RCEC (40% reduction) and HSK (77% reduction) cells. Similarly, mutations that abolish the Sp1d site in the construct α6–352/mSp1d only modestly reduced α6 promoter activity by 48% in RCECs, whereas it had no influence in the remaining cell types. On the other hand, mutating both the Sp1p and the Sp1d sites in α6–352/mSp1pd severely reduced α6 promoter activity in all types of cells (from threefold in ARPE-19 to undetectable level in HSFs). These results suggest that a single, intact Sp1 site is required to ensure basal transcription directed by the α6 promoter in all cell lines tested and that negative regulatory elements that restrict promoter activity must be present along the −84/−352 segment from the α6 promoter. 

To evaluate the respective contribution of Sp1 and Sp3 on the activity directed by the α6 Sp1p and the Sp1d sites, cotransfection experiments were conducted into Drosophila SL2 Schneider cells. These cells have been reported to be deficient in producing these transcription factors and many others expressed in higher eukaryotes, making them an ideal system for studying gene expression or transcription factors function (for a review, see Suske ⁵⁷ ). The plasmids α6–141 or α6–352 or their derivatives—which bear mutations into Sp1p (α6–141/mSp1p and α6–352/mSp1p), Sp1d (α6–352/mSp1d), or both sites (α6–352/mSp1pd)—were cotransfected into Schneider cells alone or with recombinant plasmids (pPacSp1 and pPacSp3) containing Sp1 or Sp3 cDNA under the control of the Drosophila actin gene promoter, therefore ensuring high levels of both these proteins in Schneider cells. Preliminary transfection experiments indicated that neither α6–141 nor α6–352 could sustain basal promoter activity when individually transfected into Schneider cells in the absence of Sp1 and Sp3. However, when cotransfected along with pPacSp1, 38- and 68-fold increases in promoter activity were observed with the plasmids α6–141 and α6–352, respectively (Fig. 4) . Mutations that eliminated the Sp1p site did not completely abolish Sp1 responsiveness because the α6 promoter preserved 29% of its activity with the α6–141/mSp1p construct and approximately 75% with α6–352/mSp1p. On the other hand, responsiveness toward Sp3 was much lower than that observed with Sp1 (sevenfold and threefold increases in CAT activity when pPacSp3 was cotransfected along with α6–141 and α6–352, respectively). Mutating the Sp1p site into both constructs entirely abolished promoter activity in SL2 cells, whereas disrupting Sp1d had no influence at all on the Sp3 responsiveness of the α6 promoter (Fig. 4) . Surprisingly, mutations that disrupt the Sp1d site in the α6–352/mSp1d construct did not suppress, but rather stimulated, α6 promoter function when cotransfected with pPacSp1 in SL2 cells. No such effect was observed with Sp3. Cotransfection of pPacSp1 and pPacSp3, together with the α6 promoter constructs, provided evidence that Sp1 and Sp3 do not function synergistically to modulate the transcriptional activity directed by the α6 promoter. At the most, a weak additive effect is observed when the α6–352 construct is cotransfected along with both the Sp1 and Sp3 expression vectors. 

Sp1 Binds to an Alternative Target Site when the Sp1p Site Is Mutated In Vitro

Only two Sp1 target sites were initially identified through DNAse I footprinting along the α6 promoter (Fig. 1) . However, the high residual α6 promoter activity observed in SL2 cells when the Sp1p and Sp1d sites were mutated, individually or in combination, suggested that Sp1 might have the ability to bind other alternative sequences with a lower affinity when it is prevented from interacting with its primary target sites. To investigate such a possibility, DNAse I footprinting was repeated with the Sp1p-unmutated α6 promoter fragment used in Figure 1(as a control) or with a similar fragment bearing the mutated Sp1p site as labeled probes. Incubation of the labeled probe that bore an intact Sp1p site with Sp1 yielded the same protected site on the top (Fig. 5A)and the bottom (Fig. 5B)strands as that identified previously (Fig. 1) . However, when incubated with the labeled probe bearing the mutated Sp1p site, Sp1 could still bind, though with a lower affinity, to a stretch of sequence located immediately 5′ from the Sp1p site to yield visible DNAse I protection (Sp1p_a) on both strands of the labeled probe. Examination of the Sp1p_a site revealed that it has primarily a GA-rich structure (Fig. 5C)rather than the typical GC- or GT-rich structure normally found in high-affinity Sp1 sites. Competition experiments in EMSA were conducted next to more precisely define the affinity of Sp1 toward each target site identified for this transcription factor in the α6 promoter. The oligonucleotide bearing the high-affinity binding site for Sp1 was used as the labeled probe during these assays (Fig. 6A) . Incubation of crude nuclear proteins from midconfluent RCECs with the Sp1-labeled probe yielded DNA-protein complexes characteristic of Sp1 and Sp3 binding to their GC-rich target site (Fig. 6B) . Formation of these complexes was almost entirely prevented by the addition of a 50-fold molar excess of the oligonucleotide bearing either the high-affinity Sp1 site (Sp1; 86% and 95% reduction of binding at 50- and 100-fold molar excess, respectively) or that from the α6 Sp1 proximal site (Sp1p; 91% and 94% reduction of binding at 50- and 100-fold molar excess, respectively). However, neither the Sp1d (41% and 58% reduction of binding at 50- and 100-fold molar excess, respectively) nor the Sp1p_a site (32% and 63% reduction of binding at 50- and 100-fold molar excess, respectively) was as effective as the Sp1p site for competing with the formation of the Sp1/Sp3 complexes in EMSA, as revealed by PhosphorImager analysis of the Sp1/Sp3 DNA-protein complexes. On average, Sp1d and Sp1p_a were approximately eight times less efficient in competing with the formation of the Sp1/Sp3 complexes when used at a 100-fold molar excess than Sp1p. We therefore conclude that Sp1 binds strongly to the Sp1p site but only weakly to either the Sp1d or the Sp1p_a site in the α6 gene promoter. 

Influence of Laminin on the Activity Directed by the α6 Gene Promoter

Increased secretion of LM occurs late in the process of corneal wound healing. It follows that of the temporary FN matrix, which peaks a few hours after corneal injury. We demonstrated previously that expression of the gene encoding the FN-binding integrin subunit α5 responds positively to the presence of FN when RCECs are grown on FN-coated culture plates. ³⁰ We therefore examined whether LM could similarly alter the activity directed by the α6 gene promoter in vitro in RCECs grown in the presence of LM-coated culture plates. LM deposition was accomplished after a model developed by Nguyen et al. ⁴² in which culture plates are first seeded with keratinocytes from the skin that are known for their ability to secrete substantial amounts of LM (predominantly LM-5). ^{42 58 59 60} Keratinocytes were cultured until they reached varying cell densities for various periods of time and then were removed from the culture plates before their reseeding with RCECs, which were grown to midconfluence before they were transfected. As a control, culture plates were also coated with 8 μg/cm² FN before seeding with RCECs. 

Interestingly, transfection of the α6–181 plasmid into RCECs grown on LM-coated culture plates resulted in repression of the α6 promoter activity; maximum (eightfold) repression was reached when skin keratinocytes were maintained at after confluence for 5 days before their removal, though this result cannot be considered statistically different from the other conditions used (Fig. 7A) . On the contrary, transfection of α6–181 in RCECs that have been grown on FN-coated culture plates resulted not in a reduction but rather in a nearly threefold increase in α6 promoter activity (Fig. 7A) . To test whether other types of LM will trigger different regulatory responses on the α6 promoter activity, we substituted skin keratinocytes with human colon adenocarcinoma Caco-2 cells because these cells were reported to secrete primarily LM-10. ⁴³ As shown on Figure 7B , growing RCECs on LM-10 also resulted in weaker, but significant, repression of α6 promoter function. However, this repression was dependent on the seeding density of RCECs because no repression was observed when cells were seeded at 2.5 × 10⁴ RCECs/cm², whereas a 40% reduction was observed at 7.5 × 10⁴ RCECs/cm². Although skin keratinocytes and Caco-2 cells were shown primarily to secrete LM-5 and LM-10, respectively, in the cell environment, we could not exclude the possibility that they may also secrete other components from the ECM. To validate the results obtained with the keratinocyte- and Caco-2-secreted LMs, RCECs were also grown on culture plates coated with increasing concentrations of a commercial preparation of LM type-1 (LM-1) as a control. As shown on Figure 7C , LM-1 could also downregulate, though to a much lower level than with LM from skin keratinocytes, the expression directed by the α6 promoter contained on the α6–181 construct, validating the use of keratinocyte-mediated secretion of LM as a suitable model for the remaining experiments. Extending the α6 promoter 5′ further from position −181 did not result in any further LM-mediated repression (Fig. 7D) . In addition, deletion of the α6 promoter segment located from position −181 to −84 (in plasmid α6–84) had no influence on the repression by LM, suggesting that LM-responsive sequences are located somewhere between the α6 mRNA start site and position −84. 

To exclude the possibility that the LM-mediated negative influence on the α6 promoter might be typical only of RCECs, similar experiments were conducted on human corneal epithelial cells cultured from the corneal limbus obtained from the eyes of a healthy donor (HCECs) and grown to either passage (P) 2 or P3. As with RCECs, LM resulted in a severe downregulation of α6 promoter activity in HCECs (Fig. 7E) . However, the amplitude of this LM-mediated repression varied considerably with the number of passages reached by these cells. Indeed, a dramatic 25-fold reduction of α6 promoter function was observed when P2 HCECs were grown on LM. However, expanding these cells for one more passage, at P3, considerably reduced the amplitude of this repression to fivefold, consistent with the abrupt terminal differentiation observed in HCECs; most of them cannot be cultured beyond P3 or P4. ^{37 38}  

Laminin Alters the Nuclear Level of Sp1 and Sp3 in RCECs

After we established that Sp1 and Sp3 contribute to the transcription of the α6 gene by interacting with proximal and distal Sp1p and Sp1d sites, respectively, we reasoned that LM may exert its repressive influence by altering the nuclear level of each transcription factor. Crude nuclear extracts were obtained from RCECs grown to midconfluence (near 80% coverage of the plate) on culture plates coated with BSA (which is used as a negative control) or with LM and were examined in EMSA for their ability to sustain the binding of both Sp1 and Sp3 to a labeled probe bearing a high-affinity target site for these transcription factors. As shown on Figure 8A , shifted DNA-protein complexes with electrophoretic mobilities identical to those reported for Sp1 and Sp3 could easily be observed in the nuclear extract from RCECs grown solely on BSA (−LM). Specificity of the formation of these complexes was further confirmed through competition experiments in EMSA. Indeed, only the unlabeled Sp1 oligonucleotide, but not the oligomers bearing the target sites for the unrelated transcription factor NFI or AP-1, could compete for the formation of the Sp1/Sp3 complexes (Fig. 8A , right). The identity of the nuclear proteins yielding these complexes as Sp1 and Sp3 was further verified by supershift experiments in EMSA. As shown on Figure 8B(left, −LM), adding a polyclonal antibody directed against Sp1 dramatically reduced the formation of the slowest shifted complex on gel without altering that of the faster migrating complex. On the other hand, adding an Sp3 polyclonal antibody reduced the formation of the slow-migrating complex but totally prevented the formation of the faster one. Both antibodies generated slow-migrating, supershifted complexes (SCs) that corresponded to the recognition of the Sp1 or the Sp3 protein component from the DNA-protein complexes detected on the gel by the Sp1 and Sp3 antibodies, respectively. These results suggest that Sp1 and Sp3 are constituents of the more intense, slow-migrating complex seen in EMSA, whereas the less intense, faster migrating complex is exclusively constituted of Sp3 bound to the labeled probe. Both the slow- and the fast-migrating complexes disappeared when incubated along with both the Sp1 and the Sp3 antibodies (Sp1/Sp3Abs; Fig. 8B , left). 

To eliminate the possibility that the LM influence on Sp1 might have resulted from a sporadic event, crude nuclear extracts were prepared from RCECs grown to midconfluence on either LM- or FN-coated culture plates as a control. As shown on Figure 8C(left), recognition of the Sp1-labeled probe by both Sp1 and Sp3 was substantially increased in RCECs grown in the presence of FN, as previously reported. ³⁰ On the other hand, binding of both these proteins was again dramatically reduced when cells were grown on LM-coated culture plates (Fig. 8C , right). Western blot analyses were then conducted to discern whether these alterations in the ability of Sp1 to recognize its target probe when cells were grown on FN or LM resulted from changes in the affinity of these proteins toward their target sites or from corresponding alterations in the total amount of each individual protein. As we previously reported, ³⁰ culturing cells on FN did not change the amount of Sp1 or Sp3 proteins expressed by RCECs, a clear indication that alterations in the affinity of Sp1 and Sp3 toward their target site accounted for the increased binding of Sp1/Sp3 observed in EMSA (+FN; Fig. 8D ). However, culturing RCECs in the presence of LM caused a dramatic reduction in the nuclear concentration of Sp1 and Sp3 (+LM; Fig. 8D ) that correlated perfectly with the reduced binding observed in EMSA for both these proteins. These results are consistent with the reduced α6 promoter activity observed in RCECs when cells were grown on LM-coated culture plates (Fig. 7) . 

To determine whether the LM-mediated repression of the α6 promoter also translated to similar alterations in the expression of the endogenous α6 mRNA transcript, semiquantitative RT-PCR analyses were conducted. To eliminate the possibility that saturation of the PCR reaction is reached, amplifications were performed at 24, 26, 28, 30, 32, and 34 cycles. The specific α6 PCR product became detectable on gel after 24 cycles of amplification. Its formation happened to be nonlinear at 30 cycles and reached complete saturation after 32 cycles. PCR amplification of the α6 product remained fairly linear from 26 to 30 cycles of amplification (data not shown). As shown on Figure 9A , PCR amplification using both α6 primers yielded a single α6 DNA fragment of the correct size (210 bp) when total RNA was extracted from RCECs grown on BSA (only the result at 28 cycles of PCR amplification is shown). However, a significant reduction was observed when total RNA was obtained from cells grown on LM-coated culture plates. We then isolated total RNA from RCECs grown on either BSA or LM and examined the α6 mRNA transcript by Northern blot analysis. As shown on Figure 9B , a single mRNA species of approximately 5.5 kb that perfectly matched the size previously reported for the major α6 mRNA transcript ²² was observed just below the 18S rRNA. As with RT-PCR, expression of the α6 transcript was considerably reduced when RNA was isolated from RCECs grown on LM, on normalization to GAPDH. 

As a further support of our in vitro results, binding of Sp1 and Sp3 was also examined in vivo by the ChIP assay on HCECs cultured on BSA or LM-coated culture plates (+LM). As additional controls, ChIP was also conducted on other transcription factors, such as NFI and E2F1, of which some (E2F1) are suspected to bind the α6 promoter. ⁶¹ Antibodies against these transcription factors all enriched the α6 promoter sequences in HCECs, indicating that they are bound to this genomic area in vivo (Fig. 10A) . However, among them, the signal obtained with the Sp3 antibody clearly predominated. When ChIP was conducted on cells grown on LM, all four proteins again enriched the α6 promoter but with signal intensities different from HCECs grown on BSA. Indeed, normalization of the PCR amplification products resulting from the α6 promoter segment immunoprecipitated by each antibody to the input chromatin control indicated a clear reduction in the binding of Sp3 and E2F1 and a tendency of Sp1 to decrease as well when cells were grown on LM (Fig. 10B) . Occupancy of the α6 promoter by NFI in relation to the “input” was not altered by growing cells on LM. 

Discussion

Tissue repair requires the adhesion of proliferating cells to the basement membrane and cell migration to cover the wound. ⁶² The way wound healing of the corneal epithelium proceeds is basically the same for every type of tissue epithelium (for a review, see Midwood et al. ⁶³ ). In this study, we investigated the functions of Sp1 and Sp3 in the transcription of the LM-binding α6 integrin subunit gene in various primary cultured and established tissue culture cells. Both transcription factors were found to contribute in varying degree to the transcription of the α6 gene by interacting with two distinct target sites, Sp1p and Sp1d, along the promoter of the α6 gene. Most of all, culturing RCECs in the presence of LM caused a substantial reduction in the expression directed by the α6 promoter that resulted from a coordinated reduction in the level of expression of Sp1 and Sp3 in these cells. 

Sp1 is the founding member of a Zn-finger family of transcription factors, the Sp family, which now includes nine Sp genes (Sp1-Sp9; for a review, see Zhao and Meng ⁶⁴ ). Sp1 is of a particular interest as its GC-rich target site is found in a very large number of eukaryotic genes, including many integrin subunit genes (for further details, see Zaniolo et al. ⁶⁵ ). Consequently, Sp1 is likely to mediate some, if not most, of the regulatory influences the many ECM components exert on the transcription of the integrin subunit genes. By exploiting in vitro DNAse I footprinting, two target sites for Sp1 and Sp3 were identified along the human α6 promoter; a proximal site (Sp1p, centered at position −50) located near the mRNA start site and a more distant site (Sp1d, centered at position −205). The position of the proximal Sp1 site identified in this study (−38 to −62) is consistent with that previously reported for this transcription factor, ³¹ unlike the distal site, which has never before been reported. However, the affinity of Sp1 for this more distantly located target site proved to be eightfold lower than that of the proximal site, and its mutation alone had little influence on the basal promoter function in most types of transfected cells. On the other hand, when the Sp1p site was mutated in the shorter α6 promoter-bearing construct α6–84, dramatic reductions in basal promoter activity were observed in all types of cells. However, its mutation had marginal influence considering the longer construct that included an intact Sp1d site (in α6–352). It was only when both Sp1 sites were mutated that a substantial reduction in α6 promoter activity was observed in all types of cells, indicating that both Sp1 sites are redundant with each other. Yet, despite the mutation of both Sp1 sites, basal α6 promoter activity remained relatively high in most types of cells. Remarkably, when the proximal site was mutated, Sp1 was found to bind with a lower affinity to an alternative target site located immediately 5′ of Sp1p. Examination of this sequence revealed that it contains a GA- and a GT-rich sequence. Sp1 and Sp3 have been reported to bind to GC-, GT-, or GA-rich target sites with nearly the same affinity ⁶⁶ and are therefore likely to bind the alternative Sp1p_a GT- or GA-rich Sp1 sites identified in the α6 basal promoter. Use of an alternative degenerated site by Sp1 may explain the residual promoter activity observed despite that both the Sp1p and the Sp1d sites were mutated in the α6–352 construct. This constitutes a noteworthy evolutionary advantage for Sp1. Its ability to bind and use alternative low-affinity binding sites would ensure the maintenance of a minimal promoter activity when high-affinity sites are lost during the course of evolution. 

Dramatic differences were observed in the level of α6 expression (evaluated by FACS analyses) and α6 promoter activity between primary cultured RCECs and HSKs, both of which expressed α6 to high levels, and HSFs, which barely directed any promoter activity at all (5755- and 5025-fold lower than in RCECs and HSKs, respectively, with the α6–352 construct). These results are consistent with basal corneal epithelial cells and skin keratinocytes, two epithelial-type cells, expressing high levels of the α6 integrin subunit while stromal fibroblasts from the cornea do not. ^{9 23 38 60 67 68 69 70 71} By exploiting primary RCEC cultures to study the influence played by the ECM on the expression of integrin subunit genes, we previously reported dramatic increases in the activity directed by the promoter of the α4 and α5 integrins when these cells are grown in the presence of FN. ^{30 72} In addition to α4 and α5, FN considerably increased (by approximately fivefold) the activity of the α6 promoter in in vitro cultures of RCECs (Fig. 7A) , consistent with the results reported by Grushkin-Lerner et al. ²² Interestingly, human LM did not increase but rather reduced to various degrees the expression of the α6 promoter (and that of the α4 and α5 integrin gene promoters ⁷² ), irrespective of whether corneal epithelial cells were primary cultured from rabbit or human eyes. Differences in the strength of LM repression between keratinocyte-secreted LM and commercially produced LM-1 may reside in the fact that skin keratinocytes primarily secrete type 5 LM, which is also the major LM isoform secreted during corneal wound healing. Binding of Sp1 and Sp3 to the α6 promoter primarily accounts for its transcription in the many types of cells transfected in the present study. Surprisingly, culturing RCECs in the presence of LM resulted in a reduction in Sp1 binding caused by a corresponding decrease in its expression at the protein level. The appearance of faster migrating DNA-protein complexes in EMSA (Figs. 8A 8B 8C , +LM) and much smaller protein products in Western blot (Fig. 8D , +LM), when cells are grown on LM but not on FN or BSA, indicate that both factors must be subjected to protein degradation by a yet unknown protease(s) that may become activated by the signal transduction pathway activated by the binding of LM to its integrin receptor subunit α6. Similar degradation of Sp1/Sp3 was demonstrated when RCECs reached quiescence at a high cell density. ⁵² Interestingly, Sp1 expression has been shown to predominate during the G1 phase of the cell cycle and is then subjected to proteasome-dependent degradation before the S phase, a process thought to be dependent on the level of Sp1 phosphorylation. ⁷³ Consequently, LM, by reducing the level to which Sp1 and Sp3 are expressed in RCECs in vitro, also contributes to the reduction in the transcription directed by the α6 promoter when cells are grown on this ECM component. 

Although the amount of intact Sp1 clearly decreases when corneal epithelial cells are grown on LM, only marginal reduction of α6 promoter occupation by Sp1 is observed in ChIP. However, ChIP is by no mean a reliable quantitative method. At the most, one may consider ChIP semiquantitative when repeated experiments yield reproducible results on normalization to the input DNA, as has been the case for Sp3 and E2F1 in the present study. We have shown previously that many of the Sp1/Sp3 cleavage products retain their ability to bind Sp1 target sites despite the truncations they bear in their N-terminal domain ^{49 52} (see also Figs. 8B 8C , which show the appearance of fast-migrating DNA-protein complexes as Sp1 and Sp3 disappear when cells grow on LM). Therefore, one may assume that such Sp1/Sp3 degradation products will also bind the α6 promoter in vivo when examined by ChIP assays. Without the combined use of EMSA and Western blotting, ChIP would have failed to recognize these α6 promoter-bound proteins as degradation products of Sp1/Sp3 because it does not discriminate for changes in the molecular mass of these proteins. ChIP only requires that the epitope recognized by the antibody used for the immunoprecipitation step be preserved in the truncated proteins, which is precisely the case for Sp1 and Sp3; some of their corresponding degradation products are easily recognized by the Sp1 and Sp3 Abs in Western blot ^{49 52} (also see Fig. 8D ). We must therefore consider ChIP for what it truly is: a qualitative method that tells us whether any given transcription factor binds in vivo to a particular gene regulatory area with a resolution range between 300 and 500 bp (which corresponds to the average size of the sonicated, cross-linked DNA used for the assay). 

The finding that NFI binds the α6 promoter in ChIP assays, irrespective of whether HCECs are grown on LM, suggests that members from this family of transcription factors, which comprises four proteins—NFI-A, NFI-B, NFI-C, and NFI-X—may contribute to the extinction of α6 gene expression under this culture condition. Interestingly, members of the NFI family are reported to be as efficient as repressors ^{74 75 76} as they are as activators ^{77 78} of gene transcription. Through interactions with weak binding sites, NFI may regulate gene expression by its ability to cooperate or to compete with many transcription factors. ^{74 79} NFI was reported to interact with and antagonize Sp1, resulting in the downregulation of platelet-derived growth factor (PDGF)-A gene expression. ⁷⁹ We recently reported that similar interference of Sp1 action by NFI might also account for the transcriptional repression of the PARP-1 ^{74 80} and p21 genes. ⁸¹ A search using computer databases for the identification of transcription factors target sites failed to identify any intact, prototypical NFI binding site in the proximal promoter of the α6 gene but did find a number of putative half-canonical target sites for this transcription factor (data not shown). Therefore, the possibility remains that NFI might bind these degenerated target sites because this transcription factor has been reported to bind half its canonical sequence ⁸² in the basal promoter of many genes, including those from surfactant protein-C (SP-C ⁸³ ), 3β-hydroxysteroid/dihydrodiol dehydrogenase (3β-HSD/DD, AKR1C9 ⁸⁴ ), and metallothionein-I (MT-I ⁸⁵ ). 

In addition to NFI and Sp1, the in vivo ChIP analyses evidenced the binding of E2F1 to the α6 promoter. Most interestingly, of all the transcription factors tested, E2F1 was the one whose binding to the α6 promoter was most diminished by the presence of LM on the culture plates. Members of the E2F family play a crucial role in the transactivation of G1/S transition-specific genes ⁸⁶ and are thereby required to ensure proper cell proliferation. By conducting a representational difference analysis to identify genes differentially expressed in E2F1^−/− versus wild-type undifferentiated primary keratinocytes, D’Souza et al. ⁶¹ identified four integrin gene products (α5, α6, β1, β4) of 11 distinct genes downregulated in E2F1-deficient cells, suggesting that their transcription is clearly modulated by E2F1. E2F1^−/− knockout mice had a considerably reduced wound healing response after partial tail amputation, suggesting that the α5 and α6 integrin subunits must contribute to this process. ⁶¹ D’Souza et al. ⁶¹ were unable to reveal the existence of E2F-binding elements in the 5′-flanking sequence of this integrin gene. However, the need for a perfect prototypical E2F1 site (TTTSSCGC, in which S = C or G) ⁸⁷ in the α6 promoter is not an absolute requirement because only 12% of the E2F1 target sites occupied by this transcription factor in the human genome have the canonical binding motif, as recently revealed by high-density oligonucleotide tiling arrays. ⁸⁸ Other studies have shown the importance of preserving the central SSCGC for E2F binding, ⁸⁹ whereas binding can still be retained with a single mismatch in the three Ts. ⁸⁷ Therefore, we searched for E2F1 target sites in the α6 promoter that might have diverged from the consensus by one of the three Ts. Three such sites could be identified from positions −191 to −184 on the top strand (5′-TTcGCCGC-3′) and from −113 to −106 (5′-GCGGGtAA-3′) and −79 to −72 (5′-GCGCGcAA-3′) on the bottom strand. Their true functional significance remains to be established. 

Considering the results presented in this study, the binding of the α6β4 integrin to its ligand LM is expected to trigger growth-regulatory signals (probably negative ones) completely distinct from those resulting from the binding of FN to the α5β1 integrin. Although the positive influence of FN and collagen on cell migration is well documented, that of LM remains controversial because it has been reported to promote or to suppress cell adhesion and migration. Evidence points, however, toward a major role for LM in the suppression of cell growth and migration. Indeed, the overexpression of β4 in a cell that lacks this integrin subunit slows down the migration of that cell. ⁹⁰ LM-5 densely deposited under the cell is suggested to contribute to the stabilization of the α6β4 integrin binding to this ECM ligand. Retroviral-mediated siRNA suppression of LM-5 in an established oral squamous cell carcinoma (OSCC) cell line substantially enhanced the migratory, tumorigenic, and invasive properties of these cancer cells. ⁹¹ Several breakdown products of LM-5, which is composed of three subunits (α3β3γ2), have been reported to exert different influences that may explain, at least in part, the paradoxical action of LM on cell migration. For instance, the proteolytic processing of the α3 chain of LM-5 by plasmin produces an LM-5 molecule that induces the assembly of hemidesmosomes and impedes cell migration. ⁹² On the contrary, proteolytic processing of the γ2 chain is related to the induction of cell migration. ⁹³ Therefore, the secretion of LM while FN declines during the wounding process might signal to epithelial cells the proper time to stop migrating by inducing the degradation of key transcription factors through the proteasome, such as Sp1, thereby shutting down expression of a large number of genes required for cell proliferation and adhesion. Another function LM might play, besides restricting cell migration, would be to stimulate the differentiation of the basal epithelial cells into suprabasal cells as they migrate toward the apical surface of the cornea through their vertical stratification. ⁹⁴ Recently, polycarbonic membranes coated with various ECM components were implanted on the wounded corneas of adult cats and were assessed for rapidity and for completeness and persistence of epithelial overgrowth. ⁹⁵ It turned out that FN, though efficient for wound closing, could not support persistent epithelial overgrowth, whereas LM, collagen type IV, and collagen type I were all effective in yielding multiple-layered epithelia. These results further support a role for FN in the migration of basal epithelial cells early in wound healing so that the wounded area becomes rapidly covered with epithelial cells, whereas LM, CI, and CIV would act at a later stage to favor differentiation and vertical stratification of the basal cells to reconstitute an intact corneal epithelium. 

In summary, each component from the ECM appears to trigger specific transcriptional regulatory signals when taken individually. Some, such as FN and collagen type IV, mediate positive, cell density-specific influences on gene expression (reviewed in Ref. ⁷² ), whereas others, such as LM, suppress gene transcription. Investigating how they influence gene expression when used in combination should be particularly valuable to the understanding of tissue wound healing. 

 Supported by the Canadian Institutes for Health Research (CIHR; Grant 37946) and the Vision Network of the Fonds de la Recherche en Santé du Québec (FRSQ). MG and FV hold Doctoral Research Awards from the FRSQ and the CIHR, respectively. SLG is a research scholar from the FRSQ.

 Submitted for publication January 5, 2007; revised March 29, 2007; accepted June 12, 2007.

 Disclosure: M. Gaudreault, None; F. Vigneault, None; S. Leclerc, 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.

 

DNAse I footprinting of Sp1 binding to the α6 gene promoter. (A) A preparation of recombinant Sp1 (1–10 μL) was incubated with a 217-bp α6 probe labeled on either the top (Top) or the bottom strand (Bottom) and subjected to DNAse I digestion. The digestion products were then analyzed by gel electrophoresis through an 8% sequencing gel. The position of a promoter proximal Sp1 protected site (Sp1p) is indicated for both the top and bottom strands. G, Maxam and Gilbert G sequencing ladder; C, labeled probe incubated with DNAse I but without recombinant Sp1. (B) Same as in (A) except that the labeled probe consisted of a 430-bp HindIII-KpnI DNA fragment from the α6 promoter 5′ end-labeled on either its bottom or top strand. The position of a promoter distal Sp1 protected site (Sp1d) is indicated for both the bottom and top strands. (C) Positioning of the DNAse I footprinted Sp1 sites along the α6 promoter sequence. The positions of proximal (Sp1p) and distal (Sp1d) Sp1 target sites (boxes) relative to the α6 mRNA start site (identified as +1) are provided, along with the α6 promoter 5′ end point from the recombinant constructs α6–84, α6–141, α6–181, and α6–352.

Figure 2.

 

FACS analysis of integrin α6 expression in primary cultured and established tissue culture cells. Expression of the α6 integrin subunit was monitored by FACS analyses in primary cultures of RCEC, HSK, and HSF cells and in established HeLa and ARPE19 cells. Results are expressed as low (+) to very strong (++++) expression of α6 (bottom right table). Typical histograms of ++++ (RCECs), ++ (HeLa), and + (HSFs) are also provided. Values presented are from four individual measurements from two different batches of cells.

Figure 3.

 

Transfection analyses in primary and established tissue culture cells. (A) Schematic representation of the recombinant α6/CAT constructs used in this study. Positions of proximal (Sp1p) and distal (Sp1d) Sp1 target sites identified by DNAse I footprinting are indicated along with the 5′ end of each of the α6 promoter segment contained on the recombinant construct. Mutations that disrupt binding of Sp1 to proximal or distal Sp1 sites (in plasmids α6–84/mSp1p, α6–352/mSp1p, and α6–352/mSp1d), or both (in α6–352/mSp1pd), are indicated by the removal of the corresponding Sp1 box. (B–F) Site-directed mutational analysis of Sp1p and Sp1d. Wild-type constructs α6–84 and α6–352, and their derivatives bearing mutations into either Sp1p (α6–84/mSp1p and α6–352/mSp1p) or Sp1d (α6–352/mSp1d), or both Sp1 sites (α6–352/mSp1pd), were transfected into primary cultured cells (RCECs, B; HSKs, C; HSFs, F) or established tissue culture cells (HeLa, D; ARPE19, E). Cells were then harvested, and CAT activity was determined and normalized to secreted hGH. SD is provided.

Figure 4.

 

Transfections in Drosophila SL2 Schneider cells. Both the recombinant plasmids α6–141 (which bears only the Sp1p site) and α6–352 (which bears both the Sp1p and the Sp1d sites) and their Sp1-mutated derivatives (α6–141/mSp1p, α6–352/mSp1p, α6–352/mSp1d, and α6–352/mSp1pd) were transfected alone or with pPacSp1 (Sp1), pPacSp3 (Sp3), or both (Sp1/Sp3) into Drosophila SL2 Schneider cells. Cells were harvested 48 hours later, and CAT activities (expressed as fold activity relative to the level directed by the α6–141 or the α6–352 promoter constructs alone) was determined and normalized to β-gal. SD is also provided.

Figure 5.

 

DNAse I footprinting of an Sp1 alternative target site on the Sp1p-mutated α6 promoter. Recombinant Sp1 (2, 5, or 10 μL) was incubated with the 217-bp α6 probe used in Figure 1A , which bears either an intact (wt) or a mutated (mutant) Sp1p site and 5′ end-labeled on either the top (A) or the bottom strand (B). DNA-protein complexes were then subjected to DNAse I digestion, and the products were analyzed by gel electrophoresis through an 8% sequencing gel. The position of the promoter proximal Sp1 protected site (Sp1p) identified on the wild-type promoter fragment (wt) is indicated, along with that of an alternative Sp1 target site (Sp1p_a) identified on the labeled probe bearing mutations in the Sp1p site (mutant). G, Maxam and Gilbert G sequencing ladder; C, labeled probe incubated with DNAse I but without recombinant Sp1. (C) Same as in Figure 1Cexcept that the position of the alternative Sp1 site (Sp1p_a) is indicated relative to the α6 mRNA start site (identified as +1).

Figure 6.

 

EMSA analysis of Sp1 binding to the Sp1 sites from the α6 promoter. (A) DNA sequence of the various Sp1 oligonucleotides used in the EMSA. (B) Crude nuclear proteins (5 μg) from RCECs grown on BSA were incubated with the Sp1-labeled probe either alone (C) or in the presence of double-stranded oligonucleotides bearing the sequence of either the Sp1p or Sp1d sites identified in the α6 promoter, or that of the alternative Sp1p_a site, as unlabeled competitors. Oligonucleotides bearing the high-affinity binding sites for the unrelated transcription factors NFI and E2F1 were also used as negative controls for the competition experiment. Formation of the DNA-protein complexes was then monitored by EMSA. The position of the Sp1 and Sp3 complexes is shown. P, labeled probe alone; U, unbound fraction of the probe.

Figure 7.

 

Transfection of the α6 promoter in RCECs and HCECs grown in the presence of LM. LM-producing dermal keratinocytes (A) or colon carcinoma Caco-2 cells (B) were grown to 30%, 70%, or 100% cell density for 1, 3, 5, and 10 days (for skin keratinocytes) or to 100% confluence for 25 days (for Caco-2). Keratinocytes and Caco-2 cells were then completely removed, and the culture plates were immediately reseeded with RCECs (2.5 × 10⁴/cm² on LM from skin keratinocytes and 2.5 × 10⁴, 5 × 10⁴, and 7.5 × 10⁴/cm² on LM from Caco-2 cells) that were grown to midconfluence before transfection with the recombinant plasmid α6–181. Negative controls (−LM) correspond to RCECs seeded on culture plates coated with BSA. CAT activities were measured and expressed relative to the level directed by α6–181 transfected in cells grown without LM. SD is also provided. (C) RCECs were plated on culture plates coated with BSA (used as a negative control; −LM) or with varying concentrations of a commercial preparation of LM type 1 (0.5–8 μg/cm²) before transfection with the plasmid α6–181, as in panel (A). (D) Dermal keratinocytes were grown on culture plates until they reached 100% confluence for 5 days and then were completely removed. Culture plates were reseeded with RCECs before they were transfected with the recombinant plasmids α6–84, α6–141, α6–181, α6–352, α6–430, and α6–590. CAT activities were measured and normalized as in Figure 2and are expressed relative to the level directed when transfected in RCECs without LM. (E) Same as in panel (D), except that LM-coated plates were reseeded with RCECs (as a control) and HCECs at P2 or P3 before transfection with the α6–181 recombinant construct. As negative controls, RCECs and HCECs were also seeded on BSA before transfection with α6–181 (−LM).

Figure 8.

 

Expression of Sp1 and Sp3 in RCECs grown on LM and FN. (A) EMSA analysis of Sp1 binding in RCECs cultured on LM-coated plates. Crude nuclear proteins (5 μg) from RCECs grown on BSA (−LM) or LM-coated culture plates (precultured with skin keratinocytes for 5 days before removal; +LM) were incubated with the Sp1-labeled probe (left). Where indicated, double-stranded oligonucleotides bearing high-affinity binding sites for the transcription factors Sp1, NFI, and AP-1 were added as unlabeled competitors (right). Formation of the DNA-protein complexes was then monitored by EMSA. The position of the Sp1 and Sp3 complexes is shown. P, labeled probe alone; U, unbound fraction of the probe. (B) Supershift analysis in EMSA. Nuclear proteins from RCECs grown on BSA (−LM) or on LM (+LM) were incubated with the Sp1-labeled probe in the presence of antibodies directed against Sp1 or Sp3, either individually (Sp1Ab, Sp3Ab) or in combination (Sp1/Sp3Abs). Positions of the Sp1 and Sp3 DNA-protein complexes were then revealed by EMSA, as in panel (A). SC, supershifted complexes of low electrophoretic mobility that correspond to the binding of the Sp-antibodies to the Sp1- and Sp3-DNA complexes; C, positive control in which the labeled probe was incubated with proteins but without antibodies. (C) Comparative influence of the ECM components LM and FN on the DNA-binding properties of Sp1/Sp3. Nuclear proteins (5 μg) from RCECs grown on BSA (−FN or −LM) or on culture plates coated with either LM (+LM) or FN (+FN) were incubated with the Sp1-labeled probe before analysis of the complexes formed by EMSA. (D) Approximately 30 μg nuclear proteins from the extracts used in (C) were examined in Western blot analyses using the Sp1 and Sp3 antisera. Positions of the molecular mass markers are shown.

Figure 9.

 

Influence of LM on the expression of the α6 mRNA transcript. (A) Total RNAs extracted from RCECs grown on BSA (−LM) or on LM-coated culture plates (+LM) were reverse transcribed and PCR amplified using synthetic, oligonucleotide primers specific to both the α6 and 18S ribosomal RNAs. Positions of the amplified 210-bp α6 (α6) and the 489-bp 18S fragments (18S) is indicated along with that of the most relevant markers (left). (B) Total RNA preparations used in (A) were subjected to analyses by Northern blot. Positions of the α6 5.5-kb mRNA is indicated, along with that of the 28S ribosomal RNA. As a control, the membrane was also hybridized with a 548-bp HindIII-XbaI fragment digested from the human glyceraldehyde-3-phosphate-dehydrogenase (GAPDH) cDNA and used as a labeled probe for normalization of the α6 mRNA signal to that corresponding to GAPDH (1.2 kb).

Figure 10.

 

In vivo ChIP analysis on the α6 promoter. (A) ChIP assays were performed on HCECs grown on either BSA or LM. Chromatin was isolated and immunoprecipitated with antibodies directed against the transcription factors Sp1, Sp3, NFI, and E2F1. PCR of the α6 (ITGA6) gene promoter was then carried out on the ChIP samples, along with a “no antibody” control (No Ab) that contains chromatin but no antibody, an “input” sample corresponding to 0.2% of the total input chromatin, and a “mock” sample that does not contain chromatin. PCR amplification of a gene segment located approximately 2000 bp upstream of the p21 promoter was also conducted on the same sample as a negative control for all immunoprecipitates. (B) Graph representing the amount of specific PCR products expressed as the percentage of antibody binding versus the amount of PCR product obtained using a standardized aliquot of input chromatin. The signal in the no-antibody lane corresponds to the nonspecific binding background and was subtracted from each sample.

The authors thank Claudia Fugère and Lucie Germain (Laboratoire d’Organogénèse Expérimentale [LOEX], Hôpital du Saint-Sacrement, CHA and Department of Surgery and Ophthalmology, Laval University, Québec, QC, Canada) for providing primary cultures of HSKs, HCECs, and HSFs. They also thank Vito Quaranta (Department of Cell Biology, The Scripps Research Institute, La Jolla, CA) for providing the complete α6 cDNA, and Jean-François Beaulieu (Centre Hospitalier de l’Université de Sherbrooke [CHUS], Sherbrooke, QC, Canada) for providing the Caco2/15 cells.