Induction of disease-associated keratin 16 gene expression by epidermal growth factor is regulated through cooperation of transcription factors Sp1 and c-Jun.

Overexpression of keratin 16 has been observed in keratinocytes in those skin diseases characterized by hyperproliferation such as psoriasis. Therefore, keratin 16 is usually referred to as a disease-associated keratin. In the present study, we found that epidermal growth factor (EGF) increased the expression of keratin 16 mRNA and protein synthesis in a time-dependent manner in HaCaT cells. Reporter assays revealed that the EGF response region was in the range of -162 to -114 bp. Disruption of the Sp1 site (-127 to -122 bp) and the AP1 site (-148 to -142 bp) of the keratin 16 promoter by site-directed mutagenesis significantly inhibited keratin 16 promoter activity induced by EGF. Furthermore, keratin 16 gene expression induced by Ras activation was also regulated in the same manner as the EGF response. By using the DNA affinity precipitation assay in HaCaT and SL2 cells, Sp1 directly interacted with the Sp1 site of the promoter, and c-Jun and c-Fos precipitated with the Sp1 oligonucleotide was attributable to the interaction between the Sp1 and AP1 proteins. Moreover, cotransfection assays revealed that Sp1 acted synergistically with c-Jun to activate keratin 16. The coactivators p300/CBP could collaborate with Sp1 and c-Jun in the activation of keratin 16 promoter, and EGF-induced promoter activation was blocked by the viral oncoprotein E1A. Taken together, these results suggest that Sp1 and AP1 sites in the essential promoter region are critical for EGF response, and Sp1 showed a functional cooperation with c-Jun and coactivators p300/CBP in driving the transcriptional regulation of EGF-induced keratin 16 gene expression.

The epidermis forms the external surface of the skin and consists of many layers including the stratum basale, stratum spinosum, stratum granulosum, stratum lucidum, and stratum corneum (1). Keratinocytes are the most abundant of the cell types in the epidermis. The maintenance of homeostatic balance in epidermal keratinocytes is dependent on the coordinate regulation of differentiation and activation (2). In healthy epidermis, the basal keratinocytes are mitotically active cells that differentiate sequentially from the basal to the cornified layer in skin (3). However, in response to epidermal injury or in certain skin diseases characterized by hyperproliferation, such as psoriasis and epidermal malignancies, the basal keratinocytes become activated. They are triggered to terminal differentiation and initiate their migration and hyperproliferation through the suprabasal layers (4). The activated epidermal keratinocytes become able to produce and response to growth factors and cytokines, such as epidermal growth factor (EGF) 1 and transforming growth factor-␣ (TGF-␣) (5). In psoriatic lesions, overexpression of EGF receptor (6) and TGF-␣ (7) has been reported.
The most prominent cytoskeletal proteins in keratinocytes are keratins, a large family of proteins that form the intermediate filament network in all epithelial cells. There are about 30 keratins that can be subdivided into type I (acidic) and type II (basic) according to biochemical criteria. They are usually expressed in pairs with one partner from each group (8). The programmed expression of keratins, which is determined by the location and function of the keratinocytes within the epidermis, is a commonly used phenotypic marker of epithelial development and differentiation. The basal layer of the epidermis produces keratins K5, K14, and low amounts of K15 (9), and the differentiating, suprabasal layers produce keratins K1, K2, and K10 (3), whereas in activated keratinocytes, keratins K6, K16, and K17, which are distinct from the keratins in the healthy epidermis, are expressed (10). Therefore, keratins K6, K16, and K17 are usually referred to as activation-and hyperproliferation-associated keratins.
In the skin diseases characterized by hyperproliferation such as psoriasis, keratin 16 expression has been detected (11). Furthermore, keratin 16 has been reported to be the marker of keratinocyte hyperproliferation in psoriasis in vivo and in vitro (12). Keratin 16 is absent in normal breast tissue and in noninvasive breast carcinomas, but 10% of the invasive breast carcinomas show diffuse or focal positive with the K16 antibody (13). It has been reported as having expression limited to skin and breast tissue. However, the specific function of keratin 16 is not fully understood. Mutations in the keratin 16 gene in the human population have been associated with pachyonychia congenita type I, which features hyperkeratosis and occasional blistering in the absence of keratinocyte lysis (14). Keratin 16 mutations can also be presented as focal non-epidermolytic palmoplantar keratoderma, which produce phenotypes with little or no nail changes (15).
In the studies of gene regulation of keratin 16, Jiang et al. 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 (16) reported previously that the hyperproliferation-inducing agents, EGF and TGF-␣, induce keratin 16 expression in normal human epidermal keratinocytes and found that an EGFresponsive element (EGF-RE), located at the promoter ranged from Ϫ212 to Ϫ192 bp, is required for EGF and TGF-␣ responses. However, the transcription factors acting on the EGF-RE were not identified. In the study of basal expression of keratin 16 in human epidermal keratinocyte cells, in addition to EGF-RE, an Sp1 binding site located in the promoter range from Ϫ127 to Ϫ121 bp is also essential (17). The aim of the present work is to further characterize the EGF-responsive region in the human keratin 16 gene promoter in a spontaneously immortalized keratinocyte (HaCaT) cell line. We found that the transcription factor AP1 acting on an EGF-responsive region, which was distinct from EGF-RE reported by Jiang et al. (16), cooperated with Sp1 to activate the gene promoter. Cell Culture and EGF Treatment-HaCaT cells, a spontaneously immortalized human epidermal keratinocyte cell line, were grown at 37°C under 5% CO 2 in 10-cm plastic dishes containing 8 ml of Dulbecco's modified EagleЈs medium supplemented with 10% (v/v) fetal bovine serum, 100 g/ml of streptomycin and 100 IU/ml of penicillin. In this series of experiments, cells were treated with 30 ng/ml of EGF in optimal serum-free conditions, unless stated otherwise. Drosophila Schneider line 2 (SL2) cells were maintained at 25°C in SchneiderЈs Drosophila medium supplemented with 10% (v/v) fetal bovine serum.

Materials
Reverse Transcription-PCR-Cells maintained for 2 days in serumfree medium were incubated with EGF for the indicated time period as described and harvested. Total RNA was isolated using the TRIzol RNA extraction kit, and 5 g of RNA were subjected to reverse transcription-PCR with SuperScript TM II. The keratin 16-specific primers and glyceraldehyde-3-phosphate dehydrogenase primers were used as controls. The PCR products were separated by 1% agarose-gel electrophoresis and visualized with ethidium bromide staining.
Preparation of Nuclear Extracts-Cells from 10-cm plastic dishes were washed twice with PBS and scraped in 500 l of PBS. Cells were collected by centrifuging at 7,500 ϫ g for 20 -30 s, resuspended in 400 l of buffer A (10 mM Hepes (pH 7.9), 1.5 mM MgCl 2 , and 10 mM KCl) and stood on ice for 10 min. Nuclei were pelleted by centrifugation at 7,500 ϫ g for 20 -30 s. Pellets were resuspended in 100 l of buffer C (20 mM Hepes (pH 7.9), 1.5 mM MgCl 2 , 0.2 mM EDTA, 420 mM NaCl, and 25% (v/v), glycerol) and stood on ice for 20 min. The suspension was centrifuged at 7,500 ϫ g for 2 min. The supernatants were collected and stored at Ϫ70°C until used. Buffer A and buffer C contained 0.5 mM dithiothreitol, 2 g/ml of leupeptin, 1 mM orthovanadate, 2 g/ml of pepstatin A, and 0.5 mM phenylmethylsulfonyl fluoride.
Western Blots-Analytical 10% SDS-polyacrylamide slab gel electrophoresis was performed. The cell nuclear extracts or lysates (30 -200 g of protein of each) prepared from control and EGF-treated cells were analyzed. The supernatants were mixed with 5ϫ SDS loading buffer followed by boiling for 5 min. For immunoblotting, proteins in the SDS gels were transferred to a polyvinylidene difluoride membrane by an Electroblot apparatus (Bio-Rad, Hercules, CA). After blocking with 5% non-fat dried milk in 1ϫ PBS buffer, the blots were incubated with antibodies against specific proteins as the primary antibodies, individually. Horseradish peroxidase-conjugated secondary antibodies were added, and the bands were detected by an enhanced chemiluminescence kit from Pierce Biotechnology (SuperSignal, Rockford, IL).
DNA Affinity Precipitation Assay-This assay was performed according to the method of Zhu et al. (18) with a slight modification. The binding assay was performed by mixing 200 g of nuclear extract proteins, 2 g of biotinylated keratin 16-specific Sp1 or AP1 oligonucleotides, and 20 l of streptavidin-agarose beads (4%) with 50% slurry. The mixture was incubated at room temperature for 1 h with rotating. Beads were pelleted and washed with cold PBS three times. The binding proteins were eluted by loading buffer and separated by SDS-PAGE, followed by Western blot analysis probed with specific antibodies. 5Ј-Biotinylated wild-type and mutated Sp1 sequences were Sp1-WT, 5Ј-biotin-GTTAGGAGGGCCCCGCCTTCCCCAGG-3Ј and Sp1-Mut, 5Ј-biotin-GTTAGGAGGGCCCAAACTTCCCCAGG-3Ј. 5Ј-Biotinylated wild-type AP1 sequence was AP1-WT, 5Ј -biotin-GGGGAACCTGGAG-TCAGCAGTTAGGA-3Ј.
DNA Gel Mobility Shift Assay-Double-stranded oligonucleotides prepared by the PCR method were end-labeled with [␥-32 P]ATP and T4 polynucleotide kinase as described previously (19). Labeled oligonucleotides were purified by ProbeQuant TM G-50 Micro Columns and used as probes. The binding reaction was performed in 15 l of reaction mixture containing 0.2 g of poly(dI-dC)⅐poly(dI-dC), 50 mM Hepes (pH 7.9), 0.25 mM EDTA, 25% (v/v) glycerol, 5% (w/v) polyvinyl alcohol, 2.5 mM dithiothreitol, 0.6 mM phenylmethylsulfonyl fluoride, 3 g of the cell nuclear extracts, and the radiolabeled probe (2 ϫ 10 4 cpm). The mixtures were incubated at room temperature for 30 min and loaded on a 5% (w/v) polyacrylamide gel. Electrophoresis was performed at a constant 300 V for 1.5 h. The gel was dried and autoradiographed.
Plasmid Construction-Human keratin 16 promoter regions of various lengths were prepared by the PCR-amplification method for the preparation of pXK-1, 2, 3, 4, 5, 5-1, 5-2, and 6 as described previously (20), and the mutants at the sites of Sp1 and AP1 were constructed by the site-directed mutagenesis method as described by Higuchi et al. (21). The sequences were all confirmed by DNA sequencing with the chromosome 17q12-q21 reported by The Human Genome Project. These DNA fragments were ligated into the respective polycloning sites of pXP-1.
Transfection with LipofectAMINE and Reporter Gene Assay-Cells were transfected with plasmids by lipofection using LipofectAMINE according to the manufacturerЈs instruction with a slight modification. Cells were replated 24 h before transfection at an optimal cell density in 2 ml of fresh culture medium in a 3.5-cm plastic dish. For use in transfection, 12.5 l of LipofectAMINE were incubated with 0.5 g of pXK luciferase plasmid or the indicated plasmids as described in each experiment, in 1 ml of Opti-MEM medium for 30 min at room temperature. Total DNA concentration for each transfection was matched with pcDNA3.1. Cells were transfected by changing the medium with 1 ml of Opti-MEM medium containing the plasmids and LipofectAMINE, followed by incubation at 37°C in a humidified atmosphere of 5% CO 2 for 24 h. After a change of Opti-MEM medium to 2 ml of fresh culture medium, cells were stimulated with EGF if necessary and then incubated for an additional 24 h. The luciferase activities in cell lysates were measured by the luciferase assay system and determined as described previously (22). Luciferase activity was normalized per g of extract protein.
Transfection by the Calcium Phosphate Method-Drosophila SL2 cells were replated 24 h before transfection at a density of 3.6 ϫ 10 6 cells in 8 ml of fresh culture medium in a 10-cm plastic dish. Cells were transfected with 4 g of either Drosophila expression vector pPac-Sp1 or the empty vector pPac by the calcium phosphate/DNA co-precipitation method as described (23). The DNA precipitates were left on the cells, followed by incubation at 25°C for 24 h. The medium was changed to 8 ml of fresh culture medium. After an additional 48 h, cells were washed twice with PBS, and the nuclear extracts were collected. The results are summarized in Fig. 3. Overexpression of pSV2ras significantly stimulated the keratin 16 reporter activity in a dose-dependent manner (Fig. 3A). On the other hand, transfection of pMMrasDN in cells dose-dependently inhibited EGF-induced promoter activation of keratin 16 (Fig. 3B). The results indicated that the Ras signaling pathway plays a very significant role in EGF-induced expression of keratin 16.

Stimulation of Keratin 16 Expression by EGF-EGF
Keratin 16 Promoter Analysis of EGF and Ha-ras Responses-To study the transcriptional regulation of the keratin 16 gene in HaCaT cells, the luciferase reporter vectors bearing various lengths of 5Ј-flanking regions of keratin 16 gene were con-structed as shown in Fig. 4. The transcriptional activities of luciferase-bearing vectors pXK-1, pXK-2, pXK-3, pXK-4, pXK-5, pXK-5-1, pXK-5-2 and pXK-6 were stimulated by EGF treatment and by Ha-ras overexpression. The results are summarized in Fig. 4. There was ϳ30% decrease in the stimulatory response of EGF and Ha-ras transfection observed in vectors bearing a promoter with an identical deletion from Ϫ162 to Ϫ141 bp (pXK-5-1 to pXK-5-2). A more significant decrease (ϳ70%) in EGF and Ha-ras-stimulated response was observed in a promoter region covering from Ϫ141 to Ϫ114 bp (pXK-5-2 to pXK-6). These results indicated that the 5Ј-flanking region of the keratin 16 gene ranging from Ϫ162 to Ϫ114 bp was required for EGF response as for Ha-ras response.
Sequence analysis of the keratin 16 promoter region from Ϫ162 to Ϫ114 bp indicated the presence of two putative binding sites for AP1 (Ϫ148 to Ϫ142 bp) and Sp1 (Ϫ127 to Ϫ122 bp). To further study the EGF-and Ha-ras-responsive regions in the promoter ranging from Ϫ162 to Ϫ114 bp, plasmids with sitedirected mutagenesis of the AP1 site, the Sp1 site, and both sites in the promoter region were constructed (Fig. 5A). The results are summarized in Fig. 5B. A slight decrease (ϳ30%) in the stimulatory response of EGF or Ha-ras overexpression was observed in pXK-mAP1 (5-1) with the mutated AP1 site, and the effect was equivalent to that in pXK-5-2 with a deletion of AP1 site. A more apparent decrease (ϳ70%) in EGF-or Ha-rasstimulated response was observed in pXK-mSp1 (5-1) with the mutated Sp1 site, and a complete elimination of EGF or Ha-ras response was detected in pXK-dm (5-1) with double mutations at the AP1 and Sp1 sites and pXK-mSp1 (5-2) with a deletion of the AP1 site and a mutation at the Sp1 site. These results suggested that both the AP1 site residing at Ϫ148 to Ϫ142 bp and the Sp1 site residing at Ϫ127 to Ϫ122 bp were two critical elements for EGF response and for Ha-ras response.
Effects of EGF and Ha-ras on the Expression of Sp1, c-Jun, and c-Fos-Several recent studies have implicated Sp1 proteins (25) and AP1 proteins (26), respectively, as key regulators in the transcriptional control of epidermal gene expression. To study the possible transcription factors involved in this EGF response, expression of Sp1, c-Jun, and c-Fos in cell nuclear extracts from HaCaT cells was analyzed by Western blot. As shown in Fig. 6A, EGF induced the expression of c-Jun and c-Fos proteins in a time-dependent manner. The maximum induction of c-Jun protein was observed in cells treated with EGF for 1 h and then sustained at least up to 24 h after EGF treatment. In contrast to the long-term expression of c-Jun protein, the maximum induction of c-Fos protein was also observed at 1 h after EGF treatment, but the induction then declined and almost disappeared in cells treated with EGF for 6 h. However, no change in Sp1 protein expression was found by EGF treatment. Besides the EGF response, we also examined levels of c-Jun expression in HaCaT cells transfected with Ha-Ras and found that overexpression with 0.5 to 2 g of pSV2ras induced the expression of c-Jun protein in a dose-dependent manner (Fig. 6B).
Identification of the Nuclear Factors Involved in the Expression of the Keratin 16 Gene-DNA affinity precipitation assay, which provides quantitative information of transactivator binding, was performed to examine whether any of the abovementioned nuclear factors bound to the promoter sequence from Ϫ162 to Ϫ114 bp and to evaluate the effect of EGF on the binding of these transactivators. Oligonucleotides covering the Sp1 site (Ϫ127 to Ϫ122 bp) and the AP1 site (Ϫ148 to Ϫ142 bp), respectively, were synthesized, labeled with biotin, and used as probes. Nuclear extracts from HaCaT cells treated with and without EGF as indicated were incubated with the biotinylated probe and streptavidin-agarose beads. Transcription factors present in the complex were analyzed by Western blot. No change in Sp1 binding to the 5Ј-biotinylated wild-type Sp1 sequence (Sp1-WT) was found by EGF treatment. However, a significant binding of c-Jun and c-Fos to Sp1-WT was detected in cells treated with EGF, particularly at 1 h after treatment (Fig. 7A). Additionally, EGF induced the bindings of c-Jun and c-Fos to the 5Ј-biotinylated wild-type AP1 sequence (AP1-WT) in a time-dependent manner. The maximum bindings of c-Jun and c-Fos were both observed in cells treated with EGF for 1 h, but unlike c-Fos, binding declined quickly; c-Jun binding to Values of pXK-mAP1 (5-1), pXK-mSp1 (5-1), and pXK-dm (5-1) were compared with that of pXK-5-1, and the value of pXK-mSp1 (5-2) was compared with that of pXK-5-2. AP1-WT was sustained and slightly decreased at 24 h after EGF stimulation (Fig. 7B). These results supported the notion that transcription factors Sp1 as well as c-Jun and c-Fos were all detected on the Sp1 site by EGF treatment, and those bound to the AP1 site were c-Jun and c-Fos. To confirm the specificity of the AP1 proteins bound to the Sp1 sequence of promoter, the 5Ј-biotinylated mutated Sp1 sequence (Sp1-Mut) was used. As shown in Fig. 7C, no binding of c-Jun and c-Fos to the Sp1-Mut probe was observed in cells treated with EGF for 1 h, compared with its respective binding to the Sp1-WT probe, whereas lack of Sp1 binding to the Sp1-Mut probe was also observed in cells as expected, with and without EGF treatment. These results indicated that the AP1 proteins were specifically precipitated with the Sp1 site in this probe.
To further determine whether c-Jun and c-Fos bindings to the Sp1 site were through protein-protein interaction with Sp1 or independent on Sp1, DNA gel mobility shift assay and supershift assay were performed to examine the identities and specificities of those proteins binding to the Sp1 probe. Oligonucleotide covering the Sp1 site (Ϫ127 to Ϫ122 bp) from Ϫ141 to Ϫ111 bp (SpWT) was synthesized by the PCR method and labeled with [␥-32 P]ATP and used as a probe. When nuclear extracts from HaCaT cells were allowed to react with SpWT, three retarded bands (a, b, and c) were observed (Fig. 8A, lane  1). Binding of the HaCaT cell nuclear proteins prepared from control and EGF-treated cells to the promoter probe was analyzed. No change in these nuclear protein-DNA complexes was found in cells treated with EGF for 1 and 3 h, compared with its respective control cells (Fig. 8A, lanes 1-3). For identifying whether Sp1 and Sp3 bound the DNA probe, polyclonal antibodies that specifically recognize human Sp1 and Sp3 proteins were used for the supershift assay. The retarded band b was shifted by the antibody against Sp1 (Fig. 8A, lane 4), and the retarded band c was abolished by the Sp3 antibody (Fig. 8A,  lane 5), but all of these three retarded bands were not altered by the antibodies against c-Jun and c-Fos (Fig. 8A, lanes 6 and  7). Furthermore, these three bands were not competed out by the unlabeled oligonucleotides of Sp1 site-directed mutant SpM (Fig. 8A, lane 8). These results suggested that the Sp1 site in the keratin 16 gene promoter was the DNA sequence for direct binding of nuclear proteins including Sp1 and Sp3 but not c-Jun and c-Fos.
To study directly the binding of nuclear c-Jun/Sp1 with the Sp1 site in the keratin 16 gene promoter, Sp1-deficient Drosophila Schneider SL2 cells were used as a model. The existence of Drosophila c-Jun protein that is highly homologous to members of the mammalian Jun family in both the DNA binding and leucine zipper domains has been well-documented (27). Cells were transfected with either the expression vector pPac-Sp1 (Fig. 8B, lanes 2, 4, 6, and 8) or the empty vector pPac (Fig.  8B, lanes 1, 3, 5, and 7) as control. In control cells, the presence of the endogenous Drosophila c-Jun protein expression was observed (Fig. 8B, lane 7), whereas no Sp1 protein expression was detected, as expected (Fig. 8B, lane 5). Under the control conditions, no binding of Sp1 and c-Jun to the Sp1-WT probe was observed by DNA affinity precipitation assays (Fig. 8B,  lanes 1 and 3), indicating that c-Jun could not interact with the Sp1 site directly. In the pPac-Sp1-transfected cells, no change of endogenous c-Jun was observed (Fig. 8B, lane 8). However, precipitation of c-Jun and Sp1 with the Sp1-WT probe was observed (Fig. 8B, lanes 4 and 2). These results clearly suggested that c-Jun protein binding to the Sp1 site was dependent on Sp1 protein through protein-protein interaction.
Keratin 16 Promoter Analysis of c-Jun Response-We found that overexpression of c-Jun and c-Fos, respectively, stimulated the keratin 16 promoter activity in a dose-dependent manner. However, the response of c-Fos overexpression was only one-eleventh of that of c-Jun overexpression (data not shown). Therefore, the induction of c-Jun protein might contribute more than c-Fos protein in the increase of AP1 activity induced by EGF treatment in the response of keratin 16 induction. To explore the effect of c-Jun overexpression on the promoter activation of the keratin 16 gene, luciferase reporter vectors bearing various lengths of 5Ј-flanking regions of the keratin 16 gene were used. The results are summarized in Fig.  9A. An apparent decrease in the stimulatory response of c-Jun transfection was observed in vectors bearing a promoter with a deletion from Ϫ162 to Ϫ114 bp (pXK-5-1 to pXK-6), indicating that a promoter region ranging from Ϫ162 to Ϫ114 bp was essential for the c-Jun-stimulated response of keratin 16 expression. With the aid of site-directed mutagenesis, we found that a significant decrease (ϳ63%) in c-Jun-stimulated response was observed in pXK-mSp1 (5-1), and an almost complete elimination of c-Jun response was detected in pXK-dm (5-1) and pXK-mSp1 (5-2). These results indicated that the AP1 site (Ϫ148 to Ϫ142 bp) and the Sp1 site (Ϫ127 to Ϫ122 bp), respectively, played an important role in c-Jun-induced transcription of the keratin 16 gene, as reported previously in EGF and Ha-ras responses. Transfection of pRSVjun in cells dosedependently increased keratin 16 promoter activity of pXK-5-2 bearing the Sp1 site on promoter; however, no obvious induction was detected in pXK-mSp1 (5-2) with a mutation at the Sp1 site (Fig. 9B). These results further supported the notion that the cooperation of c-Jun with Sp1 was required in the activation of the keratin 16 gene promoter.
Cooperation of Sp1 and c-Jun with p300/CBP in Up-regulation of the Keratin 16 Promoter-To determine whether c-Jun cooperates with Sp1 to activate transcription of the keratin 16 gene, HaCaT cells were cotransfected with the keratin 16 promoter constructs as indicated, along with expression plasmids encoding Sp1 and c-Jun either individually or in combination. As shown in Fig. 10A, co-expression of Sp1 and c-Jun showed a  5-8) that was similar to the effect on the empty vector pXP1 (data not shown). To explore the functional role of transcriptional coactivators p300/CBP in modulating the activity of Sp1 and c-Jun on keratin 16 promoter, the expression vectors encoding p300 and CBP were used. As shown in Fig. 10B, Sp1, c-Jun, and p300 (lanes 5-7) or CBP (lanes 9 -11) showed a dose-dependent activation on the keratin 16 promoter. Meanwhile, in the cells expressing E1A, an adenoviral oncoprotein that binds to and inactivates p300/CBP (28), the keratin 16 promoter activity induced by EGF was obviously dose-dependently suppressed (Fig. 10C).

DISCUSSION
In this study, we have provided several pieces of evidence suggesting that the Sp1 and the AP1 sites in the essential promoter region were required for EGF-induced transcription of the human keratin 16 gene as summarized in the proposed model for the transcriptional regulation of the keratin 16 gene in HaCaT cells (Fig. 11). With the aid of 5Ј-deletion constructs of the keratin 16 gene promoter, the EGF-responsive region was narrowed down to 49 bp ranging from Ϫ162 to Ϫ114 bp (Fig. 4). The EGF-RE reported by Jiang et al. (16) was not critical in our study. Differences between these two results might be attributable to the different sequences of promoters. We compared the promoter region from Ϫ323 to Ϫ249 bp in this study with that from Ϫ314 to Ϫ248 bp in the previous report by Magnaldo et al. (17) and found that the homology between these two sequences was only 74%. Indeed, the promoter sequence reported previously by Magnaldo et al. (17) does not completely match the chromosome sequence reported by The Human Genome Project. In addition, site-directed mutagenesis at both the AP1 and Sp1 sites in the essential promoter region (pXK-dm (5-1)) almost abolished the luciferase reporter activity of EGF response and correlated well with the effect on pXK-mSp1 (5-2) and pXK-6 ( Fig. 5B). Furthermore, because mutation at the Sp1 site exhibited a more significant inhibition of EGF response than mutation at the AP1 site, the Sp1 site might appear to be the primary driver of EGF-stimulated gene expression of keratin 16. In our study, we also found that the Sp1 binding site was required not only for EGF response but for basal promoter activity (data not shown), which was consistent with the finding of Magnaldo et al. (17). Lastly, the transcrip- tion factors bound to the essential binding elements for gene transcription were identified. By DNA affinity precipitation assay, we found that Sp1 was the major nuclear protein directly bound to the Sp1 site, and c-Jun and c-Fos were directly bound to the AP1 site. Moreover, EGF treatment altered the binding of AP1 proteins to AP1-WT in a time-dependent fashion (Fig. 7B), but there was no change in Sp1 protein binding to Sp1-WT after EGF stimulation (Fig. 7A). Therefore, the transcription factors Sp1 and AP1 bound to each promoter binding element were critical for the stimulatory response of EGF on keratin 16 gene regulation in HaCaT cells.
Furthermore, the regulation of keratin 16 gene expression by Ras activation, most relaying EGF signaling, was also illustrated. The results showed that overexpression of pMMrasDN dose-dependently inhibited EGF-induced keratin 16 promoter activity (Fig. 3B). The promoter region ranging from Ϫ162 to Ϫ114 bp was important for the Ha-ras response as for EGF response (Fig. 4). An apparent inhibition of Ha-ras-stimulated response was observed in pXK-dm (5-1), pXK-mSp1 (5-2), and pXK-6, in the same manner as EGF response (Fig. 5B). Evidence obtained from these data strongly suggested that EGFinduced gene expression of keratin 16 in HaCaT cells was mediated through the Ras signaling pathway.
Another interesting finding from this study is the demonstration that c-Jun and Sp1 play a functional role in EGFinduced epidermal gene expression of keratin 16 through an interaction between these two transcription factors. First, a synergistic induction of promoter activity was observed in pXK-5-1 when both Sp1 and c-Jun were overexpressed (Fig. 10A,  lanes 1-4), indicating that Sp1 might cooperate with c-Jun to activate keratin 16 transcription. Second, overexpression of c-Jun in cells activated the promoter activity of the keratin 16 gene in a similar fashion as EGF and Ha-ras responses. Interestingly, a more significant decrease in the c-Jun-overexpressed response was observed in pXK-mSp1 (5-1), which resulted in a mutation at the Sp1 site residing at Ϫ128 to Ϫ122 bp than in pXK-mAP1 (5-1) (Fig. 9A). Additionally, overexpression of c-Jun in cells dose-dependently increased keratin 16 promoter activity of pXK-5-2 carrying only the Sp1 site (Fig.  9B). Last, Sp1 as well as AP1 proteins were all detected on the Sp1 site by EGF treatment in binding assays (Fig. 7A), and the mutant of Sp1 (Sp1-Mut) blocked the binding of Sp1 and AP1 proteins to the promoter (Fig. 7C), indicating the interaction between Sp1 and AP1 proteins upon EGF treatment. Two pieces of evidence were provided to demonstrate that c-Jun could not interact with the Sp1 site on the promoter directly. First, no supershifted band of c-Jun was observed in DNA gel mobility shift assay (Fig. 8A). Second, c-Jun protein binding to the Sp1 site of the promoter was dependent on the presence of Sp1 protein in Sp1-deficient Drosophila Schneider SL2 cells (Fig. 8B, lanes 4 and 2).
In the mediation of keratin 16 gene activation by EGF, the response was not attributable to the increase in Sp1 biosynthesis, because no change of the binding of Sp1 proteins with the promoter was observed in cells treated with EGF (Fig. 7A). Therefore, induction of Sp1 protein biosynthesis triggered by the transcription factor AP1 in the response of keratin 16 gene activation seems unlikely. Sp1 belongs to a zinc finger family of transcription factors that recognize the GC-rich DNA sequence (29). It has been known to associate directly with the basal transcription machinery via TATA-binding-associated factor dTF II 110 (30). Sp1 has also been shown to cooperate through direct physical interactions with several transcription activators, including the p65 subunit of nuclear factor-B (31), E2F (32), and YY1 (33), which bind to their respective response elements in the regulation of responsive genes. In all of the above cases, the partners on target promoters that contain each binding sequence for the two proteins are usually closely spaced. In our system of the keratin 16 promoter, the synergistic mechanism of transactivation of Sp1 and c-Jun seems to be similar to the above-mentioned examples. On the other hand, we found previously that the direct interaction between Sp1 and c-Jun induced by EGF (34) and phorbol 12-myristate 13-acetate (35) cooperatively activated expression of the 12(S)lipoxygenase gene, and that Sp1 may function as a carrier bringing c-Jun to the promoter, thus activating the transcriptional activity of the 12(S)-lipoxygenase gene. This model of the 12(S)-lipoxygenase gene seems to be different from that of keratin 16, because neither the known AP1-binding sequence in the promoter region responsive to EGF nor the apparent binding between transcription factor AP1 and EGF-responsive promoter DNA was observed in the 12(S)-lipoxygenase gene (22).
A similar requirement for the Sp1 and AP1 consensus sequences for promoter activation was observed in the stimulation of keratin 16 in HaCaT cells treated with EGF, Ha-ras, and c-Jun overexpression. Therefore, induction of c-Jun biosynthesis in cells was an essential step in EGF-induced expression of keratin 16. AP1 is a nuclear transcription complex composed of dimers encoded by the fos and jun families of proto-oncogenes, which modulate transcription by binding to AP1 recognition motifs in the regulatory regions of the target genes (36). c-Jun contains a COOH-terminal basic region leucine zipper DNA-binding domain and an NH 2 -terminal transactivation domain. The transactivation activity of c-Jun can be regulated through phosphorylation of clusters of serine/threonine residues in the NH 2 terminus (37,38). Two major serine phosphoacceptor sites have been identified within the NH 2 -terminal domain at Ser-63 and Ser-73 that are activated through extracellular signal-regulated kinase and c-Jun NH 2 -terminal kinase signaling (39,40). The COOH-terminal domain at Thr-231 and Ser-249 has been found to be constitutively phosphorylated by casein kinase II (41).
Acetylation of proteins is a common principle to modify their biological activity and thus may alter protein-protein interaction, DNA recognition, and protein stability. Several experiments have suggested the importance of histone acetylation in the transcription of a variety of genes (42). p300/CBP possesses intrinsic histone acetyltransferase activities, works as a transcriptional coactivator, and cooperates with multiple transcriptional factors such as MyoD, p53, CREB, c-Jun, and Sp1 (43)(44)(45)(46)(47). In this study, the interplay of Sp1, c-Jun, and p300/CBP for the transcriptional regulation of the human keratin 16 gene in HaCaT cells was also elucidated. The coactivators p300/CBP, supposed to interact with Sp1 and c-Jun, activated the promoter activity of keratin 16 (Fig. 10B); moreover, E1A blocked EGF-induced keratin 16 promoter activity (Fig. 10C). The NH 2terminal transactivation domain of c-Jun interacts with the KIX domain of p300/CBP, whereas the COOH-terminal leucine zipper domain of ATF-2 interacts with the C/H2 domain of CBP (48). It has also been reported that Sp1 and p300 might interact indirectly for the activation of the p21 promoter (47). Analysis of whether the post-translational modification of these nuclear factors is involved in this system will clarify the spectrum of regulatory networks in the transcriptional control of keratin 16 upon EGF treatment.