Cell cycle promoting activity of JunB through cyclin A activation.

JunB, a major component of the AP-1 transcription factor, is known to act antagonistically to c-Jun in transcriptional regulation and is proposed to be a negative regulator of cell proliferation. Employing fibroblasts derived from E9.5 junB(-/-) mouse embryos we provide evidence for a novel cell cycle promoting role of JunB. Despite a normal proliferation rate, primary and immortalized junB(-/-) fibroblasts exhibited an altered cell cycle profile, which was characterized by an increase in the population of S-phase cells, while that of cells in G(2)/M-phase was diminished. This delay in G(2)/M-transition is caused by impaired cyclin A-CDK2 and cyclin B-CDC2 kinase activities and counteracts the accelerated S-phase entry. Cells lacking JunB show severely delayed kinetics of cyclin A mRNA expression due to the loss of proper transcriptional activation mediated via binding of JunB to the CRE element in the cyclin A promoter. Upon reintroduction of an inducible JunB-ER(TM) expression vector the cell cycle distribution and the cell cycle-associated cyclin A-CDK2 kinase activity could be restored. Thus, cyclin A is a direct transcriptional target of JunB driving cell proliferation.

JunB, a major component of the AP-1 transcription factor, is known to act antagonistically to c-Jun in transcriptional regulation and is proposed to be a negative regulator of cell proliferation. Employing fibroblasts derived from E9.5 junB ؊/؊ mouse embryos we provide evidence for a novel cell cycle promoting role of JunB. Despite a normal proliferation rate, primary and immortalized junB ؊

/؊ fibroblasts exhibited an altered cell cycle profile, which was characterized by an increase in the population of S-phase cells, while that of cells in G 2 /M-phase was diminished. This delay in G 2 /M-transition is caused by impaired cyclin A-CDK2 and cyclin B-CDC2 kinase activities and counteracts the accelerated S-phase entry. Cells lacking JunB show severely delayed kinetics of cyclin A mRNA expression due to the loss of proper transcriptional activation mediated via binding of JunB to the CRE element in the cyclin A promoter. Upon reintroduction of an inducible JunB-ER TM expression vector the cell cycle distribution and the cell cycle-associated cyclin A-CDK2 kinase activity could be restored. Thus, cyclin A is a direct transcriptional target of JunB driving cell proliferation.
Eukaryotic cells have developed precise and well regulated mechanisms to control progression through the cell cycle. The tight control is mediated by the interplay of sequentially activated and inactivated protein kinase complexes known as cyclindependent kinases (CDKs). 1 CDK activity is controlled by the expression levels of their respective cyclin partners acting as positive coactivators and by negative regulators, the so-called CDK inhibitors (1)(2)(3). While cyclin D-CDK4/CDK6 and cyclin E-CDK2 control the progression through G 1 -phase (4), and cyclin B-CDC2 appears to be only necessary for the entry into mitosis (5), cyclin A is a rate-limiting component required for both the initiation of DNA synthesis and entry into mitosis (6). Ablation of cyclin A results in cell cycle arrest in G 2 , as shown for Drosophila embryos (7) and somatic mammalian cells (8). Microinjection of antisense cDNA or anti-cyclin A antibodies inhibits the initiation of DNA replication (6,9), and cyclin A-CDK2 may phosphorylate and activate the cellular DNA replication factor replication protein A (10,11). Cyclin A expression is cell cycle-dependent through periodical relief of transcriptional repression. The cyclin A promoter is regulated during G 0 and G 1 by two contigous cis-acting elements, the CDE-CHR bipartite DNA element (12,13). The factors binding to these elements remain to be characterized. So far, two different repressor proteins, CDF-1 (14) and CHF (15), but also pocket proteins (pRb and p107), are discussed to regulate cyclin A expression through the CDE-CHR site (16). However, the mechanism by which the pocket proteins affect cyclin A gene expression is still a matter of controversial discussion. The cyclin A promoter region also contains potential recognition sites for ATF/CREB, AP-1, p53, Sp1, and the murine G 1 /Sspecific transcription factor Yi (17)(18)(19)(20)(21)(22). While the AP-1 and p53 sites are most likely not functional (17,18), the ATF/CREB site, CRE, is required for the transcriptional regulation of the cyclin A gene (17). It has been proposed that the CRE acts in concert with the CDE site to properly regulate cyclin A expression (23,24).
Transcription factors binding to the CRE are CREB and ATF-2. ATF-2 that is necessary for the maximal activity and serum induction of the cyclin A promoter in chondrocytes (25) is a member of the AP-1 transcription factor. AP-1 represents a heterogenous set of dimeric proteins consisting of the Jun, Fos, and ATF families. Signals affecting AP-1 activity include growth factors, cytokines, tumor promoters, carcinogens, and specific oncogenes (for reviews, see Refs. 26 and 27). Different lines of evidence have shown that AP-1 is critically involved in cell proliferation (28,29). A causal link between AP-1-dependent signaling and cell cycle regulation was provided by the analysis of c-Jun null fibroblasts, revealing several p53-dependent molecular defects in the G 1 -to S-phase transition (30) and cell cycle re-entry of UV-irradiated cells (31). While c-Jun has cell cycle promoting functions by repressing p53 and activating the cyclin D1 promoter, antagonistic functions have been assigned to JunB. At the G 1 -to S-transition JunB acts as a repressor through inhibition the cyclin D1 promoter (32) and via p16 INK4a that has been identified as a transcriptional target of JunB (33).
To assess the consequences of JunB ablation in cell proliferation and cell cycle progression, we isolated fibroblasts from JunB-deficient E9.5 embryos just prior to death due to placen-tal insufficiency (34) and generated spontaneously immortalized 3T3 fibroblast cell lines. In addition to the previously described negative function of JunB at the G 1 -to S-transition our work identifies a novel critical positive role for JunB in cell cycle progression at the S-to G 2 /M-phase as transcriptional activator of cyclin A and subsequently of the kinase activity of cyclin A-CDK2.

EXPERIMENTAL PROCEDURES
Cell Lines and Culture Conditions-Primary MEFs were isolated and immortalized as described (35). Each primary fibroblast culture was isolated from a single E9.5 mouse embryo of mixed genetic background (C57BL/6 ϫ 129/Sv), and each 3T3 fibroblast line was immortalized from individual primary cultures. Wild type and junB Ϫ/Ϫ embryos were derived from intercrosses of junB heterozygous mice (34). Genotyping of mutant embryos and fibroblasts derived thereof was performed by PCR as described (34); the presence or absence of JunB protein was confirmed by Western blot analysis. JunB-ER TM was introduced into junB Ϫ/Ϫ 3T3 fibroblasts by cotransfection with a Rous sarcoma virus-hygro plasmid encoding the gene for hygromycin B resistance. Several stable clones expressing JunB-ER TM were identified by Western blot analysis and randomly chosen for further analysis.
Fibroblasts were cultured at 37°C in 6% CO 2 in Dulbecco's modified Eagle's medium containing 10% fetal calf serum (FCS). Cells were counted and passaged at 3-4-day intervals and cumulative cell numbers determined. To obtain synchronized cells for various experiments, 3T3 fibroblasts were arrested in G 0 by culturing in medium containing 0.5% FCS for 48 h and were released into the cell cycle by stimulation. HEK293 cells were transiently transfected by the calcium phosphate method as described (36). 2.5 g of DNA of each plasmid were used. 10 h post-transfection cells were harvested, and luciferase and ␤-galactosidase activities were measured using standard chemiluminiscence procedures. Each transfection was performed at least three times, and transfection efficiency was normalized by determination of ␤-galactosidase activity of the cotransfected SV40-␤-galactosidase expression plasmid.
Construction of the junB-ER TM Expression Vector and Cyclin A Reporter Genes-For the construction of the junB-ER TM expression vector the stop codon was replaced in a junB 3Ј subclone of the intron-less junB gene (323-bp BssHII-XhoI fragment) using two oligonucleotides (oligo1, 5Ј-acggctgccagttgctgctaggggtcaagggacacgccttc-3Ј; oligo2, 5Ј-gtctggactcgaggatccccgaaggcgtgtcccttgaccc-3Ј) placed between the restriction sites AlwNI and XhoI. Subsequently, the complete junB cDNA was restricted with SmaI, EcoRI linkers were added, and the DNA fragment was cloned into pBluescript SK (Stratagene, Amsterdam). A junB-ER TM subclone in pBluescript SK was created by ligation of: the 5Ј EcoRI/ BspHI fragment of the complete junB cDNA, 3Ј BspHI/BamHI fragment containing the replaced stop codon, BamHI/EcoRI ER fragment of PMV-7-c-jun-ER (37), and EcoRI-digested pBluescript SK. Finally, this construct was digested with XbaI/SalI, and the liberated fragment was inserted into a XbaI/XhoI restricted ubi-junB expression vector (38).
To generate the luciferase reporter vector containing the mouse cyclin A promoter, a fragment spanning the region Ϫ618 to ϩ272 bp according to the murine cyclin A sequence published in the National Center for Biotechnology Information data base (accession number U57826) was cloned from genomic DNA of wild type mouse fibroblasts by PCR using the following primers: sense, 5Ј-AAACTCTGGGATTA-AAGGTATGTA-3Ј; antisense, 5Ј-TGGCGCCAGTTTTTCGGGGTTGA-3Ј. After blunt end reaction, EcoRI linkers were added to the PCR fragment and cloned into pBluescript SK (Stratagene). The correctly orientated fragment was released by double digestion with SmaI and HindIII and subcloned into the pGL3 luciferase expression vector (Promega). This resulted in the Ϫ618/ϩ272 construct. The cyclin A mCREluc (luc, firefly luciferase gene) construct was generated from the Ϫ618/ ϩ272 plasmid using the QuikChange site-directed mutagenesis kit (Stratagene) according to the manufacturer's recommendations. The putative CRE binding site at position ϩ147 was changed from GT-GACGTCA to CAGTGGTCA, thereby generating a PstI site allowing an easy screening of mutant clones. Successful mutagenesis was confirmed by sequence analysis.
Protein Analysis-Whole cell extracts were prepared as described (39). 100 g of protein extract were used for immunoprecipitation of cyclin A (generous gift from M. Pagano, New York University Medical Center, New York) or cyclin B (generous gift from I. Hoffmann, Deutsches Krebsforschungszentrum, Heidelberg). Immune complexes were incubated with histone H1 (Roche Molecular Biochemicals) as described (39). For Western blot analysis, 30 g of protein samples were separated by SDS-PAGE, blotted onto nitrocellulose, and immunodetection was performed with an enhanced chemiluminescence system (Amersham Biosciences). The following primary antibodies were used at 1:1000 dilution unless otherwise indicated: N17 (Santa Cruz, sc-43) and C11 (Santa Cruz, sc-8051) for JunB, sc-53 (Santa Cruz) for CDC2 p34, A-2066 (Sigma) for actin (1:100).
For immunoprecipitation, cell extracts were incubated with the appropriate precipitating antibodies for 3 h at 4°C, and the immune complexes were captured on Sepharose beads coated with protein A (Amersham Biosciences). The precipitates were washed three times with lysis buffer (50 mM Tris/HCl, pH 8, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, and 1 mM phenylmethylsulfonyl fluoride), resolved by electrophoresis on 8% SDS-PAGE, and transferred to nitrocellulose for Western blot analysis.
Flow Cytometry Analysis-For DNA content analysis, 10 6 ethanolfixed cells were washed twice with phosphate-buffered saline and stained with propidium iodide (50 g/ml in phosphate-buffered saline, Sigma). Data were collected and analyzed with a Becton Dickinson FACScan system using Cellquest software (BD Biosciences, San Jose, CA).
BrdUrd Labeling and Immunohistochemistry-Asynchronously growing or restimulated cells were incubated for 1 h with 60 M BrdUrd (Sigma), fixed with 4% paraformaldehyde, and analyzed by immunohistochemistry using an anti-BrdUrd mouse monoclonal antibody (Calbiochem) at a dilution of 1:200. An ABC (avidin-biotinylated enzyme complex) staining procedure (ABC Universal or rabbit IgG kit, Vector) was performed according to the manufacturer's instructions.

Normal Cell Proliferation of JunB-deficient Fibroblasts-To
analyze the role of JunB in cell proliferation and cell cycle progression, MEFs were isolated from E9.5 embryos with a targeted mutation in the junB gene (34). Although the JunBdeficient embryos were dramatically reduced in size, the proliferative potential of primary fibroblasts lacking JunB was not different from wild type fibroblasts (Fig. 1A). Following the 3T3 protocol (35), two immortalized cell lines were established from wild type and JunB-deficient MEFs, respectively. Similar to wild type fibroblasts, the mutant cells underwent a normal crisis before spontaneous immortalization and exhibited a similar proliferation rate as wild type cells with doubling times of 25 Ϯ 1 h (Fig. 1, A and B). This could be observed in both independently isolated junB Ϫ/Ϫ 3T3-like fibroblast cell lines (Fig. 1B, c1 and c2).
Altered Cell Cycle Profile of JunB-deficient Fibroblasts-Although JunB-deficient MEFs and immortalized 3T3 fibroblasts showed a normal proliferation rate, flow cytometry analysis of asynchronously growing cells revealed an aberrant cell cycle profile. In both mutant cell lines, the number of cells with a DNA content characteristic for S-phase was increased at the expense of concomitantly decreased G 0 /G 1 -and G 2 /M-phase cell populations ( Fig. 2A, c1 and c2). This phenotype was observed in two independent pairs of wild type and mutant MEFs and immortalized 3T3 fibroblast cell lines, excluding the possibility that the observed differences are due to clonal variations. In synchronized wild type and junB Ϫ/Ϫ 3T3 cells, released 48 h post-serum starvation by addition of 20% FCS to re-enter the cell cycle, comparable results were obtained. Flow cytometry profiles of wild type and junB Ϫ/Ϫ 3T3 cells taken at distinct time points throughout the cell cycle revealed an accelerated G 1 -to S-phase transition in JunB-deficient cells characterized by a higher extent of S-phase cells most prominent 14 h postrelease (Fig. 2, B and C). Analysis for the presence of S-phase nuclei via BrdUrd incorporation confirmed the observed accelerated G 1 -to S-phase transition (Fig. 2C). Despite very similar S-phase profiles (18 h post-release), JunB-deficient fibroblasts were retained in S-phase 20 h post-release, whereas the majority of wild type cells had already started to progress into G 2 /M-phase (Fig. 2B). After 24 h the cell cycle of both cell types was nearly finished, yet there were still a higher percentage of JunB-deficient cells in S-phase (20.1 Ϯ 2.63% versus 10.5 Ϯ 0.7% for wt cells). After completion of the cell cycle the proportion of junB Ϫ/Ϫ cells in S-phase remains increased (19 Ϯ 1.41%), while only 6.45 Ϯ 0.63% of wild type cells are found in S-phase (compare also Fig. 6B). Based on previous work (32,33) and our own data (42) the accelerated G 1 -to S-transition is caused most likely by the simultaneous concurrence of prematurely increased cyclin D1 and reduced p16 INK4a protein levels and hence increased cyclin D1-CDK4/CDK6 kinase activity.
Delayed Progression to G 2 /M-phase in JunB-deficient Fibroblasts-Despite the accelerated G 1 -phase and, in turn, increased S-phase population, JunB-deficient fibroblasts exhibit a normal proliferation rate (Fig. 1), suggesting a second defect in the cell cycle machinery that may compensate for the first one. To get support for this assumption, the fate of JunBdeficient cells during S-phase progression was followed using two different experimental approaches. First, asynchronously growing cells were pulse-labeled with BrdUrd and subsequently analyzed by FACS (Fig. 3A). To exclude that the observed differences in S to G 2 /M transition may be a consequence of the premature G 1 -to S-transition, cells were synchronized at the G 1 -to S-boundary by a double thymidine blockade (43), followed by restimulation. The transition from S-to G 2 /M-phase is clearly delayed in junB Ϫ/Ϫ cells, since 4 h post-restimulation, a higher fraction of JunB-deficient cells remained in the S-phase (90.25 Ϯ 1.06%) compared with wild type cells (87.5 Ϯ 2.12%). The difference in progression became significant 6 h post-restimulation when 84 Ϯ 1.41% of the JunB-deficient cells were still in S-phase compared with 78.25 Ϯ 1.76% wild type cells (Fig. 3B).
Impaired Activity of S-G 2 /M Regulators-The progression of the cell cycle from the S-to the G 2 /M-phase is mainly regulated by the activity of cyclin A-CDK2 and cyclin B-CDC2 kinase complexes (5,22). In JunB-deficient fibroblasts, both cyclin A-CDK2 and cyclin B-CDC2 activities were significantly impaired compared with wild type cells (Fig. 4A). Although the levels of cyclin A and B kinase activities in serum-starved cells were somewhat enhanced, their activity during S-to the G 2 /Mtransition was significantly decreased, and the kinetics of maximal activation of cyclin B kinase activity was delayed (Fig.  4A). Very similar results were observed in the second independently isolated JunB-deficient fibroblast clone (data not shown). Most likely, these differences are due to altered cyclin A and B protein levels. In the absence of JunB, the induction of cyclin A protein was delayed (Fig. 4B), while the protein levels of the CDK2 and CDC2 kinases were unaltered in cells lacking JunB ( Fig. 4B and data not shown). RT-PCR analysis of serumstarved and subsequently serum-treated junB Ϫ/Ϫ fibroblasts provided evidence that the maximal transcriptional activation of cyclin A is JunB-dependent (Fig. 4C).
JunB-dependent AP-1 Binding to the CRE in the Cyclin A Promoter-It has been reported that the cyclin AMP response element CRE is necessary for optimal activity and serum inducibility of the cyclin A promoter (17). Since AP-1 members are known to compete with CREB for binding to the related CRE (44), we applied in vitro protein/DNA binding studies (EMSAs) to characterize factors binding to this critical CRE element. Nuclear extracts from wild type fibroblasts contained a strong CRE binding activity that could be partially competed by an unlabeled TRE oligonucleotide harboring the AP-1 consensus binding site (Fig. 5B, wt, cyclin A, CRE). Under these conditions the amount of TRE used was sufficient to fully compete for AP-1 binding to its consensus site (Fig. 5B, wt,  TRE), suggesting that both AP-1 and CREB proteins participate in binding to the cyclin A CRE. When nuclear extracts from junB Ϫ/Ϫ cells were used, complex formation was less efficient and completely resistant to competition by the TRE (Fig. 5B, junB Ϫ/Ϫ , cyclin A CRE). These data provided strong evidence that the AP-1 component binding to the CRE is missing in extracts derived from junB Ϫ/Ϫ fibroblasts. Using a mutated CRE (mCRE) we did not observe any complex formation (Fig. 5B, mCRE). Gel shifts performed with the octamer binding site (OCT) did not reveal any differences in overall binding activities between wild type and junB Ϫ/Ϫ nuclear extracts, confirming equal quality and quantities of extracts used in the assay (Fig. 5B, OCT). Preincubation of wild type nuclear extracts with a JunB-specific antibody and the subsequent formation of supershifted complexes (ss) confirmed the presence of JunB in the complex bound to the CRE. The complex at the CRE also contained c-Jun, CREB, and ATF-2 (Fig. 5C, cyclin A  CRE, wt). In the junB Ϫ/Ϫ cells, the protein complex bound to the CRE contained only CREB (Fig. 5C, junB Ϫ/Ϫ , cyclin A CRE), while the protein complex bound to TRE contained only c-Jun (Fig. 5C, junB Ϫ/Ϫ , TRE).
JunB Potentiates Transcription from the Cyclin A Promoter through the CRE-The data described above prompted us to investigate whether JunB can directly activate the cyclin A promoter and to determine heterodimeric partners required for transcriptional activation. Due to the very low transfection efficiency of junB Ϫ/Ϫ fibroblasts, transient cotransfection experiments were performed in HEK293 cells using a luciferase reporter under the control of the Ϫ618/ϩ272 cyclin A promoter fragment containing either a wild type or mutated CRE (Fig.  5A). Coexpression of JunB resulted in a very weak transactivation of the cyclin A promoter (1.7-fold; Fig. 5D). When c-Fos  4. Impaired cyclin A-CDK2 and cyclin B-CDC2 kinase activity in junB ؊/؊ fibroblasts. Cells were synchronized and stimulated as shown in Fig. 2B. Whole cell extracts were prepared at the indicated time points (in hours) post-serum stimulation from wild type (wt) and junB Ϫ/Ϫ immortalized fibroblasts. A, induction of cyclin A-CDK2-and cyclin B-CDC2-associated histone H1 kinase activity. Immunoprecipitates containing cyclin A or cyclin B complexes were isolated from whole cell extracts, and their histone H1 phosphorylating activities were determined in vitro. B, protein levels of cyclin A and cyclin B were determined by Western blot analysis. C, RT-PCR analysis of cyclin A mRNA from wild type (wt) and junB Ϫ/Ϫ fibroblasts, respectively. Total RNA from synchronized and restimulated wild type and junB Ϫ/Ϫ cells, as indicated at the top, was prepared and subjected to semiquantitative RT-PCR analysis for expression levels of cyclin A. For normalization RT-PCR for ␤-tubulin expression was performed. Amplification products were transferred to Hybond N ϩ membrane (Amersham Biosciences) and probed with cyclin A and ␤-tubulin cDNA probes. as a potential dimerization partner was co-transfected, the transactivation of the cyclin A promoter fragment was not improved (data not shown). By contrast, coexpression of ATF-2 resulted in strong transactivation comparable with that obtained with the typical artificial JunB reporter (Fig. 5D,  5ϫTRE). Mutation of the CRE within the complete 900-bp promoter fragment resulted in strong suppression of luciferase activity (Fig. 5D, cyclin A mCRE), suggesting that JunB acts through this element. Similar results were obtained in F9 cells (data not shown).
Reintroduction of JunB Rescues the Cell Cycle Defects-To demonstrate that the observed cell cycle defects are solely due to the lack of JunB, we attempted to restore a normal cell cycle profile and cyclin A activation by exogenously expressed JunB in junB Ϫ/Ϫ fibroblasts. Independent stable clones showing constitutive expression of the inactive JunB-ER TM fusion proteins were isolated. Similar to the successful use of these clones for the reversion of JunB-dependent alterations in fibroblast/keratinocyte cross-talk (45), the impaired expression of p16 INK4a described to be a positively regulated JunB target gene (33) could be restored upon activation of the JunB-ER TM protein in the stably transfected junB Ϫ/Ϫ clones (Fig. 6A). Indeed, cell cycle distribution could be normalized, with a substantial decrease in S-phase cells and a regain of a G 2 /M population (Fig.  6B). Most importantly, cyclin A kinetics of induction, protein levels, and kinase activity could be rescued upon release of G 0 arrested cells (Fig. 6, C and D). Thus, the accelerated transition through G 1 and the delay in S-to G 2 /M-transition are caused by the specific loss of JunB, and secondary mutations that may have occurred during immortalization seem to be irrelevant for the described defects. DISCUSSION Here, we demonstrate that JunB, which so far has been considered to be a negative regulator of AP-1 (46 -48) and of the G 1 -to S-transition (32,33), has an additional, yet unrecognized, positive role in cell cycle regulation. Fibroblasts with a targeted null mutation in the junB gene exhibited a delayed entry into G 2 /M-phases, thereby compensating the expected accelerated G 1 -to S-transition resulting in an overall normal proliferation rate.
The increased G 1 -acceleration function of cyclin D1-CDK4/ CDK6 and the loss of the growth suppressor p16 INK4a as a result of JunB ablation did not result in enhanced proliferation as one may have expected based on results obtained from cyclin D1 overexpression (49,50). In fact, JunB-deficient fibroblasts exhibited overall normal cell proliferation behavior suggesting an additional impairment of a JunB-dependent function compensating for the accelerated G 1 -to S-transition. Indeed, delayed maximal kinase activities for both S to G 2 /M regulators, cyclin A-CDK2 as well as cyclin B-CDC2, were observed in junB Ϫ/Ϫ fibroblasts, which could account for this compensatory effect. The progression through S-phase toward the G 2 /Mphase is mainly regulated via the activity of cyclin A-CDK2 and cyclin B-CDC2 (5,22). In line with these data the impaired kinase activities in JunB-deficient cells can be explained by the reduced and decelerated cyclin A protein levels, which, in turn, also affect the activation and nuclear translocation of the downstream cyclin B-CDC2 complex (5).
One could assume that impaired activation of cyclin A-CDK2 is due to a premature entry of JunB-deficient fibroblasts into S-phase without having the appropriate substrates for DNA synthesis present. Unfortunately, no tool is currently available to dissect the two phenotypes, the accelerated G 1 -transition that is accompanied by elevated cyclin D1 and c-Jun levels from the delay in S-phase progression. Previous studies have shown that accelerated G 1 -to S-transition caused by cyclin D1 overexpression results in a shortened generation time, yet without a compensatory prolongation of the DNA replicative phase (49,50). Conversely, overexpression of c-Jun produces larger S/G 2and M-phase populations (29). In light of these studies, the decelerated S-to G 2 /M-transition of JunB-deficient fibroblasts is either due to the prolonged transcriptional activation of c-jun in junB Ϫ/Ϫ cells (42,51) or to a second, c-Jun-independent event. There are several arguments that favor the existence of modulators of the cell cycle machinery, which are exclusively controlled by JunB. We propose that the loss of JunB causes an impairment of the strict cell cycle-dependent regulation of cyclin A. Even in the presence of enhanced c-Jun and cyclin D1 levels and increased cyclin D1-CDK4/CDK6 kinase activity, which should result in a cyclin A activation, the levels of cyclin A were found to be reduced.
Previous studies have shown that the inducibility of the cyclin A promoter during the short window between late G 1and G 1 /S-phase is initiated by the cAMP signaling pathway leading to ATF-2 and CREB activation (17). Conversely, inhibition of protein kinase A with a specific inhibitor reduced cyclin A mRNA accumulation in late G 1 and delayed S-phase entry, demonstrating the involvement of PKA in this pathway (17). Independent reports have demonstrated that Jun/ATF heterodimers are involved in this regulation. In chondrocytes of ATF-2-deficient mice, the transcriptional regulation of cyclin A is severely impaired (25). The rat cyclin A promoter is primarily regulated by ATF and Jun proteins, predominantly JunD, binding to the CRE element (52). However, junD Ϫ/Ϫ cells show neither a cell cycle defect nor an alteration in cyclin A regulation (53), suggesting that JunD either does not play a decisive role in cyclin A regulation or that it may be functionally compensated by other AP-1 members, most likely JunB. JunB is the only Jun member that is strongly activated by the cAMPdependent PKA signal pathway (54). The in vitro binding studies and the impairment of cyclin A transcriptional activation presented in this report strongly underline that JunB is a major component in the regulation of the cyclin A promoter. Moreover, data obtained from the coexpression studies provide functional evidence that JunB, together with ATF-2, is able to transactivate the cyclin A promoter in a CRE-dependent manner. The reconstitution of proper cyclin A expression and activation, upon reintroduction of a post-translationally inducible JunB-ER TM protein, the in vitro binding studies as well as transient transfection analysis have identified cyclin A, in addition to proliferin, MMP-9 (34), and p16 INK4a (33), as one of the very few so far characterized positively regulated JunB target genes. Interestingly, an identical CRE element is implicated in the positive regulation of both cyclin D1 and cyclin A genes. The precise timing of their expression in the cell cycle thus appears to rely on their ability to bind different transcription factors. c-Jun/ATF heterodimers are responsible for cyclin D1 activation (32), while the JunB seems to be needed for its repression (Ref. 32 and our own results). By contrast, the cAMP-induced signaling pathway exerts a stimulatory effect on cyclin A (17) and an inhibitory effect on cyclin D1 expression (55). Our data suggest that a fine-tuned cell cycle-regulated adjustment of the ratio between c-Jun and JunB proteins may represent one of the decisive components.
We demonstrate that JunB has unique functions in cell cycle regulation, which cannot be compensated by other AP-1 mem-bers. Moreover, it confirms the coexistence of previously suggested opposite functions of JunB as transcriptional repressor and activator (46) that is required for one and the same process, namely cell cycle progression (Fig. 7). Our findings provide another example for the increasing complexity of positive and negative interactions among the individual AP-1 subunits embedded in a complex circuitry network.