| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Cell, Tumor, and Stem Cell Biology |
1 Laboratory of Cellular Carcinogenesis and Tumor Promotion and 2 Laboratory of Metabolism, National Cancer Institute, Center for Cancer Research, Bethesda, Maryland
Requests for reprints: Stuart H. Yuspa, Laboratory of Cellular Carcinogenesis and Tumor Promotion, National Cancer Institute, Center for Cancer Research, Bldg. 37 Rm. 4068A, Bethesda, MD 20892. Phone: 301-496-2162; Fax: 301-496-8709; E-mail: yuspas{at}mail.nih.gov or Charles Vinson, Laboratory of Metabolism, National Cancer Institute, Center for Cancer Research, Bldg. 37, Rm 3128, Bethesda, MD 20892. Phone: 301-496-8753, Fax: 301-496-8419; E-mail: Vinsonc{at}mail.nih.gov.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The mouse epidermal multistage carcinogenesis model provides a well-defined system for examining the evolution of squamous epithelial cells into benign squamous papillomas and their subsequent progression into squamous cell carcinomas (6). The majority of these tumors contain mutations in the H-ras oncogene. A function for AP-1 activity in skin tumorigenesis has been identified using knockout and transgenic mice that modulate AP-1 components. For example, papillomas in c-Fos knockout mice did not convert into malignant carcinomas (7). Deletion of c-Jun in skin produces mice that develop smaller papillomas (3). Expression of a c-Jun dominant negative in skin dramatically reduced papilloma formation in several protocols of cutaneous chemical carcinogenesis (5, 8) and reduced squamous cell cancer induction by UVB exposure (9).
Furthermore, upstream activators of the AP-1 complex also modulate tumor formation (10). AP-1 activity in the skin is regulated through a cascade of signaling events involving protein kinase C (PKC), Ras, and the p38 family of mitogen-activated protein kinases (11). Additionally, activation of AP-1 by c-Jun NH2-terminal kinase (12), PKC family members (13, 14), and tumor necrosis factor-
(15) and constitutive activation of Ras promote tumorigenesis (16–18).
Although previous studies have shown that specific AP-1 factors modify tumor incidence or progression (1, 7), studies to date have been unable to examine the effects of AP-1 modulation after tumor development. We have used the two-transgene tetracycline system (19) to regulate expression of a high-affinity AP-1 dominant-negative protein (A-FOS). A-FOS is a chimeric protein containing the c-FOS leucine zipper and a designed acidic protein sequence that replaces the DNA-binding region. A-FOS heterodimerizes with c-FOS dimerization partners, which are primarily the JUN family of proteins (c-Jun, JunB, and JunD; ref. 20) to prevent AP-1 DNA binding (21). By establishing a mouse model whereby AP-1 activity in the skin is "on" or "off," we can document the influence of AP-1 transcriptional activity on specific stages of tumor development. Here, we report the powerful influence of AP-1 on determining tumor cell lineage and premalignant progression in the mouse skin model of multistage carcinogenesis.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Mice from the keratin 5-tet-transactivator (K5-tTA) FVB/N line (23) were crossed with homozygous mice from the tetA-FOS line to produce double transgenic mice (K5/A-FOS) with tetracycline-regulated expression of A-FOS restricted to K5-expressing cells. For convenience, the tetA-FOS pups from this cross are referred to as wild type (WT). During mating and gestation, mothers were kept on doxycycline feed (20 mg/kg of feed; Bio-serve, Baltimore, MD) to repress expression of A-FOS and switched to a normal diet at weaning, which induced A-FOS expression. The TRE-Luciferase (TRE-LUC) mouse was originally described by Rincon et al. (24). This strain has 12-O-tetradecanoylphorbol-13-acetate/phorbol 12-myristate 13-acetate (PMA)–inducible luciferase expression dependent on the TRE (AP-1) site (24). K5/A-FOS mice were crossed with TRE-LUC mice to generate triple transgenic mice. Mice were sacrificed according to protocols approved by the National Cancer Institute Animal Care and Use Committee. Genotypes were determined by PCR analysis of DNA from tail biopsies. The following forward and reverse primer pairs were used: tetA-FOS, 5'-CCACGCTGTTTTGACCTCCATAG-3' and 5'-ATTCCACCACTGCTCCCATTC-3'; K5-tTA, 5'-AACAACCCGTAAACTCGCCC-3' and 5'-GCAACCTAAAGTAAAATGCCCCAC-3'; TRE-LUC, 5'-GCGGAATACTTCGAAATGTC-3', 5'-TTAGGTAACCCAGTAGATCCCC-3'.
Electrophoretic mobility shift assay. Six B-ZIP dimers, including the FOS|JUND heterodimer, CAAT/enhancer binding protein (C/EBP), cyclic AMP–responsive element binding protein (CREB), protease-activated receptor (PAR), activating transcription factor 6 (ATF6), and interleukin-3-regulated nuclear factor (NFIL3) homodimers, were mixed with 0, 1, 10, or 100 molar equivalents of A-FOS in a gel shift reaction buffer [25 mmol/L Tris-HCl (pH 8), 50 mmol/L KCl, 0.5 mmol/L EDTA, 2.5 mmol/L DTT, 1 µg/µL bovine serum albumin (BSA), 10% glycerol] to a final volume of 20 µL and incubated at 65°C for 20 minutes. The protein mix was cooled to room temperature, and 7 pg of double-stranded 32P-radiolabeled 28-mer oligonucleotide was added and incubated at 37°C for 20 minutes. The sequences of the 28-mer DNA probes are GTCAGTCAGAATGACTCATATCGGTCAG (AP-1) for FOS/JUND heterodimer, GTCAGTCAGATTGCGCAATATCGGTCAG for C/EBP, GTCAGTCAGATTACGTAATATCGGTCAG for PAR, and GTCAGTCAGATGACGTCATATCGGTCAG for CREB, ATF6, and NFIL3. Samples were separated in 7.5% acrylamide gel; the gel was dried and exposed for autoradiograghy.
To confirm the inhibition of DNA binding by A-FOS expression in transgenic mice, K5/A-FOS or WT primary keratinocytes were cultured in 0.05 mmol/L Ca2+ medium and PMA treated for 6 hours in the presence or absence of doxycycline. Nuclear extracts were prepared using the NE-PER reagent (Pierce, Rockford, IL); 10,000 cpm of labeled oligonucleotide was added to 2 µg of protein sample and then incubated in binding buffer (10 mmol/L HEPES, 80 mmol/L KCl, 0.05 mmol/L EDTA, 6% glycerol, 1 mmol/L DTT, and 1 mmol/L MgCl2) at 37°C for 20 minutes. Samples were separated on a 5% PAGE gel in 0.25x Tris-borate EDTA at 150 V for 2 hours. Oligonucleotides used contained DNA binding sites for AP-1 and CRE (25). To verify specificity of DNA binding, 50-fold unlabeled probe was added in a competition assay (26). To test the direct effect of recombinant A-FOS protein on the electrophoretic mobility shift assay (EMSA) results, WT FVB/N primary keratinocytes were cultured in 0.05 mmol/L Ca2+ media. Nuclear extracts were prepared identically except that the nuclei from one 150-mm dish (1.2 x 107 cells) were lysed in the presence of 2.5 ng pure recombinant A-FOS protein and incubated for 2 hours at 38°C before EMSA. These nuclear extracts were incubated with oligonucleotides containing binding sites for AP-1, CRE, C/EBP, and SP1 as described above.
Determination of luciferase activity. TRE-LUC and K5/A-FOS/TRE mice were placed on normal feed for 2 weeks before treatment with 5 µg PMA in 25 µL acetone on one ear, whereas the other ear was treated with acetone alone. Extracts were prepared directly in lysis buffer (Promega, Madison, WI) after 6 hours. Luminescence was quantified from 25 µL of extract in 200 µL luciferase reagent in triplicate from three mice for each group with error bars representing SE. Primary keratinocytes were cultured in 0.05 mmol/L calcium-containing media in the presence of doxycycline (15 ng/mL). Medium was changed to doxycycline-free medium 24 hours before the addition of PMA (100 ng/mL). After 4 hours, lysates were prepared (Promega), and luciferase activity was determined. For the TOP-FLASH (Upstate Biotechnology, Charlottesville, VA) ß-catenin reporter assays, cells were plated in 24-well plates and transfected with either TOP– or the mutant FOP– reporter at 0.8 µg plasmid per well with LipofectAMINE 2000 transfection reagent. After 6 hours, medium was changed to 0.05 mmol/L calcium growth medium overnight. LiCl (10 mmol/L) was added 6 hours before harvest, and assays were done as above.
Tumor initiation/promotion studies. Mice at 8 to 10 weeks of age were initiated with 7,12-dimethylbenz(a)anthracene (DMBA; 100 µg/200 µL acetone) on shaved dorsal skin. PMA (5 µg/200 µL acetone 1x/wk) promotion was started 1 week later and continued for 20 weeks, and tumor number was recorded weekly. Protocols used for dietary doxycycline are described in the Results and figure legends. Study groups included equal numbers of male and female mice for all groups. In Fig. 2C, animal numbers were WT [+Dox (n = 17), –Dox (n = 9)] and K5/A-Fos [+Dox (n = 14), –Dox (n = 10)]. In Fig. 2D, animal numbers were WT [+Dox (n = 16), –Dox (n = 15)] and K5/A-FOS [+Dox (n = 11), –Dox (n = 10)]. Some animals were sacrificed at time points after 27 weeks to examine tumor histology.
|
Immunohistochemistry. Tissue sections were deparaffinized, rehydrated through a graded series of ethanol followed by PBS, and blocked with 10% normal goat serum/3% BSA/PBS for 1 hour. Primary antibodies were applied: rabbit anti-keratin 5 or 6-FITC conjugated (Covance, Princeton, NJ), guinea pig anti-ADRP (Research Diagnostics, Flanders, NJ), rabbit anti-ß-catenin (Sigma, St. Louis, MO), goat anti-Indian Hedgehog (IHH; Santa Cruz Biotechnology, Santa Cruz, CA), and goat anti-FRP3 (R&D Systems, Minneapolis, MN) in a solution of 3% BSA/PBS. Slides were washed in PBS before application of respective FITC- or Cy3-conjugated donkey secondary antibodies (Jackson Immunologicals, West Grove, PA). Slides were stained with 4',6-diamidino-2-phenylindole for visualization of nuclei before fluorescence microscopy.
Immunoblot analysis. Total protein lysates were prepared by extracting primary keratinocyte cultures with M-PER reagent (Pierce). Protein extracts were loaded into NuPage gels (Invitrogen, Carlsbad, CA) and run according to manufacturer's protocols. Blotted extracts were blocked in TBS-Tween + 3% BSA. Primary antibodies diluted in TBS-Tween + 3% BSA were incubated at 4°C overnight. Secondary antibodies conjugated to horseradish peroxidase were applied for 30 minutes at room temperature diluted in TBS-Tween + 3% BSA. Immunoblots were developed using enhanced chemiluminescence reagents from Pierce. Bound antibodies were detected using the Pierce Pico substrate and exposing to BioMax film (Kodak, Rochester, NY). Membranes were reprobed for actin protein to confirm equivalent loading between the samples. Immunoblot detection of A-FOS protein in cultured K5/A-FOS primary keratinocytes grown in the presence or absence of doxycycline for the times indicated. Cellular lysates were collected and analyzed on a 12.5% SDS-PAGE gel. A-FOS protein was detected with a c-FOS antibody (Santa Cruz Biotechnology) that recognizes the conserved region between A-FOS and c-FOS. Keratinocyte lysates from single transgenic tetA-FOS mice or 0.2 pg recombinant A-FOS protein (rA-FOS) were included as negative and positive controls respectively (data not shown).
Reverse transcription-PCR analysis. RNA from primary keratinocytes was isolated with Trizol (Invitrogen) and DNase treated (Ambion, Austin, TX). Initial-strand cDNA was first prepared (Invitrogen) and used for reverse transcription-PCR with equal amounts of synthesized cDNA determined by equal amplification of ß-actin. For PCR amplification, the forward and reverse primer pairs are listed in the Supplementary Data.
Oil Red-O analysis. WT and K5/A-FOS skin keratinocytes were cultured in 0.05 mmol/L calcium growth media for 48 hours, and then differentiation was induced in 0.12 mmol/L calcium medium for 48 hours before fixation in 4% paraformaldehyde and staining with Oil Red-O. To quantify the number of Oil Red-O cells, 25 fields were counted for each genotype using a x10 objective, and mean ± SE was determined for each genotype.
Chromatin immunoprecipitation: ChIP on chip. Binding of phospho-c-JUN, CREB, and C/EBPß to the promoter of 13 genes shown on Fig. 6A was examined using ChIP-on-chip assay with mouse promoter microarrays from NimbleGen (Madison, WI). Primary mouse keratinocytes were treated with 1% formaldehyde for 8 minutes, and the reaction was stopped by glycine. The chromatin immunoprecipitation (ChIP) was done with Upstate ChIP reagents and protease inhibitors (Roche, Indianapolis, IN). The cells were washed, collected in cell lysis buffer, sheared by ultrasound producing DNA fragments of 700 bp average size. The lysate was centrifuged, diluted 20 times, precleared with BSA- and yeast tRNA-saturated protein A/G agarose mix (Invitrogen), and immunoprecipitated with antibodies against phospho-c-JUN (Santa Cruz Biotechnology), CREB (Upstate Biotechnology), and custom C/EBPß antibody. The ChIP with nonspecific IgG was used as negative control. The ChIP eluate was incubated with NaCl at 65°C overnight, at 95°C for 1 hour, and digested with Proteinase K and extracted with phenol. The ChIP DNA was purified using Qiagen (Valencia, CA) MinElute kit; 5 ng of ChIP DNA were randomly amplified using "RoundA/B" protocol (28) with Cy5-labeled primers in 400 µL total PCR volume. The PCR reaction consisted of 42 cycles and produced 20 µg of product. The total genomic DNA ("input") from an aliquot of the sonicated cell lysate was extracted, amplified with Cy3 primers in parallel, and used as a reference in hybridization. The PCR product was cleaned by ethanol precipitation and hybridized in presence of yeast tRNA, COT1 DNA, and poly-dATP (10 µg each) overnight in MAUI hybridization station at 42°C to mouse promoter microarrays (Nimblegen). Each array contained promoters of 21,815 unique genes. Each promoter was represented by 15 unique 50-mer oligonucleotides located from 1,300 bp upstream to 200 bp downstream of transcription start. The arrays were washed with 2x SSC, 0.1 % SDS for 2 minutes, 1x SSC for 1 minute, and with 0.2 x SSC for 15 seconds; scanned with Axon 4000B scanner; and analyzed with NimbleScan software.
|
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
7-fold in WT skin. However, in the presence of A-FOS expression, basal TRE activity was unaffected, whereas PMA-induced TRE activity was nearly eliminated (Fig. 1D). In primary keratinocyte cultures produced from triple transgenic newborn pups, A-FOS expression suppressed both basal and induced activity. The difference in A-FOS inhibition of basal TRE activity in vivo and in vitro most likely reflects the regulation of the K5 promoter that is down-regulated in differentiating epidermis but not cultured keratinocytes. Thus, A-FOS inhibits TRE-dependent transcriptional activity in both skin and cultured keratinocytes, presumably by inhibiting AP-1 binding to the TRE sequence. Previously, we have shown that the AP-1 complex in murine epidermis encompasses c-Jun, Jun B, Jun D, c-Fos, Fra 1, and Fra 2, whereas c-Fos is not detected in cultured keratinocytes (26). A series of EMSA were used to evaluate the specificity of A-FOS at inhibiting the binding of nuclear extracts to different DNA sequences. Nuclear extracts isolated from primary newborn keratinocytes expressing A-FOS for 24 hours after the removal of doxycycline do not bind either in the basal or PMA-stimulated case to a TRE-containing oligonucleotide (5'-TGAGTCA-3'). However, these nuclear extracts do bind to the closely related CRE-containing oligonucleotides (Fig. 1F). To evaluate if inhibition of TRE binding was a direct consequence of the A-FOS protein, recombinant A-FOS protein was added to nuclear extracts prepared from WT keratinocytes. The EMSA indicates that exogenous A-FOS inhibited DNA binding to TRE-containing DNA oligonucleotide but not three additional oligonucleotides containing consensus-binding sites for CREB, C/EBP, or SP1 (Fig. 1G). Furthermore, recombinant A-FOS protein specifically inhibits DNA binding of AP-1 dimers (c-FOS|JunD) but not five other B-ZIP dimers in EMSA assays at equimolar equivalents (Fig. 1H). Interference with C/EBP binding is detected only at 10- to 100-fold molar excess. Thus, A-FOS specifically and directly inhibits DNA binding and transcriptional activity of the TRE.
A-FOS expression in skin causes hyperplasia of sebaceous glands. Expression of A-FOS in mouse epidermis did not produce an obvious phenotype until late in life when mice often developed mild hair loss (alopecia), sebaceous gland hyperplasia (particularly but not exclusively on the eyelids), and focal skin erosions (Fig. 2A and B ). In the focal erosions, the hyperplastic sebaceous glands became independent of the adjacent hairs follicles that were degenerating.
cDNA microarrays were used to further characterize the consequences of expressing A-FOS in the epidermis. One of the up-regulated genes is adipocyte differentiation–related protein (ADRP), which functions in lipid sequestration (29). Immunohistochemical localization of ADRP indicates that it is a novel marker for sebocytes in developing sebaceous glands from newborn skin (Fig. 3C, inset, red ) and mature sebaceous glands (data not shown). These data further support the observation that A-FOS expression shifts the balance toward the sebaceous lineage.
|
The three groups of mice with normal AP-1 activity (WT mice on doxycycline, WT mice off doxycycline, and K5/A-FOS mice on doxycycline) developed two to three tumors per mouse during the course of two independent carcinogenesis protocols (Fig. 2D and E). These tumors were large (100-300 mm3) glistening lesions typical of squamous papillomas (Fig. 2F), and this diagnosis was confirmed with histologic examination of the tumors (Fig. 3A). As expected, these tumors had the A-to-T mutation in codon 61 of the H-ras gene characteristic of DMBA-initiated papillomas (ref. 30; Fig. 2H). In contrast, the K5/A-FOS mice continually expressing A-FOS developed approximately five times more tumors per mouse (Fig. 2D and E) that were small (2-8 mm3) and pale yellow (Fig. 2G). Histologic examination of these tumors identified 91% well-differentiated sebaceous adenomas (Fig. 3B; Table 1 ), an uncommon tumor type that arises from the sebaceous gland of the hair follicle. These tumors also contained the characteristic H-ras mutation found in control papillomas (Fig. 2H). Tumors from all groups were dependent on DMBA mutagenesis, as PMA treatment alone for 20 weeks did not produce tumors (Table 1). One third of the squamous papillomas from control (tetA-FOS) mice converted to carcinomas, whereas none of the tumors on the K5/A-FOS mice expressing A-FOS underwent malignant conversion (summarized in Table 1).
|
Inhibiting AP-1 activity in squamous tumors induces sebaceous adenomas. We used the inducible capacity of the tetracycline system to induce A-FOS expression in K5/A-FOS mice after squamous papillomas formed. The doxycycline diet was switched to a normal diet 27 weeks after the initiation of the carcinogenesis protocol when squamous papillomas had formed and tumor growth continued for an additional 22 weeks. Expression of A-FOS results in an 80% reduction in malignant conversion, indicating that inhibition of AP-1 activity after papilloma formation is sufficient to block or delay carcinoma formation (Table 1). Provocatively, approximately half of the tumors showed a profound alteration of the standard squamous papilloma histology not observed in control mice that do not express A-FOS: 26% of the tumors became mixed sebaceous/squamous lesions, 14% converted to pure sebaceous adenomas, and 53% remained squamous papillomas (see Table 1 for summary; Fig. 3D and E). These results show that continuous AP-1 activity is necessary to maintain the squamous phenotype, and after A-FOS expression, there is a conversion from a squamous to sebaceous phenotype.
Activating AP-1 activity in sebaceous adenomas induces squamous tumors. To determine if restoring AP-1 activity in sebaceous adenomas could produce a reciprocal conversion of sebaceous tumors into squamous tumors, A-FOS expression was repressed in K5/A-FOS mice with existing sebaceous adenomas to reactivate AP-1 activity. Twenty-seven weeks after the initiation of the carcinogenesis protocol, K5/A-FOS mice were switched to a doxycycline diet to restore AP-1 activity, and tumors were collected at multiple time points. Twenty percent of sebaceous adenomas converted to papillomas; 24% were mixed squamous/sebaceous tumors; and 56% remained pure sebaceous adenomas when tumors were analyzed for all time points (Table 1). Mixed tumors were detected after only 5 weeks of AP-1 activity, whereas a complete change in tumor phenotype into squamous tumors was not observed until 22 weeks.
In mixed tumors arising from either sebaceous adenomas or squamous papillomas, the topology was similar; the squamous portion was at the periphery; and the sebocytes were in a central portion of the tumor mass (Fig. 3D and E). In addition to modulating tumor lineage identity, continual suppression of AP-1 activity inhibited malignant progression of sebaceous tumors (Table 1). However, when AP-1 activity was restored to sebaceous adenomas, 5% of the sebaceous tumors converted to sebaceous carcinomas. These data are similar to the inhibition of malignant conversion of squamous papillomas expressing A-FOS, indicating that AP-1 activity is necessary for malignant conversion of both squamous and sebaceous tumors of the skin.
The cellular composition of the mixed tumors was complex. Although certain cells clearly seemed to be either squamous or sebaceous by histologic criteria, many cells seemed to have morphologic similarity to sebocytes with condensed central nuclei, while maintaining a basophilic staining akin to squamous keratinocytes (Fig. 3E). Immunohistochemical analysis of mixed tumors derived by expressing A-FOS in squamous papillomas identified individual cells that express both K6 and ADRP (Fig. 3F), suggesting transdifferentiation of squamous cells into sebaceous cells. However, in some tumors where AP-1 activity has been inhibited after papilloma formation, cells are observed that express ADRP but not K6, suggesting a complete transdifferentiation from squamous cells to sebaceous cells (Fig. 3G). Likewise, in mixed tumors derived through activating AP-1 in sebaceous adenomas, a similar colocalization of squamous and sebaceous markers in the same cells is observed (Fig. 3H, IHH and K6 are shown).
No differences in proliferation or cell death are observed in the mixed tumors compared with either the papillomas or sebaceous adenomas. Tumor regression was not detected after A-FOS induction (based on consistent tumor number and size), suggesting that sebaceous conversion was not due to death of squamous cells and selection of cells of the sebaceous lineage. All tumor types showed equivalent levels of cellular proliferation based on nuclear proliferating cell nuclear antigen staining (data not shown). Furthermore, apoptotic cells were not detected by terminal deoxynucleotide transferase–mediated nick-end labeling or active caspase-3 staining (data not shown) of papillomas, sebaceous adenomas, or mixed tumors.
Recent reports indicate that the sebaceous cell fate results from destabilization of the ß-catenin protein (33). To determine if A-FOS expression affects ß-catenin protein abundance and distribution, papillomas (Fig. 4A ), sebaceous adenomas (Fig. 4B), and mixed tumors (Fig. 4C and D) were immunostained for ß-catenin. Papillomas stain strongly for ß-catenin, with a membrane and diffuse cytoplasmic localization, whereas sebaceous adenomas are negative. In mixed tumors, intense ß-catenin staining is only evident in the peripheral region of the tumor acquiring a squamous phenotype (Fig. 4D). After restoring AP-1 activity in sebaceous adenomas, some mixed tumors had ß-catenin staining in regions with sebaceous cells, further supporting the notion that transdifferentiation is occurring. The staining pattern of sFRP-3 (previously known as Frizzled B), a secreted protein known to inhibit wnt signaling resulting in ß-catenin protein destabilization (34), has the opposite staining pattern to that seen for ß-catenin. sFRP-3 is abundant in a mixed tumor expressing A-FOS, localizing to cells with either a sebaceous or squamous phenotype (Fig. 4E) but was not detected in squamous papillomas (Fig. 4F).
|
|
To evaluate the role for AP-1 in wnt signaling, we examined the mRNA expression of canonical and noncanonical wnt family and pathway transcripts in primary keratinocytes by expressing A-FOS. Figure 6A indicates a broad regulation of wnt family transcripts by AP-1 activity with both increased and decreased expression in a complex pattern, suggesting that an intimate crosstalk exists between AP-1 and wnt pathways. This relationship is supported by several recent studies using alternative approaches (39–41).
c-JUN binds to the promoters of Wnt family members. To determine if the wnt and frizzled family members whose mRNA concentrations changed after A-FOS expression are direct targets of AP-1 transcription factors in tissue culture, we examined if phospho-c-Jun, a member of the AP-1 complex, is bound to their promoter regions using the ChIP assay. In brief, the ChIP assay was done with four different antibodies: phospho-c-JUN (specific to phosphorylated Ser73), CREB, C/EBP, and IgG. We used the Nimblegen mouse promoter array to examine binding to 21,815 promoters, each represented by 15 oligonucleotides. Phospho-c-JUN was enriched over 4-fold (for 
3 of the 15 oligos per promoter) in 1,060 promoters, CREB was enriched in 451 promoters and C/EBPß in 95 promoters. Previously identified known targets of phospho-c-JUN were also enriched (42). The genes containing multiple TRE motifs were enriched in phospho-c-JUN ChIPs.
The promoters of 31% (4 of 13) of the wnt and frizzled genes mentioned in Fig. 6A were enriched over 4-fold (for three or more oligos per promoter) for phospho-c-JUN, whereas on average, only 5% (1,060 of 21,815) of promoters are enriched for phospho-c-JUN. This suggests that A-FOS misregulation of these genes is because they are direct targets of the AP-1 complex containing phospho-c-JUN. Figure 6B presents the binding of these three antibodies to the promoter region for two genes. Phospho-c-JUN binds in the promoter region of Wnt4 and Wnt5a, but C/EBPß and CREB do not.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The AP-1 transcriptional complex is composed of multiple combinations of Jun and Fos family heterodimers. A knockin experiment that placed the JunB protein coding region into the c-Jun locus, creating a JunB protein under the regulation of c-Jun promoter, has been able to rescue the embryonic lethality associated with the c-Jun deletion (43). This shows that these proteins have redundant functions during embryogenesis and suggests that the difference in the phenotypes of c-Jun and JunB knockout mice is a consequence of their different spatial and temporal expression properties (43). The redundant properties of related Jun family members makes it difficult to identify the function of the AP-1 complex when individual members of the AP-1 family are genetically deleted from the mouse genome.
Several knockout and transgenic mice have been generated that have revealed the function of AP-1 in the skin. The tissue-specific deletion of c-Jun produces an eye-open-at-birth phenotype (EOB; refs. 3, 4). The double ablation of JunB and c-Jun in the skin produces a hyperplastic, inflamed phenotype observed in ears and paws of mice, which disappears when the mice are given antibiotics (44). A transgenic mouse that expresses an NH2-terminal truncated c-Jun protein without a transactivation domain in the skin does not produce a basal phenotype but does diminish papilloma and carcinoma formation (5, 45). This construct serves as a dominant negative outside of the DNA-binding complex and interacts with multiple pathways, including AP-1, nuclear factor-
B, and the cAMP response element (8). The unique phenotype of the A-FOS mice suggests that this transgene inhibits more of the AP-1 complex than the single or double knockout mice already described. Thus, it will be interesting to determine the phenotype of the triple deletion of all three Jun family members in the skin to determine if a sebaceous hyperplasia is observed. One conclusion is that A-FOS is doing something distinct from inhibiting only c-Jun as we do not observe an EOB phenotype.
Several additional possibilities could explain these differences in phenotypes produced by deleting Jun family members and expressing A-FOS. One possibility is the incomplete inhibition of AP-1 activity in the differentiated layers of the epidermis. As the normal epidermis differentiates, the K5 promoter is silenced; thus, A-FOS is not expressed in the differentiated strata. This is in contrast to the knockout experiment that irreversibly eliminates activity in all cells that are produced after basal layer differentiation. Furthermore, K5 expression is up-regulated and expanded into the differentiating strata of skin tumors (46); thus, A-FOS would be even more highly expressed in multiple differentiating states of tumor tissue. Thus, the hyperplastic sebaceous glands in aging K5/A-FOS mice could reflect a modification of K5 or AP-1 expression in aging skin. The mixed mouse strains used in the tissue-specific knockout experiments and subsequent crosses could also influence the phenotype relative to the pure FVB/N strain we have used. Transgenic mice expressing dominant-negative A-CREB or A-C/EBP do not produce the sebaceous adenoma phenotype in similar carcinogenesis studies3, indicating that the A-FOS targets do not overlap with the specificity of A-CREB or A-C/EBP.
These results with tumor induction in K5/A-FOS mouse skin may explain why only squamous tumors develop in the classic mouse skin carcinogenesis assay induced by DMBA, whereas tumors of other phenotypes are not observed. In chemically induced squamous tumors of the skin, H-ras mutations are frequent and result in elevated AP-1 activity (17). This favors the wnt/ß-catenin pathway and directs cells to the squamous lineage, resulting in papillomas that progress to squamous cell carcinomas (31). This signal could be further amplified by a positive feedback loop of wnt signaling activating AP-1, as ß-catenin overexpression up-regulates the expression and activity of c-Jun and Fra-1 (47).
In contrast, tumor models that prevent signaling through ß-catenin (31) may also up-regulate hedgehog signaling through the inhibition of AP-1 (this report). This would favor sebaceous tumors that rarely, if ever, progress to cancer in the absence of AP-1 signaling. The elevation of IHH by AP-1 inhibition may also contribute to antagonizing wnt signaling as shown in colonic epithelium (48). Furthermore, the up-regulation of sFRP-3 by AP-1 inhibition may contribute to suppression of wnt signaling and increased ß-catenin turnover, a finding that may have significance in other tumors of mixed phenotype. The involvement of activated ß-catenin in inducing squamous cells in glandular and secretory epithelia argues for a general mechanism that epithelia use to differentiate various cell types (49). The results from the mixed tumors and cultured keratinocytes support a dynamic link between AP-1 activity, wnt and hedgehog signaling, ß-catenin stability, and discrimination between the squamous and sebaceous cell types. The ChIP experiment that shows that phospho-c-Jun is bound to the promoters in the wnt and frizzled families suggests that the link between AP-1 and the choice between the wnt and hedgehog signaling pathways is a direct transcriptional link. A requirement for AP-1 signaling in wnt/ß-catenin–mediated intestinal cancer development has been shown in the APCmin mouse model, consistent with a broader role for AP-1/wnt signaling in cancer development (39). It will be interesting to determine if A-FOS expression can switch cell identity in other epithelial systems.
Recently, attention has focused on the potential role of a subpopulation of stem-like cells in cancers that account for the infinite replicative capacity and frequent phenotypic diversity of the tumor mass (50–53). Changes in the tumor microenvironment could regulate transcriptional pathways, including AP-1, in multi-potential cells to produce focal lineage alterations. In the K5/A-FOS tumor model described here, we detect cells that express markers of two distinct lineages as the tumor transdifferentiates from squamous to sebaceous or vice versa. Such biology suggests that tumor cells of defined terminal lineage are in fact plastic and can change lineage commitment between squamous or sebaceous end points as a result of changes in AP-1 transcriptional activity. Whether such cells also represent a population of stem cells with infinite proliferative capacity remains to be determined. However, it was recently reported that terminally differentiated pancreatic islet cells can be induced to replicate (54). Our studies reveal a previously unknown role of AP-1 transcriptional activity in determination and maintenance of cellular lineage. The reversible activation of AP-1 transcriptional potential reveals the plastic nature of tumor cell differentiation and a mechanism for balancing signaling pathways that control cell lineage determination.
|
Acknowledgments |
|---|
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Erin Kennedy and Susana Walters for their excellent assistance with carcinogenesis studies and animal husbandry, Drs. Matthew Young and Nancy Colburn (National Cancer Institute) for providing TRE-Luciferase (TRE-LUC) mice, and Drs. Jeffery Rubin and Melinda Larsen for critical reading of the article.
|
Footnotes |
|---|
3 Vinson et al., unpublished results. 
Received 4/ 5/06. Revised 5/19/06. Accepted 5/26/06.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-difluoromethylornithine accompanied by marked hair follicle degeneration and hair loss. Cancer Res 2003;63:3037–42. mediates tumour promotion via a PKC - and AP-1-dependent pathway. Oncogene 2002;21:4728–38.[CrossRef][Medline]This article has been cited by other articles:
|
|
 |
|
  K. S. Suh, M. Mutoh, T. Mutoh, L. Li, A. Ryscavage, J. M. Crutchley, R. A. Dumont, C. Cheng, and S. H. Yuspa CLIC4 mediates and is required for Ca2+-induced keratinocyte differentiation J. Cell Sci., August 1, 2007; 120(15): 2631 - 2640. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W. J. Oh, V. Rishi, A. Orosz, M. J. Gerdes, and C. Vinson Inhibition of CCAAT/Enhancer Binding Protein Family DNA Binding in Mouse Epidermis Prevents and Regresses Papillomas Cancer Res., February 15, 2007; 67(4): 1867 - 1876. [Abstract] [Full Text] [PDF]
|
|
|
|
