Activator Protein 1 Transcription Factors Are Fundamental to v-ras^Ha-induced Changes in Gene Expression in Neoplastic Keratinocytes

Mutations in ras genes occur in ∼30% of all human cancers, and the ras family of oncogenes has been shown to play a fundamental role in the early stages of epithelial neoplasms in both humans and animals. Mutated c-ras^Hagenes, resulting in a constitutively activated Ras protein, are found in nearly all squamous papillomas that arise in mouse skin after initiation with DMBA²and promotion with phorbol esters (1). Moreover,targeting a mutant ras to the epidermis can replace DMBA initiation of mouse skin, and tumors develop under these conditions after the application of promoting agents (2). Thus,mutation of ras genes is sufficient to initiate the carcinogenic process in mouse skin. Further confirmation of the role of Ras proteins in skin tumor development is derived from studies showing that squamous papillomas can be induced in skin grafts of mouse keratinocytes transduced with a viral ras gene without additional chemical treatment (3). Although activation of Ras proteins clearly plays a critical role in skin tumor formation, the mechanisms by which activated Ras contributes to changes in gene expression are not well understood.

Ras proteins participate in a widely branched signal transduction cascade that connects cell surface receptors with nuclear signaling pathways through multiple cytoplasmic phosphorylation events (reviewed in Ref. 4). One class of transcription factors that is thought to serve as a nuclear target of Ras activation is the AP-1 family (reviewed in Refs. 5 and 6). This multigene family encodes Fos (c-Fos, Fos B, Fra-1, and Fra-2) and Jun(c-Jun, Jun B, and Jun D) proteins, which form heterodimers and regulate transcription in a complex manner though binding to DNA at AP-1 sites. Ras activation can activate specific kinases that phosphorylate Jun (7, 8) and Fos (9),augmenting their transcriptional activities.

AP-1 activity contributes to malignant progression of epidermal cells,and AP-1 DNA binding activity is increased in nuclear extracts from chemically induced papillomas that have a high risk of malignant progression (10). Dong et al. (11)showed that AP-1 transactivation is required for tumor promoter-induced transformation, as measured by anchorage-independent growth in JB6 cells. Using transgenic mice with a skin-targeted c-Jun mutant (TAM67(12), Young et al. (13) could block skin tumor formation using two-stage carcinogenesis protocols. The ability of malignant mouse epidermal cells to form s.c. tumors in nude mice was similarly inhibited when the cells expressed the TAM67 c-Jun mutant (12, 14). Transduction of v-ras^Ha keratinocytes with v-fos resulted in malignant conversion of the cells(15, 16), and malignant skin tumors do not develop in response to ras gene activation and phorbol ester treatment in mice lacking the c-fos gene (17). Additionally, c-fos null v-ras^Ha keratinocytes grafted onto the backs of nude mice do not form tumors unless they are transduced with the v-fos gene (17). Thus, considerable evidence exists to indicate that Fos and Jun proteins contribute to malignant progression of epidermal cells.

Mouse keratinocytes transduced with v-ras^Ha in vitro(v-ras^Ha keratinocytes) and maintained as basal cells in culture are hyperproliferative but retain the ability to become growth arrested under conditions that induce terminal differentiation. However, the expression of differentiation-related genes is altered (18). K1 and K10, induced in control keratinocytes upon raising the calcium concentration from 0.05 to 0.12 mm (19), are suppressed in v-ras^Ha keratinocytes (18)and in chemically induced papillomas and carcinomas (20). The late markers, loricrin and filaggrin, are induced more rapidly and intensely in v-ras^Ha keratinocytes(18), reminiscent of the expression of these markers in keratinocytes treated with the phorbol ester,12-O-tetradecanoyl-phorbol acetate, an activator of PKC(19). Moreover, there is an increase in PKCα activity in v-ras^Ha keratinocytes (18),and the changes in marker gene expression seen in these cells can be reversed through the use of pharmacological inhibitors of PKC(18) or PKCα antisense oligonucleotides(10). We have shown that certain members of the AP-1 family are regulated in a PKC-dependent manner in differentiating keratinocytes (21). The correlation between PKC activation in keratinocytes and altered regulation of AP-1 proteins raises the possibility that the changes in marker gene expression in v-ras^Ha keratinocytes may be regulated by AP-1, and that these changes in gene expression may contribute to the transformed phenotype. Many keratinocyte differentiation-related marker genes contain AP-1 regulatory sequences (22, 23, 24, 25, 26, 27, 28, 29). Although AP-1 has been shown to activate the expression of many of these genes,there have been recent reports indicating that specific members of the AP-1 family can act as repressors of gene expression in various cell systems including keratinocytes (30, 31, 32, 33, 34, 35, 36, 37). Together, these studies highlight the complex nature of the activities of the AP-1 transcription factor and preclude the ability to generalize the effects of AP-1 in the regulation of gene expression. In the current study, we document changes in AP-1 protein expression between control and v-ras^Ha keratinocytes and provide evidence that AP-1 proteins are essential to the changes in marker gene expression seen in these cells. We also show that members of the AP-1 family can act as transcriptional repressors in the regulation of keratinocyte genes.

Primary keratinocytes were isolated from newborn BALB/c mice and maintained in Eagle’s Minimal Essential Medium with 0.05 mm calcium and 8% Chelex-treated fetal bovine serum(Gemini, Calabasas, CA) as described previously (38). On day 3 of culture, the cells were infected with a replication-defective retrovirus containing the v-ras^Ha gene(3) and are referred to as v-ras^Ha keratinocytes. The cells were harvested on day 8 of culture to investigate the effects of the v-ras^Ha gene on AP-1 protein expression. For the experiments involving the dominant-negative adenoviruses,v-ras^Ha keratinocytes were infected on day 6 with adenoviral constructs expressing the A-CMV, A-FOS, and A-VBP dominant-negative mutants as described below.

Frozen sections were prepared from papillomas and carcinomas excised from the backs of mice initiated with DMBA and promoted with 12-O-tetradecanoylphorbol-13-acetate. The sections were fixed in 4% formaldehyde prior to incubation with the Fos M peptide antibody (39), a biotinylated goat antirabbit secondary antibody (Vector Laboratories, Burlingame, CA) and horseradish peroxidase-conjugated avidin (Vector Laboratories). The binding of the antibodies was visualized with 3,3′-diaminobenzidine (Sigma Chemical Co., St. Louis, MO).

Poly(A)⁺ RNA was isolated from v-ras^Ha keratinocytes on day 8 after plating (day 5 after infection with v-ras^Ha; Ref. 3). One μg of poly(A)⁺ RNA was fractionated on a 1%formaldehyde/agarose gel, transferred to a nylon membrane (Nytran Plus;Schleicher and Schuell, Keene, NH), and hybridized with ³²P-labeled cDNA probes (described below). After hybridization at 42°C, the membranes were washed in 0.1× SSC(0.15 m NaCl, 0.015 m sodium citrate) at 65°C and visualized through autoradiography.

For cDNA synthesis, 5–10 μg of purified total RNA was reverse transcribed for 1 h at 42°C in a volume of 50 μl, containing 500 μm individual deoxynucleotide triphosphates, 10μ m DTT, 1.25 μm oligo-dT primer T_16–18, 20–40 units of RNAasin (Promega Corp.,Madison, WI), and 75 units of Superscript Reverse Transcriptase II(Life Technologies, Gaithersburg, MD). Reverse transcription reactions were diluted 5-fold, and for each PCR, 10% of the diluted reverse transcription reaction was used for amplification. Amplifications were performed in a volume of 20 μl in 96-well plates using a GeneAmp 9700 thermocycler in the presence of 50 μm deoxynucleotide triphosphates, 0.5 μm of each oligonucleotide primer,1.75 mm magnesium chloride, and 1 unit of AmpliTaq Gold(Perkin-Elmer Applied Biosystems). To avoid saturation or plateau effect of amplification, PCR was limited to a total of 25–30 cycles. Each reaction was performed twice using independent reverse transcription reactions to confirm reproducibility. The primers used for this analysis were: GAPDH forward sequence, TGT TCC TAC CCC CAA TGT GTC; GAPDH reverse sequence, TCT CTT GCT CAG TGT CCT TGC; K1 forward sequence, GCA AGA CCA AGA TCA ATC CCA C; K1 reverse sequence, AAA TTA AGG CGG CTC AGC G; K10 forward sequence, GAA TCG CAA GGA TGC TGA AG;K10 reverse sequence, TCT CCA GTC GGG TCT TGA TG; Loricrin forward sequence, TAC CTG GCC GTG CAA GTA AG; Loricrin reverse sequence, AAC AGG ATA CAC CTT GAG CGA C.

Nuclear extracts were isolated from v-ras^Ha keratinocytes on day 8 after plating (day 5 after infection with v-ras^Ha), according to the procedure of Schreiber et al. (40) with the modification that leupeptin, phenylmethylsulfonyl fluoride and aprotinin were added to the extraction buffer. For the preparation of SDS lysates from the same 60-mm dishes that were analyzed for nuclear protein expression,the cells were washed two times with PBS, and half the cells on the plate were scraped from the dish for making nuclear extracts. The remaining cells were lysed in 100 μl of 2× SDS lysis buffer [62.5 mm Tris (pH 6.8), 10% glycerol, 7.5% SDS, and 6% β-mercaptoethanol], scraped into a microcentrifuge tube, and boiled for 15 min. The tubes were centrifuged for 5 min, and the supernatant was transferred to a fresh microcentrifuge tube for storage at −70°C. For experiments where the entire dish was devoted to the preparation of SDS lysates, 200 μl of 2× SDS lysis buffer were added to the dish. To ensure equal loading of SDS lysates on the polyacrylamide gel, the samples were normalized through densitometry of the Ponceau S-stained membrane after transfer, and the samples were rerun a second time to obtain equal loading in each lane. Nuclear proteins (5 μg) were fractionated on a 10% SDS-polyacrylamide gel,SDS lysates were fractionated on a 7.5% SDS-polyacrylamide gel. In both cases, the proteins were transferred to a nitrocellulose membrane(Protran; Schleicher and Schuell), reacted with primary antibody(described below), horseradish peroxidase-conjugated secondary antibody(Bio-Rad, Richmond, CA), and visualized using chemiluminescence(Supersignal Reagent; Pierce, Rockford, IL).

The following antibodies were used in the Western blot analysis of AP-1 proteins in v-ras^Ha keratinocytes;anti-c-Jun sc-1694x, anti-Jun B sc-073x, anti-Jun D sc-74x, anti-Fra-1 sc-183x, anti-Fra-2 sc-604x (all from Santa Cruz Biotechnology),anti-Fos B PAI-831 (from Affinity BioReagents, Inc.), anti-Fos M peptide (from M. Iadorola, National Institute of Dental Research,Bethesda, MD), and anti-c-Fos (a kind gift from Rodrigo Bravo,Bristol-Meyers Squibb, Princeton, NJ). The antibodies against the mouse K1, K10, and K14 and loricrin have been described previously(41). The p21 Ras antibody was purchased from Transduction Laboratories (San Diego, CA). The anti-FLAG and anti-HA antibodies were purchased from Sigma and Roche Molecular Biochemicals (Indianapolis,IN), respectively.

cDNA probes were isolated by restriction digestion and purification using the Gelase enzyme (Epicenter, Madison, WI) prior to labeling with ³²P. The plasmids containing the inserts for the AP-1 proteins were obtained from the following individuals; Yusaku Nakabeppu, Kyushu University, Fukuoka, Japan (pSG5-Fos B); Tom Curran,St. Jude’s Children’s Hospital, Memphis, TN (CMVc-Fos); Donna Cohen,The Australian University, Canberra, Australia (CMVFra-2); Rodrigo Bravo, Bristol-Myers Squibb, Princeton, NJ (pFra-1); and Lester Lau,University of Illinois, Chicago, IL (pSG5c-Jun, pSG5JunB, and pSG5Jun D). For the reporter assays, the 5xTRE-CAT and pBLCAT constructs were obtained from Michael Karin (University of California, San Diego, CA). The pSG5-ΔFos B was obtained from Yusaku Nakabeppu, CMV-Fra-2 and CMV c-Fos are described above, and CMV-Fra-1 was obtained from Nancy Colburn (National Cancer Institute, Frederick Cancer Research Center,Frederick, MD). The pCMV and pSG5 control plasmids were purchased from Clontech (Palo Alto, CA) and Stratagene (La Jolla, CA), respectively.

For the generation of adenovirus, pAcCMV930 is a modification of PAcCMV.pLpA (42) created by inserting the following oligo,AAT TCC ACC ATG GAC TAC AAG GAC GAC GAT GAC AAG CAT ATG TGA TGA, into the EcoRI and HindIII sites. Insertion of the oligo replaced the original cloning site with the following sites: EcoRI, NcoI, epitope, NdeI, and HindIII. PAcCMV972 was generated by insertion of the following oligo, AAT TCC ACC ATG GCG TAT CCC TAC GAC GTG CCC GAT TAT GCC CAT ATG TGA TGA. The 930 oligo encodes a Flag epitope, and the 972 oligo contains a HA epitope. Dominant-negative vectors were generated by insertion of a cassette consisting of an NdeI sequence,phi10 sequences, an acidic amphipathic region and sequences corresponding to FOS and VBP (43, 44, 45, 46) zippers into the NdeI-HindIII site of pAcCMV930/972. Plasmids were sent to Bio Reliance Corp., and viral stocks were generated.

On day 7 after plating, the cells were transfected in triplicate with the 5xTRECAT or the parental pBLCAT constructs in serum-free medium using Lipofectamine reagent (Life Technologies). For cotransfection experiments with AP-1 expression vectors, the plasmids were administered at a ratio of 1:2, reporter:expression plasmid. After the 6-h transfection procedure, complete medium containing 0.05 mm calcium was added, and the cells were harvested 24 h later. Levels of CAT activity per μg of cell extract were determined according to the procedure of Neumann et al.(47). The results obtained with the 5xTRECAT vector are normalized to the activity of the pBLCAT plasmid to account for differences in transfection efficiency.

Nuclear extracts (3 μg) were incubated with a ³²P-labeled double-stranded probe containing the AP-1 consensus sequence (Promega) as described previously(37).

v-ras^Ha keratinocytes were infected with adenoviral constructs on day 6 after plating at MOIs ranging from 1 to 100 viral particles/cell. The cells were infected for 30 min in serum-free medium with 4 μg/ml Polybrene (Sigma) to enhance uptake. After 24 h, the medium was raised to 0.12 mm(by adding 200 mm CaCl₂),and 24–48 h later (depending on the experiment), the cells were harvested for isolation of nuclear extracts and preparation of SDS lysates (each plate was divided in half for these purposes, as described above).

Sections of chemically induced papillomas and carcinomas were analyzed for the expression of Fos family proteins using the Fos M peptide antibody (39), which is directed against the highly conserved Fos DNA binding domain (Ref. 48; Fig. 1). The ability of this antibody to detect Fos family proteins in keratinocytes has been shown previously (21). In normal skin, Fos proteins were found to be abundant in the suprabasal regions with weak expression of Fos members in the basal layer(21). Fos staining was generally weak in chemically induced papillomas except for focal areas of intense nuclear staining that were seen in the basal and suprabasal regions (Fig. 1). In the carcinomas that arose from the papillomas, intense nuclear Fos staining was seen throughout the tissue in all samples analyzed, demonstrating that Fos proteins are up-regulated in vivo during the process of skin carcinogenesis.

Because activation of AP-1 proteins can occur as a result of Ras activation (8, 49, 50), we wanted to determine the effects of v-ras^Ha activity on specific members of the AP-1 family in v-ras^Ha keratinocytes. Northern analysis of transcripts encoding AP-1 family members in control and v-ras^Ha keratinocytes (Fig. 2,A) indicates that transcripts encoding three members of the Fos family are induced in v-ras^Hakeratinocytes, the 2.2-kb message encoding c-Fos, the 5.1-kb message encoding Fos B, and the 1.8-kb message encoding Fra-1. In contrast, the 6.0-kb message for Fra-2 was significantly reduced in these cells (the fra-2 box in Fig. 2,A is overexposed ∼2-fold to enable the visualization of the fra-2 transcripts in v-ras^Ha keratinocytes). The levels of the transcripts encoding the Jun family were altered less dramatically in these cells. As shown in Fig. 2 A, the 2.1-kb message encoding Jun B and the 2.3-kb message encoding Jun D were slightly increased in v-ras^Ha keratinocytes,whereas the 3.1-kb c-Jun transcript appeared to be spliced into a 3.2-kb and a 2.7-kb message in the v-ras^Hakeratinocytes. The overall levels of the Jun transcripts were ∼25%lower in the v-ras^Ha keratinocytes relative to the levels in control cells.

Western blot analysis of nuclear extracts (Fig. 2,B)confirmed that the Fos proteins c-Fos, Fos B, and Fra-1 were significantly induced in v-ras^Hakeratinocytes, whereas Fra-2 was slightly decreased in these cells. These changes are consistent with the changes in transcript levels for c-fos, fos B, fra-1, and fra-2 shown in Fig. 2,A. However, the decrease in fra-2 message was more dramatic than the decrease in Fra-2 protein levels, suggesting that there may be posttranscriptional regulation of Fra-2 in v-ras^Hakeratinocytes. The size of the Fos B protein detected in these extracts(M_r 37,000) is consistent with the size of ΔFos B (51, 52, 53), and there is no indication that the full-length Fos B is expressed in control or v-ras^Ha keratinocytes. The M_r 48,000 Fra-2 protein represents a phosphorylated form of Fra-2 that migrates more slowly than the native protein on SDS polyacrylamide gels (54). As seen with the transcript levels, the expression levels of Jun proteins were not altered in v-ras^Ha keratinocytes with the exception of Jun B, which was slightly increased in these cells. The M_r 41,000 Jun D protein is a naturally occurring form of Jun D that is not sensitive to alkaline phosphatase treatment (55). Despite the formation of two messages for c-jun in v-ras^Ha keratinocytes,the levels of c-Jun protein remained unchanged between control and v-ras^Ha keratinocytes. Analysis of the AP-1 DNA binding activity in nuclear extracts from control and v-ras^Ha cells shows that the AP-1 DNA binding complex has a similar migration pattern and is slightly more intense in v-ras^Ha keratinocytes (Fig. 2 C).

Previously, we have demonstrated an obligate role for c-fosin the establishment of tumors by v-ras^Hakeratinocytes based on experiments in which v-ras^Ha keratinocytes that lack the c-fos gene were unable to form tumors when grafted to the backs of nude mice (17). To determine whether other Fos proteins are dependent upon the expression of c-fos and thus may also contribute to this effect, we looked at the expression of c-Fos, ΔFos B, and Fra-1 in c-fos null and wild-type keratinocytes (Fig. 2 D). These Fos proteins were examined because these are the Fos family members that are up-regulated by v-ras^Ha. This analysis demonstrated thatΔFos B and Fra-1 were induced by v-ras^Hain both genotypes, indicating that neither of these Fos proteins is dependent upon c-fos for expression, and that these factors are not critical for the transformation of v-ras^Ha keratinocytes. Moreover, this result suggests that neither of these factors alone is sufficient to cause tumor formation when v-ras^Hakeratinocytes are grafted in vivo.

Previously, we have shown that the early markers, K1 and K10, are suppressed in keratinocytes infected with the v-ras^Ha retrovirus (18). To determine whether this effect is mediated by c-Fos, we looked at the expression of K1 and K10 in control and v-ras^Ha keratinocytes maintained for 24 h in medium containing 0.12 mm calcium. These conditions are required for the induction of K1 and K10 in wild-type and c-fos null (56) keratinocytes. As shown in Fig. 3, both K1 and K10 were induced in wild-type cells and equally suppressed by v-ras^Ha in the wild-type and c-fos null keratinocytes.

A reporter construct consisting of a multimer of the AP-1 target sequence (5xTRE) cloned upstream of the CAT gene (57) was used to evaluate the levels of AP-1 transcriptional activity in control and v-ras^Ha keratinocytes. As shown in Fig. 4 A, AP-1 activity was increased in v-ras^Ha keratinocytes, indicating that the AP-1 transcription factor is activated in these cells, and that activation of the AP-1 reporter correlates with elevated levels of AP-1 proteins. The relative v-ras^Ha-mediated increases in AP-1 activity ranged from 1.7- to 5.5-fold in our experiments; such variability is likely to be a reflection of differences in the efficiency of infection with the v-ras^Ha retrovirus.

Because c-Fos, ΔFos B, and Fra-1 are induced in v-ras^Ha keratinocytes, we wanted to determine whether these factors contribute to the increased levels of AP-1 transcriptional activity. The 5xTRE reporter was cotransfected into control keratinocytes with expression vectors encoding c-Fos,Fra-1, and ΔFos B (compared with the parental expression vectors alone). The expression of c-Fos, Fra-1, and ΔFos B proteins from these vectors has been determined through Western blot analysis of nuclear extracts from transfected keratinocytes (not shown). As shown in Fig. 4 B, both ΔFos B and c-Fos caused a significant increase in AP-1 reporter activity in normal keratinocytes, suggesting that the induction of these proteins in response to the v-ras^Ha virus may contribute to the increased levels of AP-1 transcriptional activity in v-ras^Ha keratinocytes. In contrast, Fra-1 did not increase TRE activity in keratinocytes, raising the possibility that Fra-1 may act as a transcriptional silencer in keratinocytes.

Having identified changes in the expression of AP-1 proteins in v-ras^Ha keratinocytes, we wished to establish a role for AP-1 in regulating gene expression in these cells. Toward this end, we performed a dual infection of keratinocyte cultures with the v-ras^Ha retrovirus and adenoviral constructs expressing dominant-negative forms of Fos (A-FOS), VBP(A-VBP), or an adenoviral vector control (A-CMV). The A-FOS mutant dimerizes with Fos partners (presumably Jun proteins) through an intact dimerization domain. However, the heterodimer is unable to bind to DNA because of the replacement of the DNA binding region in the dominant-negative mutant with an acidic amphipathic fragment. The amphipathic fragment in A-FOS interacts with the basic region of JUN family members, which prevents DNA binding (58, 59). The effects of A-FOS on gene expression in v-ras^Ha keratinocytes were evaluated by examining the expression levels of the early differentiation markers,K1 and K10, and the late marker, loricrin, through Western blot analysis (Fig. 5,A) and of the transcripts encoding these genes through RT-PCR(Fig. 5,D). The presence of A-FOS in v-ras^Ha keratinocytes restored the expression of the protein levels of the early markers K1 and K10,indicating that these markers are suppressed in v-ras^Ha cells through an AP-1-dependent mechanism. This effect was detectable when the cells were infected at a MOI of 1 viral particle/cell but was more pronounced at an MOI of 10,an amount of virus that blocked AP-1 DNA binding activity (Fig. 5 B). A-Fos also increased the expression of the cornified envelope precursor, loricrin. The expression of loricrin was similarly induced in the cells infected with A-FOS at both an MOI of 1 and 10. A-VBP and A-CMV adenoviral constructs did not alter the expression of these keratinocyte marker proteins.

To confirm that A-FOS produced peptides capable of interfering with the DNA binding activity of AP-1, we evaluated the intensity of AP-1 DNA binding activity in control and adenovirus-infected cells. AP-1 DNA binding activity was relatively unaffected by A-FOS at an MOI of 1; however, AP-1 DNA binding was significantly reduced at an MOI of 10 and completely blocked at an MOI of 100 (Fig. 5,B). In contrast, AP-1 DNA binding activity was not affected by infection with the A-CMV construct at an MOI of 1, 10, or 100 or by the A-VBP construct at an MOI of 1 or 10. Confirmation that the adenoviruses are expressed in keratinocytes is indicated by the expression of the HA and FLAG tags in the double-infected keratinocytes, as determined by analysis of both total cell lysates (Fig. 5,B) and nuclear extracts (Fig. 5,C). Although restoration of K1 protein in the presence of A-FOS occurred in a significant number of our experiments, this was not always a reproducible finding for reasons that are unexplained. In contrast, K10 and loricrin protein levels consistently increased in the presence of A-FOS. The expression of keratin 14 was not affected in the adenovirus-treated cells, and the changes in marker gene expression seen in the presence of A-FOS were not attributable to alterations in the levels of v-ras^Ha gene, as determined through detection of the M_r 21,000 Ras gene product in extracts from these cells (Fig. 5 A).

To determine whether the effects of A-FOS on the expression of K1, K10, and loricrin occur of at the level of gene expression, we have evaluated the levels of transcripts encoding K1 and K10 and loricrin in A-CMV and A-FOS cells using RT-PCR. As shown in Fig. 5,D, the expression levels of mRNA for K1 and K10, which were low in A-CMV v-ras^Ha keratinocytes, increased substantially in the A-FOS v-ras^Hakeratinocytes. The increase in the levels of transcripts encoding K1 and K10 corresponds to the increases in K1 and K10 protein levels shown in Fig. 5 A and confirms that AP-1 proteins suppress the expression of these marker proteins at the level of gene expression. The levels of the loricrin transcripts were decreased in the A-FOS v-ras^Ha keratinocytes, indicating that the increase in the expression of the loricrin protein in A-FOS v-ras^Ha keratinocytes is likely to occur posttranslationally rather than at the level of gene expression.

Previously, we have shown that c-fos plays an obligate role in the development of skin cancers in mice, that c-fosis required for v-ras^Ha keratinocytes to form tumors when grafted to the backs of nude mice, that v-fos can restore tumorigenic potential to c-fosnull v-ras^Ha keratinocytes(17), and that the expression of v-fos together with v-ras^Ha is sufficient to transform primary mouse keratinocytes into cancer cells (15, 16). In addition, as shown in Fig. 1, the expression of Fos proteins increases during the malignant progression of skin tumors in mice. This may correlate with the induction of c-Fos, Fra-1, and ΔFos B seen in v-ras^Ha keratinocytes because the Fos M peptide antibody is capable of detecting all of these proteins. Although it is clear that c-Fos plays a causative role in the progression of skin tumors, whether ΔFos B and Fra-1 contribute to the tumorigenic potential of keratinocytes is not clear.

Here, we provide evidence that certain changes in marker gene expression that occur in v-ras^Hakeratinocytes are mediated by AP-1 proteins. Use of a dominant-negative Fos construct provided evidence that AP-1 proteins suppress K1 and K10 in v-ras^Ha keratinocytes because it was possible to restore the expression of these proteins by inhibiting AP-1 DNA binding activity. That transcripts encoding K1 and K10 also increase in A-FOS cells confirms that the effects of this dominant-negative mutant occur at the level of gene expression,consistent with the role of AP-1 as a transcriptional regulator. Although there is the potential for a large number of genes to be up-regulated by AP-1 (60), there are also examples of AP-1 proteins that suppress transcription (30, 31, 32, 33, 34, 35, 36). In general,transcriptional repression by AP-1 tends to involve specific members of the AP-1 family (JunB, Jun D, Fra-1, and Fra-2) that are distinct from c-Fos and c-Jun. Suppression of K1 was reversed by A-FOS in many by not all of the experiments performed. The AP-1 element that regulates K1 expression in response to calcium has been characterized(27) and is a composite element with a steroid hormone-response element recognition sequence. The composite element responds to both AP-1 factors and hormones such as retinoids and vitamin D. The complexity in this portion of the K1 regulatory region could modify the response to A-FOS in individual experiments. The regulatory sequences for the murine K10 gene have not been described in the literature, and thus the nature of the putative AP-1 site that regulates K10 expression is unknown. Thus, it is not known,for example, whether the K10 AP-1 site contains a deviated AP-1 binding sequence that may select for the binding of AP-1 factors that suppress transcriptional activity (such as the AP-1 factors mentioned above)because the affinities of different AP-1 heterodimers for AP-1-related sequences can vary (61). The focal up-regulation of Fos proteins in skin papillomas as shown here occurs at a time when the expression of the early markers of keratinocyte differentiation is suppressed in a focal pattern (20). In carcinomas, these early markers are completely suppressed. Our molecular data suggest that these events are linked and that the changes in early marker expression in tumors may be attributable to the activities of AP-1 factors.

Through this and previous studies, we can speculate on the role of the Fos proteins that are induced in v-ras^Hakeratinocytes (c-Fos, ΔFos B, and Fra-1) and thus are likely to help define the v-ras^Ha phenotype:

(a) In consideration of changes in marker gene expression,studies with the A-FOS adenovirus indicate that the changes in K1 and K10 expression that occur in v-ras^Hakeratinocytes are mediated by AP-1 proteins. This effect is not controlled by c-Fos because the expression of K1 and K10 is suppressed to a similar extent by v-ras^Ha in both wild-type and c-fos null keratinocytes (Fig. 3). Fra-1 is a strong candidate to mediate this suppression based on: (i)Fra-1 is undetectable in control keratinocytes but is expressed to high levels in v-ras^Ha keratinocytes (Fig. 2,B); (ii) Fra-1 does not activate an AP-1 reporter construct in keratinocytes and thus has the potential to squelch AP-1 transcriptional activity by occupying AP-1 sites without activating transcription (Fig. 4 B) as described (62, 63); and (iii) Fra-1 has been shown to act as a transcriptional suppressor in other systems (30, 31, 32, 64, 65).

(b) In the promotion of tumorigenesis and based on the studies described above, it appears that c-Fos is the essential Fos member for tumor progression. A comparison of the expression patterns of Fos proteins in keratinocytes from wild-type and c-fosnull mice indicates that ΔFos B and Fra-1 are induced to a similar extent in both genotypes (Fig. 2 D). Because wild-type keratinocytes (with a v-ras^Ha gene)support the development of malignant tumors in vivo and in vitro and c-fos null v-ras^Ha keratinocytes do not(17), ΔFos B and Fra-1 must not be sufficient to establish a tumorigenic phenotype in the absence of c-fos. Thus, we can assert that malignant potential of v-ras^Ha keratinocytes is attributable to the induction of c-Fos and that Fra-1 and ΔFos B contribute to the transformed phenotype in other ways.

The expression of loricrin protein was increased above control levels in cells infected with A-FOS. The promoter for the mouse loricrin gene contains an AP-1 site that is necessary for loricrin expression in transgenic mice expressing the β-galactosidase gene cloned downstream of the loricrin promoter (66). The effect of A-FOS to reduce loricrin transcript (Fig. 5 D) is consistent with a direct involvement of an AP-1 protein(s) in the regulation of the loricrin gene. A-FOS could also block the expression of keratinocyte transglutaminase through an AP-1-related mechanism(67). This could result in the accumulation of excess loricrin protein in the cells that, because of insufficient transglutaminase, would not be incorporated into cornified envelopes. Altogether, the changes in marker gene expression in v-ras^Ha keratinocytes reflect changes in signaling events that may lead to the reprogramming of keratinocytes as they acquire tumorigenic potential. Our results indicate that AP-1 proteins play an important role in this process.

Because AP-1 proteins and/or AP-1 activity have been shown to play an obligate role in defining the transformed phenotype of epidermal cells in various model systems, and the changes in marker gene expression in v-ras^Ha keratinocytes correlate with the onset of the transformed phenotype, developing approaches to inhibit AP-1 effects in keratinocytes could hold preventive or therapeutic potential for the treatment of skin tumors. Currently, studies are being designed to evaluate the efficiency of the A-FOS vector in vivo to restore normal patterns of gene expression and cause regression of chemically induced tumors in mice.

We thank Christina Cheng for technical assistance and Christa Walter for help in preparing the manuscript.