p51/p63 Controls Subunit α3 of the Major Epidermis Integrin Anchoring the Stem Cells to the Niche*

p51/p63, a member of the tumor suppressor p53 gene family, is crucial for skin development. We describe here identification of ITGA3 encoding integrin α3 as a target of its trans-activating function, proposing that p51/p63 allows epidermal stem cells to express laminin receptor α3β1 for anchorage to the basement membrane. When activated by genotoxic stress or overexpressed ectopically in non-adherent cells, p51/p63 transduced a phenotype to attach to extracellular matrices, which was accompanied by expression of ITGA3. Motifs matching the p53-binding consensus sequence were located in a scattered form in intron 1 of human ITGA3, and served as p51/p63-responsive elements in reporter assays. In addition to the trans-activating ability of the TA isoform, we detected a positive effect of the ΔN isoform on ITGA3. The high level α3 production in human keratinocyte stem cells diminished upon elimination of p51/p63 by small interfering RNA or by Ca2+-induced differentiation. Furthermore, a chromatin immunoprecipitation experiment indicated a physical interaction of p51/p63 with intron 1 of ITGA3. This study provides a molecular basis for the standing hypothesis that p51/p63 is essential for epidermal-mesenchymal interactions.

Germ line p51/p63 mutations in humans are associated with the ectroductyly, ectodermal dysplasia, and cleft palate syndrome (11) and other malformation syndromes (12). Immunohistochemical analyses with skin tissues revealed nuclear expression of the p51/p63 protein confined to the basal layer of epidermis (2). By a clonal analysis of human keratinocytes for proliferative potential and protein composition, it was determined that p51/p63 is specifically expressed in keratinocyte stem cells and transit amplifying cells, predominantly and less predominantly, respectively (13). Consistently, p51/p63-null mice exhibited striking epidermal defects: absence of keratinocyte stratification and differentiation, lack of normal basal cells expressing keratin 14, and exposed dermis (3,6). Thus, true biological activities of p51/p63 essential for skin development should be identified in keratinocyte stem cells.
More than 6 isoforms arise from p51/p63 by using the TAand ⌬N-type transcriptional initiation sites and by RNA splicing to form the C-terminal A/␥, B/␣, and C/␥ variants (2). Reflecting the structural similarity to p53 in the DNA binding domain, many of the genes inducible by p53 are also responsive to the p51/p63 proteins (1,2,14). Each of the TA isoforms acts more or less as a trans-activation factor, whereas the ⌬N isoforms lacking the TA domain exhibit dominant-negative activities against p53 and the p51/p63 TA isoforms in reporter assays (2). More recently, however, a report revealed that a ⌬N isoform of p73, another p53 homolog involved in neurogenesis, acts both as a positive and negative regulator of transcription (15). Furthermore, the TA isoforms of p51/p63 are unstable because of degradation by proteasome under normal conditions, but can accumulate in response to DNA damage to induce gene expression (16 -18).
p51/p63 has been implicated in transcriptional events related to cell growth and differentiation. Those include downregulation of the epidermal growth factor receptor expression (19), trans-activation of Jagged-1 encoding a Notch ligand (20), induction of ␤-globin expression indicative of erythroleukemic cell differentiation (16,17), and activation of REDD-1 implicated in redox stress responses (21). With all these findings, however, essential molecular and cellular mechanisms that are assigned to p51/p63 for epidermal stem cell regulation still remain obscure.
We report here that p51/p63 trans-activates the ITGA3 gene coding for integrin subunit ␣ 3 , also referred to as CD49c and VLA3-␣. The ␣ 3 subunit pairs with ␤ 1 to form the ␣ 3 ␤ 1 (VLA3) complex which falls into the category of major epidermis integrins with other members, ␣ 2 ␤ 1 and ␣ 6 ␤ 4 (22-24). Integrin ␣ 3 ␤ 1 is expressed predominantly in keratinocyte stem cells (23), and functions as a receptor for laminin, a major extracellular matrix (ECM) 1 protein in the basement membrane (also referred to as basal lamina). The critical interactions between the epidermal stem cells and their niche (25) may be facilitated by p51/p63 through its ability to induce ITGA3.

EXPERIMENTAL PROCEDURES
Cell Culture-U937 cell culture has been described (26). ECM binding assays were performed using culture dishes coated with laminin, fibronectin, collagen, or polylysine (BioCoat, BD Biosciences). Sixteen hours after incubation with actinomycin D, dishes were once washed with phosphate-buffered saline, and adherent and non-adherent cells were counted. Neonatal human keratinocyte (NHK) preparations from foreskin were from Cambrex. The basal medium was supplemented with human recombinant epidermal growth factor (0.2 ng/ml), insulin (10 g/ml), hydrocortisone (1 g/ml), gentamicin (0.1 mg/ml), amphotericin-B (100 ng/ml), and bovine pituitary extract (Keratinocyte Growth Medium BulletKit, Cambrex). Experiments with NHK were accomplished within three subculturing cycles, during which cells maintained a high proliferating potential: more than 80% of the cells grew to form 16-to 32-cell colonies on day 3 after attachment to the basement (day 0).
Inducible p51/p63 Expression in HEK293 Cells-We transfected HEK293 cells with pFRT/LacZeo and selected zeocin-resistant clones. The pFRT-positive cells were next transfected with tetracyclin repressor expression plasmid pcDNA6/TR, and blasticidine S-resistant clones were obtained. The pcDNA5/FRT/TO vector with each of the p51/p63 isoform cDNAs was co-transfected with pOG44 into HEK293 clones having pFRT and pcDNA5/TR. Stable hygromycin-resistant clones were maintained as cell lines that induce p51/p63 expression in response to tetracyclin (2 g/ml).
Plasmids-Overexpression of p51/p63 with the pRcCMV vector was described, as were luciferase expression constructs, pRGC-luc (18) and pGL-ES (32). To construct p(1ϩ2)luc, a synthetic double stand DNA having sequences 594 -612 and 877-901 derived from intron 1 of human ITGA3 was inserted into the multiple cloning site of a Luc expression vector, pGL3-promoter. p(3ϩ4)luc was constructed similarly, so that nucleotides 2465-2490 and 6437-6461 were inserted in tandem between the NheI and XhoI sites located 5Ј to the SV40 promoter. The (3ϩ4) segment was inserted at NheI/XhoI sites of p(1 ϩ 2) to create p(1ϩ2ϩ3ϩ4)luc. Human genomic DNA was purified from peripheral blood cells obtained from a healthy volunteer donor (Y. T.). A DNA region encompassing intron 1 of ITGA3 was amplified by PCR with primers targeting the 3Ј end of exon 1 and 5Ј end of exon 2: 5Ј-CTCCGCCTTCAACCTGGATACCCGATTCCT-3Ј and 5Ј-GGTACAC-AGCACCAGTCC GGTTGGTGTAGC-3Ј, respectively. The 7.6-kb segment containing the full intron 1 sequence was inserted into pGEM-T Easy. By BglII digestion, the 1.4-kb 5Ј terminal region was excised to be cloned in the pGL3-promoter at the BamHI site (Promega). Nucleotide mutations were introduced with a GeneTailor Site-directed Mutagenesis System (Invitrogen). PCR primers used were 5Ј-GAGAGAGG-AGGACTTTTCCCAACTCT-3Ј and 5Ј-AAGTCCTCCTCTCTCTTTCAC-CGC-3Ј for the M1 mutation, and 5Ј-ATGCCCAGGAGCATTTCCAGG-CTT-3Ј and 5Ј-ATGCTCCTGGGCATGAGCGCGTCT-3Ј for the M2 mutation. Nucleotide sequences were confirmed with the ABI 377 sequencer. Plasmids were purified from Escherichia coli DH5␣ by a GenElute Endotoxin-free Plasmid Kit (Sigma).
Indirect Immunofluorescence Microscopy-Mouse skin sections were fixed in 4% paraformaldehyde, embedded with paraffin, and sectioned at 8 m. Keratinocytes cultured on slide glass were also fixed with 4% paraformaldehyde. Deparaffinized tissue sections and cultured samples were heated at 95°C for 40 min in a buffer (10 mM sodium citrate buffer, pH 6.0; or with 50 mM glycine HCl buffer, pH 3.5, with 0.01% (w/v) EDTA) for epitope retrieval. Samples were permeabilized with 0.1% Triton X-100, blocked in phosphate-buffered saline containing 5% nonfat dry milk and 5% goat serum (30 min at room temperature), and incubated with a primary antibody (for 16 h at 4°C). Samples probed with fluorescein isothiocyanate-or rhodamin-labeled secondary antibody, for 1 h at 22°C, were washed and mounted in an anti-fade medium (Dako, Japan). Photomicrographs were obtained with a Leica DMRXA microscope equipped with an air-cooled camera (Synsys) controlled by QFISH software (Leica).
Chromatin Immunoprecipitation (ChIP)-HEK293 cells seeded on 15-cm plates were transfected with a plasmid for expression of human influenza virus hemagglutinin epitope (HA)-tagged p51A/TAp63␥ (HAp51A) (1). After 48 h, cells were fixed with formaldehyde, and the nuclei were purified. Sheared chromatin was immunoprecipitated with non-immune rabbit IgG, an anti-TFIIB antibody (Active Motif), or an anti-HA rabbit antibody (Zymed Laboratories Inc., 71-5500). Protein G-agarose blocked with salmon sperm DNA, RNase A, proteinase K, a proteinase inhibitor mixture, mini-columns for DNA purification, and buffers were supplied in the ChIP-IT kit by Active Motif. Adsorption of the antibody-protein-DNA complexes to protein G-agarose, washing, reversal of cross-links, removal of RNA, protein digestion, and DNA purification were performed following the manufacturer's protocol. A 166-base pair (bp) segment of the GAPDH gene promoter locus was amplified by PCR with primers: 5Ј-TACTAGCGGTTTTACGGGCG-3Ј and 5Ј-TCGAACAGGAGGAGCAGAGAGCGA-3Ј. For detection of the ITGA3 intronic sequences encompassing the first and second p53(p51/ p63)-binding consensus sequences, a 390-bp segment was amplified with primers: 5Ј-CTCTCCGTAATGGAAGACC-3Ј and 5Ј-GCATTAT-GAGACATCCCCAC-3Ј.

RESULTS
Induction of an ECM-binding Phenotype and Integrin ␣ 3 Expression-As we reported earlier, p51A/TAp63␥, a potent trans-activator isoform of p51/p63, accumulated in response to DNA damage to induce p21 waf1 in a mouse cell line (16,26). In this study, we propagated U/p51A and U/p53 cell lines derived from p53-deficient U937 human lymphoid cells after transfection with an Rous sarcoma virus promoter-driven, low-level expression vector for p51A/TAp63␥ and p53, respectively (16). In U/p51A cells, p51A/TAp63␥ accumulated 8 -24 h after exposure to 5 nM actinomycin D without an alteration in the mRNA level (Fig. 1A). The RT-PCR and Western blot analyses also showed activation of the p21 waf1 and GADD45 expression with an increase in the p21 waf1 protein amount. U/p53 cells incu-bated with actinomycin D also stabilized p53, inducing p21 waf1 . Control U937 cells caused a slight up-regulation of p21 waf at 16 h, probably by a p53-independent mechanism (33).
We examined those cells for attachment to basements coated with laminin, fibronectin, or collagen (Fig. 1B). More than 90% of the U/p51A cells incubated with actinomycin D attached to laminin-coated dishes within 16 h. Approximately 55 and 16% of the cells bound to fibronectin and collagen, respectively, with little evidence of their binding to polylysine. In contrast, only a 9% fraction of the drug-treated U/p53 cells bound to laminin, and a yet smaller fraction to fibronectin. Control U937 cells To find the molecular basis of the p51A-caused cell attachment to ECM, cellular RNA was analyzed for expression of integrin genes by RT-PCR (Fig. 1C). With two different primer pairs, ␣ 3 (#1), and ␣ 3 (#2), an increase in ITGA3 mRNA was evident at 8 -24 h after exposure to actinomycin D. By contrast, expression of the ␣ 4 , ␣ 5 , ␣ 6 , and ␣ M genes remained constant through the time course. Although ␤ 1 gene expression increased slightly, reaching maximum at 8 h, the response was not specific to U/p51A. Consistent with RT-PCR analysis, the 145-kDa ␣ 3 (Ref. 34) content increased in U/p51A cells during the period (Fig. 1C, bottom). Because the ␣ 3 subunit associates with ␤ 1 to form the ␣ 3 ␤ 1 heterodimer whose major and minor ligands are laminin and fibronectin/collagen (35), respectively, we hypothesized that p51/p63 promotes ITGA3 expression directly or indirectly to cause cell-ECM attachment.
Transient or Drug-controlled p51/p63 Expression also Activates ITGA3-We performed transient overexpression of p51/ p63 in U937 cells using a cytomegalovirus promoter vector with a liposome-based transfection carrier. The p51A/TAp63␥ expression vector, but not the vector-only plasmid, caused laminin binding activity (Fig. 2, A and B). Reflecting the transfection efficiency, 10%, determined by a G418-resistant colony formation assay, a ϳ10% fraction of the cells bound to laminin (Fig. 2B). Although the apparent increases in the mRNA and protein of ␣ 3 were not so great (2-3-fold) when the entire culture was analyzed ( Fig. 2A), we could speculate that there was a greater increase in ␣ 3 expression in the attached cell population (10%). Furthermore, ⌬Np51A/⌬Np63␥ lacking the trans-activation domain also generated cells adherent to laminin (Fig. 2B), which was accompanied by ITGA3 activation as detected by mRNA and protein analyses ( Fig. 2A). In contrast to the poor retention of 57-kDa p51A/TAp63␥ having the Nterminal sequences that determine its fate for degradation by proteasome (16,18), ⌬Np51A/⌬Np63␥ was stable enough to form an intense band at 52 kDa in Western blotting ( Fig. 2A). Thus, neither the laminin-adherent phenotype nor the ␣ 3 induction shown in Fig. 2 required a cellular signaling event caused by actinomycin D. When expressed to a certain level, the TA and ⌬N proteins seemed to induce ITGA3.

FIG. 3. Finding of p51/p63-responsive sequences in intron 1 of ITGA3.
A, schematic presentation of the 5Ј end of human ITGA3. The promoter-enhancer region, exon 1 (279 bp), intron 1 (7431 bp), and the 5Ј part of exon 2 are shown. Closed triangles at positions 597, 881, 2475, and 6422 (relative to the 1st nucleotide of intron 1) indicate putative p53(p51/p63)-responsive sites. The (1ϩ2) and (3ϩ4) segments inserted 5Ј to the promoter of the luc vector are depicted in coordination with their original locations in intron 1. Nucleotide sequences matching the p53(p51/p63)-binding consensus sequences are underlined. B, luc-expression assays with p(1ϩ2)luc, p(3ϩ4), and p(1ϩ2ϩ3ϩ4)luc. The core structure comprising the enhancer-type insert, the promoter (P), and luc coding region (luc) is schematically shown for each reporter construct. Co-expressed activators, p53, p51A/TAp63␥, and ⌬Np63␥, are indicated below the columns. Basal level luciferase activities were obtained by co-transfection with the vector (pRcMCV). Luminescence intensity is presented in relative light units ( p51/p63 Induces Integrin ␣ 3 and ⌬Np51A/⌬Np63␥ caused a 2.5-fold activation of (␣3-1)luc, implying an interaction between the TA and ⌬N isoforms. When a single mutation, G603T, and a double mutation, G603T/G887T, were introduced into (␣ 3 -1)luc to render M1(␣ 3 -1)luc and M1M2(␣ 3 -1)luc, respectively, the efficiency of transactivation by p51A/TAp63␥ dropped to 1.4-and 1.2-fold. These results indicated that the 1st and 2nd half-site motifs cooperatively play an essential role in response to p51A/TAp63␥. The ⌬N isoform seemed to act on the same sites to cause moderate activation. On the other hand, (␣ 3 -4)luc, in which the 1.4-kb segment was placed in the 5Ј to 3Ј orientation, produced a high background of luciferase activity, and was less sensitive to p53 and p51/p63 (Fig. 3D), implying that G/C-enriched sequences in the 5Ј terminal region of the insert affected the heterologous viral promoter activity in the vector.
Concurrent Expression of p51/p63 with ␣ 3 in Epidermis Development-Immunostaining of skin sections from mouse embryos on day 14 (E14) showed p51/p63 protein localization in the inner layer of the double-layered surface ectoderm or the periderm (Fig. 4A, far left). The p51/p63 nuclear stain intensified in the basal layer of epidermis on E16 when epidermis stratification was in progress. In newborn mice, however, the overall p51/p63 stain significantly decreased, leaving p51/p63positive cells in clusters that corresponded to the patches of keratinocyte stem cells (13). The double immunofluorescence analysis showed that nuclear p51/p63 stain (fluorescein isothiocyanate) coincided with peripheral ␣ 3 (Rhodamin) stain in the basal cells of the E14, E16, and newborn tissues (Fig. 4A,  three right panels). Furthermore, the ␣ 3 label was also markedly weakened at birth. In their temporal and spatial expression profiles, p51/p63 and ␣ 3 were closely related to each other in mouse skin development.
The ␣ 3 Expression Is Associated with p51/p63 in Keratinocyte Stem Cells-We cultured NHK to analyze p51/p63 and integrin ␣ 3 expressions. More than 90% of the cells in the NHK culture were able to replicate to form a colony of 16 or 32 descendant cells within 3 days after plating with medium containing 0.1 mM Ca 2ϩ (low Ca 2ϩ ). As judged by the high growth potential, epidermal stem cells were predominant in the culture. By changing the extracellular calcium concentration, we could control growth and differentiation of the epidermal cells (41)(42)(43). When the Ca 2ϩ concentration was raised to 1 mM (high Ca 2ϩ ), those cells ceased replicating and underwent differentiation as described. Seven days after incubation with high Ca 2ϩ , we detected mRNA for involucrin, a keratinocyte differentiation marker (Fig. 4B).
In NHK (day 0), 145-kDa full-length ␣ 3 appeared predominant in the Western blot with the antibody reactive with the cytoplasmic peptide. On day 4 of Ca 2ϩ induction, the 30-kDa light chain of ␣ 3 (44), instead of the 145-kDa protein, formed an intense band, indicating that the ␣ 3 maturation process had became active (Fig. 4B). We did not detect a decrease in the ITGA3 expression by RT-PCR for at least 4 days after the Ca 2ϩ input. On day 7, when the p51/p63 protein content had mark- FIG. 4. Changes in p51/p63 and integrin ␣ 3 during keratinocyte differentiation. A, p51/p63 immunostain developed with 5-bromo-4-chloro-3-indolyl phosphate (far left panels) and double immunofluorescent staining for p51/p63 with fluorescein isothiocyanate and integrin ␣ 3 with rhodamin (three right panels). Skin sections from E14 and E16 embryos and newborn were stained simultaneously on the same glass slide. Bars indicate 50 m. B, analyses for mRNA and protein compositions in NHK. An NHK culture that had been maintained with 0.1 mM Ca 2ϩ (day 0) was incubated with 1 mM Ca 2ϩ (high Ca 2ϩ ) in the medium to days 4 and 7. mRNAs examined by RT-PCR are indicated (left panels). Western blots (right panels) show p51/p63 isoforms, immature (145 kDa) and mature (30 kDa) forms of integrin ␣ 3 and ␤-actin (41 kDa). Dots indicate protein standards (sizes in kDa) in lane M.
edly decreased, the ␣ 3 mRNA and protein levels had declined markedly. In contrast, p21 waf1 gene expression was up-regulated on days 4 and 7 compared with day 0, possibly because of the decline in the ⌬N isoforms that negatively regulate p21 waf1 . The level of integrin ␤ 1 mRNA did not change significantly up to day 7. Thus, p51/p63 protein decrease preceded suppression of ITGA3 during keratinocyte differentiation in vitro.
By immunocytostaining, more than 99% of the cells were positive in p51/p63 (Fig. 5A, upper panels), assuring that the NHK culture was enriched in keratinocyte stem cells (13). Double staining indicated the presence of ␣ 3 in perinuclear regions as observed in ␣ 3 -transfected Chinese hamster ovary and NIH3T3 cells (45,46). Seven days after incubation with high Ca 2ϩ , we detected morphological features of in vitro keratinocyte differentiation: cell flattening and formation of cellto-cell contacts (Fig. 5A, lower panels). The nuclear p51/p63 label was significantly weakened, supporting the Western blot analysis (Fig. 4B). The ␣ 3 label had not only faded, but had also changed its localization to the cell-cell borders as found in Madin-Darby canine kidney cells (45), probably by forming a complex with proHB-epidermal growth factor and DRAP27/ CD9 proteins (47).
Furthermore, we introduced siRNA-1 and -2 targeting the DNA binding domain-encoding sequences of p51/p63 into NHK using an electrical transfection system (Fig. 5B). Twenty-four hours later, ϳ40% of the cells in the culture transfected with siRNA1 or siRNA2 displayed a decrease in p51/p63. Obviously, ␣ 3 also decreased in the cells that underwent suppression of p51/p63. Unlike the differentiation process with high Ca 2ϩ (Fig. 5A), the siRNA experiment showed no evidence of the ␣ 3 protein shift from the perinuclear regions to the cell-cell borders. Control double labeling with an anti-pan-keratin antibody indicated that the siRNA did not affect the total level of cytokeratins. Thus, the integrin ␣ 3 -inducing ability of p51/p63 seemed vital in keratinocyte stem cells.
Detection of an Interaction of p51/p63 with Intron 1 of ITGA3-To assess whether p51/p63 can directly interact with the 1st intron of ITGA3 on the chromosome, a ChIP experiment (48) was carried out with HEK293 cells transfected with an HAp51A expression vector and control untransfected cells. Expression of HAp51A was confirmed by Western blotting with anti-HA antibody and 4A4 anti-p51/p63 antibody (Fig. 6A). An increase in 145-kDa integrin ␣ 3 was also detectable in the HAp51A-expressing cells.
For immunoprecipitation, we used three different antibodies: control non-immune IgG, an anti-HA antibody, and an antibody against TFIIB, a general transcription factor. 4A4 was not reactive with either of the p51/p63 isoforms unless proteins were fully denatured, and was not useful for ChIP. DNA fractions recovered from the immunoprecipitates were examined by PCR for a GAPDH promoter locus and an ITGA3 intron 1 segment (553-942) encompassing the 1st and 2nd half-sites. From both the HAp51A-transfected and untransfected cells, the GAPDH promoter segment was more abundantly precipitated by the anti-TFIIB antibody than by the control and anti-HA antibodies (Fig. 6B, left panels). Constant transcription of the housekeeping gene was thus detectable by this ChIP experiment. On the other hand, the ITGA3 segment was more enriched in the anti-TFIIB and anti-HA precipitates from the HAp51A-expressing cells than in the control non-immune IgG precipitates. Neither the anti-TFIIB or anti-HA precipitate FIG. 5. Suppression of ␣ 3 coincides with p51/p63 depletion in human keratinocytes. A, NHK and the Ca 2ϩ -incubated culture on day 7 were doubly stained for p51/p63 and ␣ 3 . B, NHK cultures transfected with p51/p63 siRNA1 or -2 were analyzed for alterations in p51/p63 and ␣ 3 (upper panels). Large and small arrows indicate apparently complete and substantial suppression of p51/p63, respectively. Also shown is a double stain for p51/p63 (fluorescein isothiocyanate) and pan-keratins (rhodamin) in a siRNA2-transfected culture. from untransfected cells was enriched for the ITGA3 sequences (Fig. 6B, right panels). These results not only indicated a physical interaction of p51/p63 with intron 1 of ITGA3, but also verified that the gene transcription is induced by p51/p63. DISCUSSION We have shown (i) induction of integrin ␣ 3 by DNA damagecaused p51/p63 protein activation and transient and controlled p51/p63 overexpression, (ii) transduction of an ECM-binding phenotype relevant to integrin ␣ 3 ␤ 1 in a non-adherent leukemic cell line, (iii) presence of p51/p63-responsive sites in the 1st intron of human ITGA3, (iv) suppression of ␣ 3 by p51/p63 knockdown with siRNAs and by differentiation in the NHK culture, and (v) association of p51/p53 with intron 1 of ITGA3 on the chromatin. Transcriptional activation of ITGA3 may be one of the pivotal roles of p51/p63 in epidermis development.
Integrin ␣ 3 ␤ 1 is a major epidermis laminin receptor whose expression is confined to the stem cells (23). Although Itga3null mice caused occasional skin blisters where the epidermis separated from the dermis (49), conditional ␤ 1 knock-out mice displayed extreme skin blistering, hair defects, and basement membrane disassembly, indicating more critical roles for the partners of ␤ 1 in skin development (50). The terminal differentiation program was, however, preserved even in the absence of ␤ 1 (50). A double mutation in ␣ 3 and ␣ 6 impaired basement membrane assembly and epidermal cell compaction in the apical ectodermal ridge, and thereby caused abnormal limb patterning (51). Defects in the urogenital tracts and other organs were also observed in the Itga3 Ϫ/Ϫ Itga6 Ϫ/Ϫ mice. Thus, most of the abnormalities in the integrin ␣ or ␤ knock-out embryos are closely related to the phenotypes of the ectroductyly, ectodermal dysplasia, and cleft palate syndrome (11,12) and p51/p63 deletion (3,6). Our p51/p63 knockdown experiment with siRNA provided evidence that p51/p63 is a dominant factor inducing the ␣ 3 expression in keratinocyte stem cells. Because of the localized, abundant expression in undifferentiated keratinocytes (23), the p51/p63 proteins may control ITGA3 efficiently in those cells. However, the more striking skin phenotypes observed in the p51/p63-null mice imply cooperation of ␣ 3 with other signaling and/or structuring molecules under control of p51/p63.
Most of the known p53-responsive promoters have a full-site that forms a stable complex with a p53 tetramer (38). However, a half-site can accommodate a p53 dimer (39), and serves as a minimal p53-responsive element when placed adjacent to a Sp1 site (40). Furthermore, a full p53-binding site with a 33-base spacer has recently been identified (52). The half-sites at 597 and 881 were active not only in the full-site context at the enhancer position in p(1ϩ2)luc, but also in the original half-site context placed downstream of luc in (␣ 3 -1)luc. It remains to be determined whether the p51/p63 proteins also form functional tetramers, although the peptide-peptide interaction in the oligomerization domain was detectable by a yeast two-hybrid assay, indicating dimer formation (53). The mechanism of ITGA3 regulation by p51/p63 may involve topological arrangement of the separate half-sites in intron 1, interactions of p51/p63 with other transcriptional regulatory factors, and association between different p51/p63 isoforms.
The mouse ␣ 3 gene also contains three half-site motifs in intron 1, in which the 1st and 2nd motifs are located in the 5Ј terminus, suggesting a similar regulatory mechanism in intron 1 by p51/p63. The promoter/enhancer region extending to Ϫ4 kb has been extensively studied (32,54) to determine that the Ets binding site at Ϫ133 is critical for gene expression in MKN1 gastric carcinoma cells. However, the 4-kb promoter/ enhancer region lacked a putative p51/p63-binding site, and did not respond to p51/p63 in our assay (data not shown).
When assayed with the promoters of p53 target genes, the ⌬N isoforms suppressed the trans-activating ability of p51A/ TAp63␥ (2), by which the ⌬N proteins were originally determined to be dominant-negative type isoforms. A recent study demonstrated that a ⌬N isoform of p73 is able to both positively and negatively regulate p53 target genes (15). p53 activates the human BAX promoter using Sp1 as a cofactor (40), whereas p51A/TAp63␥ suppresses epidermal growth factor-receptor gene expression by an interaction with Sp1 (19). Gene regulation by the TA and ⌬N forms of p51/p63 now appears more elusive than originally characterized. To exert the positive effect, the ⌬N proteins may interact with an activation factor whose expression could be cell-type specific.
␣ 3 ␤ 1 is essential for cell spreading on the basement membrane, ECM assembly, hemidesmosome stability, establishment/maintenance of the cytoskeletal organization, epidermal proliferation, and control of cell migration (24,25,50,(55)(56)(57). It may be through the ␣ 3 ␤ 1 -caused epidermal-mesenchymal interactions that p51/p63 facilitates generation of an undifferentiated basal cell population. In addition to the epidermis, p51/ p63 expression occurs in the basal cells of epithelia of the mammary gland (58), oral tissues (59), uterocervix (60), and bladder (30), where the ␣ 3 -inducing role of p51/p63 may also function. Furthermore, the overexpression of p51/p63 in squamous cell carcinomas of head, neck, and lung (7,9) might contribute to the high-level ␣ 3 expression that determines cellular capability of invasion and metastasis (61)(62)(63). It is interesting to envisage that p51/p63, a hypothetical ancestor gene to tumor suppressor p53 (64), had evolved as an inducer of the critical interaction between growing epithelial cells and their niche.