The POU Domain Factor Skin-1a Represses the Keratin 14 Promoter Independent of DNA Binding A POSSIBLE ROLE FOR INTERACTIONS BETWEEN Skn-1a AND CREB-BINDING PROTEIN/p300*

The genes encoding keratin 5 and 14 are highly expressed in the basal cell layer keratinocytes of the epidermis, but both genes are silenced when keratinocytes move into the suprabasal compartment. The POU homeodomain factors Skn-1a and Tst-1, which are expressed in epidermis, may play a role in the suprabasal repression of the keratin 5 and 14 genes because keratin 14 mRNA expression persists in suprabasal cells in Skn-1/ Tst-1 double knockout mice. In transfection experiments, both Skn-1a and Tst-1 repress the keratin 14 promoter, with the POU domain being sufficient for repression. The region of the keratin 14 gene sufficient and required for repression by Skn-1a is a 100-base pair sequence lacking POU-binding sites adjacent to the transcription start site. DNA-binding defective mutants of Skn-1a and Tst-1 are as effective at mediating repression as the wild type proteins, suggesting that protein-protein interactions rather than direct DNA binding are important for repression. We also show that CREB-binding protein (CBP)/p300 co-activators are strong activators of

The mammalian epidermis is a stratified squamous epithelial tissue characterized by differential expression of keratin genes in its different compartments. The basal cell layer, which is in contact with the basal lamina and contains proliferative cells, expresses keratin (K) 1 5 and 14 at high levels. In fact, these proteins constitute up to 30% of the protein content of keratinocytes in this layer (1). As keratinocytes move out of the basal cell layer into the suprabasal compartment, expression of the K5 and K14 genes is extinguished, whereas the K1 and K10 genes are activated (1). The cells moving progressively through the different layers of the suprabasal epidermis: stratum spinosum, granulosum and corneum, express other distinct keratinocyte markers, including loricrin and filaggrin, which are expressed in the most differentiated cells of the epidermis just prior to their death and the formation of the stratum corneum (2). Keratinocytes moving out of the basal cell layer into the suprabasal compartment also stop proliferating, suggesting a connection between differential keratin expression and the switch regulating proliferation/differentiation decisions in the epidermis (2,3). Therefore, greater understanding of the molecular mechanisms regulating differential keratin expression may provide insight into the control of epidermal development and homeostasis.
One of the genes highly suitable for such analysis is the K14 gene, whose regulation has been extensively studied both in vitro in cell culture models and in vivo in mice. Fuchs and co-workers (4 -7) have shown that ϳ2.2 kb of the K14 promoter is sufficient to drive expression of reporter genes in transgenic mice in a pattern that mimics the restricted expression of the endogenous gene in basal cells of epidermis and internal epithelia. Detailed mapping of the 2200-bp K14 promoter has suggested that the region from Ϫ2000 to Ϫ1300 linked to a minimal promoter may be sufficient to correctly direct expression into the epidermis of transgenic mice (5). This 700-bp enhancer region of the K14 gene contains binding sites for Ets, AP-1, AP-2, and GATA factors (5). Studies in keratinocyte cell lines indicate that simultaneous mutations of the AP-1, Ets, and AP-2 sites may have a particularly deleterious effect on the expression directed by the K14 enhancer (5). There is also evidence that the promoter-proximal region, as well as the enhancer, is important for cell-specific expression of the K14 gene in vivo. Thus, while replacing the K14 minimal promoter with the heterologous TK promoter allowed expression in the epidermis of transgenic mice, the normal down-regulation of K14 in suprabasal cells of the epidermis was disrupted (5). These experiments suggest that, although dispensable for directing selective expression to epidermis, the minimal K14 promoter is important for establishing the precise expression pattern involving the physiological repression of the K14 gene in keratinocytes of the suprabasal compartment.
POU domain factors, a family of transcription factors with 16 known mammalian members as well as representatives found in other species, including Drosophila, Xenopus, and zebrafish, are characterized by a conserved DNA-binding domain referred to as the POU domain (8,9). The bipartite POU domain con-tains a variant homeodomain joined at its N terminus via a nonconserved linker to another conserved domain, the POUspecific domain (10). Both the POU homeodomain and the POU-specific domain make direct DNA contacts and are required for efficient DNA binding (11). In addition, POU domain factors contain N-and C-terminal sequences that are responsible for transcriptional activation and/or repression. At least three POU domain factors, Oct-1, Skn-1a, and Tst-1 (Oct-6), are highly expressed in normal mammalian epidermis (12)(13)(14)(15). Oct-1, which is ubiquitously expressed, is found in both proliferating and differentiated epidermal keratinocytes, whereas Skn-1a and Tst-1 proteins are primarily expressed in differentiated, postmitotic cells (13). Skn-1a is restricted to the epidermis, but Tst-1 is found in other epithelia such as mouth, esophagus, and vagina and is highly expressed in the nervous system, including glial cells, where it is responsible for regulating myelination (12,(15)(16)(17). The location of Skn-1a in the suprabasal compartment suggested an involvement in the activation of the K1 and K10 genes, and in vitro studies in cell lines have shown that Skn-1a can activate the K10 promoter (12,18). However, Skn-1 gene-deleted mice express normal levels of K1 and K10, implying that if Skn-1a is involved in K10 expression in vivo, redundant mechanisms prevent an abnormality in Skn-1 gene-deleted mice.
In contrast to normal expression of the K1 and K10 genes, alterations to K14 gene expression were observed in the migrating wound front of Skn-1 (Ϫ/Ϫ) mice and in transplanted skin from Skn-1/Tst-1 double mutant mice (13). During wound healing, a thickened front of epithelial cells migrates toward the center of the wound, ultimately closing it (19). Gene expression studies indicate that there are at least two loosely defined compartments found in this front. In the suprabasal cells of the wound front, the Skn-1 gene repressed along with other markers for normal suprabasal cells such as K1, K10, loricrin, and filaggrin (13). However, cells in this compartment also express genes not normally found in suprabasal cells including K5, K14, and the small proline-rich protein Spr-1 (13). Thus, in wound healing, down-regulation of Skn-1a in the migrating wound front correlates with the "ectopic" activation of the K14 gene in this location. Further, in Skn-1 (Ϫ/Ϫ) mice the upregulation of K14 in the wound front is enhanced, supporting the notion that Skn-1a might play a role in repression of the K14 gene (13).
Further evidence for the role of Skn-1a in repression of the K14 gene comes from studies in which epidermis from Skn-1/ Tst-1 double mutant mice was transplanted onto nude mice (13). In these studies, K14 mRNA expression persisted in the suprabasal layer, whereas in mice deleted for either gene alone it was normally down-regulated. This suggests that in vivo Skn-1 and Tst-1 repress the K14 gene in suprabasal cells in a redundant manner.
Skn-1a has previously been shown to function as a transactivator of keratin-specific genes such as K10 and HPV1A in transient transfection assays (12,14,18,20). As suggested by the in vivo studies, the experiments presented here demonstrate that Skn-1a is capable of acting as a repressor on the K14 promoter. However, in contrast to its role in activation of the K10 promoter, this effect is independent of DNA binding. The POU domain, which is normally responsible for DNA binding, is the primary mediator of the repression. The repression maps to the proximal promoter region of the K14 gene. This is the same region that mediates activation by the co-activators CREB-binding protein (CBP) and p300, both of which can rescue the Skn-1a-mediated repression. Co-immunoprecipitation studies also show that the POU domain of Skn-1a can interact directly with CBP. These data suggest the possibility that POU domain repression of the K14 promoter may be due to interference with the function of co-activators such as CBP and p300.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfections-Normal human epidermal keratinocytes (NHEK) were obtained from Clonetics and cultured in serum-free defined medium (keratinocyte growth medium) from the same company. For transfections, the keratinocytes were cultured in 6-well dishes and transfected at 50 -80% confluency. Transfections were performed with N-[1-(2,3-dioleoyloxy)-propyl]-N,N,N-trimethylammonium methylsulfate liposomal transfection reagent (Roche Molecular Biochemicals) according to the manufacturer's recommendations. In a standard reporter assay, we transfected 1.25 g of reporter plasmid/ well and the same amount of expression plasmid. The cells were harvested, and luciferase activity was measured 24 h after transfections. The keratinocyte cell lines HaCat and Scc-25 were obtained from American Type Culture Collection and cultured in Dulbecco's modified Eagle's medium containing 10% fetal calf serum. These cell lines were also transfected with N-[1-(2,3-dioleoyloxy)-propyl]-N,N,N-trimethylammonium methylsulfate liposomal transfection reagent as described for NHEK. HeLa, CV-1, and 293 cells were cultured and transfected with the calcium precipitation method as described previously (13). Plasmids were transfected into duplicate or triplicate wells. The results are expressed as the means and S.E. (or range of a duplicate determination). Transfection efficiency was monitored with a reporter plasmid expressing ␤-galactosidase under control of the ␤-actin promoter.
DNA Construction-The plasmids were created using standard recombinant techniques. Reporter plasmids were based on the luciferasecontaining pGL2-Basic vector (Promega). The K14 luciferase plasmid was created by inserting a 2.2-kb EcoRI/BamHI K14 promoter fragment from the vector pG3ZK14 Cassette (a gift from Dr. Elaine Fuchs) upstream of the luciferase gene in pGL2-Basic. 5Ј deletions were created using convenient restriction sites, and these constructs were named based on the distance from the middle of the TATA box of the promoter.
Immunostaining of Epidermis-HaCat and 293 cells were fixed in 4% paraformaldehyde for 10 min ϳ24 h after transfection. Expression from the pCDNA HASkn-1a N244A plasmid was detected with a polyclonal anti-HA antiserum (BAbCO), and expression of CMV Tst-1 W383C/ F384S was detected with a polyclonal Tst-1 antiserum described previously (22). Detection was with peroxidase staining.
Gel Mobility Shift Assays-Gel mobility shift assays were performed as described previously (12), using in vitro translated wild type Skn-1a and Skn-1a N244A from pCDNA vectors using T7 polymerase. The sequence of the octamer-heptamer binding site was described previously (13).
Protein-Protein Interaction Assays: Co-immunoprecipitations-HEK293T cells grown on 15-cm dishes were transfected with pCDNA HASkn-1A POU plasmid, using the calcium precipitation method (13). After 36 h, the cells were scraped into ice-cold phosphate-buffered saline and pelleted, followed by quick freezing at Ϫ80°C. The cells were then allowed to thaw on ice for 20 min in 1 ml of lyses buffer containing 0.5% Nonidet P-40, 20 mM Tris-HCl, pH 8.0, 300 mM NaCl, 1 mM EDTA, and protease inhibitors. After spinning at top speed in a microcentrifuge to remove cellular debris, the supernatant was transferred to a new tube and precleared for 1 h with 40 l of protein A-agarose. The appropriate antiserum was added, followed by incubation for 2 h at 4°C. We used purified anti-CBP IgG (Santa Cruz) and, as a control, the same amount of nonspecific IgG. Immune complexes were precipitated by adding 40 l of protein A/G-agarose and incubating for 1 h at 4°C. The immunoprecipitated material was washed five times with 1 ml of lyses buffer, and the pellet was resuspended in SDS sample buffer. Following SDS-polyacrylamide gel electrophoresis the gel was transferred to nitrocellulose and blotted with antiHA (Babco).
GST Pull Down Assays-GST fusion proteins were expressed in Escherichia coli using standard procedures and stored on glutathione agarose beads (25). 35 S-Labeled in vitro translated proteins were incubated with the GST proteins on a rotating wheel at 4°C for 30 min in a buffer containing 20 mM Hepes, pH 7.9, 100 mM NaCl, 1 mM EDTA, 4 mM MgCl 2 , 1 mM dithiothreitol, 0.02% Nonidet P-40, 10% glycerol, albumin 1 mg/ml, and 0.5 mM phenylmethylsulfonyl fluoride. After washing five times in the same buffer, the glutathione-agarose beads were resuspended in SDS sample buffer, boiled, and analyzed on 10% SDS-polyacrylamide gels (26).

Specific Repression of the K14 Promoter by POU Domain
Factors-To test whether the POU domain factor Skn-1a could repress the K14 promoter in a cell autonomous manner, we co-transfected into a variety of cells a luciferase reporter construct containing 2.2 kb of the proximal part of the human K14 promoter (Ϫ2200 K14 luciferase) and expression plasmids encoding Skn-1a. We selected this region of the K14 promoter because it faithfully replicates the expression pattern of the endogenous K14 gene in transgenic mice (4 -7). Transfection of Skn-1a resulted in greater than 10-fold repression of the 2.2-kb K14 promoter in NHEK and in two immortalized squamous keratinocyte cell lines, HaCat and Scc-25 (Fig. 1A, left panel). Skn-1a-mediated repression of the K14 promoter was also observed in three cell lines, HeLa, CV-1, and 293, which do not normally express the K14 gene at a high level, suggesting that the mechanisms underlying the repression are not keratinocyte-specific (Fig. 1A, left panel). It should be noted that more robust repression is observed in NHEK and keratinocyte cell lines than in the heterologous cell lines. Consistent with a previous study (15) using the SCC-13 keratinocyte cell line, Tst-1 also represses the K14 reporter plasmid (Fig. 1A, left panel) in NHEK, HeLa, and CV-1 cells. Pit-1, a POU domain factor not normally expressed in keratinocytes, is also capable of repressing the K14 promoter (data not shown), suggesting that the ability to repress the K14 promoter may be a common feature of POU domain factors. This repression is specific because expression plasmids encoding the transcriptional co-ac-tivators CBP/p300 activated the K14 promoter (see Fig. 6). Skn-1a and Tst-1 (15) will also repress transcription of the other major basal cell keratin, K5, as shown with a luciferase reporter construct containing 0.8 kb of the bovine K5 promoter. This repression was particularly high in NHEK (Fig. 1A, right  panel).
Repression of the K14 promoter by Skn-1a is dose-dependent (Fig. 1B, left panel) and highly specific because Skn-1a does not repress the K10 promoter (Fig. 1B, right panel), the CMV enhancer/promoter (Fig. 1C), the TK promoter (see Fig. 4), or the minimal prolactin promoter (see Fig. 4). In fact, Skn-1a strongly activates the K10 promoter in several different heterologous cell lines, including HeLa and CV-1 (12). The lack of robust activation of the K10 promoter in normal human epidermal keratinocytes may be due to endogenous levels of Skn-1a increasing the basal activity of the promoter in these cells. Collectively, these data suggest that both Skn-1a and Tst-1 are capable of repressing the genes encoding the two major keratins of the basal cell layer in a cell autonomous manner. Further, although the repressive effect of these factors is promoter-specific, it is observed in all tested cell lines, indicating that the repression does not require keratinocyte-specific components.
The POU Domains of Skn-1a and Tst-1 Mediate Repression of the K14 Promoter-The ability of Skn-1a and Tst-1 to activate transcription has been well described with definition of distinct transactivation domains in the N and C termini (12,14,18,20). Yet, under certain conditions, both Tst-1 (15,27) and Skn-1a can repress transcription, including that of the keratinocyte-specific genes involucrin (28) and profilaggrin (29). Therefore, we were interested in understanding further the bifunctional transcriptional role of POU domain factors in regulation of gene expression, focusing mainly on the effect of Skn-1a on the well characterized K14 promoter.
First, we defined the region of the Skn-1a protein that was responsible for the repression of the K14 promoter. Plasmids expressing distinct domains of Skn-1a were co-transfected with Ϫ2200 K14 luciferase into NHEK. Skn-1a lacking the C terminus was able to repress the K14 promoter to the same extent as full-length Skn-1a, and removing the N terminus only slightly decreased the ability to repress. The isolated POU domain of Skn-1a was still able to repress about 3-fold ( Fig. 2A). These A, K14 and K5 luciferase reporter plasmids were co-transfected with CMV Skn-1a and CMV Tst-1 into the indicated cell lines. The activity of the keratin luciferase reporter genes co-transfected with an empty CMV vector was arbitrarily set at 100% in each cell type. Expression of both keratin reporter genes was highest in NHEK and keratinocyte cell lines but easily detected in heterologous cell lines. Both Skn-1a and Tst-1 repressed these promoters in all cell lines tested. B, in NHEK, CMV Skn-1a represses the K14 promoter in a dose-dependent fashion (left panel), whereas the same amount of Skn-1a has little effect on the K10 promoter (right panel). C, in NHEK, Skn-1a has no effect on transcription from the CMV promoter, indicating that the repression of the K14 promoter by Skn-1a is specific.

Skn-1a Repression of the Keratin 14 Promoter
data suggest that the POU domain of Skn-1a is sufficient for repression of the K14 promoter but that full repression requires the N terminus, which is known to contain strong transactivation domains (14). Interestingly, Skn-1i, a form of Skn-1 that contains an inhibitory domain in the N terminus that blocks transactivation of the K10 promoter by Skn-1 (12), is less efficient in repressing the K14 promoter (data not shown). These data are consistent with previous observations that Skn-1i was less efficient than Skn-1a in repressing the involucrin (28) and profilaggrin (29) promoters. Thus, the same domain that interferes with transactivation is also capable of interfering with repression mediated by Skn-1a. Similar mapping results were obtained with Tst-1; full repression required N-and C-terminal sequences, but the Tst-1 POU domain alone was sufficient for 10-fold repression of the K14 promoter (Fig.  2B). From these data we conclude that although the POU domains of Skn-1a and Tst-1 are sufficient for repression of the K14 promoter, regions outside the POU domain may contribute to the full repressive effect.
The Proximal Promoter Region of K14 Mediates Repression by Skn-1a and Tst-1-Having mapped the repressive effect of Skn-1a to the POU domain, we next defined the regions of the K14 promoter through which Skn-1a can act to repress gene expression. A series of 5Ј deletions were made to the K14 promoter and linked to luciferase (Fig. 3). When transfected into NHEK the basal activity of Ϫ1008 K14 luciferase and Ϫ420 K14 luciferase was similar to that of the Ϫ2200 K14 luciferase reporter construct, but basal activity fell by ϳ50 and 95% when the promoter was truncated down to Ϫ274 and Ϫ70, respectively (data not shown). The K14 5Ј deletion reporter plasmids were co-transfected with Skn-1a and Tst-1 expression plasmids to map the region of the K14 promoter responsible for repression. All tested reporter plasmids in this experiment, including the Ϫ70 K14 luciferase construct, were repressed by both Skn-1a and Tst-1, indicating that the proximal 70 bp surrounding the promoter of the K14 gene can mediate repression by POU domain factors (Fig. 3).
Because the mapping data indicated that the K14 minimal promoter can mediate repression by POU domain factors, we tested whether replacing the native K14 promoter with heterologous promoters from the prolactin (P36) and TK genes could obliterate repression (Fig. 4). Although the Ϫ2200 K14 luciferase construct was repressed ϳ5-fold in these experiments, the K14-TK luciferase hybrid reporter construct was repressed less than 2-fold (Fig. 4). Further, the K14-P36 luciferase hybrid reporter was activated rather than repressed by Skn-1a (Fig.  4). Although the mechanism for the activation of the K14-P36 luciferase reporter plasmid is not known, these experiments clearly demonstrate the promoter specificity of the repression effect of POU domain factors. Taken together, our experiments indicate that the minimal promoter of the K14 gene is both required and sufficient to mediate POU domain factor repression.
POU Domain Factors Repress the K14 Promoter Independent of Direct DNA Binding-As previously reported, the K14 promoter region contains no obvious octamer-binding sites (15), suggesting that the repressive effect of Skn-1a and Tst-1 on the K14 promoter may be independent of direct DNA binding. To test this hypothesis directly, we mutated amino acid 244, located in the DNA-contacting third helix of the the POU homeodomain of Skn-1a, from an aspargine to an alanine. Selection of this amino acid was based on analyses of the crystal structure of Oct-1 bound to DNA. The analogous residue in Oct-1 makes direct base contacts with octamer DNA-binding sites (11), and so this mutation is likely to interfere with DNA binding of Skn-1a. When expressed in vitro, the Skn-1aN244A mutant protein was incapable of binding to an octamer site to which the comparably expressed wild type protein bound avidly (Fig. 5A). However, this mutant protein was just as effective as the wild type protein in repressing expression of Ϫ2200 K14 luciferase in transfected NHEK cells (Fig. 5B). Similarly, an expression plasmid encoding a mutated form of Tst-1, Tst-1 W383C/F384S, which was previously shown to lack DNA binding activity (22), was equally effective in repressing the K14 promoter as wild type Tst-1 (Fig. 5B). Because mutations in the POU domain of Tst-1 have been found to interfere with nuclear translocation (22), we studied the location of mutated proteins in transfected cells by immunohistochemistry. Both wild type and mutated Skn-1a and Tst-1 proteins were localized to the nucleus, indicating that no abnormality in nuclear transport was introduced with these mutations (Fig. 5C). Collectively, these data indicate that POU domain factors repress the K14 promoter independent of DNA binding, suggesting that protein-protein interaction mechanisms rather than direct protein-DNA interaction are involved in POU factor-mediated repression of the K14 gene.
The Co-activators CBP and p300 Activate the K14 Promoter-Because our data suggested that POU domain factors inhibited transcription of the K14 gene via protein-protein interactions, we considered the possibility that POU domain factors might interfere with the activity of transcriptional coregulators involved in stimulating expression of the K14 gene. The co-activators CBP and the highly related p300 both possess histone acetylase activity and are important in multiple tran- The indicated reporter constructs were transfected with an empty CMV plasmid as a control (activity arbitrarily set at 100%) and two different amounts of CMV Skn-1a. When the native K14 promoter is replaced with the TK promoter, Skn-1a can only repress the promoter by 50%, and when replaced by the minimal prolactin promoter, a mild activation is observed, indicating that the native K14 promoter is required for full repression by Skn-1a.
scriptional complexes on a variety of genes (30). Although CBP/ p300 co-activators have not been implicated in activation of K14 gene expression, we considered them potential candidates because CBP has been shown to act on numerous different genes. Immunohistochemical analyses of mouse epidermis using a specific antibody shows that CBP expression overlaps with K14 protein but is more widespread (Fig. 6, A-D). CBP is present in all layers of the interfollicular epidermis (Fig. 6A) as well as in matrix cells and the outer root sheath of hair follicles (Fig. 6, B-D) but is not detected in the dermal papilla, inner root sheath, or cuticular cells of the hair follicle (Fig. 6, B-D). Because expression of CBP is robust in the basal cell layer of the interfollicular epidermis and in the outer root sheath of hair follicles, locations where K14 is highly expressed, the distribution of CBP is compatible with its involvement in K14 gene expression in vivo.
To test the potential role of CBP/p300 in regulation of K14 gene expression, we co-transfected the Ϫ2200 K14 luciferase reporter plasmids with vectors expressing CBP and p300 into NHEK (Fig. 6E) and heterologous cell lines (data not shown). These studies showed that both factors could activate the Ϫ2200 K14 reporter plasmid and that the minimal promoter region (Ϫ70 K14 luciferase) was sufficient to mediate activation (Fig. 6E). Further support for the role of CBP/p300 in K14 gene expression came from experiments with vectors expressing the adenoviral oncoprotein E1A, which is known to interact with and block the activity of CBP/p300 (30). In NHEK, wild type E1A strongly inhibited the Ϫ2200 K14 luciferase plasmid, whereas an E1A lacking the first 36 amino acids, which are critical for interaction with CBP/p300 (30), had little effect on expression mediated by the K14 promoter (Fig. 6F). Collectively, these data suggest that CBP/p300 co-activators are important for activation of the K14 gene in normal human keratinocytes and that the minimal promoter region of K14 is sufficient for mediating a stimulatory role of CBP.
CBP Interacts Functionally and Physically with the POU Domain of Skn-1a-Studies indicate that CBP/p300 may be present in limiting amounts in cells (30 -32), suggesting the possibility that POU domain factors could repress the K14 promoter by sequestering CBP/p300, thus making it unavailable as a co-activator. If this model is correct, then CBP/p300 may be able to overcome and rescue the Skn-1a-mediated repression of K14. To test this possibility, we co-transfected a fixed amount of Skn-1a expression plasmid with increasing amounts of CBP expression plasmid and measured the activity of the Ϫ2200 K14 luciferase reporter plasmid. 1-2.5 g of CMV CBP expression plasmid overcame the 5-fold repression observed with 0.5 mg of CMV Skn-1a expression plasmid (Fig.  7A). Expression of p300 also partially reduced the repression by Skn-1a (data not shown). Furthermore, in the presence of E1A, Skn-1a could not further repress K14 activity, which is consistent with the idea that POU factor-mediated repression of K14 depends on functional CBP (data not shown).
The transient transfection experiments demonstrate functional antagonistic interactions between Skn-1a and CBP in regulation of K14 gene expression. However, a sequestration model would only be plausible if Skn-1a can directly interact with CBP. To test this possibility, we transfected an expression plasmid encoding the HA-tagged POU domain of Skn-1a into 293 cells and isolated nuclear extracts. These extracts were then subjected to immunoprecipitation with CBP antisera or control IgG, followed by detection with anti-HA on Western blots. Experiments demonstrate that the POU domain of Skn-1a specifically co-immunoprecipitates with CBP (Fig. 7B), indicating that CBP and Skn-1a can interact in a cellular environment and further that the POU domain of Skn-1a is sufficient for interaction with CBP. To map the domains of CBP responsible for interaction with Skn-1a, we expressed various regions of CBP as fusion proteins with GST and incubated these with 35 S-labeled in vitro translated Skn-1a protein (Fig.  7C). These experiments show that Skn-1a interacts strongly with two regions of CBP located between amino acids 1068 -1459 and amino acids 1458 -1890. In addition, we detected lower affinity interactions with CBP regions 1-450 and 1890 -2400, both of which were shown to interact with the POU factor Pit-1 (24). The N244A mutation in the Skn-1a POU domain has no effect on the interaction with CBP (Fig. 7D). In summary, these experiments show that the POU domain of Skn-1a interacts directly with CBP. DISCUSSION In the mammalian epidermis the K14 gene is highly expressed in the basal cell layer. But as keratinocytes move into the suprabasal compartment, K14 mRNA expression is completely extinguished (1,3). Studies have demonstrated that 2200 bp of the proximal promoter region of the K14 gene is sufficient to correctly target expression of reporter genes to epithelial tissues of transgenic mice (4 -7) and that no expres- 5. DNA binding-independent repression of the K14 promoter. A, the top panel shows 35 S-labeled Skn-1a proteins, N244A mutant, and wild type, which were used to test for binding in gel mobility shift assays. When analyzed in the gel mobility shift assay with a 32 P-labeled octamer DNA sites, the wild type Skn-1a protein binds avidly, whereas the N244A mutant of Skn-1a is incapable of binding to the DNA site. B, the Ϫ2200 K14 luciferase reporter plasmid was co-transfected with the indicated expression plasmids. The W383C/F384S mutation of Tst-1 obliterates its binding to DNA (22). C, immunohistochemical analyses of cells transfected with plasmids expressing wild type and mutated Skn-1a and Tst-1 proteins, showing that all proteins localize to the nucleus of transfected cells. Collectively, these data indicate that DNA binding is not required for repression of the K14 promoter by POU domain factors.

Skn-1a Repression of the Keratin 14 Promoter
sion is detected from transgenes containing only 450 bp of proximal K14 sequence (7). A 700-bp enhancer, located between Ϫ2000 and Ϫ1300, is particularly important for tissuespecific expression; this region alone linked to a heterologous promoter is sufficient to drive expression of a reporter gene in the epidermis of mice (5). A cluster of binding sites for transactivators, including AP-1, AP-2, Ets, and GATA factors, can be found in a 125-bp fragment within the enhancer (5). Mutations of these binding sites lead to decreased activity of the enhancer in transfected keratinocytes in vitro (5).
A simple explanation for the layer-specific regulation of the K14 gene is that transactivators that are normally important for driving expression of the K14 gene in the basal cell layer are absent in the suprabasal compartment. However, an alternative model is possible. According to this model, the transactivation program persists as keratinocytes move into the suprabasal compartment. Transcriptional repressors, limited to the suprabasal compartment, then override the positively acting factors and silence K14 gene expression. There is precedence for such a model in the epidermis because studies on the regulation of the loricrin gene have implicated negative gene regulatory elements as important for restricting expression of this gene to the granular layer in mouse epidermis (33,34).
Several lines of evidence indicate that transcriptional repression may be an important mechanism in developmental regulation of the K14 gene. First, the transactivating factors thought to be important for stimulating expression of the K14 gene, including AP-1 (35) and Ets factors (36 -38), are prominently expressed in differentiating cells of the suprabasal compartment. Second, and more importantly, when the K14 enhancer is linked to the heterologous TK promoter, aberrant reporter gene expression is found in suprabasal cells of the mouse epidermis (5). This finding suggests that cis-acting sequences within the proximal promoter region are important for mediating repression of the K14 gene as keratinocytes move into the suprabasal compartment. The experiments presented in this paper suggest that transcriptional repression, a commonly employed strategy in developmental control (39 -42), could be important for spatial-specific expression of the K14 gene within the epidermis.
Some insight into possible transacting factors responsible for the suprabasal repression of the K14 promoter came from our . E, the Ϫ2200 K14 luciferase (left panel) and Ϫ70 K14 luciferase (right panel) reporter plasmids were co-transfected with vectors expressing p300 and CBP. Both p300 and CBP can activate the K14 promoter, and the minimal K14 promoter is capable of mediating activation. F, the Ϫ2200 K14 luciferase plasmid was co-transfected with the indicated expression plasmids. Wild type E1A inhibits expression of the K14 promoter, whereas an N-terminal deletion mutant of E1A (⌬36 E1A) that does not interact with CBP/p300 has little effect of K14 expression, consistent with an activation role for CBP/p300 in K14 gene expression.
previous studies of POU domain factors expressed in epidermis. Simultaneous inactivation of the POU domain genes Skn-1a/i and Tst-1 in mice leads to derepression of the K14 gene in the suprabasal compartment of transplanted epidermis (13), suggesting that POU domain factors may play a biologically important role in transcriptional repression of the K14 promoter. The experiments presented in this manuscript suggest that Skn-1a represses the K14 promoter in a cell autonomous fashion similar to that demonstrated for Tst-1/Oct-6, another POU domain factor highly expressed in epidermis (15). The POU domain seems to be sufficient for repression, and N-and C-terminal sequences add little to this function. This repression function is distinct from the transactivation function of both Skn-1a and Tst-1 where the POU domain alone has little effect, but N-and C-terminal sequences are crucial for transactivation (14,18). Although the POU domain mediates DNA-binding, it also plays important roles in protein-protein interactions (8), and in the case of K14 repression, the latter mechanism is most likely crucial. This is supported by the observation that the K14 promoter contains no obvious POU binding sites (15) and that DNA-binding inactivating mutations in the POU domains of Skn-1a and Tst-1 have no effect on repression (Fig. 5).
Because our studies suggested that protein-protein interactions might be important for repression of the K14 promoter, we sought to find positively acting co-activators that might play a role in K14 gene expression and discovered that CBP and p300 are important activators of K14 gene expression. Apparently, these factors act via the promoter region of the K14 gene, the same region that is required for transcriptional repression of the K14 promoter. CBP and p300 are closely related proteins that act as transcriptional co-activators (30). Both proteins contain highly conserved regions and are thought to carry out similar functions, one of which is histone acetyl transferase activity consistent with a role in chromatin remodeling. CBP was initially described as a co-activator for the CREB (CREbinding protein), but it is now known that both CBP and p300 interact with multiple transcriptional activators in various coactivator complexes, including nuclear receptors, p53, and SMADs (30). In addition, CBP has been implicated as a coactivator for the POU homeodomain factor Pit-1 (24).
Although CBP and p300 are generally thought to be recruited to enhancer regions by DNA-binding transcription factors, there is ample evidence that CBP/p300 associate with and have roles in the core transcription machinery (43). CBP and p300 have been shown to complex with the TATA-binding protein TBP (44) and with TFIIB (45). The human RNA polymerase II complex contains the histone acetyl transferases CBP and pCAF in addition to the Swi-Snf complex and Srb proteins (46,47). Given these findings, combined with our results that the core promoter of K14 can mediate repression by Skn-1a and that CBP overexpression can rescue this repression, we suggest that Skn-1a mediates repression at the core by antagonizing CBP/p300 function.
The POU domain factor Pit-1 has previously been shown to recruit CBP (24,48), and here we demonstrate that Skn-1a can also interact with CBP via the POU domain. One of the possible mechanisms by which Skn-1a can repress the K14 promoter is by partitioning CBP away from the core K14 promoter. There are many lines of evidence in support of this model. First, gene knockouts of CBP and p300 have haploinsufficiency, demonstrating that there are limiting amounts of these factors within a cell (32). Second, repression independent of DNA binding has been demonstrated for several classes of transcription factors and in some cases the repression can be rescued by overexpression of CBP/p300. Examples include repression of AP-1 (31,49,50), NF-B (51,52), and CREB by nuclear receptors in a manner independent of nuclear receptor DNA binding, and in each case the repression can by rescued by overexpressing CBP (49). In addition, other classes of transcription factors display a similar form of repression. For instance Janus kinase/STAT can antagonize AP1 (53), and Pit-1 can antagonize nuclear receptors (24). Thus, there are currently numerous examples of DNA binding independent repression mediated by interference with the cofactors CBP/p300 (49).
The present study together with previous investigations suggests that Skn-1a may be a bifunctional transcription factor in its regulation of keratin gene expression. On certain promoters containing octamer-like DNA sites, such as the K10 and HPV1A promoters, Skn-1a binds to the gene regulatory region and acts as a transcriptional activator (12,14,18,20). On other promoters lacking octamer sites, such as the K14 promoter, Skn-1a may act as a transcriptional repressor without binding to the gene regulatory region. Studies of the POU domain factor Pit-1 have demonstrated that this protein also performs both FIG. 7. Functional and physical interactions between Skn-1a and CBP. A, the Ϫ2200 K14 luciferase plasmid was co-transfected with CMV Skn-1a, without or with increasing amounts of an expression vector for CBP. CBP is capable of overcoming the repression observed by Skn1a. B, a CMV plasmid expressing the HA-tagged POU domain of Skn-1a was transfected into 293 cells. After isolation of protein extracts, immunoprecipitation (IP) was performed with control IgG (lane 2) and CBP antibody (lane 3). Following SDS-polyacrylamide gel electrophoresis and transfer to nitrocellulose, the blot was probed with HA antibody, demonstrating that CBP antibody can specifically immunoprecipitate the Skn-1a POU domain. C, the parts of CBP were fused to GST and expressed in bacteria. 35 S-Labeled in vitro translated Skn-1a protein was incubated with the indicated GST fusion proteins, and interactions were detected by precipitation with glutathione-agarose beads followed by SDS-polyacrylamide gel electrophoresis. Skn-1a interacts strongly with two regions of CBP. D, the indicated GST-CBP fusion proteins were incubated with 35 S-labeled, in vitro translated wild type Skn-1a (lanes 2 and 3) and Skn-1aN244A (lanes 4 and 5). The Skn-1a interaction with CBP is not affected by the N244A mutation. activator and repressor functions. Pit-1 binds to and activates the thyroid-stimulating hormone ␤ subunit gene promoter (54,55) but is also capable of inhibiting activation of the SF-1 gene, which lacks Pit-1-binding sites, by DNA-independent mechanisms (54). However, although Skn-1a and Pit-1 may exhibit similar features in this respect, the underlying mechanisms are quite different. In the case of Pit-1, the inhibition of SF-1 gene expression is mediated by interaction of the POU domain with the DNA-binding transcription factor GATA-2. In contrast, Skn-1a seems to repress the K14 promoter by interacting with a co-activator protein that does not bind DNA.
In summary, we propose the following model for the regulation of K14 gene expression in the epidermis. The 700-bp enhancer located between Ϫ2000 and Ϫ1300 is responsible for directing expression to the epidermis, and the proximal promoter region is important for refinement of K14 expression within this tissue to allow normal repression of the K14 gene in suprabasal cells (5). We hypothesize that the co-activators CBP and p300, acting, at least in part, on the proximal promoter region, are important for maintaining K14 expression in the basal cell compartment. In the suprabasal cell compartment POU domain factors can interfere with the function of CBP/ p300, possibly through direct protein-protein interactions and thus silence the K14 gene in suprabasal cells. However, it is clear that neither Skn-1a or Tst-1 alone nor the combination of the two can function as a master switch to repress the K14 gene as cells move from the basal cell layer into the suprabasal compartment. It is more likely that there are additional mechanisms involved in K14 repression. For example, although we did observe ectopic K14 gene expression in skin transplanted from Skn-1/Tst-1 double knockout mice, K14 gene expression was normal in the skin of newborn Skn-1/Tst-1 double knockout mice. These data suggest either that the ectopic expression of the K14 gene occurred later during postnatal development or that the added stress of transplantation contributed to the occurrence of the phenotype. In addition, experiments with a K14 promoter transgene lacking sequences between Ϫ1120 and Ϫ450 show general epidermal expression with limited suprabasal compartment expression, suggesting the possibility that repression may also be mediated by the region between Ϫ450 and Ϫ1120 (7). Other in vitro studies show that the K14 promoter can be repressed by the glucocorticoid receptor acting via monomeric binding sites located at Ϫ79 to Ϫ49 (56). Thus, repression of the K14 promoter in suprabasal cells likely involves control by multiple parallel mechanisms, one of which includes the interaction of POU domain factors with CBP/p300.