Characterization of the Regulatory Domains of the Human Skn-1a/Epoc-1/Oct-11 POU Transcription Factor*

The Skn-1a POU transcription factor is primarily expressed in keratinocytes of murine embryonic and adult epidermis. Although some POU factors expressed in a tissue-specific manner are important for normal differentiation, the biological function of Skn-1a remains unknown. Previous in vitro studies indicate that Skn-1a has the ability to transactivate markers of keratinocyte differentiation. In this study, we have characterized Skn-1a’s transactivation domain(s) and engineered a dominant negative protein that lacked this transactivation domain. Deletional analysis of the human homologue of Skn-1a with three target promoters revealed the presence of two functional domains: a primary C-terminal transactivation domain and a combined N-terminal inhibitory domain and transactivation domain. Skn-1a lacking the C-terminal region completely lost transactivation ability, irrespective of the promoter tested, and was able to block transactivation by normal Skn-1a in competition assays. Compared with full-length, Skn-1a lacking the N-terminal region demonstrated either increased transactivation (bovine cytokeratin 6 promoter), comparable transactivation (human papillomavirus type 1a long control region), or loss of transactivation (human papillomavirus type 18 long control region). The identification of a primary C-terminal transactivation domain enabled us to generate a dominant negative Skn-1a factor, which will be useful in the quest for a better understanding of this keratinocyte-specific gene regulator.

The Skn-1a POU transcription factor is primarily expressed in keratinocytes of murine embryonic and adult epidermis. Although some POU factors expressed in a tissue-specific manner are important for normal differentiation, the biological function of Skn-1a remains unknown. Previous in vitro studies indicate that Skn-1a has the ability to transactivate markers of keratinocyte differentiation. In this study, we have characterized Skn-1a's transactivation domain(s) and engineered a dominant negative protein that lacked this transactivation domain. Deletional analysis of the human homologue of Skn-1a with three target promoters revealed the presence of two functional domains: a primary Cterminal transactivation domain and a combined N-terminal inhibitory domain and transactivation domain. Skn-1a lacking the C-terminal region completely lost transactivation ability, irrespective of the promoter tested, and was able to block transactivation by normal Skn-1a in competition assays. Compared with fulllength, Skn-1a lacking the N-terminal region demonstrated either increased transactivation (bovine cytokeratin 6 promoter), comparable transactivation (human papillomavirus type 1a long control region), or loss of transactivation (human papillomavirus type 18 long control region). The identification of a primary C-terminal transactivation domain enabled us to generate a dominant negative Skn-1a factor, which will be useful in the quest for a better understanding of this keratinocyte-specific gene regulator.
The epidermis is a regenerative tissue composed of several layers of keratinocytes with each layer representing an increased state of differentiation. The basal layer consists of proliferating undifferentiated keratinocytes that migrate outward into the nondividing suprabasal compartment and terminally differentiate. To ensure homeostasis in normal epidermis, a strict balance is maintained between the proliferation of keratinocytes in the basal layer and the rate of progression through the differentiated suprabasal layers. The end result of this tightly regulated differentiation process is the external cornified layer of the epidermis which provides a protective barrier (1).
Morphologically distinct epidermal layers can be identified by the expression of genes which are only expressed at specific stages of keratinocyte differentiation (1)(2)(3). Examples of these molecular markers include the keratin (K) 1 intermediate filaments (4), structural proteins such as profilaggrin (5), and the substrates of keratinocyte transglutaminase (12), the cornified cell envelope proteins (small proline-rich proteins (SPRR) (6 -8), involucrin (9), and loricrin (10,11)). The proliferating basal keratinocytes express K5 and K14, and as keratinocytes move suprabasally and begin to differentiate, K5 and K14 expression is lost and K1 and K10 expression begins. The expression of keratinocyte transglutaminase as well as structural proteins such as involucrin, loricrin, profilaggrin, and SPRR also becomes evident at later stages of keratinocyte differentiation (1,2). Detailed analysis of the regulatory regions of these genes have demonstrated the importance of a number of transcription factors (TFs), such as AP1 and AP2 (13)(14)(15)(16). However, these TFs are also expressed in other organ systems and are not likely to ultimately determine which genes are uniquely expressed in keratinocytes. This has prompted a search for TFs that are only expressed in keratinocytes and are responsible for regulating expression of keratinocyte-specific genes during differentiation. One TF that may play a role in regulating keratinocytespecific gene expression is the POU TF Skn-1a/Epoc-1/Oct-11 (17)(18)(19). POU TFs are of particular interest because they are usually expressed in a tissue-specific manner and play an important role in the establishment of cell identity and cellular differentiation during development (20 -24). POU TF family members, for instance, are essential for proper maturation of anterior pituitary cells (Pit-1) (25), B-cells (Oct-2) (26), and neuronal cells (Unc-86) (27). Since Skn-1a is primarily expressed in the epidermis, it may have an important regulatory role in both epidermal development and keratinocyte differentiation (17,19). During mouse embryogenesis, a stratified epidermis appears by day 15 to 16. Skn-1a is expressed in the ectoderm underlying the periderm just prior to the sloughing off of the periderm and emergence of the stratified epidermis (17). In adult murine skin, Skn-1a expression is present primarily in differentiating suprabasal keratinocytes (17,28). Correspondingly, in vitro studies have demonstrated that Skn-1a binds to and transactivates the promoter regions of keratinocyte genes involved in epidermal differentiation, such as human K10 (17) and SPRR-2A (29). Skn-1a also transactivates the long control region (LCR) of the human papillomaviruses (HPV) types 16, 18 (30), and 1a (31) and the expression * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  pattern of the early HPV gene products E6 and E7 parallels that of Skn-1a, with very low detection in basal keratinocytes and increased expression suprabasally (30).
Because of Skn-1a's potential importance as a regulator of keratinocyte differentiation, our goal was to characterize Skn-1a's primary transactivation domain(s) in order to generate a dominant negative protein that could be used to interfere with Skn-1a function in different biological systems. Unlike the POU domain, which is highly conserved and responsible for binding target DNA sequences, the location of the transactivation domain in POU TFs is not predictable (23,32). The POU TFs Oct-1, Oct-2, and Oct-3/4, for instance, have multiple transactivation domains, which reside on either end of the proteins (24,(33)(34)(35)(36)(37). In this paper we report the cloning of the human homologue of murine Skn-1a (hSkn-1a) and the identification of two novel functional domains, a C-terminal transactivation domain and a N-terminal inhibitory domain, that reside outside of the complex POU domain. C-terminal deleted hSkn-1a is able to effectively compete with full-length hSkn-1a and is therefore a good candidate dominant negative protein.
Characterization of the functional domains of Skn-1a may provide insights regarding the biological function of this epidermal-specific POU TF.

MATERIALS AND METHODS
Isolation and Characterization of hSkn-1a cDNA Clone-A human primary keratinocyte gt11 cDNA library (CLONTECH) was screened with two murine Skn-1a-derived probes generated by RT-PCR. The probes corresponded to the first 310 bp of the ORF and 451 bp spanning the POU domain (GenBank TM accession number L23862). Double nylon membrane lifts were performed (GeneScreen Plus, NEN Life Science Products) and hybridized with one of the two probes according to the supplier's guidelines (CLONTECH). Plaques that cross-hybridized to both probes were evaluated further. Briefly, the membranes were hybridized overnight at 42°C with randomly primed probe and washed to a final stringency of 1 ϫ SSC, 0.1% SDS for 1 h at 50°C. Phage DNA from double-positive plaques was obtained by conventional phenolchloroform extraction and sequenced. A 1.0-kb partial hSkn-1a cDNA spanning the 5Ј-UTR and two-thirds of the ORF was cloned. The remaining third of the ORF and 3Ј-UTR were obtained by 3Ј rapid amplification of cDNA ends (Life Technologies, Inc.). Full-length hSkn-1a cDNA was synthesized by standard RT-PCR on total RNA derived from human foreskin epidermal sheets. Briefly, 1 g of total RNA was used as template for reverse transcription with SuperScript II RT (Life Technologies, Inc.) as specified, followed by Pfu-based PCR (Stratagene). Thirty cycles of PCR were performed as follows: 45 s at 94°C, 30 s at 50°C, and 2 min at 72°C. PCR cycles were followed by 10 min at 72°C. Four independent RT-PCR products were cloned and sequenced. All sequencing reactions were performed by the Sanger method (38).
Establishment of Deletion Constructs and Expression Vectors-hSkn-1a cDNA was subjected to HindIII/XbaI double digestion and cloned into pcDNA3 (Invitrogen), referred to as FL-Skn-1a. ⌬NH 3 -Skn-1a was generated by digesting FL-Skn-1a with XhoI/NotI, which eliminates a 672-bp fragment, including the N-terminal 120-amino acid region of the ORF. This portion of the cDNA was replaced with a PCR-generated fragment spanning from the XhoI site to an upstream primer that aligns starting with nucleotide 475 (codon 121). This adapter primer includes an internal translation initiation sequence (39), a FLAG tag, and multiple unique restriction sites (sense primer: 5Ј-acgcggatccgcggccgcaggatggactacaaggacgacgatgacaagGGTCTGCAG-CCAAATCTCCTCCCC-3Ј and antisense primer: 5Ј-CTGCTTGAAGGT-CTTGGCAAAC-3Ј). The PCR fragment was digested with XhoI and NotI and directionally cloned into the excised region of FL-Skn-1a. ⌬COOH-Skn-1a was engineered by digesting FL-Skn-1a with MscI (restriction site located just downstream of the POU domain) and subcloning a DNA fragment that included an in frame FLAG tag sequence along with a premature termination codon (CCGACTACAAGGACGA-CGATGACAAGTGA). The lowercase bases correspond to adaptor sequences.
Establishment of Reporter Plasmids-bK6-CAT, corresponding to 5.5 kb of upstream sequence from the translational start site derived from the bovine K6 gene, was a generous gift from H. Herrmann (40,41). HPV-18-CAT reporter plasmid (p18P-4321) was a generous gift from C. Baker. A SmaII/BamHI 1.0-kb fragment of the HPV-18 genome spanning the LCR and encompassing the nucleotides 6929 -119 was subcloned into the HindIII site of pSB1 in the sense orientation. The HPV-1a LCR (ATCC), corresponding to nucleotides 3395-4370, was PCR amplified and subcloned into pCR2.1-TOPO (Invitrogen) and subsequently digested with KpnI and ligated into the KpnI site of pCAT3basic Promega. The HPV-1a PCR primers utilized were as follows: 5Ј-ggggtaccTTAGTATATATTATATATAACTATATTT-3Ј (sense) and 5Ј-tcccccgggTGTGCAGAGTCTTACCTGTGTATTT-3Ј (antisense).
Northern Blot Analysis-Epidermal sheets were obtained by incubating neonatal human foreskins with dispase (25 units/ml Hanks' balanced salt solution) at 4°C overnight, homogenized with 2 ml/100 mg tissue of RNAzol TM B (Cinna/Biotecx), and total RNA was prepared as described by the manufacturer and resuspended in 10 mM Tris-HCl (pH 7.4), 1 mM EDTA. Ten micrograms of total RNA was subject to standard 1% agarose gel electrophoresis with 660 mM formaldehyde as a denaturant. Transfer of the RNA was performed with Hybond N (Amersham Pharmacia Biotech) according to the company's specifications. The membrane was hybridized with a 1.7-kb 32 P-labeled hSkn-1a probe at 42°C overnight with 50 l/cm 2 of Hybrizol™ (Oncor), and the membrane was washed to a final stringency of 0.2 ϫ SSC, 1% SDS at 55°C for 5 min. mRNA was visualized by autoradiography.
FISH Analysis-The assay was performed on lymphocyte-derived chromosomes with a 1.7 kb hSkn-1a cDNA probe according to established procedures (DNA Biotech, Inc., York University, North York, Ontario (42,43)).
CAT Assay-The assay was performed according to established protocols (44). The quantification was performed with a Storm Phospho-rImager (Molecular Dynamics). Assays were performed in duplicate, and experiments were repeated (n ϭ 2).
EMSA conditions were based on standard procedures and performed according to the manufacturer's specifications (Amersham Pharmacia Biotech). Samples were subjected to electrophoresis in 5% polyacrylamide nondenaturing gels and analyzed with a Storm PhosphorImager (Molecular Dynamics). Target DNA sequence and irrelevant DNA sequence were as follows, respectively: 5Ј-ATCCTGCTCAATGCCAGT-CATGGATAAATTTGCATCTGGCT-3Ј and 5Ј-AATTGAGCTCGGTA-CCCGGGGATCCTATCTGGGTAGCATATGCTATCCTAATGGATCC-TCTAGAGTCGACCTGCAGGCATGC-3Ј. Highlighted (in boldface) regions correspond to the binding sites for Skn-1a and EBNA-1, respectively. The underlined area corresponds to the octamer consensus sequence.
Western Blot Analysis-Total cell extracts of NIH-3T3 cell lines overexpressing either ⌬NH 3 -Skn-1a or ⌬COOH-Skn-1a were obtained as described above. One and a half g of protein per sample was subject to fractionation by standard SDS-polyacrylamide gel electrophoresis on a 10% polyacrylamide gel (Novex) and transferred to an Immobilon-polyvinylidene difluoride membrane ((Millipore) as recommended by the manufacturers. Immunoblotting was performed with a goat anti-FLAG polyclonal IgG (Santa Cruz Biotech.) against the FLAG epitope using the SuperSignal TM CL-HRP enhanced chemiluminescent detection system (Pierce).
Transient Transfection Assays-CaPO 4 transfection assays were performed using the Modified Bovine Serum Transfection Kit (Stratagene). A total of 10 g of plasmid DNA/100-mm tissue culture dish was utilized. Typically, 5 g of any one of the Skn-1a expression vectors was co-transfected with 5 g of one of the CAT reporter plasmids described above. For the control experiments, 5 g of reporter plasmid was cotransfected with 5 g of empty pcDNA3 vector. Constant amounts of transfected DNA were also used for the competitive CAT assays. Cells were harvested 2-3 days post-transfection.

RESULTS
Isolation of the hSkn-1a cDNA-A combination of cDNA library screening along with 3Ј-rapid amplification of cDNA ends was used to clone the full-length human Skn-1a cDNA. A human primary keratinocyte gt11 cDNA library (CLON-TECH) was screened with two probes derived from murine Skn-1a: one spanning a unique region within the 5Ј portion of the ORF and the other spanning the POU domain (17). The resulting 2.4-kb cDNA demonstrated minimal divergence from the murine Skn-1a cDNA (Fig. 1A). A comparison of the ORF for both human and murine Skn-1a reveals that the human protein (hSkn-1a) is slightly larger than its murine counterpart by 6 amino acid residues (436 versus 430), but still shares approximately 90% overall identity. As anticipated, sequences within the POU domain only diverge within the hypervariable linker region (Fig. 1B). Sequence analysis of four independent clones of hSkn-1a revealed a polymorphism at residue 152 (Fig.  1, A and C). The nucleotide sequence for this particular codon was CAC or CGC with equal frequency. This A to G transition FIG. 1. hSkn-1a sequence and expression. A, the amino acid sequence is denoted in single letter codes under the nucleotide sequence. Regions corresponding to the unique 120 N-terminal amino acids, POU-specific domain, and POU homeodomain are depicted under the thick line, thin line, and parallel lines, respectively. The polymorphic codon for the G 3 A transition is in bold type. Bold type numbers to the right of every line correspond to the last amino acid residue for that particular line. B, identical amino acid residues are boxed. The N-terminal POU-specific domain and C-terminal POU-homeodomain are depicted under the solid lines, while the hypervariable linker region is depicted under the dashed line. The numbers to the left of every line corresponds to total residues for that particular line. Human Skn-1a is 436 residues long, while murine Skn-1a is 430 residues long. C, sequence analysis demonstrating the A 3 G transition in codon 152, which leads to an arginine (R) to histidine (H) amino acid substitution. Letters above each lane designate the nucleotides, while letters on the left side or on the right side of each panel designate the sequence flanking the polymorphic region. The nucleotide difference is highlighted by an asterisks. The 5Ј to 3Ј orientation is noted for each sequence. D, 10 g of total RNA from human foreskin epidermal sheets and HaCaT cells was electrophoresed in a 1% agarose gel, blotted onto a nylon membrane, and hybridized with a 1.7-kb hSkn-1a probe. Hybridization revealed two hSkn-1a transcripts at approximately 3.5 and 2.4 kb. The third and slowest migrating band corresponds to the 28 S ribosomal subunit RNA. HaCaT cells do not express hSkn-1a and emphasize the nonspecific hybridization. lead to an amino acid substitution of histidine to arginine (Fig.  1C).
Northern blot analysis of total RNA derived from human foreskin epidermal sheets revealed that keratinocytes expressed two hSkn-1a transcripts estimated to be 2.4-and 3.5-kb in length (Fig. 1D). Our screening of a human cDNA library with murine Skn-1a yielded clones of rRNA cDNA in addition to Skn-1a. Thus, the 4.4-kb band on the Northern blot (which migrates at the same level as 28 S rRNA) most likely corresponds to nonspecific hybridization to 28 S rRNA.
Attempts to clone a human homologue of the previously described Skn-1i splicing variant were unsuccessful. All cDNAs obtained with a unique Skn-1i probe contained sequences homologous to Skn-1i, but did not contain a significant ORF due to frequent termination codons. Additionally, in vitro transcription and translation of the cDNAs in both orientations did not result in the synthesis of any detectable protein (data not shown). Northern blots also failed to reveal specific Skn-1i transcripts (data not shown). Therefore, a human homologue to murine Skn-1i may not exist.
Chromosomal Localization of hSkn-1a-To determine whether hSkn-1a was located at a chromosomal site where other keratinocyte-specific genes are located, FISH mapping and 4Ј,6Ј-diamidino-2-phenylindole (DAPI) fluorescent dye multibanding assay was performed on human chromosomes with a 1.7-kb hSkn-1a cDNA probe (42,43). The DAPI banding pattern localized the hSkn-1a gene to the long arm of human chromosome 11 ( Fig. 2A). Detailed mapping was determined by compiling the alignment of 10 independent photos of DAPIstained chromosomes and FISH signals, which further localized the probe to human chromosome 11, region q23.3 (Fig. 2B).
Functional Domains of hSkn-1a-To localize the functional domains of hSkn-1a, we generated three constructs (Fig. 3). These constructs did not alter the conserved POU domain that is responsible for DNA binding, nuclear localization, and oligomerization (17, 23). The first construct contains an unaltered hSkn-1a cDNA (Fig. 3A). In the second construct, ⌬NH 3 -Skn-1a, the N-terminal 120 amino acids were deleted and replaced with a FLAG tag (Kodak IBI) and an in-frame AUG at codon 121 (Fig. 3B). The third construct, ⌬COOH-Skn-1a, lacks the C-terminal 92 amino acids that have been replaced with a FLAG tag and a stop codon immediately following the POU domain at codon 345 (Fig. 3C). To verify the integrity of the constructs, an in vitro transcription and translation assay with rabbit reticulocyte lysates was performed. All three constructs generated proteins that migrated at the predicted molecular masses of 48 kDa (FL-Skn-1a), 39 kDa (⌬COOH-Skn-1a), and 36 kDa (⌬NH 3 -Skn-1a) in SDS-polyacrylamide gel electrophoresis (data not shown). To localize the transactivation domain of hSkn-1a, the ability of the deleted hSkn-1a constructs to transactivate the bovine K6 (bK6) promoter (equivalent to the human K10 promoter) (40,41) was assessed in CAT cotransfection studies performed in both an epithelial cell line (HeLa S3) and nonepithelial cell line (NIH-3T3) (Fig. 4). Cotransfection of HeLa S3 cells (Fig. 4A) or NIH-3T3 cells (Fig.  4B) with the C-terminal deleted ⌬COOH-Skn-1a and bK6-CAT plasmids did not result in transactivation of bK6-CAT. By comparison, the full-length FL-Skn-1a transactivated bK6-CAT approximately 2-fold (HeLa) to 5-fold (NIH-3T3) above background. This indicated that the transactivation domain resides within the C-terminal 92 amino acids of the protein, downstream of the POU domain. In contrast, the N-terminaldeleted ⌬NH 3 -Skn-1a protein demonstrated much greater transactivation of bK6-CAT than the full-length FL-hSkn-1a (Fig. 4). ⌬NH 3 -Skn-1a was able to transactivate bK6-CAT approximately 6.5-fold (HeLa) to 12-fold (NIH-3T3) above background, suggesting the presence of a negative regulatory domain within the N-terminal region of the protein.
⌬COOH-Skn-1a Retains DNA Binding Ability-Although the inability of ⌬COOH-Skn-1a to transactivate the bK6 promoter indicates the presence of a C-terminal transactivation domain, it is possible that the deleted construct was not able to bind DNA target sequences in the bK6 promoter because of conformational changes. To determine whether ⌬COOH-Skn-1a was able to bind DNA, EMSAs were performed with all three constructs and an octamer consensus sequence DNA probe (Fig. 5A (29)). Radiolabeled double-stranded DNA probe was incubated with either FL-Skn-1a, ⌬NH 3 -Skn-1a, or ⌬COOH-Skn-1a derived from cell lines overexpressing these factors. Lanes 2, 5, and 8 correspond to band shifts of FL-Skn-1a, ⌬NH 3 -Skn-1a, and ⌬COOH-Skn-1a, respectively. The full-length and deleted Skn-1a TFs both formed a complex with the DNA probe. The different band shift migration patterns resulted from the different molecular masses of the three constructs, with FL-Skn-1a followed in size by ⌬COOH-Skn-1a and ⌬NH 3 -Skn-1a. A second, larger molecular mass shift is seen with the double-stranded DNA probe and is due to binding with the ubiquitously expressed Oct-1 protein (Fig. 5A). The EMSA demonstrated that deletion of the C-terminal portion of the TF did not affect the ability of the protein to bind DNA. Additionally, the EMSA indirectly demonstrated that ⌬COOH-Skn-1a was present in cells at comparable levels to FL-Skn-1a and ⌬NH 3 -Skn-1a. In order to directly assess the cellular accumulation of ⌬COOH-Skn-1a protein relative to ⌬NH 3 -Skn-1a (which demonstrated increased transactivation of the bK6 promoter), a Western blot was performed on total cell extracts from NIH 3T3 cells overexpressing these two constructs (Fig. 5B). This assay detected bands of similar intensities at the predicted molecular masses for both ⌬COOH-Skn-1a (39 kDa) and ⌬NH 3 -Skn-1a (36 kDa), which, in conjunction to the EMSA results, conclusively demonstrate that the loss of transactivation by ⌬COOH-Skn-1a is due to the absence of a transactivation domain.
FL-Skn-1a and ⌬NH 3 -Skn-1a Have Similar DNA Binding Affinities-The enhanced transactivation potential of ⌬NH 3 -Skn-1a might result from deletion of an N-terminal negative regulatory domain or reflect increased affinity for DNA target sequences. To determine whether increased DNA-protein complex stability contributed to the greater transactivation of bK6 by ⌬NH 3 -Skn-1a, an affinity EMSA was performed. A constant amount of hSkn-1a protein was preincubated with increasing amounts of unlabeled octamer consensus DNA and a fixed amount of labeled octamer DNA probe. The relative affinity of both full-length and truncated hSkn-1a constructs for a known Skn-1a DNA target proved to be very similar (Fig. 5C). Thus, the increased transactivation of bK6 by ⌬NH 3 -Skn-1a is not due to increased DNA binding affinity and suggests the presence of an N-terminal negative regulatory domain.
hSkn-1a constructs were co-transfected into the CV-1 green monkey kidney-derived cells along with a CAT reporter plasmid driven either by the HPV-1a LCR (Fig. 6A) or the HPV-18 LCR (Fig. 6B). CV-1 cells were chosen because they do not express Skn-1a and are suitable for HPV LCR CAT assays (31). FL-Skn-1a transactivated both LCRs by 3. 2-fold (HPV-1a) and FIG. 4. CAT assay with bK6-CAT in different cell lines. HeLa S3 cells (A) or NIH-3T3 cells (B) were co-transfected with a bK6-CAT reporter plasmid and one of the three hSkn-1a expression vectors (Fig. 3). Activation of the bK6 promoter was tested by measuring the CAT activity of each co-transfection relative to that of transfection of the reporter plasmid alone (control groups). x axis, hSkn-1a constructs; y axis, fold transactivation above background control. Each bar graph represents one experiment with duplicate samples and is representative of additional experiments. Results demonstrating a statistically significant difference from control at p Յ 0.01 (*) are indicated. 7.5-fold (HPV-18), comparable with what was initially observed with the bK6 promoter. Unlike the FL-Skn-1a construct, the ⌬COOH-Skn-1a construct did not transactivate the two viral promoters (Fig. 6, A and B), consistent with the bK6 promoter transactivation studies (Fig. 4). In contrast, the ability of ⌬NH 3 -Skn-1a to transactivate the HPV-1a and HPV-18 LCRs depended on which viral promoter is being tested. For HPV-1a LCR (Fig. 6A), ⌬NH 3 -Skn-1a and FL-Skn-1a had similar transactivating activity (3.5-fold versus 3.2-fold, respectively); however, ⌬NH 3 -Skn-1a is unable to transactivate the HPV-18 LCR (Fig. 6B), similar to the results obtained with ⌬COOH-Skn-1a. This is dramatically different from the increased transactivation of the bK6 promoter by ⌬NH 3 -Skn-1a (approximately 12fold). In summary, FL-Skn-1a was consistently able to transactivate all three promoters and ⌬COOH-Skn-1a consistently lacked transactivation potential, while ⌬NH 3 -Skn-1a variably transactivated the three promoters (Figs. 4 and 6).
Functional Effect of ⌬COOH-Skn-1a by Competitive CAT Assay-Because ⌬COOH-Skn-1a did not transactivate the bK6 promoter, a competitive CAT assay was performed to determine whether ⌬COOH-Skn-1a could interfere with FL-Skn-1a transactivation of the bK6 promoter. NIH-3T3 cells were transiently co-transfected with FL-Skn-1a and ⌬COOH-Skn-1a in varying molar ratios along with the bK6-CAT reporter plasmid (Fig. 7). While ⌬COOH-Skn-1a is able to interfere with bK6 transactivation by FL-Skn-1a when both constructs are present in equimolar amounts (approximately 32% decrease in CAT activity relative to unchallenged FL-Skn-1a), the ability of FL-Skn-1a to transactivate the bK6 promoter was completely blocked by 3-fold molar excess of ⌬COOH-Skn-1a. This indi- FIG. 5. EMSA, affinity EMSA, and Western blotting. A, total cell extracts derived from NIH-3T3 cells overexpressing FL-Skn-1a, ⌬NH 3 -Skn-1a, or ⌬COOH-Skn-1a were incubated with a 32 P-labeled octamer consensus sequence DNA probe. Samples were subjected to polyacrylamide gel electrophoresis under nondenaturing conditions. Competitions were performed with 30-fold molar excess of cold octamer consensus DNA (Cold Oct-1) or irrelevant EBNA-1 consensus DNA. Lane 1, octamer DNA probe. Lanes 2,5,8,and 11, octamer DNA probe and FL-Skn-1a, ⌬NH 3 -Skn-1a, or ⌬COOH-Skn-1a, respectively. Lanes 3, 6, 9, and 12, octamer DNA probe and 30-fold molar excess cold octamer DNA along with protein extracts in the same order as above. Lanes 4, 7, 10, and 13, octamer DNA probe and 30-fold molar excess of EBNA-1 irrelevant DNA along with protein extract in the same order as above. B, Western blot analysis of total cell extracts derived from NIH-3T3 cells overexpressing ⌬COOH-Skn-1a (lane 1) and ⌬NH 3 -Skn-1a (lane 2) was performed with a goat anti-FLAG polyclonal antibody. The two truncated TFs are present at similar amounts and migrated at the predicted molecular masses of 39 kDa (⌬COOH-Skn-1a) and 36 kDa (⌬NH 3 -Skn-1a). Total cell extract from NIH-3T3 cells overexpressing ␤-galactosidase was used as a negative control (lane 3). C, total cell extracts derived from NIH-3T3 cells overexpressing FL-Skn-1a (q) or ⌬NH 3 -Skn-1a (E) were preincubated on ice for 20 min with unlabeled octamer DNA and subsequently incubated at room temperature with a fixed amount of labeled octamer DNA probe. Cold DNA was added as indicated. Data points were normalized to the uncompeted band shift control for each construct. Each data point represents the average of two separate experiments.
cates that ⌬COOH-Skn-1a is able to effectively compete with FL-Skn-1a for bK6 DNA target sites and prevent it from transactivating the bK6 promoter. Thus, ⌬COOH-Skn-1a may be useful in exerting a dominant negative effect if overexpressed in in vivo model systems. DISCUSSION POU factors are implicated in cell lineage progression during early embryogenesis and tissue-specific cell maturation in later developmental stages (20). Skn-1a POU TF is primarily expressed in the epidermis (17,19) and is a strong candidate gene for regulating keratinocyte-specific gene expression and keratinocyte differentiation. However, a biological function and mode of action for Skn-1a in epidermal differentiation has not yet been established. In vitro assays have demonstrated that Skn-1a can activate (K10) or inhibit (involucrin) the expression of the keratinocyte-specific genes (17,45). Yet, Skn-1a/i null mice develop normally and do not exhibit any overt phenotypic alterations (28), perhaps reflecting functional redundancy between Skn-1a/i and other ubiquitous POU TFs in skin. Consequently, the precise role of this keratinocyte-specific POU TF in epidermal differentiation still needs to be determined.
Elucidation of Skn-1a's mechanism of action requires identification of functional domains by testing the effects of truncated Skn-1a constructs on different target promoters. To facilitate future in vitro studies with human keratinocyte organ cultures, we cloned the human Skn-1a homologue (hSkn-1a). hSkn-1a proved to be virtually identical to its murine counterpart. Interestingly, a human homologue of the previously described murine Skn-1i splice variant could not be detected. Deletion analysis of hSkn-1a revealed that it contained a primary transactivation domain downstream of the POU domain, within the C-terminal 92 amino acids. While FL-Skn-1a transactivated all promoters tested, regardless of which cell line was assayed, ⌬COOH-Skn-1a did not transactivate any of the promoters. Conversely, the Nterminal 120 amino acids of hSkn-1a contains either an inhibitory domain or a secondary transactivation domain, depending on the target DNA sequences in the different promoters. For the bK6 promoter, ⌬NH 3 -Skn-1a demonstrated greater transactivation than FL-Skn-1a. However, deletion of the N-terminal 120 amino acids did not increase transactivation of the other promoters. ⌬NH 3 -Skn-1a was incapable of transactivating the HPV-18 LCR, suggesting that either the N-terminal region contains a necessary transactivation domain for this promoter or that interaction of ⌬NH 3 -Skn-1a with the nonconsensus DNA target site of HPV-18 LCR induces a conformational change in the POU domain preventing it from interacting with accessory factors necessary for proper promoter activation (46,47). An earlier study, using a Skn-1a-LexA fusion protein approach, reported that the primary transactivation domain for the murine Skn-1a resides within the N-terminal 182 amino acids (31). Our data suggest FIG. 6. CAT assay of hSkn-1a constructs on different promoters. The ability of the hSkn-1a constructs to transactivate two target promoters was evaluated by performing co-transfections of FL-Skn-1a, ⌬NH 3 -Skn-1a, or ⌬COOH-Skn-1a with CAT reporter constructs being driven by HPV-1a LCR (A) or HPV-18 LCR (B). CAT activity was measured by determining the percent conversion of unacetylated chloramphenicol to its acetylated forms for each individual sample relative to control transfections of the respective reporter plasmids alone. x axis, constructs used; y axis, fold transactivation above background control. Each bar graph represents one experiment with duplicate samples and is representative of additional experiments. Results demonstrating a statistically significant difference from control at p Յ 0.01 (*) and at p Յ 0.05 (ϩ) are indicated. that the critical domain for hSkn-1a transactivation lies in the C terminus, while a secondary transactivation domain in the N terminus is required for some promoters. This apparent discrepancy may reflect conformational changes resulting from fusion of Skn-1a fragments with the LexA DNA-binding domain or use of the LexA promoter as a target for the fusion protein.
POU TFs often have multiple transactivation domains (33, 35, 48 -50). Oct-3/4, for instance, has two transactivation domains residing outside of the POU domain: a primary N-terminal transactivation domain and a secondary, cell-specific, C-terminal transactivation domain that is dependent on the POU domain itself (33)(34)(35). This dependence is lost if Oct-3/4's C-terminal transactivation domain is fused to the heterologous POU domain of Pit-1, indicating that the activity of the Oct-3/4 C-terminal depends on a particular POU domain and the specific co-factors interacting with that POU domain (33,35). Inhibitory domains have also been observed in other POU factors. Oct-2 harbors, in addition to two transactivation domains, an N-terminal, cell line-specific inhibitory domain (32,51).
Our results suggest that hSkn-1a has two distinct transactivation domains: 1) an invariant primary C-terminal domain and 2) a promoter-specific N-terminal domain. Additionally, we also demonstrate that the N-terminal 120 amino acids of hSkn-1a contain a promoter-specific inhibitory domain. The C-terminal truncated protein lacking the primary transactivation domain is capable of effectively competing with FL-Skn-1a in a dominant negative manner. Therefore, ⌬COOH-Skn-1a is a good candidate construct to effectively block and disrupt the function of endogenous Skn-1a in keratinocytes. The identification of important functional domains of hSkn-1a may help define its biological function in keratinocyte differentiation and epidermal homeostasis. Constructs with or without these defined domains can be inappropriately overexpressed or used to block endogenous Skn-1a in keratinocyte organ cultures and transgenic animal models to assess phenotypic and biochemical effects.