Keratin 10 Gene Expression During Differentiation of Mouse Epidermis

Developmental Biology 216, 164–181 (1999) Article ID dbio.1999.9460, available online at http://www.idealibrary.com on

CORE Metadata, citation and similar papers at core.ac.uk

Provided by ElsevierKeratin - Publisher Connector 10 Gene Expression during Differentiation of Mouse Epidermis Requires Transcription Factors C/EBP and AP-2

Edward V. Maytin,*,† Julia C. Lin,†,‡ Ramachandran Krishnamurthy,§ Nikoleta Batchvarova,¶ David Ron,¶ Pamela J. Mitchell,§ and Joel F. Habener†,‡ *Department of Dermatology and †Laboratory of Molecular Endocrinology, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts 02114; ‡Howard Hughes Medical Institute and ¶Skirball Institute of Biomolecular Medicine, New York University Medical Center, New York, New York 10016; and §Department of Biochemistry and Molecular Biology, Pennsylvania State University, University Park, Pennsylvania 16302

The epidermis forms a vital barrier composed of stratiﬁed keratinocytes and their differentiated products. One of these products, keratin K10, is critical to epidermal integrity, because mutations in k10 lead to abnormal blistering. For the normal expression of k10, differentiation-associated transcription factors C/EBP␣, C/EBP␤, and AP-2 are well positioned to play an important role. Here, regulation of the k10 gene is examined in keratinocytes in the skin of normal mice and in transgenic mice carrying targeted deletions of c/ebp␤ and ap-2␣. In cultured cells, C/EBP␣ and C/EBP␤ are each capable of activating the k10 promoter via three binding sites, identiﬁed by site-directed mutagenesis. In a given epidermal cell in vivo, however, the selection of C/EBP␣ versus C/EBP␤ for k10 regulation is determined via a third transcription factor, AP-2. This novel regulatory scheme involves: (1) unique gradients of expression for each transcription factor, i.e., C/EBP␤ and AP-2 most abundant in the lower epidermis, C/EBP␣ in the upper; (2) C/EBP-binding sites in the ap-2␣ gene promoter, through which C/EBP␤ stimulates ap-2␣; and (3) AP-2 binding sites in the c/ebp␣ promoter, through which AP-2 represses c/ebp␣. Promoter-analysis and gene-expression data presented herein support a regulatory model in which C/EBP␤ activates and maintains AP-2 expression in basal keratinocytes, whereas AP-2 represses C/EBP␣ in those cells. In response to differentiation signals, loss of AP-2 expression leads to derepression of the c/ebp␣ promoter and activation of k10 as cells migrate upward. © 1999 Academic Press Key Words: keratinocytes; cell differentiation; transcription factors; skin.

INTRODUCTION negative mutations in the heritable disorder epidermolytic hyperkeratosis, leads to epidermal disruption and blistering The skin is the largest organ in the body and provides a (Cheng et al., 1992; Rothnagel et al., 1992; Syder et al., 1994). vital barrier against fluid loss, chemical and infectious Mechanisms that control k10 expression are poorly un- agents, and UV light (Goldsmith, 1991). This protection is derstood. In the basal epidermal layer, k10 transcription is provided by the epidermis, a stratified tissue composed of suppressed and becomes activated upon entering the supra- progenitor cells in the basal layer and differentiated kera- basal compartment (Roop et al., 1988). We are interested in tinocytes in the suprabasal layers. The keratinocytes pro- factors that regulate k10 expression. Because of its abun- duce differentiation-specific proteins, including keratins, dance in the skin and restriction to the basal layer, we 1 involucrin, and loricrin (Eckert et al., 1997; Fuchs, 1990), considered the transcription factor AP-2 as one such factor that ultimately become cross-linked during cornification, a (Byrne et al., 1994; Leask et al., 1991; Mitchell et al., 1991). ␣ specialized form of apoptosis (Haake and Polakowska, AP-2 (also called AP-2 ) is a member of a family that now 1993). Keratin K10 is expressed early during the differentiation of keratinocytes and is essential for epidermal integ- 1 Abbreviations used: C/EBP, CCAAT/enhancer binding protein; rity. Loss of normal k10 expression, through dominant- AP-2, activator protein 2.

0012-1606/99 $30.00 Copyright © 1999 by Academic Press 164 All rights of reproduction in any form reserved. C/EBP and AP-2 Transcription Factor Network in Skin 165 includes AP-2␤ (Moser et al., 1995) and AP-2␥ (Oulad- late during keratinocyte differentiation in vitro (Maytin and Abdelghani et al., 1996); these proteins dimerize via a Habener, 1998). C/EBPs are expressed in the skin in a helix-span-helix domain (Williams and Tjian, 1991) and spatially organized manner that corresponds to the timing bind to the consensus sequence 5Ј-GCCN3(4)GGC-3Ј in of expression seen in cultured keratinocytes in vitro, i.e., DNA (Mitchell et al., 1987). One or more proteins of the C/EBP␤ and CHOP appear first in the lower undifferenti- AP-2 family are important for the transcriptional regulation ated epidermis, followed later by C/EBP␣ in the upper of K14, a keratin expressed specifically in the basal epider- differentiated epidermis (Maytin and Habener, 1998). These mal layer (Byrne et al., 1994). findings imply that C/EBPs may exert functions in epider- Transcription factors of the C/EBP (CCAAT-enhancer mal differentiation. binding protein) family are also candidates for regulation of Here, the functions of C/EBPs in skin development and k10 transcription, because of their links to cellular differ- differentiation are examined more directly, emphasizing entiation (see below). C/EBPs are members of the leucine their role in k10 regulation. A segment of the mouse k10 zipper or “bZIP” family of transcription factors (Haas et al., promoter was used for transactivation and mutagenesis 1995; Lamb and McKnight, 1991; Vinson et al., 1993), work in BALB/MK cells, a cell culture model of epidermal recognized early on for their involvement in cellular differ- differentiation. Changes in k10 expression were found to be entiation (McKnight, 1992). C/EBPs bind to degenerate mediated via three C/EBP binding sites within the k10 nucleotide sequence sites consisting of a consensus motif promoter. Gene expression was examined in mice with a A C C /GTTNNG /TAA /T (McKnight, 1992; Osada et al., 1996; targeted disruption of the c/ebp␤ gene (cebp␤ null). The Ryden and Beemon, 1989; Ubeda et al., 1996); the pseudo- c/ebp␤ null mice displayed a marked ectopic increase of palindromic bases (in italic) separated by four intervening C/EBP␣ and K10 expression in the basal epidermis. Based basepairs are invariable. A role for C/EBPs in cellular on the known suppression of C/EBP␣ by AP-2 in adipocytes, differentiation has been established in adipocytes in which we hypothesized that increase of C/EBP␣ in the skin of differentiation is accompanied by unique temporal changes c/ebp␤ null mice might represent a release of the c/ebp␣ in the levels of individual C/EBP members. Upregulation of gene from transcriptional repression and that the repressor C/EBP␦ and C/EBP␤ precedes expression of C/EBP␣ (Cao et might be AP-2. Experiments in c/ebp␤ null mice showing al., 1991; MacDougald and Lane, 1995; Yeh et al., 1995) and loss of AP-2 expression, and other experiments showing of fat-specific genes (Friedman et al., 1989). Direct trans- that ap-2␣ null mice reproduce the skin phenotype of the genic overexpression of C/EBP␣ (Freytag et al., 1994; Lin c/ebp␤ null mice, support this hypothesis. Overall, the data and Lane, 1994) or of C/EBP␤ (Yeh et al., 1995) accelerates suggest a model of regulation in which C/EBP␤ activates or the differentiation of adipoblasts into adipocytes. These maintains AP-2 expression, and AP-2 represses C/EBP␣,in observations suggest that expression of C/EBP␤ initially, undifferentiated keratinocytes. In response to differentia- and C/EBP␣ at a later time, is a prerequisite for correct tion signals, a loss of AP-2 expression leads to derepression expression of the differentiation-specific genes. of the C/EBP␣ promoter and contributes to activation of the Both C/EBPs and AP-2 display broad regulatory potential, k10 gene. This novel system of crosstalk between C/EBPs through the existence of multiple isoforms and autoregula- and AP-2 appears to contribute to the regulation of the tory loops. Dominant-negative isoforms of the C/EBPs differentiation-specific gene k10 in the mammalian epider- (Descombes and Schibler, 1991; Lin et al., 1993; Ossipow et mis. al., 1993; Ron and Habener, 1992) or of AP-2 (Meier et al., 1995) heterodimerize with other isoforms in the same class, thus counteracting stimulatory effects. Autoregulation oc- MATERIALS AND METHODS curs because the promoters for c/ebp␣ (Christy et al., 1991; Lin et al., 1993), c/ebp␤ (Chang et al., 1990), and chop Preparation of Skin Specimens from Mice (Fawcett et al., 1996; Sylvester et al., 1994) each contain Nullizygous for c/ebp␤ and ap-2 C/EBP-binding motifs that should be available for interac- Mice in which the bZIP region of the c/ebp␤ gene has been tions with both stimulatory and inhibitory C/EBP isoforms. deleted (Screpanti et al., 1995) were a gift from Dr. Valeria Poli. The notion of counterregulation among C/EBP family These mice were backcrossed into an FVB/n background (Charles genes, at the level of their individual promoters, was River, Wilmington, MA), and the line was maintained as heterozy- recently extended by the observation of a system of cross- gous breeding pairs. ap-2␣ null mice, lacking a genomic segment talk between AP-2 and C/EBPs. CUP (C/EBP␣ undifferenti- containing exon 5 of the ap-2␣ gene (Schorle et al., 1996), were also ␣ ated protein), a repressor of the C/EBP␣ promoter in pre- used. Because ap-2 null mice die perinatally, embryos from adipocytes (Tang et al., 1997) which disappears as the cells matings of heterozygotes were harvested shortly before birth (e17.5 to e18.5). The genotype of each individual was determined by PCR differentiate, is identical to AP-2 (Jiang et al., 1998). of tail DNA or yolk-sack DNA, for c/ebp␤ or ap-2␣, respectively. C/EBPs have only recently been examined in the epider- For histological studies with c/ebp␤ knockout mice, pieces of mis (Maytin and Habener, 1998; Oh and Smart, 1998; Swart skin from pairs of adult mice (c/ebp␤ ϩ/Ϫ and c/ebp␤ Ϫ/Ϫ, et al., 1997). Earlier, we examined the expression of C/EBPs matched for age and sex) were embedded side by side in OCT in keratinocytes of the mouse and found that C/EBP␤ and freezing compound (Miles, Inc., Elkhart, IN) along with some CHOP are expressed early, whereas C/EBP␣ is expressed extraneous tissue as an orientation marker. Frozen blocks were

then cut into 5-␮m sections and stained as described under “Im- A region of the mouse ap-2␣ gene containing C/EBP-binding munocytochemistry.” For ap-2␣ studies, wild-type or ap-2␣ ϩ/Ϫ motifs at Ϫ710 and Ϫ760 bp relative to the ATG codon in exon 1a embryos were embedded side by side with ap-2␣ homozygous null (GenBank No. U17289) was synthesized as an 87-bp double- embryos from the same litter. In this approach, paired tissue stranded oligonucleotide with NotI and BamHI ends, then ligated samples were exposed to identical conditions at every step in the into an SKLuc-based reporter containing a minimal promoter (ϩ1 processing for immunochemical analyses. to Ϫ86 bp of k10). This construct was named pWtAP2Luc. A similar construct, pMutAP2Luc, harbored mutations in both Cell Culture and Differentiation Protocol C/EBP-binding motifs (see Fig. 6 for sequence). BALB/MK cells (Weissman and Aaronson, 1983) were grown in EMEM containing low calcium (0.07 mM; Biofluids, Inc., Rock- DNase I Footprint Analysis of the k10 Promoter ville, MD) to which 5 ng/ml EGF (Collaborative Research, Bedford, To create a probe suitable for analysis of site-specific binding of MA) and 10% Chelex-treated serum (Marcelo et al., 1978) were proteins to DNA, the sense and antisense strands of the dsDNA added. For calcium step-up experiments, EMEM supplemented probe were individually labeled using [␥-32P]ATP and T4 polynucle- with calcium to 0.12 mM (Yuspa and Dlugosz, 1991) was added to otide kinase to end-label the plasmid K10-pCRII that had been confluent cells, and this “high-calcium” medium was changed cleaved at a restriction site adjacent to the 5Ј end of the K10 insert. daily for the duration of the experiment to minimize effects from Subsequently, the insert was released by restriction enzyme cleav- nutrient depletion. age at the unlabeled end. A truncated form of C/EBP␤ called C/EBP␤-LIP, translated from an internal methionine and contain- Cell Transfection and Luciferase Assays ing the bZIP binding domain (Poli et al., 1990), was used for ␤ Eukaryotic expression plasmids containing the C/EBP␣, C/EBP␤, footprinting. The C/EBP -LIP protein was produced as a histidine- and CHOP coding sequences in pCDNA1 (InVitrogen) have been tagged fusion protein in pRSET-A (InVitrogen). After transforma- described (Ron and Habener, 1992). These plasmids, along with tion of Escherichia coli strain BL21(DE3)pLysS and induction with luciferase reporter constructs (see “Cloning,” below), were trans- IPTG (Studier et al., 1990), cells were lysed and the supernatant fected into subconfluent BALB/MK cells using a standard calcium- was applied over a nickel–NTA resin column (Qiagen, Inc., Chats- phosphate precipitate (Ausbel et al., 1992) or using Lipofectamine worth, CA) and the protein eluted in a pH step gradient according (Gibco BRL, 6 ␮l per 60-mm dish, 5 h exposure). At 48 h after to instructions from the manufacturer. The DNase I protection assay was performed by incubating the His-tagged C/EBP␤ with transfection, cells were rinsed with cold PBS, lysed in glycyl- 32 glycine buffer with 0.1% Triton X, and assayed for luciferase 50,000 cpm of P-labeled K10 probe at 4°C for 1 h prior to digestion activity in an AutoLumat LB 953 luminometer (Berthold Analyti- with DNase I, essentially as described (Galas and Schmidt, 1978). cal Instruments, Nashua, NH) as described (Brasier et al., 1989). Electrophoretic Mobility Shift Assays (EMSA) Cloning and Analysis of the Mouse k10 and ap-2␣ Promoters Double-stranded DNA probes for EMSA experiments were produced by designing matched sense and antisense oligonucleotides A fragment of the mouse K10 gene promoter was obtained by that produced BamHI/BglII 4-bp overhangs that could be filled in utilizing a highly conserved sequence of nucleotides at Ϫ245 bp, with E. coli polymerase Klenow fragment and [32P]dATP. Oligonu- apparent on cross-species examination of bovine and human K10 cleotides were synthesized in the MGH Biopolymer Facility and genes. A 19-base oligonucleotide corresponding to this conserved purified by electrophoresis on urea-acrylamide gels. Probe se- region was used as the 5Ј primer for PCR amplification of mouse quences are given below. Sense strands are shown, with uppercase genomic DNA from NIH 3T3 cells; the matched 3Ј primer corre- letters denoting keratin K10 sequences and lowercase letters unre- sponded to the known coding sequence of k10 at ϩ185 bp. The lated sequences, as they would appear after the Klenow fill-in resulting 430-bp PCR product, containing 245 bp of promoter and reaction. For Fig. 3C, wild-type probes (26 bp): Site A, gatcc TAA 185 bp of k10 coding sequence, was ligated into plasmid pCRII (TA TTT ATG CAA TCA Tagatc; Site B, gatcc GTT ATT ACT GAA cloning kit; InVitrogen, San Diego, CA) to create the construct GAT Aagatc; Site C, gatcc AAA ACC ATG CAA GTA Aagatc. For K10-pCRII. To obtain a fragment containing only the promoter, Fig. 3C, mutant probes (26 bp) were identical to the wild-type two PCR primers were designed to anneal at positions Ϫ245 bp and probes except for the basepair substitutions underlined, as follows: ϩ1 bp adding extra bases to create restriction sites as follows: Mut A, gatcc TAA TTT ATG ATA TCA Tagatc; Mut B, gatcc GTG 5Ј-KpnI primer, CGG GGG TAC CCA TTT TCA ATT TC; 3Ј-BglII ATC ACT GAA GAT Aagatc; Mut C, gatcc AAA AGG ATG CCC primer, GAA GAT CTG GTG GTG TAG TGT AGC CC. Using GTA Aagatc. (For Fig. 4C, the same mutant bases underlined above these two primers and the K10-pCRII construct as templates, an in Mut A, Mut B, or Mut C were introduced into the wild-type K10 ϳ250-bp fragment was obtained by PCR and cloned into the promoter in various combinations by site-directed mutagenesis.) In KpnI/BglII sites of the pGL2-Basic vector (Promega, Madison, WI) Fig. 3C and 3D, the “Standard” probe was a high-affinity C/EBP site and thence into the SmaI–HindIII site of SKLuc to place the from the angiotensinogen gene, also known as M6 (Brasier et al., wild-type k10 promoter directly upstream of the luciferase gene. 1990): gatc CAC AGT TGT GAT TTC ACA ACC TGA CCA This new construct was named K10-Luc. The orientation and Gagatc. For Fig. 3D, the K10 probe was 44 bp, consisting of native integrity of all inserts were confirmed by DNA sequencing. To k10 sequence encompassing Sites A and B: attaCT GAA GAT AAT produce the mutants described in Figs. 4C–4E, site-directed mu- TTA TGC AAT CAT AAG CCA AAG ATG Ctatg. tagenesis was performed using the QuickChange kit from Strat- Gel-shift assays (EMSA) were performed as described (Vallejo et agene. The sequence of mutations introduced at Sites A, B, and C al., 1992). Briefly, 20-␮l reaction mixtures containing 1 ml of cell in the k10 promoter was identical to that of oligonucleotides Mut extract, 4 ml of albumin (10 mg/ml), 1 ml of poly(dIdC) (1 mg/ml), A, Mut B, and Mut C (see Fig. 3C and “EMSA,” below). and1mlof32P-labeled dsDNA probe (20,000 cpm) were prepared at

a 50 mM final salt concentration and allowed to equilibrate for 20 from Dr. Dennis Roop, Baylor College of Medicine, Houston, TX) min at 25°C. Complexes were separated on 5% Tris/borate acryl- that could be employed as a positive control for PCR. As a control, amide gels, and the dried gel was exposed to Kodak X-Omat AR each cDNA sample was amplified using primers to mouse ␤-actin film overnight. (Clontech, Palo Alto, CA). Southern blots were probed, after transfer of RT-PCR products to nitrocellulose membranes, using a mixture of 32P-end-labeled oligonucleotides complementary to the Immunocytochemistry internal region of the amplified K10 product. Mouse skin was frozen in OCT compound (Miles, Inc.), cut into 5-␮m sections, and permeabilized in methanol prior to staining. For Western Analysis the immunostaining of cells, BALB/MK cells were grown in four- well plastic chamber slides (Nunc, Inc., Naperville, IL), fixed 5 min Western immunoblotting was performed using polyclonal K10 in 1% paraformaldehyde, and permeabilized in methanol followed antisera at a 1:5000 dilution (gift from Dr. Stuart Yuspa, NCI, by 0.1% Triton X-100 to promote access to nuclei (Collier and Bethesda; also available from Berkeley Antibody Co.) as previously Schlesinger, 1986). For paraffin-embedded skin, 5-␮m sections were described (Maytin and Habener, 1998). Films were quantified with deparaffinized by successive immersion in xylene, graded alcohols, a 2-D laser scanning densitometer (Molecular Dynamics, Sunny- and saline. All specimens were blocked in 3% normal donkey vale, CA). serum, incubated 12–18 h with primary antisera at 4°C, rinsed in saline, and incubated4hatroom temperature with the secondary antibody. Slides were mounted in fluorescence mounting medium RESULTS (Kirkegaard & Perry Laboratories, Gaithersburg, MD) and viewed under a Nikon Optiphot-2 microscope with epifluorescence attach- AP-2, C/EBPs, and Keratin K10 Are Differentially ments. For most experiments, rabbit polyclonal antisera were used Expressed in Stratifying Epidermis as follows (dilution, commercial source): C/EBP␣, C/EBP␤, gadd153/CHOP (1:50; Santa Cruz Biotechnology, Santa Cruz, CA), The epidermis consists of morphologically distinct cell AP-2 (1:500, Santa Cruz, sc-184), and keratin K10 and keratin K14 layers, notable for their unique patterns of expression of (1:500 and 1:2000, Berkeley Antibody Co., Berkeley, CA). Goat K10 and of the transcription factors AP-2, C/EBP␣, and polyclonal antisera against AP-2␣ (1:100, Santa Cruz, sc-6312) or a C/EBP␤ (Fig. 1). Architecturally, a single proliferative layer mouse monoclonal antibody against keratin K10 (1:10, Sigma clone of basal cells is surmounted by differentiating suprabasal 8.6) were used in Fig. 1 only. The secondary antibody for most cells, which eventually undergo apoptosis and become experiments was Cy3-conjugated donkey anti-rabbit IgG (1:1500). cornified squames (see Fig. 1A and diagram). AP-2 is highly For double staining, Cy2-conjugated donkey anti-goat or anti- mouse IgG were used (Jackson ImmunoResearch, West Grove, PA). expressed in the basal layer (Fig. 1B), with most of the ␣ Antiserum sc-184 against AP-2 protein is not entirely specific for immunofluorescent signal coming from AP-2 (Mitchell et AP-2␣. On Western blots loaded with roughly equimolar amounts al., 1991; and see Fig. 1EЈ) although some contribution from of recombinant, 35S-labeled AP-2␣, AP-2␤, or AP-2.2 (AP-2␥), the AP-2␤ and AP-2␥ cannot be excluded (see Materials and sc-184 antibody recognizes all three proteins, although its affinity Methods). In contrast, K10 appears only in the suprabasal for AP-2␣ is severalfold higher than for AP-2␤ or AP-2␥ (P. J. layers (Fig. 1C). C/EBP␣ is found in the upper suprabasal Mitchell, data not shown). layers (Fig. 1D), whereas C/EBP␤ is found in low amounts in Images were captured with a digital CCD camera (Optronics the basal layer and increases at the transition zone between Model TEC-470 camera) attached to a Macintosh 7100 PowerPC basal and suprabasal layers (Fig. 1E). C/EBP␤ and AP-2 are computer with RasterOps framegrabber, running IPLab Spectrum both expressed in the basal layer, but levels of these two software (Signal Analytics Corp., Vienna, VA). Integration of signal intensities, background subtraction, and normalization were per- factors change in a reciprocal manner; i.e., as cells migrate ␤ formed as described (Maytin and Habener, 1998). upward into the cornified epithelium, C/EBP increases Of particular note, the immunofluorescent staining patterns and AP-2 decreases (Figs. 1E and 1EЈ). obtained with C/EBP-specific antisera used here were previously Immunocolocalization studies can establish putative re- shown to accurately reflect the relative distribution of each C/EBP lationships between the expression of transcription factors protein within cells, both for keratinocytes in vitro and for the and potential target genes such as k10. Thus, AP-2 colocal- epidermis in vivo, using Western blotting techniques (Maytin and izes with C/EBP␤ in the basal layer (Figs. 1E and 1EЈ), where Habener, 1998). K10 expression is absent. At the basal/suprabasal boundary, K10 expression begins; cells expressing K10 display an RT-PCR and Southern Analysis increase in C/EBP␤ relative to the basal layer (Figs. 1F, 1FЈ, and 1FЉ). In the high suprabasal (granular) layers, where ␮ Equal amounts of total RNA (5 g), prepared from BALB/MK expression of C/EBP␤ begins to decline, K10 expression cells or from newborn mouse skin (guanidinium/phenol reagent, colocalizes with C/EBP␣ (cf. Figs. 1D and 1F). TelTest “B”, Inc., Friendswood, TX) were reverse-transcribed using 1 ␮l oligo(dT) primers and the Superscript II kit (Gibco BRL), following the manufacturer’s instructions. As a negative control, The Promoter of the k10 Gene Contains Three reverse transcriptase was omitted from the reaction (data not Functional C/EBP Binding Sites shown). Aliquots of cDNA were then analyzed by PCR, using primers to amplify a 220-bp region at the 3Ј end of the K10 mRNA. To determine whether C/EBPs might be involved in the This region corresponded to a K10 cDNA insert in pGEM-3 (gift upregulation of k10 expression during keratinocyte differ-

FIG. 1. Spatial relationships of C/EBPs, AP-2, and K10 in the stratifying epidermis. Frozen sections from the skin of newborn mice (5 days old) were immunostained as described under Materials and Methods. Relative locations of the basal (B), suprabasal (SB), and cornified (C) layers of the epidermis are shown in the line drawing (upper left) and in (A), a phase-contrast image. In A–FЉ, the boundary between epidermis and dermis is indicated by dashed lines, between living epidermal layers and the dead cornified layer by dotted lines, and between basal and suprabasal epidermal layers by horizontal arrowheads. Specificities of the antibodies used are indicated on the left, where recombinant C/EBP␣ or C/EBP␤ expressed in cos cells, recombinant AP-2␣ expressed in reticulocyte lysates, and native K10 expressed in differentiating mouse keratinocytes are seen as bands of the correct size on Western blots. (B) AP-2 in the basal layer, using the sc-184 antibody. (C) K10 in the suprabasal layers. (D) C/EBP␣, abundant in the highest suprabasal layers. (E) C/EBP␤, moderately expressed in the basal layer and highly abundant in the suprabasal layers. (EЈ) The same field shown in E, but now stained for AP-2␣, using the sc-6312 antibody. Arrows in E and EЈ are for orientation. ns, nonspecific staining in the cornified layers. (F–FЉ) A single specimen double stained for C/EBP␤ (F), and for K10 (FЈ), with the two images superimposed by digital overlay in (FЉ). Scale bar, 20 ␮m.

entiation, a k10 promoter fragment was isolated for use in ties were found with the human (Rieger and Franke, 1988) binding and transactivation studies in BALB/MK cells, an and bovine (Rieger et al., 1985) k10 promoters. An identity immortalized keratinocyte line (Weissman and Aaronson, of 95–100% nucleotides is present in the first 170 bp, a 1983) that reproduces many of the differentiation-related sequence containing three C/EBP binding sites (designated changes that occur in the epidermis, including the sequen- A, B, and C). All three sites (Site B oriented in the reverse tial upregulation of C/EBP␤ and C/EBP␣ (Maytin and Ha- direction) closely resemble the consensus binding motif for bener, 1998). The BALB/MK cells, exposed to high-calcium C/EBPs (Fig. 3A, bottom). To determine whether these sites (differentiation-promoting) medium, respond with an in- bind C/EBPs, DNase I protection analyses of the k10 crease in K10 expression within 24 h, measurable at the promoter were performed (Fig. 3B). Recombinant C/EBP␤ protein (Fig. 2A) and mRNA (Fig. 2B) levels. These results protected a region extending from Ϫ140 to Ϫ176 bp (Fig. agree with earlier findings in which K10 increases in 3B). A hypersensitive region was found within this foot- primary murine keratinocytes at 1–2 days of differentiation print, between Sites A and B (Fig. 3B, asterisk). Site C (Yuspa et al., 1989). corresponded to a location that showed weak protection The cloned segment of the 5Ј flanking region of mouse from DNase I, but was flanked by two strong hypersensitive k10 revealed several features (Fig. 3A). Nucleotide similari- sites consistent with a possible C/EBP␤-induced conforma-

C/EBP␤ Is Present in Keratinocyte Nuclear Extracts and Binds to the k10 Promoter in a Differentiation-Specific Manner To show that C/EBPs are present within nuclei and bind to the k10 promoter in a differentiation-specific manner, proteins were extracted from the nuclei of BALB/MK cells at various times during calcium-induced differentiation and analyzed for C/EBP-binding activity via gel-shift analysis. Nuclear extracts were mixed with a 44-bp probe containing Sites A and B from the k10 promoter. Two specific complexes which increased in intensity over the first 3 days of differentiation, and then decreased by 7 days, were observed (Fig. 3D). The composition of these complexes was examined via immunosupershift and oligonucleotide competition experiments (Fig. 3D). Specific complexes were disrupted by antisera against C/EBP␤ (lane 6), but not by antisera to C/EBP␣ nor CHOP. DNA competition with the M6 oligonucleotide, a high-affinity C/EBP-binding site (Brasier et al., 1990), selectively abolished the specific complexes (lanes 9 and 10), whereas competition with the unlabeled k10 probe itself competed both the specific and FIG. 2. Expression of the k10 gene is increased in a the nonspecific complexes (lanes 11 and 12). These results differentiation-associated manner in calcium-induced keratino- implicate C/EBP␤ as a predominant factor responsible for cytes. Confluent cultures of BALB/MK cells were switched to differentiation-dependent binding activity at the C/EBP- high-calcium medium for the times indicated and harvested for total cellular proteins and for RNA. (A) Western immunoblot binding sites within the k10 promoter during the differen- analysis of BALB/MK proteins, using an antibody to keratin K10; tiation of BALB/MK cells. molecular size (kilodaltons) shown at left. (B) Analysis of K10 mRNA expression; molecular sizes (kilobases) shown at left. (Top) C/EBP␤ Introduced into Keratinocytes Activates RT-PCR using primers to a region at the 3Ј end of the K10 mRNA; the k10 Promoter through a Mechanism That Is ethidium-stained agarose gels are shown. (Bottom) Southern blots C/EBP-Specific and DNA Binding Site-Specific from the same gels, probed with a 32P-labeled K10 probe. Controls, PCRs were run with a plasmid containing a K10 cDNA insert To test whether C/EBPs can regulate k10 transcription in (pK10) and with cDNA prepared from mouse epidermal RNA vivo, a reporter plasmid containing the 245-bp wild-type (skin). (C) RT-PCR products from the same RNA samples used in k10 promoter fused to the luciferase reporter gene (B), using primers specific for mouse actin. pActin, a control from (pWtK10Luc) was cotransfected into BALB/MK cells with PCR amplification of a plasmid containing mouse actin sequences. expression plasmids for C/EBP␣, C/EBP␤, or CHOP (Fig. 4). Expression of both C/EBP␣ and C/EBP␤ activated pWtK10Luc by 6- to 12-fold relative to the empty expression vector, pCDNA (Fig. 4A). C/EBP␣ was more effective tional change in the DNA (Fig. 3B, upper two asterisks). The than C/EBP␤ in activating the k10 reporter. CHOP, as individual DNA sequence motifs were tested in gel-shift expected from its inability to bind DNA (Ron and Habener, assays for their ability to bind C/EBPs (Fig. 3C). 32P-labeled 1992), failed to activate the k10 reporter. The empty lucif- oligonucleotides, corresponding to each of the three motifs, erase reporter vector alone, pOLuc, was not affected by any bound recombinant C/EBP␤ in a C/EBP-specific manner, as of the expression constructs, confirming that transactiva- demonstrated via competition with a high-affinity C/EBP- tion effects observed were attributable to k10 promoter binding oligonucleotide and via supershift studies with sequences. antisera against C/EBP␤ (Fig. 3C). Oligonucleotides with Additional transfection/transactivation experiments were mutations in each of the sites (Mut A, Mut B, or Mut C) performed to test the specificity of interactions between k10 failed to compete for C/EBP␤ binding and failed to bind and C/EBP. Cotransfection of full-length 32-kDa C/EBP␤ C/EBP␤ when used as probes (Fig. 3C). The affinity of (LAP) and pWtK10Luc resulted in transactivation of the k10 binding to C/EBP␤ differed for each of the sites, with the promoter (Fig. 5B, left side). A fivefold excess of the 20-kDa affinity at Site A being 5- to 10-fold higher than for sites B inhibitory form of C/EBP␤ (LIP), introduced into the cells and C, based upon differential film exposure times (data not along with LAP, attenuated the transactivation of k10. Simi- shown). However, the affinities of the sites measured in larly, overexpression of the dominant-negative inhibitor, isolation (Fig. 3C) do not address the issue of cooperative CHOP, also inhibited transactivation. These effects depended binding effects, observed when the sites lie adjacent to one on the presence of intact C/EBP-recognition sites in DNA, another in native k10 DNA (see below). because a k10 promoter construct (pMutABC) containing

FIGURE 3

FIG. 3—Continued

Copyright © 1999 by Academic Press. All rights of reproduction in any form reserved. 172 Maytin et al. point mutations at Sites A, B, and C failed to respond to reduction in C/EBP␤-stimulated activity of the k10 pro- C/EBP␤, either in the presence or in the absence of excess moter. Elimination of either Site B or Site C caused smaller inhibitory proteins (Fig. 5B, right). reductions in k10 activity. Double mutants confirmed the relative importance of Site A; constructs with a mutation in B and A or C and A, but not B and C, showed significant Full Activation of the k10 Promoter by C/EBP␤ decreases in k10 activation by C/EBP␤. Importantly, how- Requires Three Intact C/EBP-Binding Sites ever, only the triple mutation (Mut ABC) showed 100% loss within the Promoter of C/EBP␤-inducibility. Therefore, despite the higher affin- The role of each C/EBP-binding site in the k10 promoter ity of Site A relative to Sites B or C when tested in isolation was examined by creating point mutations within Sites A, (Fig. 3C), all three sites appear to be important in the native B, and C (Figs. 4C–4E). Each mutation abolished binding of promoter, i.e., all must be intact to achieve full activation C/EBP␤ at the individual sites; see Fig. 3C. Mutations were of the k10 promoter in response to C/EBP␤. introduced into the k10 promoter either individually or together in all combinations, to yield a panel of mutant k10 Targeted Disruption of the c/ebp␤ Gene Results transcriptional reporters (Fig. 4C). These reporters were in Ectopic Increases in Expression of C/EBP␣ tested in cotransfection experiments with a C/EBP␤ expres- and K10 in the Basal Layer of the Epidermis sion vector (Fig. 4D). The effects of these mutations upon transcriptional activity proved to be complex, with loss of Having established a role for C/EBP␤ in the regulation of any one of the sites leading to derepression of basal tran- the k10 promoter in vitro, we investigated its role in the scription and a concomitant loss in relative C/EBP regulat- skin in vivo. A targeted disruption of the c/ebp␤ gene in ability. The latter point is best seen in Fig. 4E, in which the mice, which led to the development of lymphocyte abnor- mutants are displayed in decreasing order of ability to malities in aged mice, was described previously; no pheno- respond to C/EBP␤. Thus, in comparison to the fourfold typic changes were reported in the skin (Screpanti et al., induction of the wild-type k10-luciferase reporter after 1995). When these mice were backcrossed into a FVB/n C/EBP␤, mutation of Site A alone caused only a 50% background, a skin phenotype appeared (Fig. 5). The c/ebp␤

FIG. 3. The mouse k10 promoter contains three C/EBP binding motifs which specifically bind C/EBPs, including C/EBP␤ found in keratinocyte nuclear extracts. (A) Comparison of the murine k10 proximal promoter region with the corresponding regions of human and bovine k10 genes; locations of C/EBP-binding motifs. Nucleotides conserved between at least two of the three species are shaded. The TATA-binding site, the transcription initiation site (TRX) and the translation start site (TRL) are boxed. Sites A, B, and C, representing candidate sites for binding of C/EBPs, are delineated by boxes and are also shown at the lower right beneath the consensus motif for C/EBP binding (Osada et al., 1996). The sequence labeled “mK10 in 1985” denotes the 5Ј end of the published sequence of the murine k10 gene (Krieg et al., 1985); bases that differ from our sequence are shown, while identical bases are indicated by asterisks. (B) DNase I protection analysis of the K10 promoter. A 245-bp fragment of the murine k10 promoter was labeled with 32P at the 5Ј terminus and incubated with increasing amounts (0, 50, 100, 200, 400, and 800 ng) of a recombinant 20-kDa isoform (LIP) of C/EBP␤ and then separated on an acrylamide gel and exposed to film. Footprinted regions (vertical arrows, on right) encompass the C/EBP␤-binding sites predicted from A (vertical boxes). Hypersensitive sites are indicated by asterisks. Nucleotide locations, derived from the sequencing lane (G-ladder), are given on the left. (C) Gel-shift analyses demonstrating the affinity and specificity of binding for each of three sites (A, B, and C) in the K10 promoter to a recombinant C/EBP␤. In the three gels on the right, three different 32P-labeled 26-bp oligonucleotides derived from the wild-type k10 promoter sequence (wildtype Site A, B, or C; see A and Materials and Methods) or 26-bp oligonucleotides with mutations at sites A, B, or C (mA, mB, or mC; also see Fig. 4C and Materials and Methods) were incubated with 1 ␮l of a gst-C/EBP␤ bacterial lysate or with the probe alone. In the gel on the left, a standard high-affinity C/EBP-binding site (Standard; the M6 mutant from the angiotensinogen gene) (Brasier et al., 1990) served as the probe. Locations of DNA–protein complexes (C/EBP␤ϩDNA) and free probe are indicated. In all gels, numbered lanes indicate the following successive additions (also see labeling above the gels): (1) probe alone, (2) gst-C/EBP␤ protein, (3) 100-fold excess of unlabeled C/EBP␤ standard (S) oligonucleotide, (4) 100-fold excess of unlabeled probe itself (A, B, or C), (5) 100-fold excess of unlabeled mutant oligonucleotide (mA, mB, or mC), (6) antibody against C/EBP␤. In lane (7), gst-C/EBP␤ protein was incubated with a 32P-labeled mutant probe (mA, mB, or mC) to demonstrate absence of binding at the mutated sites. Film exposure time was 40 h for the Site A probes, 7 days for the Site B and Site C probes. (D) Characterization of native C/EBP-binding activity in differentiating BALB/MK cells. Time course: Nuclear extracts, 1 ␮l, from cells exposed to 0.12 mM calcium for 0, 3, or 7 days were mixed with a radiolabeled 40-bp oligonucleotide (K10 probe) containing Sites A and B from the K10 promoter (see Materials and Methods). Changes in two specific complexes (SC) and a nonspecific complex (NS) were observed. In the subsequent experiments, 1 ␮l of 3-day nuclear extract was used in all lanes. Supershift analysis with antisera: Addition of either 1 ␮l of anti-C/EBP␣ antisera (lane 5), or anti-CHOP antisera (lane 7), had no effect, but anti-C/EBP␤ antisera (lane 6) perturbed the specific complexes and formed supershifted complexes instead. Oligonucleotide competition: Addition of a 10-fold (lane 9) or 100-fold (lane 10) excess of unlabeled standard oligonucleotide (M6) eliminated binding only of the specific C/EBP-containing complexes. The K10 probe itself (K10, lanes 11 and 12) competed both the C/EBP-specific and the nonspecific bands, as expected. Supershift (SS) controls: Positive controls to show that the antisera are capable of supershifting recombinant gst-C/EBP␣ (lanes 13 and 14) or C/EBP␤ (lanes 15 and 16). The M6 probe (Standard) was described in C.

FIG. 4. A k10 promoter–luciferase reporter can be transactivated by different C/EBP isoforms, in a manner requiring three C/EBP-binding motifs in the DNA. (A) Expression constructs (0.5 ␮g) containing the protein-coding sequences for C/EBP␣, C/EBP␤, or CHOP, or the empty pCDNA vector (Ϫ), were cotransfected into BALB/MK cells along with 5 ␮g of a luciferase reporter containing either the wild-type K10 promoter (pWtK10Luc) or reporter vector alone (pOLuc). Cells were harvested at 48 h for luciferase assay. Data are expressed as relative luciferase activity (arbitrary units). Induction ratios, relative to empty pCDNA, are shown above some of the bars. Values represent the mean Ϯ SEM for three experiments, each done in triplicate and normalized to the empty-vector controls. (B) The wild-type reporter (pWtK10Luc) or a mutant reporter in which all three C/EBP-consensus sites have been altered (pMutABCLuc) was cotransfected with vectors encoding full-length C/EBP␤ (LAP), truncated C/EBP␤ (LIP), or CHOP (Chop) or with pCDNA alone (Ϫ). By itself, C/EBP␤ induced significant K10 promoter activity, but when C/EBP␤ was cotransfected with a five-fold excess (2.5 ␮g) of either LIP or CHOP, the C/EBP␤-induced activation was abolished. All values are from the same experiment. Each bar represents the mean Ϯ SD, for dishes in triplicate. (C) Nomenclature (left side) and arrangement of mutated sites (boxes), for the k10–luciferase reporters used in these studies. Each k10 construct was 245 bp in length, with mutations in the k10 promoter introduced at Sites A, B, and/or C as indicated; see Materials and Methods for sequence. (D) Transactivation of wild-type or mutant k10–luciferase constructs; pOLuc, empty SKLuc reporter vector. Reporter constructs (5 ␮g) were cotransfected into BALB/MK cells together with 0.5 ␮g full-length C/EBP␤ in pCDNA (solid bars), or with pCDNA alone (open bars), and harvested at 48 h for luciferase assay. Raw data were pooled from three experiments, each condition in duplicate or triplicate. Error bars represent the means Ϯ SEM. For each K10 mutant, the relative ratio (fold increase) due to addition of C/EBP␤ is given above the bar. (E) Induction ratios for the wild-type and mutant k10–luciferase constructs, displayed in order of decreasing ability to respond to C/EBP␤. Error bars represent means Ϯ SEM. Asterisks, significant difference from the wild-type construct, WtK10Luc, by one-sided Student t test (*P Ͻ 0.01; **P Ͻ 0.001). null mice were slightly smaller than controls and mani- 5A and 5AЈ). However, substantial (ϳ50%) thinning of the fested a scruffy appearance of their coats (Fig. 5, top left). subcutaneous fat layer was observed (see Fig. 5AЈ and figure The phenotypic changes in the fur allowed identification of legend), consistent with the decreases in interscapular and homozygous animals before confirmation by genotyping. epididymal adipose tissue previously reported in c/ebp␤ The histomorphology of hematoxylin- and eosin-stained null mice (Tanaka et al., 1995). skin in the c/ebp␤ null-mice appeared similar to that of Closer examination of skin sections using a panel of controls; no obvious changes in the epidermis and dermis, antisera directed against transcription factors and keratins nor in the size and number of hair follicles, were seen (Figs. revealed major differences between c/ebp␤ null and control

FIG. 5. Gross appearance, standard histology, and immunohistochemical analysis of transcription factor and keratin expression in the skin of c/ebp␤ null mice. (Top) Photograph of a pair of littermates (6 weeks old, both females), heterozygous (ϩ/Ϫ, left) or homozygous (Ϫ/Ϫ, right) for the targeted deletion in the c/ebp␤ gene. A “scruffy” appearance of the coat is evident in the c/ebp␤ Ϫ/Ϫ mouse. (A, AЈ), H&E stains of dorsal skin from a c/ebp␤ ϩ/Ϫ and a c/ebp␤ null animal, respectively. The thickness of each of the indicated layers was measured digitally, from the skin of 5 heterozygous versus 5 nullizygous mice, as follows (mean Ϯ SEM): e, epidermis (18.2 Ϯ 1.4 ␮mvs 19.4 Ϯ 1.6 ␮m); d, dermis (315 Ϯ 34 ␮mvs291Ϯ 24 ␮m); a, adipose layer (170 Ϯ 24 ␮mvs95Ϯ 13 ␮m); m, muscle (not measured). The adipose layer was significantly thinner in the c/ebp␤ Ϫ/Ϫ mice, Student t test, P Ͻ 0.0005. Scale bar, 100 ␮m. (B–G) Frozen skin sections from pairs of mice (left, c/ebp␤ ϩ/Ϫ, right, c/ebp␤ Ϫ/Ϫ), mounted on the same slide and analyzed by immunofluorescence using the specific antiserum indicated. Photographic exposure times were identical for each pair of images. Dashed lines, dermal–epidermal junction. Dotted lines, junction between living epidermis and stratum corneum. Horizontal arrows, top of the basal layer. Scale bar, 20 ␮m.

mice (Fig. 5). C/EBP␤ protein was absent in the epidermis of creased in both concentration (fluorescent intensity) and c/ebp␤ null mice (Figs. 5B and 5BЈ). Although CHOP geographic area (Fig. 5DЈ). These changes in expression of expression remained unchanged (Figs. 5C and 5CЈ), a C/EBP␣ were significant when the skins of multiple ani- marked increase in the distribution of C/EBP␣ expression mals were studied by computer-assisted image analysis was observed within the epidermal layers (Figs. 5D and (Table 1). Notably, in the epidermis of c/ebp␤ null mice, the 5DЈ). Compared to the normal pattern of C/EBP␣ expression expression of K10 was also markedly increased (Figs. 5E and (in the high suprabasal layers), C/EBP␣ in c/ebp␤ null mice 5EЈ). The normal suprabasal pattern of K10 now expanded extended downward to include all of the cells in the downward into the basal layer (Fig. 5EЈ). Upregulation of suprabasal and basal layers. C/EBP␣ expression was in- K10 appeared to be specific, because the expression levels of

TABLE 1 Integrated Fluorescence Intensitiesa of Epidermal Transcription Factorsb and Keratins in the Skin of Adult c/ebp␤ Knockout Mice

CHOP C/EBP␣ K14 K10 Genotype Pair I.D. No. Sex (cebp␤) Mean Ϯ SD Mean Ϯ SD Mean Ϯ SD Mean Ϯ SD

110Fϩ/Ϫ 527 Ϯ 66 343 Ϯ 88 4551 Ϯ 810 3462 Ϯ 414 11 F Ϫ/Ϫ 539 Ϯ 73 574 Ϯ 78** 4278 Ϯ 1433 8521 Ϯ 976*** 212Mϩ/Ϫ 536 Ϯ 66 322 Ϯ 31 2976 Ϯ 548 2668 Ϯ 491 14 M Ϫ/Ϫ 507 Ϯ 134 705 Ϯ 82** 3167 Ϯ 479 3785 Ϯ 547** 315Fϩ/Ϫ 477 Ϯ 89 311 Ϯ 107 3023 Ϯ 287 3531 Ϯ 380 17 F Ϫ/Ϫ 566 Ϯ 119* 603 Ϯ 60** 3153 Ϯ 208 4691 Ϯ 711** 419Mϩ/Ϫ 502 Ϯ 44 347 Ϯ 72 2091 Ϯ 402 2476 Ϯ 1194 20 M Ϫ/Ϫ 457 Ϯ 101 618 Ϯ 135** 2006 Ϯ 474 4718 Ϯ 796**

a Histologic specimens (5-␮m frozen sections) from the skin of paired littermates, laid adjacent to one another on microscope slides, were immunostained with primary antisera to the proteins indicated, followed by Cy3-conjugated secondary antisera, and fluorescence intensity was analyzed as described under Materials and Methods. Values shown represent the means Ϯ SD (n ϭ 6 fields/animal) of integrated fluorescence intensity per unit length of epidermis. Unit length was used for normalization rather than unit area, since epidermal thickness can vary considerably. Statistical comparison by the Student t test (P value) is shown for each pair of animals. b No values are listed for C/EBP␤ because quantitation was not required to demonstrate the all-or-none difference between strong C/EBP␤-specific signals in controls versus the complete absence of signal in the c/ebp␤ Ϫ/Ϫ mice. * P Ͻ 0.01. ** P Ͻ 0.005. *** P Ͻ 0.0005. another epidermal keratin, K14, remained unchanged (Figs. binding motifs by selective mutation completely abolished 5F and 5FЈ). C/EBP␤-inducibility of the AP-2 promoter (Fig. 6).

AP-2 Resides in a Pathway between C/EBP␤ ap-2␣ Nullizygous Mice Reproduce the Skin and C/EBP␣, Linking the Increase of C/EBP␣ Phenotype of c/ebp␤ Null Mice, with Increased to the Loss of C/EBP␤ in the Epidermis Epidermal C/EBP␣ and K10 ␤ of c/ebp Null Mice To determine whether c/ebp␤ and ap-2␣ may lie in a As an explanation for the increase in C/EBP␣ expression pathway upstream of c/ebp␣ and k10, epidermis from mice in c/ebp␤ null skin (e.g., active stimulation versus removal lacking the ap-2␣ gene was examined (Schorle et al., 1996). of repression), we considered repression of c/ebp␣ gene Experiments were limited to late-stage fetal skin, since expression by AP-2. Such an explanation would have to ap-2␣ null mice die perinatally with severe cranial and include a role for C/EBP␤ in the maintenance of AP-2 abdominal wall defects (Schorle et al., 1996). However, expression, in order to explain derepression of C/EBP␣ in these mice develop intact skin over the upper thoracic the skin of c/ebp␤ null mice. Two experiments were region (Zhang et al., 1996). Frozen sections from 17.5- to performed, the results of which support such a role. First, 18.5-day embryos were tested using the panel of antisera immunostaining demonstrated a near complete disappear- shown in Fig. 7. As with c/ebp␤ null mice, the ap-2␣ null ance of AP-2 in the epidermis of c/ebp␤ null mice, consis- mice showed an overall increase in C/EBP␣ staining inten- tent with the notion that C/EBP␤ is essential for AP-2 sity (Figs. 7B and 7BЈ and Table 2) and a marked increase in expression (Figs. 5G and 5GЈ). Second, an examination of C/EBP␣ expressed ectopically in basal cell nuclei (Figs. 7C regulatory sequences upstream of the AP-2␣ gene (Meier et and 7CЈ). K10 expression was increased in the suprabasal al., 1995) revealed several potential C/EBP-binding site layers (Figs. 7E and 7EЈ and Table 2), although no expansion motifs. Two of these motifs, located at nucleotide positions into the basal layer was noted in this case. No changes were 315 and 365 (GenBank Accession No. U17289), were tested seen in the hair follicles. The levels of expression of C/EBP␤ for their activity as C/EBP␤-responsive enhancers of tran- (Figs. 7D and 7DЈ), K14 (Figs. 7F and 7FЈ), and CHOP (data scription. An 87-bp sequence of the AP-2 promoter contain- not shown) were unchanged. ing these two motifs was linked to a transcriptional reporter comprising the K10 minimal promoter fused to the lucif- DISCUSSION erase gene (see Materials and Methods). When cotransfected into keratinocytes in the absence or presence of an expres- An important goal in epidermal biology is to identify sion plasmid for C/EBP␤, the AP-2 promoter was activated differentiation-speciﬁc transcription factors that determine in a C/EBP-dependent manner (Fig. 6). Disruption of the the expression of tissue-speciﬁc target genes during kera-

al., 1988; Yuspa et al., 1989), the regulation of the k10 promoter has not been examined in detail. We identified a cluster of three C/EBP-binding sites within the proximal k10 promoter via footprint and gel-shift analysis. All three sites were shown by site-directed mutagenesis studies to be essential for C/EBP responsiveness of the k10 promoter (Figs. 3 and 4). The presence of these sites appears to maintain the k10 promoter in a relatively suppressed state at baseline, since mutation of any one of the three sites increases basal promoter activity in keratinocytes (Fig. 4D). Whether this reflects a physical interaction between the sites through bending of the chromatin (e.g., hairpin forma- tion in the presence of C/EBPs and/or other binding proteins), as shown for the case of multiple CREB-responsive elements (Spiro et al., 1995), is an interesting hypothesis for future study. For purposes here, characterization of the C/EBP-responsive sites was essential to confirm that the k10 promoter can be regulated by C/EBPs at the transcriptional level. Multiple C/EBP-responsive motifs have been demonstrated experimentally in only a few other genes to date, ␤ FIG. 6. C/EBP -responsive enhancer elements are found in the including keratin K6 (Blumenberg et al., 1998) and ␣-1 acid ␣ upstream regulatory region (exon 1a) of the ap-2 gene. Luciferase glycoprotein (Lee et al., 1993). For the k10 promoter, such reporter constructs were constructed by placing an 87-bp region of redundancy may preserve transcriptional expression of this the ap-2␣ promoter containing two putative C/EBP-enhancers at Ϫ710 and Ϫ760 bp (pWtAP2Luc) in front of the luciferase gene vital intermediate filament gene. The disease epidermolytic driven by a minimal promoter (LUC); see Materials and Methods. hyperkeratosis illustrates the importance of proper k10 In a mutant version of the construct, both AP-2 enhancers were expression in cutaneous biology. K10 normally assembles disrupted by site-directed mutagenesis (pMutAP2Luc; mutant into cytoskeletal filaments as a heterodimer with its spe- bases in lowercase italic). Each of these constructs, or the empty cific partner, keratin K1; together, K1 and K10 constitute reporter (pOLuc), was cotransfected with C/EBP␤-expression plas- ϳ85% of the protein in fully differentiated keratinocytes mid (hatched bars) or empty pCDNA (open bars) into BALB/MK (Cheng et al., 1992). Perturbation of K1:K10 filaments via cells, and the cells were harvested at 48 h for luciferase assay. Data dominant-negative mutations in K10, either introduced were pooled from three experiments, each performed in triplicate, experimentally into transgenic mice (Fuchs et al., 1992) or and normalized using the baseline (empty vector) values; error bars naturally occurring in the heritable disease epidermolytic represents the means Ϯ SEM. hyperkeratosis (Cheng et al., 1992; Rothnagel et al., 1992; tinocyte differentiation (reviewed in Fuchs, 1993). Here we Syder et al., 1994), leads to epidermal cytolysis and blister- show that C/EBP and AP-2 transcription factors together ing, with severe clinical consequences. regulate the transcriptional activity of the keratin k10 gene in the epidermis. C/EBPs act directly upon the k10 pro- AP-2 Acts Indirectly upon k10 Expression moter. Although C/EBP␣ and C/EBP␤ can both activate the via Activation or Repression of the c/ebp␣, k10 gene, they are differentially expressed in the stratified c/ebp␤, and ap-2␣ Gene Promoters epidermis. C/EBP␤ acts early and C/EBP␣ acts later during ␣ ␤ upward migration and differentiation of keratinocytes. To determine how C/EBP and C/EBP regulate k10 AP-2 influences k10 transcription indirectly as a repressor expression in the skin in vivo, we examined the skin of ␤ ␤ of C/EBP␣ expression in the basal layer of the epidermis, mice lacking c/ebp .Inthec/ebp null animals, the k10 where C/EBP␤ and AP-2 are both expressed. In the high phenotype of increased expression was initially difficult to ␣ suprabasal layers, AP-2 expression is extinguished and explain. However, it was noted that C/EBP transactivates ␤ C/EBP␤ expression is superceded by expression of C/EBP␣. the k10 promoter (Fig. 4A). In the c/ebp null mice, loss of ␤ A novel aspect of the findings is that c/ebp␣, c/ebp␤, and C/EBP is compensated for by an increase in epidermal ␣ ␣ ap-2␣ genes are functionally linked in a regulatory network C/EBP . It is reasonable to propose that C/EBP activates ␤ that ultimately determines the expression of the k10 gene transcription of the k10 gene in c/ebp null mice because ␣ in the skin in vivo. C/EBP and K10 are both expressed ectopically in cells of the basal layer, a place where neither gene is normally expressed. C/EBPs Directly Regulate k10 Gene Transcription Why is C/EBP␣ upregulated in the skin of c/ebp␤ null Although the suprabasal expression pattern of keratin mice? Among several possible mechanisms, a release from k10 in the epidermis has been well characterized (Roop et transcriptional repression was considered. Analogous to the

FIG. 7. Analysis of transcription factors and epidermal keratins in the skin of ap-2␣ null mice. Frozen sections from the skin of wild-type (ap-2␣ ϩ/ϩ) or nullizygous (ap-2␣ Ϫ/Ϫ) mice, analyzed by immunoﬂuorescence using the antisera indicated. Dashed lines, dermal– epidermal junction. Dotted lines, top of living epidermis. Horizontal arrows, top of the basal layer. Magniﬁcation: The white bar in AЈ (scale bar, 20 ␮m) applies to all images except C and CЈ, for which the scale bar in C represents 10 ␮m. b, basal cell.

role of AP-2 as a repressor of the c/ebp␣ promoter in mitotic TABLE 2 preadipocytes (Jiang et al., 1998), we hypothesized that Integrated Fluorescence Intensitiesa of Epidermal C/EBP␣ and AP-2 may be a repressor of C/EBP␣ in the basal epidermis. K10 in the Skin of Embryonic ap-2␣ Knockout Mice This hypothesis is supported by our findings that in the C/EBP␣ K10 skin of ap-2 null animals, expression of C/EBP␣ is increased Genotype (Fig. 7). Also, because AP-2 expression is lost in the epider- Pair I.D. No. (ap-2␣) Mean Ϯ SD Mean Ϯ SD mis of c/ebp␤ null mice, loss of AP-2-mediated repression ϩ Ϫ Ϯ Ϯ can explain the high C/EBP␣ levels seen in those animals 1a3 / 2519 350 2683 121 Ϫ Ϫ Ϯ Ϯ (Fig. 5). Why is AP-2 expression lost in c/ebp␤ null mice? At a2 / 3652 419** 3470 189* 2 a17 ϩ/ϩ 2485 Ϯ 246 2826 Ϯ 137 least part of the mechanism may be transcriptional, since Ϫ Ϫ Ϯ Ϯ ␣ a16 / 3151 242** 3846 192** we have shown that the ap-2 gene promoter contains at 3 a22 ϩ/Ϫ 2314 Ϯ 163 2774 Ϯ 193 least two C/EBP-responsive enhancer elements (Fig. 6). a20 Ϫ/Ϫ 3178 Ϯ 232** 4276 Ϯ 267** Collectively, the above observations suggest a model in 4 a25 ϩ/Ϫ 2598 Ϯ 171 2351 Ϯ 287 which c/ebp␤, ap-2␣, and c/ebp␣ genes are linked in a a23 Ϫ/Ϫ 3764 Ϯ 463* 4053 Ϯ 231** regulatory network (Fig. 8). Within a given cell (e.g., a a keratinocyte in the basal layer), C/EBP␤ activates or main- Paired histologic specimens were immunostained with primary ␣ tains transcription of the ap-2␣ promoter (Fig. 8A). AP-2, in antisera to either C/EBP or K10, followed by secondary antisera conjugated to Cy3, and were then quantitated by digital image turn, represses C/EBP␣ expression. In c/ebp␤ null mice (Fig. processing as described under Materials and Methods and the legend 8B), the elimination of AP-2 expression unmasks a C/EBP- to Table 1. Values (means Ϯ SD, n ϭ 6 microscopic fields/animal) repre- ␣ binding site in the C/EBP promoter (Tang et al., 1997), sent integrated fluorescence intensity per unit length of epidermis. allowing positive autoregulation of C/EBP␣ and subsequent The Student t test results (P values) are shown for each pair of animals. upregulation of k10. Elimination of ap-2␣ by gene targeting * P Ͻ 0.001. yields a skin phenotype similar to that in Fig. 8B, with the ** P Ͻ 0.0005.

FIG. 8. Model for regulatory interactions between C/EBPs and AP-2 factors in the epidermis. (A and B) Network of interactions at the gene-promoter level, in either (A) wild-type or (B) cebp␤ null mice. Consensus elements for binding of C/EBP-family (solid boxes), AP-2 family (hatched boxes), or unidentiﬁed factors (gray boxes) are shown. (C) Spatial features of the cross-regulation among transcription factors and the k10 target gene, at different cell layers in the epidermis. B, basal layer. SB, suprabasal layer. See text for details.

exception that C/EBP␤ remains unaltered, since it lies Potential for Additional Regulatory Interactions upstream of ap-2␣. The geographic distribution of C/EBP␣, C/EBP␤, and Our data raise several additional questions. For example, why does the expression of keratin K14 remain unaltered in AP-2 within the various epidermal layers plays a role in k10 the ap-2␣ null and the c/ebp␤ null mice, despite reports expression in vivo, as summarized in Fig. 8C. Normally, that AP-2 is intimately involved in k14 transcription (Byrne C/EBP␤ is present in low amounts in the basal cells, then et al., 1994; Leask et al., 1991)? It is possible that multiple undergoes a surge of expression as the cells enter the AP-2 family genes are expressed in the skin. Epidermal suprabasal layers. C/EBP␣ is normally repressed in the basal expression of AP-2␤ (Moser et al., 1997) and of AP-2␥ layers by the presence of AP-2. Once AP-2 expression is (Buehner et al., 1998; Oulad-Abdelghani et al., 1996) has ␤ ␣ lost, in either the c/ebp or the ap-2 knockout models, been reported in mouse skin at the mRNA level. In our ␣ C/EBP expression is activated in the lower epidermis, with work, expression of several AP-2 family proteins in murine Ј qualitative increases in the basal layer (Figs. 7C and 7C ) epidermis is suggested by the persistence of faint immuno- and quantitative increases in the low suprabasal layers staining of the basal layer of the epidermis in ap-2␣ null ␣ (Tables 1 and 2). Increased C/EBP expression leads to mice using the sc-184 antibody from Santa Cruz, Inc. (data augmented k10 expression at the locations noted (Fig. 8C). not shown). The sc-184 antibody is not speciﬁc for AP-2␣, While we have shown that either C/EBP␣ or C/EBP␤ can since it also detects AP-2␤ and AP-2␥ (albeit to a lesser contribute to k10 expression (Fig. 4A), C/EBPs are clearly extent) when tested against in vitro-translated proteins on not the only factors involved. For example, k10 expression Western blots; see Materials and Methods. Thus, in mice is normally off in the basal layer, even in the presence of lacking AP-2␣, the presence of AP-2␤ and AP-2␥ may be low amounts of C/EBP␤. Therefore, additional unknown sufﬁcient to maintain k14 gene expression. In contrast, factors (designated “Z” in Fig. 8) must play a role in the AP-2␤ and AP-2␥ (assuming they are present) appear to be activation of k10 at the basal/suprabasal transition. much less effective than AP-2␣ in repressing the c/ebp␣

Copyright © 1999 by Academic Press. All rights of reproduction in any form reserved. C/EBP and AP-2 Transcription Factor Network in Skin 179 gene, since C/EBP␣ is strongly upregulated in the epidermis IL-1␣ induced, NFkB mediated transactivation of the angio- of ap-2␣ null mice. tensinogen gene acute-phase response element. EMBO J. 9, Another question is the possible involvement of alterna- 3933–3944. tive isoforms of c/ebp␣, c/ebp␤, and ap-2␣ in the regulation Brasier, A. R., Tate, J. E., and Habener, J. F. (1989). Optimized use of k10. For example, LIP (a C/EBP␤ isoform) or the 30-kDa of the firefly luciferase assay as a reporter gene in mammalian isoform of C/EBP␣ might counteract stimulatory effects of cell lines. Biotechniques 7, 1116–1122. ␥, the relevant activator C/EBP isoforms. We have demonstrated that the Buehner, A., Stolz, W., and Buettner, R. (1998). AP-2 AP-2 isoform for skin differentiation? J. Invest. Dermatol. 110, long and short protein isoforms of both C/EBP␣ and C/EBP␤ 539. are expressed in the epidermis (Maytin and Habener, 1998). Byrne, C., Tainsky, M., and Fuchs, E. (1994). Programming gene A major role for these short isoforms seems unlikely, expression in the developing epidermis. Development 120, however, because the amount of the short isoforms relative 2369–2383. to the full-length proteins is small (Ͻ20%) and because no Cao, Z., Umek, R. M., and McKnight, S. L. (1991). Regulated significant changes in the ratios of LIP to full-length expression of three C/EBP isoforms during adipose conversion of C/EBP␤ or of C/EBP␣–30 kDa to full-length C/EBP␣ were 3T3-L1 cells. Genes Dev. 5, 1538–1552. observed (data not shown). Likewise, no major changes in Chang, C.-J., Chen, H.-Y., Lei, D.-S., and Lee, S.-C. (1990). Molecu- levels of CHOP were observed in this study. However, lar cloning of a transcription factor, AGP/EBP, that belongs to because CHOP (a dominant-negative repressor of C/EBPs) is members of the C/EBP family. Mol. Cell. Biol. 10, 6642–6653. expressed in the epidermis along a gradient similar to that Cheng, J., Syder, A. J., Yu, Q. C., Letai, A., Paller, A. S., and Fuchs, of C/EBP␤ (Maytin and Habener, 1998), we cannot exclude E. (1992). The genetic basis of epidermolytic hyperkeratosis: A a counterregulatory role for CHOP that might be demon- disorder of differentiation-specific epidermal keratin genes. Cell strable in a different experimental context. In the case of 70, 811–819. Christy, R., Kaestner, K., Geiman, D., and Lane, M. (1991). AP-2, a role for multiple isoforms must also be considered. CCAAT/enhancer binding protein gene promoter: Binding of Four AP-2 isoforms, produced by alternative RNA splicing, nuclear factors during differentiation of 3T3-L1 preadipocytes. are expressed in murine skin (Meier et al., 1995). In particu- Proc. Natl. Acad. Sci. USA 88, 2593–2597. lar, a splicing variant which can form heterodimers but Collier, N. C., and Schlesinger, M. J. (1986). The dynamic state of lacks the N-terminal transactivation domain could func- heat shock proteins in chicken embryo fibroblasts. J. Cell Biol. tion as a dominant-negative inhibitor (Meier et al., 1995). 103, 1495–1507. The large number of isoforms, each with a potential to Descombes, P., and Schibler, U. (1991). A liver-enriched transcrip- participate in autoregulation and cross-regulation among tional activator protein, LAP, and a transcriptional inhibitory the C/EBP-family and AP-2 family genes, adds to the protein, LIP, are translated from the same mRNA. Cell 67, potential complexity of k10 regulation. 569–579. Eckert, R. L., Crish, J. F., and Robinson, N. A. (1997). The epidermal keratinocyte as a model for the study of gene regulation and cell ACKNOWLEDGMENTS differentiation. Physiol. Rev. 77, 397–424. Fawcett, T. W., Eastman, H. B., Martindale, J. L., and Holbrook, We thank Dr. Valeria Poli (University of Dundee, Scotland, UK) N. J. (1996). Physical and functional association between for the generous gift of the C/EBP␤ knockout mice. We also thank gadd153 and CCAAT/enhancer-binding protein ␤ during cellular Dr. Dan Drucker (University of Toronto) for the SKLuc vector and stress. J. Biol. Chem. 271, 14285–14289. Dr. Ulupi Jhala (Joslin Diabetes Institute, Boston, MA) for recom- Freytag, S. O., Paielli, D. L., and Gilbert, J. D. (1994). Ectopic ␤ binant C/EBP protein used in some of the gel shift studies. We expression of the CCAAT/enhancer-binding protein ␣ promotes thank Dr. Melissa Thomas and Dr. Doris Stoffers for helpful the adipogenic program in a variety of mouse fibroblastic cells. discussions. This work was supported by grants from the National Genes Dev. 8, 1654–1663. Institutes of Health (E.V.M., D.R., J.F.H.) and the Swiss National Friedman, A. D., Landschulz, W. H., and McKnight, S. L. (1989). Science Foundation (P.J.M.). E.V.M. is a recipient of an NIH CCAAT/enhancer binding protein activates the promoter of the Clinical Investigator Award. D.R. is a Stephen Birnbaum Scholar of serum albumin gene in cultured hepatoma cells. Genes Dev. 3, the Leukemia Society of America. J.F.H. is an Investigator of the 1314–1322. Howard Hughes Medical Institute. Fuchs, E. (1990). Epidermal differentiation: The bare essentials. J. Cell Biol. 111, 2807–2814. REFERENCES Fuchs, E. (1993). Epidermal differentiation and keratin gene expression. J. Cell Sci. Suppl. 17, 197–208. Ausbel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seidman, Fuchs, E., Esteves, R. A., and Coulombe, P. A. (1992). Transgenic J. G., Smith, J. A., and Struhl, K. (1992). “Current Protocols in mice expressing a mutant keratin 10 gene reveal the likely Molecular Biology,” Wiley, New York. genetic basis for epidermolytic hyperkeratosis. Proc. Natl. Acad. Blumenberg, M., Komine, M., Rao, L., Ma, S., and Freedberg, I. M. Sci. USA 89, 6906–6910. (1998). Blueprint to footprint to toeprint to culprit: Regulation of Galas, D., and Schmidt, A. (1978). DNase footprinting, a simple K6 keratin gene promoter by extracellular signals and nuclear method for the detection of protein–DNA binding specificity. transcription factors. J. Invest. Dermatol. 110, 494. Nucleic Acids Res. 5, 3157–3170. Brasier, A., Ron, D., Tate, J., and Habener, J. (1990). A family of Goldsmith, L. A. (1991). “Physiology, Biochemistry, and Molecular constitutive C/EBP-like DNA binding proteins attenuate the Biology of the Skin,” 2nd ed. Oxford Univ. Press, New York.

Haake, A. R., and Polakowska, R. R. (1993). Cell death by apoptosis Moser, M., Ruschoff, J., and Buettner, R. (1997). Comparative in epidermal biology. J. Invest. Dermatol. 101, 107–112. analysis of AP-2␣ and AP-2␤ gene expression during murine Haas, N. B., Cantwell, C. A., Johnson, P. F., and Burch, J. B. E. embryogenesis. Dev. Dyn. 208, 115–124. (1995). DNA-binding specificity of the PAR basic leucine zipper Oh, H. S., and Smart, R. C. (1998). Expression of CCAAT/enhancer protein VBP partially overlaps those of the C/EBP and CREB/ATF binding proteins (C/EBP) is associated with squamous differen- families and is influenced by domains that flank the core basic tiation in epidermis and isolated primary keratinocytes and is region. Mol. Cell. Biol. 15, 1923–1932. altered in skin neoplasms. J. Invest. Dermatol. 110, 939–945. Jiang, M.-S., Tang, Q.-Q., McLenithan, J., Geiman, D., Shillinglaw, Osada, S., Yamamoto, H., Nishihara, T., and Imagawa, M. (1996). W., Henzel, W. J., and Lane, M. D. (1998). Derepression of the DNA binding specificity of the CCAAT/enhancer-binding pro- C/EBP␣ gene during adipogenesis: Identification of AP-2␣ as a tein transcription factor family. J. Biol. Chem. 271, 3891–3896. repressor. Proc. Natl. Acad. Sci. USA 95, 3467–3471. Ossipow, V., Descombes, P., and Schibler, U. (1993). CCAAT/ Krieg, T. M., Schafer, M. P., Cheng, C. K., Filpula, D., Flaherty, P., enhancer-binding protein mRNA is translated into multiple Steinert, P. M., and Roop, D. R. (1985). Organization of a type I proteins with different transcription activation potentials. Proc. keratin gene: Evidence for evolution of intermediate filaments Natl. Acad. Sci. USA 90, 8219–8223. from a common ancestral gene. J. Biol. Chem. 260, 5867–5870. Oulad-Abdelghani, M., Bouillet, P., Chazaud, C., Dolle, P., and Lamb, P., and McKnight, S. L. (1991). Diversity and specificity in Chambon, P. (1996). AP-2.2: A novel AP-2-related transcription transcriptional regulation: The benefits of heterotypic dimeriza- factor induced by retinoic acid during differentiation of P19 tion. Trends Biochem. Sci. 16, 417–422. embryonal carcinoma cells. Exp. Cell Res. 225, 338–347. Leask, A., Byrne, C., and Fuchs, E. (1991). Transcription factor AP2 Poli, V., Mancini, F. P., and Cortese, R. (1990). IL-6DBP, a nuclear and its role in epidermal-specific gene expression. Proc. Natl. protein involved in interleukin-6 signal transduction, defines a Acad. Sci. USA 88, 7948–7952. new family of leucine zipper proteins related to C/EBP. Cell 63, Lee, Y. M., Tsai, W. H., Lai, M. Y., Chen, D. S., and Lee, S. C. (1993). 643–653. Induction of liver alpha-1 acid glycoprotein gene expression Rieger, M., and Franke, W. W. (1988). Identification of an ortholo- involves both positive and negative transcription factors. Mol. gous mammalian cytokeratin gene: High degree of intron se- Cell. Biol. 13, 432–442. quence conservation during evolution of human cytokeratin 10. Lin, F.-T., and Lane, M. D. (1994). CCAAT/enhancer binding J. Mol. Biol. 204, 841–856. protein ␣ is sufficient to initiate the 3T3-L1 adipocyte differen- Rieger, M., Jorcano, J. L., and Franke, W. W. (1985). Complete tiation program. Proc. Natl. Acad. Sci. USA 91, 8757–8761. sequence of a bovine type I cytokeratin gene: Conserved and Lin, F.-T., MacDougald, O. A., Diehl, A. M., and Lane, M. D. (1993). variable intron positions in genes of polypeptides of the same A 30 kDa alternative translation product of the CCAAT/ cytokeratin subfamily. EMBO J. 4, 2261–2267. enhancer binding protein alpha message: Transcriptional activa- Ron, D., and Habener, J. F. (1992). CHOP, a novel developmentally tor lacking antimitotic activity. Proc. Natl. Acad. Sci. USA 90, regulated nuclear protein that dimerizes with transcription fac- 9606–9610. tors C/EBP and LAP and functions as a dominant-negative MacDougald, O. A., and Lane, M. D. (1995). Transcriptional regu- inhibitor of gene transcription. Genes Dev. 6, 439–453. lation of gene expression during adipocyte differentiation. Annu. Roop, D. R., Krieg, T. M., Mehrel, T., Cheng, C. K., and Yuspa, S. H. Rev. Biochem. 64, 345–373. (1988). Transcriptional control of high molecular weight keratin Marcelo, C. L., Kim, Y. G., Kaine, J. L., and Voorhees, J. J. (1978). gene expression in multistage mouse skin carcinogenesis. Can- Stratification, specialization, and proliferation of primary kera- cer Res. 48, 3245–3252. tinocyte cultures. J. Cell Biol. 79, 356–370. Rothnagel, J. A., Dominey, A. M., Dempsey, L. D., Longley, M. A., Maytin, E., and Habener, J. (1998). Transcription factors C/EBP␣, Greenhalgh, D. A., Gagne, T. A., Huber, M., Frenk, E., Hohl, D., C/EBP␤, and CHOP (gadd153) expressed during the differentia- and Roop, D. R. (1992). Mutations in the rod domains of keratins tion program of keratinocytes in vitro and in vivo. J. Invest. 1 and 10 in epidermolytic hyperkeratosis. Science 257, 1128– Dermatol. 110, 238–246. 1130. McKnight, S. L. (1992). CCAAT/Enhancer binding protein. In Ryden, T., and Beemon, K. (1989). Avian retroviral long terminal “Transcriptional Regulation” (S. L. McKnight and K. R. repeats bind CCAAT/enhancer-binding protein. Mol. Cell. Biol. Yamamoto, Eds.), Monogr. 22, pp. 771–795. Cold Spring Harbor 9, 1155–1164. Laboratory Press, Cold Spring Harbor, NY. Schorle, H., Meier, P., Buchert, M., Jaenisch, R., and Mitchell, P. J. Meier, P., Koedood, M., Philipp, J., Fontana, A., and Mitchell, P. J. (1996). Transcription factor AP-2 essential for cranial closure and (1995). Alternative mRNAs encode multiple isoforms of tran- craniofacial development. Nature 381, 235–238. scription factor AP-2 during murine embryogenesis. Dev. Biol. Screpanti, I., Romani, L., Musiani, P., Modesti, A., Fattori, E., 169, 1–14. Lazzaro, D., Sellitto, C., Scarpa, S., Bellavia, D., Lattanzio, G., Mitchell, P. J., Timmons, P. M., Hebert, J. M., Rigby, P. W., and Bistoni, F., Frati, L. R, Guilino, A., Cilberto, G., Costantini, F., Tjian, R. (1991). Transcription factor AP-2 is expressed in neural and Poli, V. (1995). Lymphoproliferative disorder and imbalanced crest cell lineages during mouse embryogenesis. Genes Dev. 5, T-helper response in C/EBP␤-deficient mice. EMBO J. 14, 1932– 105–119. 1941. Mitchell, P. J., Wang, C., and Tjian, R. (1987). Positive and negative Spiro, C., Bazett-Jones, D. P., Wu, X., and McMurray, C. T. (1995). regulation of transcription in vitro: Enhancer-binding protein DNA structure determines protein binding and transcriptional AP-2 is inhibited by SV40 T antigen. Cell 50, 847–861. efficiency of the proenkephalin cAMP-responsive enhancer. Moser, M., Imhof, A., Pscherer, A., Bauer, R., Amselgruber, W., J. Biol. Chem. 270, 27702–27710. Sinowatz, F., Hofstadter, F., Schule, R., and Buettner, R. (1995). Studier, W. S., Rosenberg, A. H., Dunn, J. J., and Dubendorff, J. W. Cloning and characterization of a second AP-2 transcription (1990). Use of T7 polymerase to direct expression of cloned genes. factor: AP- 2 beta. Development 121, 2779–2788. Methods Enzymol. 185, 60–89.

Swart, G. W., van Groningen, J. J., van Ruissen, F., Bergers, M., and binding: Prediction and rational design. Genes Dev. 7, 1047– Schalkwijk, J. (1997). Transcription factor C/EBP␣: Novel sites of 1058. expression and cloning of the human gene. Biol. Chem. 378, Weissman, B., and Aaronson, S. (1983). BALB and Kirsten murine 373–379. sarcoma viruses alter growth and differentiation of EGF- Syder, A. J., Yu, Q.-C., Paller, A. S., Giudice, G., Pearson, R., and dependent BALB/c mouse epidermal keratinocyte lines. Cell 32, Fuchs, E. (1994). Genetic mutations in the K1 and K10 genes of 599–606. patients with epidermolytic hyperkeratosis: Correlation between Williams, T., and Tjian, R. (1991). Characterization of a dimeriza- location and disease severity. J. Clin. Invest. 93, 1533–1542. tion motif in AP-2 and its function in heterologous DNA-binding Sylvester, S. L., ap Rhys, C. M. J., Luethy-Martindale, J. D., and proteins. Science 251, 1067–1071. Holbrook, N. J. (1994). Induction of GADD153, a CCAAT/ Yeh, W.-C., Cao, Z., Classon, M., and McKnight, S. L. (1995). enhancer binding protein (C/EBP)-related gene, during the acute Cascade regulation of terminal adipocyte differentiation by three phase response in rats. J. Biol. Chem. 269, 20119–20125. members of the C/EBP family of leucine zipper proteins. Genes Tanaka, T., Akira, S., Yoshida, K., Umemeoto, M., Yoneda, Y., Dev. 9, 168–181. Shirafuji, N., Fujiwara, H., Suematsu, S., Yoshida, N., and Kishi- Yuspa, S., and Dlugosz, A. A. (1991). Cutaneous carcinogenesis: moto, T. (1995). Targeted disruption of the NF-IL6 gene discloses its essential role in bacteria killing and tumor cytotoxicity by Natural and experimental. In “Physiology, Biochemistry, and macrophages. Cell 80, 353–361. Molecular Biology of the Skin,” 2nd ed. (L. A. Goldsmith, Ed.), Tang, Q. Q., Jiang, M.-S., and Lane, M. D. (1997). Repression of pp. 1365–1402. Oxford Univ. Press, New York. transcription mediated by dual elements in the CCAAT/ Yuspa, S., Kilkenny, A. E., Steinert, P. M., and Roop, D. R. (1989). enhancer binding protein ␣ gene. Proc. Natl. Acad Sci. USA 94, Expression of murine epidermal differentiation markers is tightly 13571–13575. regulated by restricted extracellular calcium concentrations in Ubeda, M., Wang, X.-Z., Zinszner, H., Wu, I., Habener, J. F., and Ron, vitro. J. Cell Biol. 109, 1207–1217. D. (1996). Stress-induced binding of the transcription factor CHOP Zhang, J., Hagopian-Donaldson, S., Serbedzija, G., Elsemore, J., to a novel DNA control element. Mol. Cell. Biol. 16, 1479–1489. Plehn-Dujowich, D., McMahon, A. P., Flavell, R. A., and Wil- Vallejo, M., Penchuk, L., and Habener, J. F. (1992). Somatostatin liams, T. (1996). Neural tube, skeletal and body wall defects in gene upstream enhancer element activated by a protein complex mice lacking transcription factor AP-2. Nature 381, 238–241. consisting of CREB, Isl-1-like, and alpha-CBF-like transcription factors. J. Biol. Chem. 267, 112876–12884. Received for publication July 8, 1999 Vinson, C. R., Hai, T., and Boyd, S. M. (1993). Dimerization Revised August 16, 1999 speciﬁcity of the leucine zipper-containing bZIP motif on DNA Accepted August 16, 1999