Three Activator Protein-1-binding Sites Bound by the Fra-2·JunD Complex Cooperate for the Regulation of Murine Laminin α3A (lama3A) Promoter Activity by Transforming Growth Factor-β*

Several lines of evidence suggest a role for laminin-5 in skin wound healing. We report here that transforming growth factor-β (TGF-β), which elicits various responses during cutaneous healing, stimulates transcription of the mouse laminin α3A (lama3A) gene. To identify the TGF-β-responsive elements (TGFβ-REs) on the lama3A promoter, we have generated a series of 5′-deletions of the promoter upstream of the β-galactosidase reporter gene. Transient cell transfection assays using mouse PAM212 keratinocytes revealed that TGFβ-REs lie between nucleotides −297 and −54 relative to the transcription start site. Insertion of the TGFβ-RE in front of the unresponsive minimal SV40 promoter conferred TGF-β inducibility. Computer analysis of the promoter sequence identified three canonical activator protein-1 (AP-1) sites located at nucleotides −277 (AP-1A), −125 (AP-1B), and −69 (AP-1C). Site-directed mutagenesis of either the AP-1A or AP-1C site did not drastically alter the basal activity of the lama3Apromoter, but reduced TGF-β responsiveness by 50%. Simultaneous mutation of these two AP-1 sites resulted in a 65% decline in the response to TGF-β, suggesting a cooperative contribution of each site to the overall promoter activity. In contrast, mutation of the AP-1B site markedly reduced the basal activity of the lama3Apromoter, indicating that this AP-1 site is essential for gene expression. Mobility shift assays demonstrated specific binding of Fra-2 and JunD to the AP-1 sites, suggesting for the first time a possible regulatory function for the Fra-2·JunD AP-1 complex in a basal keratinocyte-specific gene.

Several lines of evidence suggest a role for laminin-5 in skin wound healing. We report here that transforming growth factor-␤ (TGF-␤), which elicits various responses during cutaneous healing, stimulates transcription of the mouse laminin ␣3A (lama3A) gene. To identify the TGF-␤-responsive elements (TGF␤-REs) on the lama3A promoter, we have generated a series of 5-deletions of the promoter upstream of the ␤-galactosidase reporter gene. Transient cell transfection assays using mouse PAM212 keratinocytes revealed that TGF␤-REs lie between nucleotides ؊297 and ؊54 relative to the transcription start site. Insertion of the TGF␤-RE in front of the unresponsive minimal SV40 promoter conferred TGF-␤ inducibility. Computer analysis of the promoter sequence identified three canonical activator protein-1 (AP-1) sites located at nucleotides ؊277 (AP-1A), ؊125 (AP-1B), and ؊69 (AP-1C). Site-directed mutagenesis of either the AP-1A or AP-1C site did not drastically alter the basal activity of the lama3A promoter, but reduced TGF-␤ responsiveness by 50%. Simultaneous mutation of these two AP-1 sites resulted in a 65% decline in the response to TGF-␤, suggesting a cooperative contribution of each site to the overall promoter activity. In contrast, mutation of the AP-1B site markedly reduced the basal activity of the lama3A promoter, indicating that this AP-1 site is essential for gene expression. Mobility shift assays demonstrated specific binding of Fra-2 and JunD to the AP-1 sites, suggesting for the first time a possible regulatory function for the Fra-2⅐JunD AP-1 complex in a basal keratinocyte-specific gene.
Laminin-5 is the major adhesion ligand present in the basement membranes of stratified squamous epithelia (1,2). In the skin, this adhesive protein is secreted by the basal keratinocytes and colocalize with the anchoring filaments of the lamina lucida of the dermal epidermal junction (3)(4)(5). Laminin-5 binds to integrin ␣ 3 ␤ 1 in focal adhesions and interacts with hemides-mosomes via ␣ 6 ␤ 4 to form a stable anchorage complex (6,7). Laminin-5 is a heterotrimeric glycoprotein composed of the ␣3A, ␤3, and ␥2 polypeptide chains that are products of different genes. Mutations in the genes encoding laminin ␣3 (LA-MA3), ␤3 (LAMB3), and ␥2 (LAMC2) have been shown to underlie the Herlitz or non-Herlitz forms of junctional epidermolysis bullosa, characterized by blister formation and erosions of the skin and mucosas that frequently lead to neonatal death (8 -13).
Several lines of evidence suggest a role for laminin-5 in the re-epithelialization of wound skin repair. Laminin-5 is found at the epidermal-dermal junction at sites and times that coincide with actively migrating or rapidly proliferating basal keratinocytes (14,15). Moreover, enhancement of laminin-5 transcription is observed at low cell densities in vitro in migrating and proliferating keratinocytes, similar to what happens at the wound edge (15). Although the function of laminin-5 in wound healing remains to be clarified, its major dual contribution would be to allow migration and adhesion of keratinocytes during the wound process. In addition, accumulating data suggest that laminin-5 might be associated with growth and migration of cancer cells (16 -19). Recently, cloning of the cDNAs encoding the ␣3 chain of mouse laminin-5 has identified two distinct polypeptides (␣3A and ␣3B) that derive from a single alternatively spliced gene (20 -22). Genomic organization analysis of the murine lama3 gene revealed that the ␣3A chain is transcribed by an independent internal promoter (21).
In this study, we investigated the transcriptional regulation of the laminin ␣3A (lama3A) gene by TGF-␤. 1 Indeed, it has been shown earlier that TGF-␤ up-regulates the adhesion protein laminin ␣3A mRNA level (23). TGF-␤, a member of a large superfamily of cytokines, is generally acknowledged to be the cytokine with the broadest range of activities in injured tissue repair and tumor progression (24). It affects nearly every aspect of tissue repair. Indeed, TGF-␤ is the most potent known stimulator of chemotaxis since it promotes the migration of the majority of cell types that participate in the repair processes. TGF-␤ regulates the transcription of a wide spectrum of matrix proteins, increasing their production while decreasing their proteolysis and modulating their interactions with cellular integrin receptors (25). Matrix production by keratinocytes is also regulated by TGF-␤ (26 -28).
Using transient transfection, site-directed mutagenesis, and nuclear protein binding assays, we have mapped a TGF-␤responsive region between nt Ϫ297 and Ϫ54 of the lama3A promoter. Computer analysis of this sequence identified three * This work was supported in part by EEC Biomed-2 Grant BMH4-CT972062, by the ssociation pour la Recherche sur le Cancer Grant 5003, and by grants from the Ligue du Var contre le Cancer, INSERM, and the Fondation pour la Recherche Médicale. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  canonical AP-1 sites. Insertion of this region upstream of the unresponsive minimal SV40 promoter conferred TGF-␤ inducibility. Site-directed mutagenesis of either AP-1 site suggested a cooperativity among these AP-1 sites in the laminin ␣3A promoter. Mobility shift assays demonstrated specific binding to the three AP-1 sites. Moreover, antibody supershift analyses have identified JunD and Fra-2 proteins binding to these sites.

EXPERIMENTAL PROCEDURES
Cell Cultures-Mouse PAM212 keratinocytes (kindly provided by Dr. S. H. Yuspa, NCI, National Institutes of Health, Bethesda, MD) were cultured in Eagle's minimal essential medium supplemented with 10% fetal calf serum (Life Technologies, Inc.).
Transient Transfections-Plasmid DNAs were purified using silica columns from QIAGEN (Hylden), and transfections were carried out in 24-well dishes using LipofectAMINE TM (Life Technologies, Inc.) as detailed elsewhere (21). Each DNA construct was tested in triplicate wells in at least three separate experiments. The transfection efficiency of pGalA and its derivative constructs was determined by cotransfection of the plasmid pGL2-control (Promega, France). The values of ␤-galactosidase expression data were normalized to the measured luciferase activity. The enzyme activities were measured using a luminometer (Berthold Biolumat LB9500C). The activity of the promoterless vector pGalbasic, which contain no insert, was measured to determine background activity. In the TGF-␤ induction assays, 4 h post-transfection, the cells were treated with or without 100 pM recombinant TGF-␤1 (R&D Systems) for an additional 16 -20 h.

TGF-␤ Stimulates lama3A
Transcription-The structure of the murine lama3A promoter and several of its regulatory DNA elements is depicted in Fig. 1. Full-length pGalA, which contains 1 kb of promoter region upstream of the ␤-galactosidase reporter gene, was transfected into PAM212 cells and examined for responsiveness to TGF-␤. A dose-responsive increase in lama3A activity was observed with maximal stimulation (4.5-fold) at 100 pM TGF-␤ (Fig. 2). A series of 5Ј-deletions of the lama3A promoter were examined to delineate the minimal region responsive to TGF-␤. TGF-␤ responsiveness was significant (3.0 -4.5-fold) for pGalA (nt Ϫ1096), ⌬2.4M (nt Ϫ839), ⌬4.2M (nt Ϫ619), ⌬6.5M (nt Ϫ505), and ⌬SM (nt Ϫ297) constructs, but decreased to 1.5-fold upon deletion from nt Ϫ297 to Ϫ244 (⌬PA20) (Fig. 3, A and C). This slight induction was lost from nt Ϫ244 to Ϫ94 (⌬SAP). It therefore appeared that an essential TGF-␤-responsive element (TGF␤-RE) was located between nt Ϫ297 and Ϫ244 of the lama3A promoter. However, since the basal activity drastically decreased for ⌬SAP, it remained to be determined whether the loss of stimulation was due to the low basal activity or to the presence of additional sequences allowing secondary TGF-␤ responsiveness between nt Ϫ244 and Ϫ94.
Delineation of TGF-␤-responsive Elements within the lama3 Promoter-To more precisely map the potential TGF␤-RE that should reside between ⌬SM and ⌬PA20, additional 5Ј-deletions of ⌬SM constructs were created (Fig. 3, B and C). Similar induction with TGF-␤ treatment was obtained with the ⌬SM (nt Ϫ297), ⌬PA23 (nt Ϫ291), and ⌬PA24 (nt Ϫ279) constructs, whereas a decrease in induction was observed with ⌬PA25 (nt Ϫ265). This result indicates that a TGF-␤-responsive element is located between nt Ϫ279 and Ϫ265 of the lama3A promoter. Examination of the deleted region that reduced TGF-␤ responsiveness revealed the presence of one potential AP-1-binding site (Figs. 1 and 3B).
To assess whether the region between nt Ϫ279 and Ϫ265 was sufficient for activation by TGF-␤, four copies of this sequence were linked to a heterologous unresponsiveness SV40 promoter (construct SV-RE4) (Fig. 4). Since no TGF-␤ induction was observed in SV-RE4, we conclude that the 15-base pair sequence containing AP-1 (nt Ϫ279) is necessary but not sufficient to confer TGF-␤ responsiveness.
As suggested previously, additional TGF␤-RE sequences must reside downstream of the AP-1-binding site. To assess whether the region between nt Ϫ297 and the TATA box was sufficient for activation by TGF-␤, the MscI/SmaI DNA fragment from pGalA (from nt Ϫ297 to Ϫ54) was linked to a heterologous SV40 promoter (construct SV-MS). This construct was able to fully restore the TGF-␤ induction of ␤-galactosidase transcription in PAM212 cells as compared with ⌬SM, whereas no induction was observed with the pGal-promoter vector alone (Fig. 4). Thus, the sequence of this region of the lama3A promoter is sufficient to confer TGF-␤ responsiveness to a previously non-inducible promoter. Examination of the TGF-␤-responsive region indicated the presence of three potential AP-1-binding sites located at nt Ϫ277 (AP-1A), Ϫ125 (AP-1B), and Ϫ69 (AP-1C) (Fig. 1). Site-directed mutagenesis was used to sequentially modify these putative AP-1 sites within the context of the pGalA promoter fragment because this region retains high basal activity and TGF-␤ responsiveness (Fig. 3A). The mutated constructs were then used in transient transfection experiments in parallel with their wild-type counterparts. As shown in Fig. 5, the mutation of AP-1B (mutB) drastically reduced the promoter activity, whereas mutation of each of the other AP-1 sites (mutA and mutC) still maintained 60% of the pGalA basal activity. However, elimination of each AP-1 site reduced TGF-␤ responsiveness to half the level seen with the wild-type promoter construct pGalA. Simultaneous mutation of both AP-1A and AP-1C (mutAC) abated by 65% the inducible activity by TGF-␤, whereas simultaneous mutations of the three AP-1 sites (mutABC) completely abolished TGF-␤ responsiveness, as did dual mutants including the AP-1B mutation (mutAB and mutBC) (Fig. 5). These results indicate that the integrity of each AP-1 site of the lama3A promoter is necessary for full transcriptional induction. Furthermore, since half the response to TGF-␤ is retained within single AP-1 mutants, the full transcriptional induction of the lama3A promoter by TGF-␤ requires cooperation among the three AP-1 sites.
AP-1 Proteins Bind to the lama3 TGF-␤-responsive Element-To confirm the interaction of nuclear proteins with the AP-1 sites, electrophoretic gel mobility shift assays were performed using radiolabeled lama3A TGF␤-RE sequences corresponding to each of the three AP-1 sites and flanking regions, as detailed in Table I. AP-1Am, AP-1Bm, and AP-1Cm are mutants of the AP-1A, AP-1B, and AP-1C sites, respectively. Fig. 6 shows a gel mobility shift experiment in which unstimulated and TGF-␤-stimulated mouse keratinocyte total cell ex- FIG. 4. Functional analysis of putative TGF␤-responsive elements using the heterologous promoter. A synthetic oligonucleotide spanning four identical copies of the region from nt Ϫ279 to Ϫ265 of the lama3A promoter was cloned upstream of the SV40 promoter linked to the ␤-galactosidase gene, generating the plasmid SV-RE4. The construct SV-MS carries the MscI/SmaI fragment (Ϫ297/Ϫ54) in front of the SV40 promoter-␤-galactosidase plasmid. The ␤-galactosidase activities are represented as -fold stimulation of TGF-␤-treated cells over untreated cells. The pGL3-control cotransfectant served as an internal transfection control. Other conditions were the same as described in the legend of Fig. 2. pGalprom, pGal-promoter. tracts were incubated with 32 P-labeled AP-1A, AP-1B, or AP-1C oligonucleotides in the absence or presence of excess nonradioactive competitors. Our results indicate that, although these regions bound protein complexes isolated from both control and TGF-␤-treated keratinocytes, TGF-␤ treatment induced a significant increase in DNA binding within 1 h for the AP-1A, AP-1B, and AP-1C oligonucleotides (Fig. 6, A-C). Protein binding was specifically competed in a dose-dependent manner by addition of 20-and 50-fold excesses of homologous DNA, but not by a 50-fold excess of the corresponding homologous mutant AP-1 sequence or heterologous Sp1 oligonucleotide competitors. These data indicate that AP-1 complexes bind to each of the three TGF␤-RE elements of the lama3A promoter. Although the AP-1A and AP-1C oligonucleotides were specifically competed by 20-fold (data not shown) and 50-fold (Fig. 6, A and C, respectively) molar excesses of every AP-1 oligonucleotide, the AP-1B band shift was not displaced by a 50-fold molar excess of either AP-1A or AP-1C competitor (Fig. 6B).
To characterize further the protein complexes binding to the TGF␤-RE region of the murine lama3A promoter, total cell extracts from untreated and TGF-␤-treated PAM212 cultures were incubated prior to DNA/protein interactions with antibodies specific for each member of both the Jun (c-Jun, JunB, and JunD) and Fos (c-Fos, FosB, Fra-1, and Fra-2) family proteins and visualized by gel mobility shift assay. Since identical re-sults were obtained from untreated and TGF-␤-treated cells, we illustrated the data from treated cells. As shown in Fig. 7, the antibodies against JunD and Fra-2 (and to a much less extent, JunB) induced a supershift of the labeled DNA probes (brackets). These results indicate that JunD and Fra-2 participate in the formation of the complex that binds to the three AP-1 sites of the lama3A promoter. DISCUSSION This study was designed to investigate the molecular mechanism of lama3A transcriptional induction by TGF-␤. TGF-␤ is known to regulate many different gene transcripts of the extracellular matrix, thereby modulating cell adhesion, migration, and proliferation, under both physiological and pathological conditions (24). Our results define three AP-1-binding sites in the mouse lama3A promoter that cooperate to confer TGF-␤ responsiveness since (a) their presence is needed in front of unresponsive SV40 promoter sequences to fully restore TGF-␤ inducibility; (b) mutations of the AP-1A, AP-1B, and AP-1C sites significantly abate the TGF-␤-promoter transactivation in PAM212 cells; and (c) simultaneous mutations of the three AP-1 sites abolish this transactivation. Furthermore, the AP-1B site is crucial for lama3A promoter activity since mutating its sequence almost completely abolished the basal transcriptional activity of the lama3 promoter. Since the AP-1B site FIG. 5. Mutation of AP-1 sites modulates the lama3A promoter response to TGF-␤. A, full-length pGalA and variants in which one (mutA, mutB, and mutC), two (mutAB, mutAC, and mutBC), or three (mutABC) AP-1 sites have been inactivated by mutation were transfected into PAM212 cells in the presence or absence of TGF-␤. Constructs were tested for activity exactly as described in the legend of Fig. 2. B, the relative induction of each construct by TGF-␤ is represented and compared with that obtained by pGalA. ␤gal, ␤-galactosidase.
is not cross-competed on band shift assays by a 50-fold molar excess of either AP-1A or AP-1C oligonucleotide competitors, the AP-1 protein heterodimer binds the AP-1B site with higher affinity than the two other AP-1 sites. These results suggest that the AP-1B site is preferentially bound by the AP-1 proteins under basal activity and must cooperate with the two other AP-1 sites to allow full transcriptional basal and TGF-␤-stimulated activities.
As reviewed recently (32), many promoters of keratinocytespecific genes contain cis-acting elements that are capable of binding AP-1 heterodimer complexes. In the epidermis, the AP-1 heterodimer serves as an activator of gene expression in all the keratinocyte layers, i.e. keratin K5 in the basal layer; human papillomavirus in the suprabasal compartment; loricrin and profilaggrin in the granular layer; and finally, keratin K1, involucrin, and transglutaminase 1 in the spinous/granular layer (32). Therefore, it is intriguing how the cell coordinately regulates genes through identical DNA elements. Since AP-1 consists of proteins from the Jun and Fos families that associate as homodimers (Jun⅐Jun) or heterodimers (Fos⅐Jun), one clue might be the composition of the AP-1 dimers, meaning that a certain combination of Fos⅐Jun complexes may be necessary to provide different levels of transcriptional activity, depending on the stage of keratinocyte differentiation. It should be noted that laminin ␣3A transcripts are produced only by proliferating basal keratinocytes in vivo (33) and in vitro (15). We have shown here that the lama3A promoter AP-1 sites are preferentially bound by the Fra-2⅐JunD complex and that the binding is significantly increased by TGF-␤ stimulation of PAM212 keratinocytes as compared with untreated cells. Recently, immunohistological studies on Fos/Jun factors have shown that the distribution of AP-1 proteins in the mouse epidermis is compartmentalized in vivo, suggesting that distinct AP-1 complexes act in the various layers of the skin (34). In accordance, Fra-2 and JunD are coexpressed only in the basal layer of mouse skin, but are absent in the spinous (for Fra-2) and granular (for JunD) layers of the epidermis, implying that the particular Fra-2⅐JunD heterodimer is selectively down-regulated in the early stages of epidermal differentiation (34). All the epidermal genes investigated so far are regulated by different AP-1 family members, i.e. JunB/JunD for HPV18 (35), JunB/Fra-1 and JunD/Fra-1 for involucrin (36), and c-Jun/c-Fos for profilaggrin (37). These keratinocyte genes are ex-FIG. 6. Binding of nuclear proteins to the lama3A AP-1-binding sites. A-C, binding of nuclear proteins to the AP-1A, AP-1B, and AP-1C binding sites, respectively. 32 P-Labeled oligonucleotides were incubated with nonstimulated (NS) or TGF-␤-stimulated total cell extracts in the absence (0x) or presence of a 20-or 50-fold molar excess of unlabeled competitors. none indicates a lane in which the total cell extract was omitted. Arrows indicate the free probe (bottom) and the specific retarded bands (top). For details, refer to Table I  pressed in the suprabasal compartments of the epidermis. Since the lama3A gene is expressed exclusively in the basal cells of most stratified epithelia, it will be important to determine the potential role of Fra-2 and JunD in the regulation of basal keratinocyte-specific genes in general and of the lama3A gene in particular.
In conclusion, we have used a combination of techniques to identify the regulatory elements important for the transcriptional induction of the lama3A gene by TGF-␤. We have demonstrated a cooperative contribution of three canonical AP-1binding sites to the stimulation of the promoter activity and that the AP-1B site is essential for basal gene expression. These AP-1-binding sites contain almost exclusively the JunD and Fra-2 proteins, which suggests for the first time a possible regulatory function of the Fra-2⅐JunD AP-1 complex in a basal keratinocyte-specific gene. Our results therefore strongly support the idea that AP-1 proteins play a central role in the transcriptional regulation of epidermal gene expression (32). As for many other genes, the same regulatory elements are involved in both basal and stimulated transcriptional activities. These studies represent the initial steps toward the identification of the signaling pathways involved in TGF-␤-mediated transcriptional activation of laminin ␣3A under physiological and pathological conditions, such as wound healing and carcinogenesis. I Sequence of 1ama3A AP-1 sites and double-stranded oligonucleotides employed in gel mobility shift assays AP-1Am, AP-1Bm, and AP-1Cm are mutants of AP-1A, AP-1B, and AP-1C in which the A at position 3 and the C at position 6 of the consensus sequence were changed to C and T, respectively (underlined in each oligonucleotide). Sp1 has commercial (Promega) consensus binding sites. Consensus binding sites of all oligonucleotides are underlined.