ABSTRACT
Ets proteins have been implicated in the regulation of gene expression during a variety of biological processes, including growth control, differentiation, development and transformation. More than 35 related proteins containing the `ets domain' have now been found which specifically interact with DNA sequences encompassing the core tetranucleotide GGAA. Although ets responsive genes have been identified in the epidermis, little is known about their distribution and function in this tissue. We have now demonstrated that epidermis and cultured epidermal keratinocytes synthesize numerous ets proteins. The expression of some of these proteins is regulated as a function of differentiation. Among these is a novel ets transcription factor with a dual DNA-binding specificity, which we have called jen. The expression of jen is not only epithelial specific, but it is the only ets protein so far described, and one of the very few transcription factors whose expression is restricted to the most differentiated epidermal layers. We show that two epidermal marker genes whose expression coincides with that of jen are transregulated by this protein in a complex mode which involves interactions with other transcriptional regulators such as Sp1 and AP1.
The epidermis is a multi-layered epithelium in which differentiation proceeds outwards. As cells migrate from the basal layer, the tightly regulated program of epidermal differentiation is reflected by the synthesis of sets of biochemical markers such as specific keratins and their associated proteins, cell envelope precursors and processing enzymes (1 ,2 ). The pattern of expression for most of these genes is controlled at the level of transcription (3 ). Consequently, exploring the control of the expression of these marker genes may provide useful models for elucidating the mechanisms underlying the cell type and differentiation specific regulation. Recent studies have revealed an essential role in these processes for a group of transcription factors termed ets (4 ,5 ). The ets proteins have been implicated in such fundamental processes as cell growth, differentiation, developmental programs, viral infectious cycles and malignant transformation (6 -8 ). A number of human cancers are associated with specific chromosomal translocations which involve ets genes (6 ,9 ,10 and refs therein). The common feature of all ets proteins is the ets domain, a winged helix-loop-helix motif (11 , see 7 ,8 ) that is required to bind in the major groove of specific DNA sequences centered over a conserved core GGAA/T motif (11 ,12 ). A subset of ets proteins share a homologous region N-terminal to the ets domain which has been referred to as B (13 ), pointed (14 ) or HLH domain (15 ). Recently Jousset et al. (10 ) have demonstrated that the B domain regulates the transcriptional activity of tel by mediating homo-oligomerization.
In this paper we report the identification of a novel ets protein which we have termed jen. Among the ets factors jen is distinguished by its dual DNA binding specificity and by the epithelium-restricted pattern of expression. In the epidermis jen is unique among the other ets proteins in that its expression is detected exclusively in the most differentiated cell layers. We show that jen can transregulate the transcription from epidermal promoters which are active during the late stages of epidermal differentiation.
All recombinant DNA work was done according to standard procedures (16 ,17 ).
Jen positive plaques were obtained by screening a [lambda]gt11 human foreskin keratinocyte cDNA library (Clontech) with an oligonucleotide probe, ets-d, designed from a conserved sequence in the 3' region of the ets domain (Stauffer and Buonanno, submitted). The jen cDNA inserts were amplified by PCR and subcloned into pCR2.1 (Invitrogen). DNA was sequenced using the PRISM Ready Reaction Dye Deoxy Terminator Cycle Sequencing Kit (Perkin Elmer) according to manufacturer's specifications. The nucleotide sequence of jen was submitted to GenBank under accession number U97156. Sequence homologies were analyzed by MacVector, BLAST, BLITZ and FASTA programs using GenBank, EMBL, DDBJ, and PDB data bases.
The 5' and 3' ends of jen were obtained using the MarathonTM cDNA Amplification Kit (Clontech) according to the manufacturer's specifications. Foreskin epidermal poly(A)+ mRNA and the following nested primers specific for jen cDNA were used: 5-1: 5'-TCCCAGTACTCTTTGCTCAGCTTT-3'; 5-2: 5'-CTCGTTTCCGCTTCCCGTGCTT-3'; 3-1: CTGGACTGTTCC-A-3'; 3'-2: 5'-GCTGATGCTGGGACCTTAGGAT-3'. A full length jen cDNA was amplified from double stranded human foreskin cDNA using the primers: J1: 5'-TGGCGGAACTGGATTTCTCT-3'; J2: 5'-CATAGTCTCCGAGTCCCAGGA-3'. The amplified product was cloned into the eukaryotic TA expression vector (Invitrogen) and into the bacterial expression system PinPoint (Promega) and was sequenced to ensure fidelity during the amplification.
RNAs were extracted from frozen foreskin epidermis or from submerged normal human epidermal keratinocytes (NHEK) using Trizol (Life Technologies). Poly(A)+ mRNA was isolated with magnetic oligo-dT beads according to the manufacturer's protocol (Boehringer Mannheim). An aliquot of1 µg of poly(A)+ mRNA was resolved on methylmercury denaturing gels, transferred to nitrocellulose and hybridized to ets-specific DNA probes (ets-d, ets-2, elf1, jen) in ExpressHyb solution (Clontech). Commercially available northern blots (Clontech) were hybridized with the same probes.
Sections of paraffin embedded formalin fixed adult human skin were digested with pepsin for 30 min at room temperature. Two of the sections on each slide were treated with RNase-free DNase (Boehringer Mannheim) overnight at 37oC. The reverse transcription and the in situ PCR were performed in one step using GeneAmp EZ rTth RNA PCR kit (Perkin Elmer), digoxigenin labeled dUTP (Boehringer Mannheim) and jen specific primers (A: 5'-GAAGGCGTCTTCAAGTTCCT-3'; B: 5'-CATAGTCTC-CGAGTCCCAGGA-3') (18 ). The amplified transcripts were detected using DIG Nucleic Acid Detection Kit (Boehringer Mannheim).
Rabbit polyclonal antibodies against the ets proteins were purchased from Santa Cruz Laboratories. Sections of frozen human foreskin were consecutively incubated with the primary antibodies (1:1000 dilution), biotinylated anti-rabbit IgG (1:5000) and Streptavidin-Peroxidase (KPL Laboratories) and developed with AEC chromagen (Biogenex).
Total cellular proteins and nuclear subfractions were extracted from NHEK cells by procedures described previously (4 ).
Proteins were resolved on 10% Tricine polyacrylamide gels (Novex), transferred electrophorectically to nitrocellulose filters and processed according to protocols detailed in the ECL western blotting detection kit (Amersham). All primary antibodies (Santa Cruz) and the horseradish peroxidase-conjugated secondary antibody (Amersham) were diluted in PBS containing 0.5% Triton X100, 5% non-fat dry milk (Carnation) and 10% fetal calf serum (Life Technologies, Inc.). Chemiluminescence was detected using Pierce ECL detection kit.
Cryopreserved NHEK cells were obtained from Clonetics and grown as previously described (4 ). Third passage cells were used for transfection experiments. HeLa cells were purchased from American Tissue Culture Collection (ATCC) and were grown following recommended procedures. Neuroblastoma cells (SK N-AS) were a gift from Dr C. Thiele and were grown in RMPI 1640 medium supplemented with 10% fetal calf serum. HaCaT cells (a gift from Dr N. Fusenig) were grown in Dulbecco's modified Eagle medium supplemented with 4.5 g/l glucose, 10% fetal bovine serum and non-essential amino acids (Life Technologies).
The TGM3 and profilaggrin constructs in pCAT-Basic vector (Promega) and the transient transfection experiments were as previously described (4 ,27 ). Reporter constructs (1 µg) were co-transfected with 0.5 µg of expression vectors containing Sp1 (a gift from Dr R. Tjian), c-jun (a gift from Dr M. Karin), ets1 (a gift from Dr J. Dittmer) and jen cDNA or with the same amount of the corresponding empty expression vectors. Extracts of cells transfected with the pCAT-Basic vector were used as a negative control, while the activity of CAT gene driven by SV40 early promoter in the pCAT-Promoter vector served as a positive control. The CAT values are the average of at least three independent experiments, each with duplicate samples.
Mobility shift experiments were performed with 2 µg of partially purified bacterially expressed jen (PinPoint system, Promega) and 4 × 104 c.p.m. of gel-purified end-labeled double-stranded oligonucleotides carrying ets or HMG recognition motifs as already described (4 ). The sequences of the oligonucleotides were as follows: TGM3-ets: CTACAGGAATGACCTGGGTG; TGM3-etsm: CTACATACGTGACCTFGGGTG; PF ets: GGATCTAGGTTTGGTTAGGAATGA; PF-etsm: GGATCTAGGTTTGGTTATACGTGA; PF-HMG: GGATCTAGGTTTGGTTA; PF-HMGm: GGATCTAGCGAGCGTTA; cHMG: GGATCTA-ACAAAGTGGATCT. The flanking sequences of all oligonucleotides were identical.
Recent studies have indicated that ets proteins may be important regulators of epidermal transcription, especially during the late stages of differentiation (4 ,5 ). However, little is known about the presence and the distribution of these proteins in the epidermis. To explore this we incubated frozen sections of human foreskin epidermis with a panel of antibodies specific for individual ets proteins. The epidermis stained positive with ets1, ets2, elf1, elk1, sap1a, erg and fli1 antibodies (Fig. 1 a). We could not ab
We compared the immunohistochemical data from the frozen epidermal sections with the profiles of the same ets proteins in nuclear and total cellular extracts of NHEK cells. On the western blots (Fig. 1 b) ets2, sap1a and ets1 antibodies detected single protein bands of the expected molecular weight (6 ) thus confirming the specificity of the antibodies. The elf1 antibody recognized a band which migrated like a protein of >90 kDa (Fig. 1 b). This anomalous mobility of elf1, which is a 69 kDa protein, has been previously observed (19 ). Elk1, erg and fli1 antibodies detected multiple protein bands (Fig. 1 b). To determine whether the faster migrating bands represented phosphorylated forms of these proteins, the extracts were preincubated with alkaline phosphatase prior to western analysis. Only the slow migrating bands corresponding to the unphosphorylated forms of elk1 and fli1 were preserved while the erg1/erg2 protein pattern was not affected (data not shown). Importantly, it appeared that the nuclear extracts contained predominantly unphosphorylated or weakly phosphorylated forms while the highly phosphorylated forms were localized in the cytoplasm (Fig. 1 b).
In concordance with the immunostaining results, erg and fli1 were confined predominantly to the cytoplasm, since the signal for these proteins was barely detectable in the nuclear extracts (Fig. 1 b). It is unlikely that this was an experimental artifact since the same antibodies detected strong staining in HeLa nuclear extracts (data not shown).
Bandshift, southwestern and transient transfection analyses have implicated ets2, elf1 and at least one other, so far unidentified ets protein in the activation of TGM3 and profilaggrin promoters (4 and unpublished results). Accordingly, we screened a keratinocyte cDNA library for new ets-like sequences using a probe spanning 50 nt of the conserved ets domain (ets-d, kindly provided by A. Buonanno). Consistent with the immunohistochemical data, numerous clones corresponding to ets1, ets2, elf1, erg and fli1 were obtained. Five overlapping clones were detected that encoded a sequence with a high degree of homology only to the ets domain. 5'- and 3'-RACE on poly(A)+ foreskin mRNA were used to obtain the 5' and the 3' portions of the cDNA. The coding sequence was extended into human genomic DNA to obtain the transcription initiation site and the 5' regulatory region. The 1924 nt full length cDNA encompassed an open reading frame of 1113 nt flanked by 125 nt of 5'-untranslated and 687 nt of 3'-untranslated sequences. A canonical TATA box was found 35 bp upstream of the mRNA start site (Fig. 2 ).
The first in-frame initiator codon resided in a sequence which conforms well to the translation initiatior consensus (20 ). A stop codon was found 108 nt upstream, in frame with this initiator site. The coding sequence could be translated into a protein of 371 amino acids with a predicted molecular mass of ~42 kDa (Fig. 2 ). The protein, which we called jen, is highly hydrophilic, with acidic amino acids predominating in the N-terminal half, and a more basic C-terminus. A domain rich in leucine and serine can be discerned in the central portion of the protein. Three putative MAP kinase phosphorylation sites (SP, amino acids 163-164, 177-178, 190-191; ref. 21 ) are present in this region. Although a region with a weak homology to the docking site of the Jun kinases (JNKs) can be discerned in the N-terminus of jen between amino acids 25 and 55, the sequences flanking the putative phosphoacceptor sites lack the crucial arginine residues at position P+5 and thus are not a likely target for the JNK/p38/ERK enzymes (22 ). One of these phosphoacceptor sites (SPY, amino acids 177-179) is a potential target for the dual-specificity kinases (23 ). One potential PKC (TEK, amino acids 53-55), as well as two sites for cAMP and cGMP-dependent protein kinases (RKLS, amino acids 251-254; KKNS, amino acids 319-322) are present. Five putative casein kinase II phosphorylation sites are clustered in the serine rich central region. The highly basic region between amino acids 236 and 255 is a potential target for specific acetylation.
Outside of the ets domain, jen shares only limited homology with the known ets proteins. Within the ets domain the homology ranges from 30% with PU.1 to 67% with E74. Although jen could be classified with the major subclass of ets proteins with a C-terminal ets domain (6 -8 ), it also shares a similar degree of homology with ets proteins belonging to the other two subclassess (N-terminal, e.g. Elk1, 69%; or central, e.g. Elf1, 67%). The ets domain of jen contains most of the characteristic amino acids, including Arg331, Arg334 and Tyr335 that have been shown to be critical for the recognition between human fli1 and its DNA (11 ). In addition, jen contains the N-terminal B domain which is shared by most members of the ets1/ets2 subclass.
A remarkable feature of jen is a basic region that precedes the ets domain (Fig. 2 ) and has a potential for direct binding to DNA. It is homologous to the so called A/T hook, a domain known to directly interact with the minor groove of AT rich DNA stretches (24 ).
The B and the A/T hook domains are separated by a region with a high serine content which is homologous to the Xenopus and mouse homologs of the segment polarity protein dishevelled (25 ), and to the nuclear period clock protein (26 ), proteins known to be involved in signal transduction. Notably, several of the putative jen phosphorylation sites are found within this region, suggestive of its involvement in the transduction of extracellular signals through phosphorylation cascades.
The tissue specificity of jen expression was established in northern blot analysis. A jen specific probe (see Materials and Methods) recognized a major transcript of the expected length of 1.9 kb only in epithelium-containing tissues such as placenta, liver, kidney, pancreas and lung (Fig. 3 a). The faint signal of 1.7 kb and the longer bands were detected by several ets probes (e.g. ets-d and ets-2, Fig. 3 a) and most probably originated from hybridization with homologous but different ets transcripts. Accordingly, a single jen transcript of the expected length was detected by RT-PCR of liver and placenta mRNA but could not be amplified in brain, skeletal muscle and dermal fibroblast mRNA preparations (data not shown). A dot blot that contained poly(A)+ mRNAs from 49 different fetal and adult tissues also showed positive hybridization only with the tissues containing cells of epithelial origin (Fig. 3 b). This suggests that jen expression is epithelial specific and this specificity is imposed early during development.
Human foreskin and NHEK poly(A)+ mRNAs hybridized to the jen-specific probe at stringent conditions to reveal a single 1.9 kb transcript (Fig. 4 a). The same transcript could be detected with the probe ets-d, recognizing the core ets domain, but was not observed upon hybridization with the ets2-specific probe. We assessed the stratum specificity of jen by in situ RNA PCR on sections of adult human skin. This approach offers the highest possible specificity of detection since it eliminates the cross hybridization between the homologous ets transcripts. We detected jen transcripts exclusively in the granular layer (Fig. 4 b). This is in agreement with northern blot experiments (Fig. 4 c) in which a 5-10-fold induction of jen specific transcripts was detected when NHEK cells were switched from growth medium containing 0.05 mM Ca2+ to 1.2 mM Ca2+ and 20 nM TPA, conditions favoring keratinocyte differentiation in general and expression of the late differentiation markers, such as TGM3 and profilaggrin, in particular (4 ,27 and references therein). Significantly, the expression of elf1, which is uniformly present in all epidermal layers (Fig. 1 ), was not affected under these conditions (Fig. 4 c). a.
We assessed the dual DNA binding properties of jen in bandshift assays. A full length jen fused to a biotinylated tag was expressed in bacteria and partially purified on a streptavidin resin. The bacterially expressed protein was recognized by the antibody ets1/ets2 which is directed against the ets domain and reacts with most ets proteins (data not shown). Incubation of jen with a labeled oligonucleotide encompassing the ets binding site of the TGM3 promoter (4 ) (Fig. 5 a, lane 1) produced a specific complex (lane 2) which was supershifted by the ets1/ets2 antibody (lane 8). The complex was successfully competed by the ets binding site of the profilaggrin promoter (lane 3) (27 ), whereas oligonucleotides with mutations in the GGAA core of the ets motifs (lanes 4 and 5) or controls using irrelevant oligonucleotides (data not shown) failed to compete for the binding. Significantly, the complex was affected by oligomers containing either a consensus (lane 6) or profilaggrin HMG binding motifs (lane 7). Therefore, we performed the bandshift experiments with a labeled oligonucleotide carrying the consensus HMG binding site (28 ) and detected a complex which migrated similarly to that of the ets motif (compare Fig. 5 a and b, lanes 2). Oligomers of the profilaggrin HMG-like site (Fig. 4 b, lane 3) and TGM3 ets site (lane 5) competed for the binding whereas their counterparts carrying mutations in critical nucleotides failed to compete (lanes 4 and 6). Together, these data indicate that jen has both ets and HMG binding activity.
The stratum specificity of jen expression suggested that the protein might be involved in transcriptional control of genes expressed during the late stages of the epidermal differentiation. To test this, an ets-responsive TGM3-CAT construct (4 ) was co-transfected with a jen expression vector into epithelial and non-epithelial cells. The activity of the TGM3 promoter was not affected by the presence of jen in dermal fibroblasts and in neuroblastoma cells (Fig. 6 a). In contrast, a 60-70% increase in the activity of CAT, which depended on an intact ets binding motif, was detected in the simple epithelial-like HeLa cells, in the immortalized keratinocyte cell line HaCaT and in the normal human epidermal keratinocytes grown in 0.05 mM Ca2+ (Fig. 6 a). The effect of jen was most pronounced in NHEK cells grown under conditions favoring differentiation such as 1.2 mM Ca2+ (2.5-fold, Fig. 6 b). We have previously shown that the transcriptional activity of TGM3 promoter requires interactions between ets and Sp1 transcription factors (4 ). Accordingly, we co-transfected the TGM3-CAT constructs with expression vectors for both jen and Sp1 and observed an additional 2-fold increase in the activity (Fig. 6 b). The transactivation required binding of jen and Sp1 to their respective recognition sites, since mutations in either of them reduced the activity by >5-fold. The effect of jen alone or together with Sp1 over the activity of the TGM3 promoter was similar but more pronounced compared with the previously reported effect of ets1 (4 ), another ets factor whose expression is increased in the most differentiated epidermal layers (Fig. 1 a).
There is accumulating evidence that the expression of TGM3 (4 ), SPRR2A (5 ) and profilaggrin genes (unpublished results), all of which are markers of terminally differentiating epidermal cells, depends critically on interactions at ets binding sites. In this paper we have demonstrated for the first time that numerous ets proteins are synthesized in the epidermis. They show distinct patterns of subcellular localization and stratum specificity. None of them, however, is expressed exclusively in the most differentiated epidermal layers. In the search for the particular ets proteins which are involved in the regulation of the late differentiation markers we have isolated a new member of this family. The protein, which we have named jen, binds specifically to ets motifs (Fig. 5 a) and in epidermal cells cooperates with Sp1 and AP1 to transactivate the TGM3 and profilaggrin promoters, respectively (Fig. 6 b and c).
A surprising feature of jen is its ability to bind specifically to HMG recognition motifs (Fig. 5 b) most probably through a highly basic region that precedes the ets domain (Fig. 2 ), and is homologous to the DNA binding domains A/T hook (24 ) of the HMG-I(Y) proteins (29 ,30 ), the HMG-like transcription factors such as LEF/TCF-1 (28 ,31 ), sry (32 ), tramtrack (33 ), MML1 (34 ); the yeast origin binding protein (35 ), and other chromatin modifying factors (26 ,36 ). It has been postulated that the interactions of these proteins in the minor groove of AT tracks of DNA (24 ,30 ) produce bends which drastically change the chromatin structure (32 ,37 ,38 ). Several HMG-like transcription factors have been shown to repress transcription (e.g. 38 ). A juxtaposition of ets and HMG binding sites, with a possible functional significance, has been found in the proximal regulatory region of the mouse myelin basic protein (38 ). Physical association between elf1, HMG-I(Y) and NF-[kappa]B has been recently reported as crucial for the cell type specific and inducible IL-2Ra expression (39 ). Motifs with high degree of homology to the HMG-binding sites reside in close proximity to the ets binding motifs both in TGM3 and the profilaggrin promoters and were found to specifically interact with jen (Fig. 5 ). Preliminary experiments have shown that these motifs interact in bandshift experiments with NHEK nuclear proteins both in single- and in double-stranded form and that mutations in critical nucleotides lead to increased promoter activity (unpublished results). These findings suggest that an interplay between ets and HMG proteins can be a component of the regulation of these epidermal genes. Depending on whether an HMG-like protein or jen has bound, the same HMG motif may mediate either repression or transactivation. Furthermore, depending on whether it has bound to an ets or an HMG binding site, a dual specificity protein such as jen might act either as an activator or repressor. Further experiments are needed to elucidate whether jen encompasses distinct ets and HMG transactivation domains.
The individual ets proteins display distinct binding specificity determined by the contacts with the sequences flanking the core motif in the minor groove (12 ,40 ). Some ets factors have broader specificity than others, and different ets motifs appear to vary in their selectivity for factors (8 ). However, the differences are more often quantitative rather than qualitative (7 ,8 ). One of the most intriguing questions then is how is the specificity of the regulation achieved, especially in tissues and cells which contain a variety of ets proteins.
By far the easiest way would be the presence of tissue-specific ets factors. However, with a few exceptions ets proteins are broadly expressed in different tissues and cells (6 -8 ). Our data suggest that the major mature mRNA species derived from the jen primary transcript is epithelial specific. This finding may lead to important implications not only for the epidermis but for other stratified squamous and simple epithelia. It would be interesting to determine the role of jen during embryogenesis and whether translocations associated with jen may be implicated in the carcinogenesis in human epithelial tissues.
Stratum specificity provides another level on which a specificity of ets regulation can be achieved. The epidermis is a particularly attractive target for such studies in view of the simple and stratified pathway of epidermal differentiation. The data presented in this paper provide for the first time information about the distribution of the ets proteins in the epidermis and in epidermally derived cells. We show (Fig. 1 ) that numerous ets proteins are synthesized with a varying degree of stratum specificity. In this respect jen again represents an intriguing discovery. It is the only ets transcription factor so far described whose expression is restricted to the most differentiated epidermal layers and this makes it, along with the recently characterized homeodomain protein Dlx3 (41 ,42 ), a candidate for a key regulator of genes expressed during the late stages of epidermal differentiation.
Specificity of regulation by ets proteins can also be determined by the ability of a particular ets protein to recruit and to associate with other transcriptional regulators (43 ,44 , see 6 -8 ). In the epidermis, three genes which have been found to be regulated by ets factors, require association of ets protein(s) with Sp1 (TGM3; 4 ), POU domain and IRF factors (SPRR2A; 5 ) or AP1 (profilaggrin; 27 ). The specificity of the ets contacts in the major groove opens the possibility for interactions with a protein bound to the minor groove via the same recognition motif. The latter has been described for the LyF1 protein whose recognition site encompasses an ets site in the Tdt promoter (45 ). In fact, this may be an example of a mechanism by which an HMG and an ets protein could simultaneously control the TGM3 and the profilaggrin promoters.
Posttranslational modifications offer another mechanism to achieve specificity. Selective phosphorylation at particular sites could activate or deactivate only particular ets proteins and not affect the status of others (7 ,8 ). Potential phosphorylation sites for several different types of protein kinases are present within the jen sequence, thus indicating that the protein could be a target for multiple signal transduction pathways. The homology between the lysine rich A/T hook domains of jen and HMG proteins opens the possibility that jen may be a target for acetylation, a process that appears to be a check point in the transcriptional control (46 ,47 ). A consequence of the posttranslational modifications is the possibility of a rapid exchange of ets proteins over the target sites in response to differential stimuli. The cytoplasmic localization of erg and fli1 in the epidermal keratinocytes and the predominance of the unphosphorylated forms of ets proteins in the nuclei (Fig. 1 b) suggest that the epidermal ets proteins might be a target for a quick transient nuclear translocation to occupy targets that are normally silenced (48 ). Such a rapid exchange has been demonstrated for elk1 (49 ).
In addition to the transregulation via binding at the promoters of the target genes, the ets proteins can regulate transcription by affecting the expression of other transcriptional regulators, including autoregulation. c-Fos protein is one of the most well studied examples of an ets regulated transcription factor (15 ,16 ). The expression of another member of the AP1 complex, junB, is also regulated by ets proteins (50 ,51 ). This fact is particularly interesting in view of the critical involvement of AP1 activity in the regulation of epidermal transcription (27 ,52 ).
In conclusion we have demonstrated that epidermal cells express a variety of ets transcription factors, among which is the heretofore unidentified epithelial specific ets protein jen. While the ets proteins exhibit a broad spectrum of differentiation specificity, jen is exclusively expressed in the terminally differentiating epidermal cells. A differential responsiveness of the epidermal promoters to several ets transregulators, including jen, offers a unique opportunity for exploring the interactions of the different ets family members among themselves and with other transcriptional regulators over the target promoters during the terminal differentiation of the epidermis.
While this manuscript was under review, the cloning of an epithelial specific ets member, ESE-1, was reported (53 ). Comparison of the two nucleic acid sequences reveals that they are identical with exception of two silent substitutions.
We thank V. DeLaurenzi for the help with the initial screening of the keratinocyte cDNA library and G. Poy for the synthesis of the oligonucleotides. We are grateful to C. Thiele for the neuroblastoma cells and to N. Fusenig for the HaCaT cells, to R. Tjian, J. Dittmer and M. Karin for the Sp1, ets1 and c-jun expression vectors, respectively, and to A. Buonanno for the ets-domain probe. We are especially grateful to M. Simon for her support and helpful discussions.
REFERENCES