Transcriptional analysis of the 5'-noncoding region of the human involucrin gene.

Human involucrin whose gene transcription is directed by a 2456-nucleotide (nt) 5'-noncoding region is a structural component of the epithelial cornified layer. Transient transfection assays demonstrated that this region is transcriptionally active in multiplying keratinocytes and is enhanced by 2 mM CaCl2 treatment. Calcium-independent transcriptional activity and the interaction with the AP-1 transcriptional factor was located on the proximal part (nt -159 to -1) of the 5'-noncoding region. However, CaCl2 responsiveness was mapped to a distal 1185-nt fragment (nt -2456 to -1272). Moreover, this fragment potentiated the Herpes simplex thymidine kinase promoter in normal keratinocytes and is responsive to calcium treatment in a cell type-specific manner. Interestingly, the absence of a 491-nt fragment located between the two enhancer domains (nt -651 to -160) resulted in transcriptional activation in multiplying keratinocytes. This fragment interacts with AP-1 and the YY1 transcriptional silencer. It is concluded that human involucrin 5'-noncoding region contains at least three regulatory domains, a distal CaCl2-responsive enhancer, a putative transcriptional silencer (that interacts with AP-1 and YY1), and a proximal enhancer/promoter (that interacts with AP-1). Thus, this study demonstrates the presence of particular transcriptional factors can potentially regulate the human involucrin expression.

The differentiation of stratified epithelia requires the harmonious expression of several structural and regulatory proteins. The complex regulatory pathways that direct the transcription of epithelial differentiation-related genes are of particular importance in human disease.
Involucrin, a precursor of the cornified envelope of terminally differentiated keratinocytes (1,2), is apparently limited to primates (3). The involucrin protein has a molecular mass of 68 kDa and possesses a central glutamyl-rich domain formed with 39 repeats of a 10-amino acid cassette (4) which is required for the cross-linking activity of the calcium-dependent epithelial transglutaminase during cornified envelope formation (5)(6)(7)(8).
The human involucrin gene is about 6000 nt in size composed of two exons of 43 and 2107 nt, respectively, separated by an intron of 1188 nt (4). A 2456-nt noncoding sequence located 5Ј of the first involucrin exon has transcriptional regulatory elements that control its transcriptional activity (9 -11). Analysis of in vitro and in vivo results show that the involucrin gene activation depends on the interaction of transcriptional factors present in the keratinocyte nucleus (9 -11).
In vitro and in vivo experiments correlate the presence of involucrin transcripts and protein following the progression of keratinocytes from the basal layer to terminal differentiation state (12)(13)(14). Thus, the transcriptional factors required for specific involucrin gene transcription may also be necessary for expression of other epithelial terminal differentiation-related genes (15)(16)(17). Interestingly, several genes related to terminal differentiation, such as involucrin, profilaggrin, and loricrin, are located on chromosome 1q21 (18).
The 5Ј-noncoding region controlled the expression of the involucrin gene in transient transfection of cultured human keratinocytes (9). This region was divided functionally into two portions: the proximal 900-nt promoter region with the putative TATA box and an upstream 1600-nt region with necessary elements for the proper expression of the involucrin gene (9). Interestingly, the entire 2456-nt 5Ј-noncoding segment activity is tissue-specific in transgenic mice, suggesting that the basic regulatory elements of the involucrin gene are widespread in mammals (10,19).
The reported sequence of the 900-nt proximal promoter region active in keratinocytes contains putative target sites for the AP-1 family of transcriptional factors (9,11). Moreover, the addition of TPA, 1 an AP-1 activator, moderately activates this region in transient transfection assays using cultured rat cells. The latter suggests that AP-1 could be necessary for involucrin expression. Furthermore, the proximal 900-nt enhancer was activated by overexpression of c-fos and c-jun oncogenes, components of AP-1 (11). Treatment of normal keratinocytes with calcium, TPA, or vitamin A depletion (20 -26) are able to increase involucrin mRNA levels. However, how these compounds directly regulate the involucrin promoter region is not clear.
To investigate the transcriptional regulation of the involucrin gene, several constructs of the 2456-nt 5Ј-noncoding region were transfected into cultured human keratinocytes during multiplication (0.1 mM CaCl 2 ) or differentiation (2 mM CaCl 2 ) conditions. Involucrin transcription is shown to be regulated by several functional elements: a distal cell type-specific 1100-nt upstream enhancer (nt Ϫ2456 to Ϫ1272) responsive to calcium stimulation and a possible transcriptional silencer (nt Ϫ651 to Ϫ160) which in turn is coupled to a proximal enhancer/promoter (nt Ϫ159 to Ϫ1/ϩ1) unaffected by calcium concentration. Further DNA-protein characterization of the silencer and proximal enhancer/promoter regions established that AP-1 and YY1 are the main transcriptional factors interacting with these elements.
The complete nucleotide sequence from p2.6CAT insert is recorded in GeneBank™ (accession number U23404). All constructs were sequenced using Sequenase (Amersham Corp.) or the chemical degradation method (28). All oligonucleotides (Table I) were synthesized in an Applied BioSystems 391 DNA synthesizer.
Cell Culture-HeLa cervical carcinoma cells and MRC-5 human fibroblasts were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. The cervical carcinoma cell line C-33A was grown in Dulbecco's modified Eagle's medium/F-12 (1:1) medium supplemented with 7% fetal bovine serum. Secondary cultures of neonatal human foreskin keratinocytes were obtained essentially as described (29) and grown in keratinocyte-SFM medium (Life Technologies, Inc.) in a 6% CO 2 atmosphere. Calcium stimulation of differentiation was performed on confluent cultures by addition of 2 mM CaCl 2 for 4 days to keratinocyte-SFM medium lacking epidermal growth factor and bovine pituitary extract.
Transient Transfections and CAT Assays-Normal human keratinocytes cultures, 60 -70% confluent in 100-mm tissue culture dishes, were transfected with 10 g of total plasmid DNA using Lipofectin (Life Technologies, Inc.) as described previously (30). MRC-5 fibroblasts were transfected using Lipofectamine (Life Technologies, Inc.) using the same protocol. C-33A cells in 60-mm dishes were transfected using the calcium phosphate method as described previously (31). For keratinocyte differentiation conditions, the culture medium was replaced with keratinocyte-SFM lacking epidermal growth factor and bovine pituitary extract with 2 mM CaCl 2 12 h post-transfection. Cells were harvested 48 h post-transfection in TEN buffer (40 mM Tris-HCl, pH 8.0, 1 mM EDTA, 15 mM NaCl) and lysed with three freeze-thaw cycles in 0.25 M Tris-HCl, pH 8.0; protein was quantified employing the Bradford method (32). Standardized amounts of lysate protein were incubated with 0.25 Ci of [ 14 C]chloramphenicol (50 mCi/mmol, Amersham Corp.) and 0.66 mM acetyl-CoA (Sigma) in a final volume of 115 l. The acetylation reactions were carried out at 37°C for up to 4 h. The samples were extracted with ethyl acetate (J.T. Baker Inc.) and loaded in TLC plates (Sigma). Chromatography was developed with chloroform:methanol (19:1) and exposed to Hyperfilm radiographic films (Amersham Corp.). Radioactive spots were quantified in a Beckman LS6000SC scintillation counter. When comparing different cell lines and transfection methods, CAT activities from MRC-5 and C-33A cells were standardized relative to the SV40 enhancer/promoter that is active in all cell types tested. Otherwise, CAT activities are expressed as the acetylated fraction corrected for the activity of the pCAT-basic vector.
Gel-shift Assays-Nuclear extracts from keratinocytes or HeLa cells were incubated on ice with 0.5-1 g of poly[d(I-C)] (Pharmacia Biotech Inc., Alameda, CA) and 1 ng of 32 P-end-labeled DNA in 2 ϫ BDG buffer (24 mM HEPES, pH 7.8, 20% glycerol, 0.1 mM EDTA, 8 mM MgCl 2 , 20 mM KCl, 2 mM dithiothreitol, 4 mM spermidine). The reaction mixtures were electrophoresed in low ionic strength 0.5 ϫ TBE buffer (44.5 mM Tris-HCl, pH 7.9, 44.5 mM boric acid, 1 mM EDTA) 4 or 6% polyacrylamide gels at 150 V. The gels were dried and exposed to Kodak X-Omat radiographic films. For competitive studies, the reaction mixtures were preincubated with different amounts of unlabeled competitor oligonucleotide before the addition of labeled DNA. For gel supershift experiments, reactions with the DNA-protein complexes were incubated at 4°C with anti-c-jun/AP-1 (sc-44; Santa Cruz Biotechnology, Santa Cruz, CA) or anti-HPV16 E7 rabbit polyclonal antibodies for 6 h prior electrophoresis.
DNase I Footprinting-EcoRI-HindIII fragments from pIN220 and pIN630 plasmids ( Multiplying normal keratinocytes were transfected with 10 g of total plasmid DNA from different deletion constructs as described under "Materials and Methods." The average CAT activities relative to the promoterless vector pCAT-basic were obtained from at least three independent experiments 48 h post-transfection. The transcriptional start site is represented by an arrow. The various deleted constructs, the putative TATA box, the restriction sites for ApaI, HindIII, PstI, RsaI, and XbaI, as well as the SV40 minimal promoter (SV) and the herpes simplex type 1 thymidine kinase promoter (TK) are indicated. B, cell type-specific enhancer activity of the human involucrin distal enhancer region. pTKM and p1.1TKM plasmids (10 g) were transfected into multiplying human keratinocytes, MRC-5 fibroblasts, and C-33A cell line. Cells were harvested 48 h post-transfection. Because of the different transfection efficiencies, the activities are plotted relative to the SV40 enhancer/promoter. empirically determined concentrations and stopped using 600 g/ml proteinase K (Boheringer Mannheim) 1 h at 42°C. The nucleic acids were phenol-extracted and precipitated with ethanol. The pellets were washed with 70% ethanol and dissolved in gel loading buffer (80% formamide, 0.1% bromphenol blue, 0.1% xylene-cyanol) and denatured 5 min at 95°C prior electrophoresis through 6% polyacrylamide, 7 M urea sequencing gels.

The Human Involucrin Gene 5Ј-Noncoding Region Contains
Several Transcriptional Regulatory Domains-The transcriptional regulation of the human involucrin gene was analyzed employing deletion constructs derived from p2.6CAT plasmid which express the CAT reporter gene under the control of the intact 5Ј-noncoding region (2456 nt). p2.6CAT activity in multiplying normal keratinocytes was significatively higher to that obtained with p97CAT and p827CAT constructs which contain the minimal promoter and the proximal 784 nt from the human involucrin 5Ј-noncoding region, respectively (Fig. 1A). These results suggest the presence of a distal enhancer element located in the 1185-nt HindIII-RsaI fragment of p2.6CAT. Cloning of this fragment in front of the herpes simplex TK promoter (p1.1TKM) and transfection of multiplying keratinocytes caused a 20-fold increase in CAT activity relative to pTKM plasmid, thus confirming the presence of a transcriptional enhancer (Fig. 1A). Interestingly, the relatively high activity observed in p1.1TKM compared with that obtained for p2.6CAT suggests the presence of a potential transcriptional silencer

FIG. 2. Activity of the involucrin gene 5-noncoding region in differentiation-induced keratinocytes. A, differentiation induction
does not stimulate human involucrin 5Ј-noncoding proximal 784-nt fragment. Normal keratinocytes were transfected with 10 g of total plasmid DNAs from p2.6CAT, p827CAT, p220CAT, and p97CAT constructs. Transfected cells were grown in multiplying (0.1 mM CaCl 2 ) or after differentiation induction (2 mM CaCl 2 ) conditions as described under "Materials and Methods." Cell extracts were obtained 48 h posttransfection. Representative CAT chromatograms from three independent experiments are shown. pCAT-basic and pCAT-control vectors were used as negative and positive controls, respectively. B, calcium responsiveness resides in the distal enhancer. pTKM and p1.1TKM plasmids (10 g) were transfected into keratinocytes and processed as described above. The average CAT activities relative were obtained from three independent experiments.

FIG. 3. Human involucrin 5-noncoding region.
A, complete nucleotide sequence of the involucrin 2456-nt regulatory region. Nucleotide position number Ϫ2456 corresponds to the HindIII site of p2.6CAT plasmid. The sequence segment from nt Ϫ784 to 49 was reported previously (12). Predicted consensus sequences for several transcriptional factors within the proximal 784 nt (underlined) and restriction sites for ApaI, HindIII, and PstI are shown. TATA box and transcription start site are double-underlined. B, plasmid constructs employed for DNase I footprint analysis. The ApaI-XbaI and HindIII-ApaI fragments from p827CAT plasmid are cloned in pIN220 and pIN630 plasmids, respectively. Thick black lines show the position of oligonucleotides employed in gel-shift assays (H1, H2, H3, H4, H4 2072, and H4 2126). The TATA box (vertical box) and ApaI, HindIII, PstI, and XbaI restriction sites are shown. (Fig. 1A).
Confirmation of the above observation was obtained with the p220CAT plasmid containing the ApaI-XbaI fragment from p827CAT, a 6-fold increase in CAT activity relative to p827CAT was observed (Fig. 1A). Therefore, the region upstream the ApaI site, probably has a negative regulatory element. To explore this possibility, the PstI-PstI fragment from p827CAT was cloned in the p610CP plasmid before the SV40 promoter and transfected to multiplying human keratinocytes. However, the observed CAT activity of p610CP was similar to that obtained with the SV40 promoter alone, denotating that other elements present in p827CAT construct are associated with the inhibitory function (Fig. 1A). None of the constructs presented activity when transfected into HeLa cells, which do not express involucrin (data not shown). The p1.1TKM construct was transfected in MRC-5 fibroblasts and C-33A cells to establish the cell type specificity of this enhancer. A mild relative activity of p1.1TKM was noticed in C-33A cells, whereas this same construct remained silent in MRC-5 fibroblasts (Fig. 1B). These results suggest that the elements regulating transcription from the distal enhancer are specific of epithelial-derived cells.
The p2.6CAT plasmid displayed 5-fold activation in calciuminduced differentiation conditions when transfected in keratinocytes stimulated to differentiate by increasing the CaCl 2 concentration to 2 mM in the absence of epidermal growth factor and bovine pituitary extract ( Fig. 2A). These conditions stimulate 3-5-fold the transcription of the involucrin gene (34). However, activity of p97CAT, p220CAT, and p827CAT remained unchanged ( Fig. 2A). Therefore, the calcium responsiveness should reside in the distal 1185-nt enhancer. To test this, the p1.1TKM construct was also transfected in normal keratinocytes under multiplying and differentiation conditions resulting in a significant increase on the p1.1TKM activity in  (Table I) to the binding reaction for 10 min before incubation with DNase I. The relative location of the HP-2 and HP-3 footprints is indicated by brackets. Triangles show recovered sites. Pu, purine sequence ladder. calcium-treated keratinocytes (Fig. 2B).
Thus, three transcriptional regulatory domains are established: a distal enhancer with calcium responsiveness located between nt Ϫ2456 and Ϫ1272, a possible transcriptional silencer located between nt Ϫ651 and Ϫ160, and a proximal enhancer/promoter located between nt Ϫ160 and Ϫ1/ϩ1.
Footprinting Analysis of the Human Involucrin Gene Proximal Promoter/Enhancer-The complete nucleotide sequence of the 2456-nt 5Ј-noncoding region revealed several putative transcriptional factor binding sites (Fig. 3A). AP-1, YY1, and TBP binding sites were located within the proximal enhancer/promoter and the putative transcriptional silencer domains (Fig.  3A). As a part of the analysis of the involucrin transcriptional regulation, DNase I footprinting assays were performed to detect nuclear factors capable of specifically interacting with these domains. Nuclear extracts from multiplying or 2 mM CaCl 2 treated human keratinocytes and HeLa cells were incubated with end-labeled EcoRI-HindIII fragments from pIN220 plasmid, which spans the proximal enhancer/promoter domain (Fig. 3B). The DNase I digestion patterns had several protected regions designated HP-1/HP-2 (nt Ϫ116 to Ϫ26), HP-3 (nt Ϫ139 to Ϫ119), and HP-4 (Ϫ156 to Ϫ143) separated by sites of enhanced DNase I sensitivity readily observable in both strands (Fig. 4A, black arrows). No noticeable difference was seen in the footprint patterns produced by nuclear extracts from multiplying or 2 mM CaCl 2 treated keratinocytes or from HeLa cells.
HP-1 footprint includes the putative TATA box and a consensus sequence for the Sp-1 transcriptional factor 5Ј-GGAGGG-3Ј (35). HP-2 overlaps HP-1 and is located on two putative AP-1 binding sites. The HP-3 footprint is localized over a third AP-1 binding site meanwhile HP-4 is associated to a putative Myb protein binding sequence 5Ј-CCTAAAG-3Ј (6). Footprint assays of pIN220 employing different amounts of a competitor oligonucleotide containing a bona fide AP-1 site from the human papillomavirus type 18 (HPV-18; Ref. 31), and nuclear extracts from multiplying keratinocytes resulted in a dose-dependent competition of HP-2 and HP-3 footprints, suggesting that the nuclear factor involved is AP-1 (Fig. 4B).
AP-1 Binds to the Human Involucrin Proximal Enhancer/ Promoter-Sequence analysis of the footprints produced by keratinocyte nuclear proteins indicates the presence of three potential 5Ј-TGAC/GTCA-3Ј AP-1 binding sites coincident with HP-2 and HP-3 footprints (11,36,37). To examine whether the nuclear factors associated with HP-2 and HP-3 footprints indeed correspond to AP-1-related proteins, gel-shift competition assays were done with nuclear extracts from human keratinocytes cultures and with and without 2 mM CaCl 2 and the EcoRI-HindIII fragment from pIN220 as a probe. The complexes were efficiently competed by a 100-fold molar excess of nonlabeled wild-type AP-1 competitor oligonucleotide (AP-1) but not by a 200-fold molar excess of a mutated AP-1 (AP-1M) or adenovirus NF-1 (NF-1) binding sites ( Fig. 5A and Table I). Similar results were obtained using nuclear extracts from HeLa cells (data not shown).
In addition, gel supershift experiments were performed with nuclear extracts from multiplying keratinocytes to confirm the identity of the observed AP-1-specific complexes in the pIN220 fragment. A decrease in the intensity of the specific DNAprotein retarded complex was observed in the presence of a specific rabbit polyclonal anti-c-jun/AP-1 antibody with the appearance of a clear supershifted band (Fig. 5B). In contrast, a heterologous rabbit polyclonal antibody directed against the human papillomavirus type 16 E7 protein did not affect the retarded complexes (Fig. 5B). Similar results were obtained with nuclear extracts from HeLa cells and CaCl 2 -treated keratinocytes (data not shown). Thus, it is concluded that AP-1 is the nuclear factor from normal keratinocytes associated with the proximal 159-nt enhancer/promoter. Footprint Analysis of the Human Involucrin Transcriptional   H1  5Ј-CTAGCATCACAGGAATAGTTGAGCTACCAGAATCCTCTGG-3Ј  H2  5Ј-CTAGAGGATGCCTCACTCAGCTCCCTGTGCCTAGGTC-3Ј  H3  5Ј-CTAGGGAAAAATTGAGATTCAGAGCAGAAGTTTGACTAAGGTCACAAAACAGT-3Ј  H4 5Ј-CTAGCATGCCCTTGAAAGTGAATCTTTTGGCCTAATAATGAGAA-3Ј a Consensus binding sequences for various transcriptional factors are underlined (as indicated in Figure 3A). b Base changes relevant to transcriptional factor binding are in boldface.
Silencer-The nature of the nuclear factors associated with the transcriptional silencer region found in p827CAT construct was investigated using the cloned HindIII-ApaI 624-nt fragment in the pIN630 plasmid ( Fig. 3B) and nuclear extracts from multiplying or CaCl 2 -treated keratinocytes and HeLa cells. Footprinting analysis revealed four protected regions, H1 (nt Ϫ222 to Ϫ166), H2 (nt Ϫ287 to Ϫ235), H3 (nt Ϫ313 to Ϫ292), and H4 (nt Ϫ387 to Ϫ317), respectively (Fig. 6). The overall footprinting pattern obtained with nuclear extracts from keratinocytes with and without CaCl 2 treatment was similar, although a distinct reproducible difference occurred in the upper strand H4 footprint (Fig. 6). Interestingly, the sequence protected by H4 footprint contains a sequence track homologous to binding sites for the YY1 factor (38,39). Distinct differences between keratinocytes and HeLa cells nuclear extracts were also noticed in H3 and H4 footprints (Fig. 6). H1 and H2 footprints displayed similar patterns for all nuclear extracts employed but with slight differences in the size of the protected zone for both strands (Fig. 6). The location of the H2 footprint in a DNA segment containing a putative AP-1 site suggests that AP-1 could also interact with this region. Differential Nuclear Factor Binding in the Involucrin Gene Putative Transcriptional Silencer-Synthetic oligonucleotides containing the H1, H2, and H3 footprints from the pIN630 plasmid ( Fig. 3B and Table I) were used to identify the associ-ated nuclear factors as well as possible cell type-and stagespecific variations in standard gel-shift assays. To facilitate analysis, the H4 footprint was spliced into three different oligonucleotides H4 2072, H4, and H4 2126 (Table I). Several retarded complexes were noticed using both multiplying and calcium-induced keratinocytes nuclear extracts. The specificity of the DNA-protein interactions was tested by preincubating the nuclear extracts with a 100-fold molar excess of unlabeled homologous probe (Fig. 7). For H1 and H4 oligonucleotides, no specific competition was noticed when using the binding sites of AP-1, NF-1, and YY1 (data not shown). H1 and H2 oligonucleotides presented single band-specific retarded complexes, suggesting that only one factor may be implicated in these interactions (Fig. 7, panels H1 and H2). The specific retarded complex from H1 displayed a 2-3-fold increase in intensity with nuclear extracts derived from CaCl 2 -treated keratinocytes, but no noticeable difference was found for the H2-retarded complex (Fig. 7, panels H1 and H2).
H3 and H4 oligonucleotides produced a more elaborated gelshift pattern. H3 presented at least two specific retarded complexes which were increased in nuclear extracts from CaCl 2treated keratinocytes (Fig. 7, panel H3). H4 oligonucleotide had two specific DNA-protein complexes with either nuclear extract, suggesting the interaction of multiple nuclear factors with this sequence (Fig. 7, panel H4).
Cell type specificity was tested using nuclear extracts from HeLa cells. The similarity of the gel-shift pattern with H1 and FIG. 7. DNA-binding proteins interaction with the human involucrin transcriptional silencer. Gel-shift assays were performed incubating nuclear extracts from multiplying (Ker) and 2 mM CaCl 2treated (Ki) keratinocytes with 1 ng of end-labeled oligonucleotides containing the H1, H2, H3, and H4 footprint sequences (Table I) from  pIN630 (panels H1, H2, H3, and H4, respectively). The binding reactions were done as described in the legend to Fig. 5A. Competitions were performed with 100 molar excess of competitor oligonucleotide before electrophoresis 6% nondenaturing low ionic strength polyacrylamide gels. Arrows indicate the position of specific retarded complexes.
FIG. 8. Differential DNA-protein binding between keratinocytes and HeLa cells. Nuclear extracts of multiplying (Ker) and CaCl 2 -treated keratinocytes (Ki) or HeLa cells were used in gel-shift assays with the H1, H2, H3, and H4 oligonucleotides as described in the legend to Fig. 5A (panels H1, H2, H3, and H4, respectively). The keratinocyte (black arrows) and the differential HeLa cells (open arrows) complexes are shown.
H2 suggests that nuclear factors are shared by HeLa cells and keratinocytes (Fig. 8, panels H1 and H2). In contrast, differences were observed between HeLa cells and keratinocytes with the H3 and H4 oligonucleotides. H3 had an extra upper DNA-protein complex with HeLa nuclear extracts. The common complexes seem to be produced by a nuclear protein more abundant in HeLa cells than in keratinocytes (Fig. 8, panel  H3). For H4 oligonucleotide, at least one DNA-protein complex was absent from keratinocyte nuclear extracts, suggesting that in HeLa cells additional factors may interact with this region (Fig. 8, panel H4).
AP-1 and YY1 Transcriptional Factors Interact with the Human Involucrin Putative Transcriptional Silencer-As with pIN220, several potential AP-1 binding sites were found within the pIN630 fragment (Fig. 3A). Three of them coincide with the observed footprints H1, H2, and H3. Gel-shift competition experiments using nuclear extracts from multiplying keratinocytes demonstrated that only the H2 DNA-protein complex is efficiently competed by a 30-fold molar excess of AP-1 homologous competitor (Fig. 9A). Similar results were obtained with HeLa and CaCl 2 -treated keratinocytes (data not shown). A 100-fold molar excess of either AP-1M or NF-1 competitor oligonucleotides had no effect on the H2 complex (Fig. 9A).
Gel supershift assays confirmed the AP-1 identity of the H2 complex using nuclear extracts from multiplying and CaCl 2 -treated keratinocytes and HeLa cells nuclear extracts. The H2 complex intensity was simultaneously reduced with the appearance of a supershifted complex only after addition of an anti-AP-1 antibody to the binding mixture, verifying that the H2 footprint indeed corresponds to AP-1 (Fig. 9B).
Two different potential YY1 binding sites coincide with the position of H4 footprint, 5Ј-TTTCCATTTCA-3Ј and 5Ј-TCAT-TTTGAA-3Ј at nt Ϫ383 and Ϫ329, respectively (Fig. 3A). These sequences share homology with the 5Ј-CAT-3Ј motif present in the YY1 binding sites from several genes (Fig. 10A). To test if these sequences indeed bind the YY1 transcriptional factor, competitive gel-shift assays were performed using the endlabeled H4 2072 and H4 2126 oligonucleotides (flanking the H4 footprint) and the P5ϩ1 and P5ϩ1 mutant from AAV (38) and YY1 (40,41) nonlabeled competitors with nuclear extracts from HeLa cells and multiplying keratinocytes. Specific complexes for H4 2072 and H4 2126 were efficiently competed with both P5ϩ1 and YY1 oligonucleotides, but not with a P5ϩ1 mutant, which does not bind YY1 (Fig. 10B). Additionally, cross-competition between H4 2072 and H4 2126 oligonucleotides indicates that YY1 interacts with both sequences.

DISCUSSION
The human involucrin 5Ј-noncoding region contains several binding sites for transcriptional regulatory proteins. The present results describe the presence of three functional domains, one enhancer/promoter and a transcriptional silencer domains located within a 784-nt fragment proximal to the transcription start site as well as a far upstream 1185-nt enhancer domain (Fig. 11).
The full-length 2456-nt involucrin upstream regulatory region displayed significant activity in multiplying normal keratinocytes, while the proximal 784-nt fragment did not. Interestingly, only the p2.6CAT construct with the intact 5Јnoncoding region and the distal 1185-nt enhancer were activated after calcium induction of differentiation, suggesting that factors associated with calcium activation interact within this last region. The distal enhancer was ineffective in fibroblasts indicating cell type specificity of this enhancer function. Nevertheless, the proximal enhancer/promoter and the transcriptional silencer were not altered by differentiation induction. No transcriptional activity was noticed with any of these functional regions in non-involucrin expressing cells, such as HeLa or fibroblasts, indicating that cell type specificity could be dependent either on several factors or a single transcriptional factor associated with all the three regulatory domains.
The interaction of AP-1 with the proximal enhancer/promoter and the putative transcriptional silencer reported here suggests that this transcriptional factor may be primarily re- H4 2072 and H4 2126 are referred to oligonucleotides containing the 5Ј and 3Ј ends of the H4 footprint, respectively. B, YY1 interacts with the human involucrin putative transcriptional silencer. Gelshift assays were done employing 1 ng of end-labeled H4 2072 or H4 2126 oligonucleotides (containing the putative YY1 sites from H4 footprint) and nuclear extracts from multiplying keratinocytes as described in the legend to Fig. 5A. For specific competition, 100-and 200-fold molar excess of the indicated competitor oligonucleotide was used prior electrophoresis through 4% nondenaturing low ionic strength polyacrylamide gels. sponsible for specific involucrin transcriptional activity in normal keratinocytes. The similarity between the footprint and gel-shift patterns observed with both regions independent of the differentiation state of normal keratinocytes supports this notion. AP-1, a transcriptional factor integrated by dimerization of products from fos and jun oncogene families (42), activates genes with the 5Ј-TGANTC/AA-3Ј consensus motif in response to compounds such as TPA that activate the protein kinase C (36,37). The results presented here agree with previous reports showing that TPA treatment or fos and jun overexpression activates transcription from the proximal 784-nt fragment (11). A recent report (43) demonstrates that AP-1 sites present in the proximal enhancer/promoter (nt Ϫ124 to Ϫ118) and in the transcriptional silencer (nt Ϫ288 to Ϫ282) are important for the TPA responsiveness of the human involucrin gene. Accordingly, these AP-1 sites coincide with the position of HP-3 and H3 footprints. Furthermore, the present results provide evidence of the presence of an extra AP-1 site (H2) at positions Ϫ263 to Ϫ255.
Calcium-induced differentiation of transfected keratinocytes did not affect the activity of either the proximal promoter/ enhancer or the transcriptional silencer regions, both being capable of interacting with AP-1. Additionally, no activity was registered with any reporter construct in fibroblasts, a cell type that contains AP-1. Furthermore, the p220CAT construct was inactive in transfected HeLa cells despite the interaction of AP-1 (data not shown), suggesting that a particular combination of AP-1 may be implicated in involucrin gene transcription. In agreement with this, Welter et al. 1995 (43) established that Fra1, JunB, and JunD are the factors associated to the enhancer/promoter region. Thus, the sum of the results suggests the existence of two different control mechanisms for involucrin gene transcription, one dependent on AP-1 activation and the other associated with calcium-dependent pathways.
Consistent with this hypothesis, the intact 2456-nt noncoding region is more efficient in the presence of 2 mM CaCl 2 . The enhanced activity requires the far upstream 1648 nt that includes the distal 1185-nt enhancer domain described in this work. Thus, the AP-1-and calcium-dependent involucrin regulatory pathways are apparently functionally and physically separable within the 5Ј-noncoding region.
Several putative binding sites for transcriptional factors are located within the 1185-nt distal enhancer. A detailed analysis is needed to establish the interaction and functional value of each of these factors in the context of calcium-induced differentiation.
Ying-Yang 1 or YY1 is a zinc finger protein related to the Krü ppel family of transcriptional regulators of Drosophila melanogaster, with the unusual property of being able to activate or repress transcription initiation depending on the cellular context. Moreover, YY1 binding sites vary among cellular and viral promoters (44). On one hand, YY1 activates transcription of c-myc (45), ribosomal proteins L30 and L32 (40), and cytochrome c oxidase genes (46) and the leaky late promoter of herpes simplex (47) and the P6 promoter of B19 parvovirus (48). On the other hand, YY1 represses the regulatory regions from c-fos (49), the skeletal ␣-actin (50), human immunodeficiency virus type 1 (51), HPV-18 long control region (52), and the human cytomegalovirus major immediate early enhancer/promoter (53).
The ambivalent nature of YY1 as an activator or a silencer led to a hypothesis concerning the importance of this factor for human involucrin transcription. Although the abundance and function of YY1 in human keratinocytes are not known, the current results suggest that this factor may repress human involucrin transcription in multiplying and calcium-treated keratinocytes. Both, the 1185-nt distal enhancer and the 159-nt proximal enhancer/promoter are active, lacking the 624-nt fragment that contains the YY1 binding sites. Because YY1 physically interacts with other proteins (39,54), it is possible that the mechanism of YY1 repression in the involucrin gene could be the association of YY1 with the Sp-1 basal transcription factor, whose putative binding site is present within the 159-nt enhancer/promoter. Accordingly, the substitution of the native involucrin TATA box with the SV40 promoter in the p610CP plasmid (that contains several Sp-1 sites in the 21-nt repeats) showed no increase in activity when compared with the control despite the presence of four AP-1 sites (Fig. 1A). Site-directed mutagenesis experiments will be required to verify such interaction.
The association between two apparently antagonistic transcriptional factors such as AP-1 and YY1 with the involucrin 5Ј-noncoding region resembles the epithelial-specific HPV-18 long control region, which also interacts with both factors in similar tissue-specific enhancer (AP-1) and transcriptional silencer (YY1) functions (31,52). A particular combination of AP-1 containing junB is responsible for the HPV-18 tissuetrophism (31). Interestingly, oligonucleotides containing an AP-1 site from HPV-18 efficiently competed for the involucrin AP-1 complexes from the 159-nt promoter/enhancer. It has been shown that JunB is associated with involucrin transcription (43). Therefore, functional association between YY1 and junB can be proposed as a possible regulatory mechanism for epithelial expressed genes.