Transformation by Ras Modifies AP1 Composition and Activity

Transformation by ras modi®es AP1 composition and activity

Fatima Mechta1, Dominique Lallemand, Curt M Pfarr2 and Moshe Yaniv

UniteÂ des Virus OncogeÁnes; URA 1644 du CNRS, DeÂpartement des Biotechnologies, Institut Pasteur, 25, rue du Dr Roux, 75724 Paris Cedex 15, France.

The Ras proteins play a central role in regulating cell Ras proteins play a central role in signal transduc- growth and their mutation can lead to abnormal tion (Feig and Cooper, 1988; Mulcahy et al., 1985; proliferation. To analyse the potential link betwen AP1 Smith et al., 1989). These proto-oncogenes, frequently activity, encoded by members of the jun and fos gene mutated in human tumors, connect receptors as the cell families, and Ras-mediated cellular transformation, we membrane with a cytoplasmic cascade of protein have studied several NIH3T3 clones which overexpress kinases followed by downstream nuclear events the Ha-Ras or Ki-Ras oncogenes. These transformed including induction of transcription and DNA synth- ®broblasts accumulated higher levels of cJun, JunB, esis. The proteins (p21Ras) encoded by the ras genes Fra1 and Fra2 proteins relative to their normal counter- (Harvey-ras/Ha-ras, Kirsten-ras/Ki-ras and N-ras) bind parts. They also displayed increased AP1 DNA binding guanine nucleotides, possess an intrinsic GTPase activity which was predominantly composed of cJun and activity and cycle between an active (GTP bound) Fra1 containing dimers. Following serum stimulation of form and an inactive (GDP bound) state (Barbacid, Ras clones, the elevated levels of cJun and Fra1 1987). Activated p21Ras proteins stimulate a cascade remained steady, while the induction of JunB and Fra2 of cytoplasmic protein kinases including Raf, MEK, was partially attenuated. Moreover, deregulated Ras MAPK (Mitogen Activated Protein Kinase) which in signaling resulted in a complete loss of the serum turn culminates in the phosphorylation of nuclear inducibility of cFos and FosB. Ectopic co-expression of transcription factors resulting in transcriptional activa- cJun and Fra1 in NIH3T3 ®broblasts led to a tion of growth-mediating genes (Bollag and McCor- transformed phenotype, attenuation of cFos serum mick, 1991; Boulton et al., 1991; de Vries-Smit et al., inducibility, increased AP1 activity and Cyclin D1 1992; Dent et al., 1992; DeÂ rijard et al., 1994; Hibi et accumulation, all characteristics of oncogenic Ras al., 1993; Howe et al., 1992; Kyriakis et al., 1992; expressing cells. These results demonstrate that cJun Lowenstein et al., 1992; Wood et al., 1992). and Fra1 are crucial mediators of the Ras-transforma- The induction of c-jun and c-fos transcription is one tion process. of the earliest nuclear responses to a wide variety of extracellular stimuli and is thought to be crucial in Keywords: AP1; cJun; Fra1; Ras; serum inducibility; mediating cellular proliferating (Curran and Franza, transformation 1988; Herschman, 1991). The mitogen stimulation of c-fos transcription is a paradigm for a gene regulated via the Ras pathway. The SRE (Serum Response Element) in the c-fos promoter is continuously bound by two nuclear proteins SRF and Elk/TCF (Herrera Introduction et al., 1989; KoÈ nig, 1991; Norman et al., 1988; Prywes and Roeder, 1987; SchroÈ ter et al., 1990; Shaw et al., The precise mechanism by which speci®c oncopro- 1989; Treisman, 1987). Upon Ras activation Elk1 is teins cause malignant transformation of cells by rapidly phosphorylated by MAPK-1 and MAPK-2 transmitting signals for aberrant growth is not fully (also called ERK1 and ERK2) resulting in a strong understood. Proto-oncogenes commonly participate in increase in its transcriptional activity (Hipskind et al., the regulation of normal cellular growth through 1994; Prywes et al., 1988; Zinck et al., 1993). A signal transduction pathways which convert extra- dierent mechanism has been proposed for c-jun cellular stimuli into a program of gene expression. transcriptional activation which involves a cJun/

Many of these genes are also activated during the G0 ATF2 heterodimer stably bound to a TRE element to G1 transition when quiescent cells are stimulated in the c-jun promoter (van Dam et al., 1993). Ras by mitogens to re-enter the cycle. The rapid activation promotes phosphorylation of the cJun induction of such genes, called `immediate early protein augmenting its transcriptional activity (BineÂ - genes' is followed by the transcription of down- truy et al., 1991; Smeal et al., 1991). stream target genes that are critical for the The jun and fos families of genes are composed of proliferative response of the cell. A number of these three jun members (c-jun, junB, and junD) (Bohmann et immediate early genes encode transcription factors, al., 1987; Hirai et al., 1989; Maki et al., 1987; Ryder et including Jun and Fos. al., 1988, 1989) and four dierent fos partners (c-fos, fosB, fra1 and fra2) (Cohen and Curran, 1988; Nishina et al., 1990; Zerial et al., 1989). These genes encode Correspondence: M Yaniv components of the AP1 transcription factor formed by 1Present addresses: 1UniteÂ INSERM 368, Biologie moleÂ culaire du homo- or hetero-dimerization of the Jun and Fos DeÂ veloppement, ENS, 46 rue d'Ulm. 75230 Paris Cedex 05 France proteins. Dimerization occurs through a leucine-repeat 2Department of Cell Biology and Biochemistry, Texas Tech University Health Science Center, 3601 4th Street, Lubbock, Texas located in the highly conserved C-terminal part of the 79430 USA Jun proteins or in an internal position in the Fos API composition and activity in Ras-transformed fibroblasts FMechtaet al 838 proteins (Landschulz et al., 1988; Sassone-Corsi et al., dimers exhibit similar DNA binding speci®cities but 1988a). As the Fos proteins do not form dimers among dier in their transactivation eciencies (Chiu et al., themselves, AP1 is a set of low anity Jun/Jun and 1989; Hirai et al., 1990; Suzuki et al., 1991). The high anity Jun/Fos dimers (Nakabeppu et al., 1988). binding anities for dierent dimer combinations is This complex set of dimers recognizes and binds to the dependent on the speci®c TRE sequence and on the TRE element (TPA Responsive Element or TGACT- promoter context (Halazonetis et al., 1988; Ryseck and CA) through the basic domain of both proteins, Bravo, 1991). Moreover, the jun and fos immediate adjacent to the Leucine-zipper. AP1 recognition early genes are dierentially expressed during develop- sequences are found in many promoters and enhancers ment and in adult tissues (Dony and Gruss, 1987; Hirai of cellular and viral genes, including Collagenase, et al., 1989; Wilkinson et al., 1989). They also respond Stromelysin, Metallothionein IIA, Interleukin 2, dierently to mitogenic stimuli, genotoxic or differ- Transforming Growth Factor b, Simian Virus 40, entiating agents (Almendral et al., 1988; de Groot et Polyomavirus and Papillomavirus. The dierent AP1 al., 1990; Devary et al., 1991; Lau and Nathans, 1987; 3

Clone Ras Protein — NIH3T —R5 —R4 —R3 —R2 —R1

NIH3T3 L

Ras L

Figure 1 Characterization of Ras expressing ®broblasts. Dierent neomycin-resistant clones were isolated after stable transfection of an activated Ha-ras gene in NIH3T3 ®broblasts. The level of Ras expression was analysed by immuno¯uorescence and immunoblot staining with a speci®c anti-Ras monoclonal antibody (see Material and methods). The plating eciency of soft agar clones was de®ned as the ratio of the number of macroscopic colonies after 15 days of growth to the number cells initially seeded. The plotted numbers represent data of one typical experiment. There was a good correlation between the amount of activated Ras and transformation as measured by a soft agar assay (ie: normal NIH3T3 ®broblasts or clone R5 that did not show accumulation of Ras protein do not grow in agar, whereas clones expressing the higher levels of Ras produced the higher number of colonies) API composition and activity in Ras-transformed fibroblasts FMechtaet al 839 3 3 — NIH3T —R5 —R4 —R3 —R2 —R1 —R5 —R4 —R3 —R2 —R1 — NIH3T L L cJun L

cFos L L

FosB L L L L L L L ∆ L JunB L FosB

Fra1 L

JunD L

Fra2

Figure 2 Expression pattern of the Jun and Fos proteins after transformation by Ras. The Jun and Fos protein levels were determined in the various Ras overexpressing clones by immunoblots probed with speci®c anti-Jun and anti-Fos antibodies as described in Material and methods. Identical amounts of whole-cell extracts were loaded in each lane, transferred to nitrocellulose, incubated with the indicated primary antibody and developed using enhanced chemiluminescence. The same extracts were used to performed each blot. The arrows or brackets indicate the position of the speci®c immuno-reactive band(s). Multiplicity of bands observed for JunB, Fra1 and Fra2 is caused by phosphorylation (DL, unpublished results). Note that cFos, FosB and DFosB are undetectable in these extracts. The weak bands reacting with the anti-cFos or anti-Fra1 antibodies in the NIH3T3 cell extracts are non speci®c. The dierences observed for the migration of JunB and Fra2 between the DT (R1) and the R2-R5 clones concerns probably their degree of phosphorylation and may depend on the nature of the Ras protein (Kirsten versus Harvey ras)

Mechta et al., 1989; Ryseck et al., 1988; Lallemand et Results al., 1997). These observations suggest that, under various conditions, the members of the Jun and Fos Increased expression of cJun, JunB, Fra1 and Fra2 in families present dierent properties and have distinct Ras transformed cells functions. Several lines of evidence suggest that AP1 couples To study the consequences of oncogenic Ras expression signals from activated Ras proteins to the transcription on AP1 composition, we isolated a set of Ras- of target genes in transformed cells. Blocking AP1 transformed clones by transfection of NIH3T3 fibro- activity using dominant negative cJun or cFos mutants blasts with an activated Ha-ras gene isolated from a reverts the transformed phenotype of Ras-overex- human bladder carcinoma (Shih and Weinberg, 1982) pressed NIH3T3 ®broblasts (Lloyd et al., 1991; Suzuki together with a plasmid conferring neomycin resistance. et al., 1994). A correlation was also observed between Four clones R2, R3, R4 and R5 were selected and Ras transforming eciency and cJun transcriptional expanded for further analysis. In addition, a well activity (Alani et al., 1991; Westwick et al., 1994). This characterized NIH3T3-derived cell line (DT) trans- is consistent with the idea that AP1 complexes formed by incorporation of two copies of a Ki-ras containing cJun are essential downstream eectors of gene (Noda et al., 1983) was also used in this study and Ras. In a recent study, we reported opposing eects of renamed for convenience R1. Ras expression, the cJun or JunD overexpression on transformation by transformed phenotype, and the Jun and Fos protein Ras (Pfarr et al., 1994). While cJun cooperated with levels were monitored for each clone. The level of Ras Ras, excess JunD partially reversed the transformed expression was determined both by immuno¯uorescence phenotype. These experiments suggested that dierent and Western blotting and the extent of transformation Jun and Fos proteins might modulate Ras signals was evaluated by scoring colony formation in agar. As dierently. Furthermore, it remained unknown which shown in Figure 1, the frequency of agar colony Fos proteins heterodimerize with cJun in these cells formation was roughly proportional to the level of and which AP1 dimers may be necessary for Ras protein accumulation. In this experiment, we did transformation by Ras. Therefore, we set out to not distinguish between the endogenous Ras and the measure the level of the Jun and Fos proteins and overexpressed mutant Ras, but rather estimated the the AP1 DNA binding activity in dierent Ras total level of Ras protein. transformed cell lines and attempted to establish a The levels of the Jun and Fos proteins were relationship between AP1 composition and the determined by Western blot analyses. A set of speci®c transformed phenotype. rabbit polyclonal antibodies with similar ranges of API composition and activity in Ras-transformed fibroblasts FMechtaet al 840 reactivity was used to determine the level of accumula- eciency of Ras transformation and the half-life of tion of the dierent Jun and Fos proteins (Pfarr et al., cJun was observed, with higher stability of cJun 1994; Lallemand et al., 1997). As shown in Figure 2, each occurring in the most severely transformed Ras of the three Jun proteins were detected in growing clones (R1 and R2) (data not shown). NIH3T3 cells, while Fra2 is the only Fos member detected in these cells. Transformation by Ras resulted in Ras transformation results in constitutive expression of an increase in the levels of cJun and JunB proteins cJun and Fra1 and a loss of serum-inducibility of cFos relative to the parental NIH3T3 cells. JunD expression and FosB was markedly lower in the more transformed clones (R2 and R1). Fra1 and Fra2 protein levels were increased in Growth factors present in serum (e.g. PDGF and all of the transformed cell lines. Since Fra1 protein levels EGF) stimulate cell proliferation by binding to tyrosine in normal NIH3T3 cells were lower than those of Fra2, kinase receptors which activate the Ras signalling the fold induction of Fra1 in the Ras clones was pathway. This in turn activates the immediate early relatively higher. The R5 clone was an exception in genes, a step thought to be essential for mitogenesis. It that it expressed little if any exogenous Ras protein, did was therefore surprising that constitutively expressed not grow in agar, yet still showed increased JunB, Fra1 oncogenic Ras did not promote the synthesis of either and Fra2 expression. While transient expression of Ras the c-fos or fosB immediately early genes. We therefore oncoprotein results in increased levels of cFos and FosB checked the response of Ras-transformed cells to high (Stacey et al., 1987), these proteins were not detected in concentrations of serum. our stable Ras transformed cells. cFos, FosB, and DFosB As shown in Figure 3, cFos, FosB, and DFosB were each detected in serum-stimulated NIH3T3 protein synthesis was stimulated by serum treatment ®broblasts which were used as a positive control of normal NIH3T3 cells whereas they were largely (Figure 3). refractory to such treatment in Ras-transformed In the case of cJun we further analysed the events clones (R3, R2, R1). JunB and Fra2 were stimulated responsible for its overexpression. Northern blot by serum in both NIH3T3 and R3 clone. However, analysis demonstrated that c-jun mRNA levels were their serum-inducibility was decreased in the R2 and increased roughly fourfold in the most transformed lost in the R1 clone. As previously reported, JunD clones (R3, R2, R1) indicating heightened c-jun gene underwent a rapid and partial degradation after serum expression (due to c-jun transcription and/or c-jun stimulation of NIH3T3 cells. The levels of JunD in mRNA stability) (data not shown). To determine the Ras-transformed ®broblasts were rather insensitive to half-life of cJun protein in Ras-transformed fibro- serum-stimulation. Finally, and in contrast to the blasts, we performed pulse/chase experiments with other AP1 members, cJun and Fra1 protein levels [35S]methionine. The stability of the cJun protein was remained constitutively elevated after serum addition. estimated by measuring the incorporated label The constitutive levels of cJun in Ras clones was following immunoprecipitation of cJun from these comparable to the level of cJun detected in serum- extracts. The half-life of cJun protein was enhanced treated NIH3T3 while Fra1 protein levels in Ras-

2.3-fold between NIH3T3 (t1/2=1.4 h) and the R1 overexpressing ®broblasts was markedly higher than in

cell line (t1/2=3.3 h). A correlation between the serum-stimulated 3T3 ®broblasts.

NIH R3 R2 R1 NIH R3 R2 R1 serum – + – + – + – + – + – + – + – + L L L

cJun L cFos L

FosB L L L L L L L

∆ L JunB L FosB

Fra1 L L

JunD L L

Fra2

Figure 3 Serum-inducibility of Jun and Fos proteins. Cells were grown in 7% fetal calf serum and half of them were stimulated for 2 h with fresh medium containing 15% serum. Total proteins were then extracted either from exponentially growing cells (7)or from serum-stimulated cells (+). The amount of Jun and Fos products was measured by immunoblot analyses with the indicated speci®c anti-Jun and anti-Fos antibodies. Arrows or brackets indicate the positions of the immuno-reactive proteins. The lower band observed with the anti-JunD antibody is probably due to internal initiation of translation (CP, unpublished observation). Multiplicity of bands observed for JunB, Fra1 and Fra2 is caused by phosphorylation (D Lallemand, 1997) API composition and activity in Ras-transformed fibroblasts FMechtaet al 841 Ras enhances AP1 activity the mutant Ras proteins modi®ed AP1 activity, we measured the AP1 DNA binding activity in nuclear To examine whether the quantitative and qualitative extracts by gel retardation assays (Garner and Revzin, variations of Jun and Fos protein levels induced by 1981; Piette et al., 1988). The level of AP1 DNA

a NIH R5 R3 R2 R1

AP1 L

Figure 4 Level and composition of AP-1 binding complexes. (a) Gel retardation assays were performed with nuclear extracts from normal NIH3T3 ®broblasts (NIH), the R5 clone and the Ras-overexpressing cell lines (R1, R2, R3). A 32P labeled doubled stranded oligonucleotide containing a consensus TRE (TPA Responsive Element) from the collagenase promoter was used as probe. Equal amounts of nuclear protein extracts were used for each cell line. The AP1 binding activity was quanti®ed by the 32P labeled signals of the retarded bands with a phosphoimager system and the bar graph in the right of (a) represents the average value and standard deviation for three independent experiments. (b) The composition of AP1 proteins bound to DNA was probed by addition of speci®c puri®ed polyclonal antibodies. The anti-Jun and anti-Fos antibodies decrease AP1-DNA complex formation. The degree of inhibition of AP1 DNA binding activity was estimated by measuring with a phosphoimager the intensity of the residual complex after addition of each antibody in the binding mixture. The symbols used in this ®gure represents the degree of inhibition of binding: ± sign corresponds to 0 ± 5%, +/±:10 ± 20% ++: 20 ± 30%, +++: 30 ± 50% and ++++: 50 ± 60% of inhibition API composition and activity in Ras-transformed fibroblasts FMechtaet al 842 binding activity was measured with equal amounts of transformed ®broblasts. The highly transformed R2 nuclear extracts obtained from normal NIH3T3, the clone was isolated from the same parental NIH3T3 neomycin control clone (R5), and each of the

previously described Ras-transformed clones; R1, R2 1 a and R3. Figure 4a shows that AP1 DNA binding activity was enhanced roughly four- to ®vefold in Ras transformed ®broblasts relative to NIH3T3 parental cells. This experiment also showed a correlation Fra1 cJun cJun+Fra Ras between Ras-transforming potential and enhanced

AP1 activity, as the highest AP1 binding was detected NIH F1 F2 C1 CF1 CF2 CF3 R2 L in the most transformed clones (R1 and R2). The Fral L composition of AP1 DNA binding activity was probed by addition of speci®c Jun and Fos antibodies to the b binding mixture (Figure 4b). Polyclonal antibodies Soft agar Plating abolish the AP1-DNA complex formation. Addition Phase contrast clones Efficiency (%) of puri®ed antibodies directed against cJun and Fra1 strongly decreased AP1 binding activity in Ras- NIH3T3 0 transformed ®broblasts. Fra2 was the main Jun partner in NIH3T3 cells whereas Fra1 became the major active Fos component in Ras-transformed ®broblasts. Nevertheless, JunB and Fra2 proteins F1 3 also participated in AP1 binding activity in Ras- transformed ®broblasts though less abundantly than cJun and Fra1. Corroborating with increased DNA binding activity, we have also observed an increased AP1 transcriptional activity by transient transfection F2 0.5 assay using a chimeric construct carrying three TRE sites upstream of a minimal tk promoter (data not shown). C1 6.75

cJun and Fra1 are key mediators in Ras transformation To determine if high concentrations of AP1 heterodimers composed of cJun and Fra1 were CF1 19 sucient for transformation, we tested if overexpression of cJun and Fra1 could replace oncogenic Ras. We isolated a set of Fra1 overexpressing clones either from the parental NIH3T3 cells or from CF2 27.8 ®broblasts stably expressing cJun (C1 clone). The C1 clone was isolated previously from NIH3T3 cells (Pfarr et al., 1994). Fra1-overexpressing cells were obtained by transfecting a plasmid carrying the CF3 32 mouse fra1 cDNA driven by the SRa promoter along with a hygromycin B selectable marker contained in the pcDEBD vector (Nakabeppu et al, 1993). Following selection, individual clones were R2 74.3 screened for overexpression of Fra1 protein by immunoblot (Figure 5a). Two clones (F1 and F2) isolated from NIH3T3 and three clones (CF1, CF2, Figure 5 Transformed properties of cJun and Fra1 overexpress- CF3) isolated from C1 ®broblasts expressed consti- ing cell lines. (a) Characterization of Fra1 overexpressing clones. These clones were isolated either from normal NIH3T3 ®broblasts tutively high levels of Fra1 protein and were used transfected with a fra1 expression vector (F1, F2) or from a cJun for further analysis. Using RT-PCR with one primer overexpressing clone (C1). The three double transfectants in the SV40 leader of the transfected plasmid and expressing both cJun and Fra1 proteins were named CF1, CF2 another in the Fra1 coding sequence, we con®rmed and CF3. The Fra1 protein level was measured by immunoblot that the exogenous Fra1 gene was expressed in F with a speci®c polyclonal antibody. (b) Morphology and growth in soft agar of Fra1 overexpressing clones. The left hand column are and CF clones (data not shown). The Fra1 protein phase contrast micrographs of the indicated clone in culture. The level was markedly higher in clone C1 than in scale bar represents 100 mm. The dierent clones display various normal NIH3T3 cells and Fra1 protein accumulation morphologies: cJun and Fra1 overexpression leads to a refractile seemed to be facilitated in the CF clones compared phase contrast image, fuciform shape, and compact nuclei approaching the phenotype of Ras-transformed cells. All of these to F1 and F2. This may result from both positive cell lines were also assayed for growth in soft agar. To the right of autoregulation of the endogenous Fra1 promoter by each phase contrast image is a micrograph of a representative AP1 (Busslinger and Bergers, 1994) and stabilization colony of cells in soft agar. The same magni®cation was used for phase contrast images of cells and micrographs of colonies. 102 or of the Fra1 protein by the heterodimerization with 3 cJun. 10 cells were plated in 5 cm dishes and the number of visible colonies were scored after 15 days in culture. This number de®nes We tested the transformed characteristics of these the plating eciency (as described in Figure 1) for each clone and clones and compared them to the properties of Ras- is indicated on the right API composition and activity in Ras-transformed fibroblasts FMechtaet al 843 cells and was used in this comparative study. The cJun and Fra1 overexpression abolishes cFos induction morphology of the clones and their ability to grow in and increases cyclin D1 levels soft agar and in low serum are documented in Figures 5b and 6, respectively. Phase contrast images show the As shown above, deregulated expression of both cJun distinct morphologies presented by the various clones and Fra1 resulted in a transformed phenotype. As Ras (Figure 5b). Fibroblasts overexpressing cJun alone (C1) transformation resulted in attenuation of cFos serum and Fra1 overexpressing cells (F1, F2) were smaller inducibility and enhanced AP1 DNA binding activity, than NIH3T3. F1 and F2 clones also grew overlapping we checked if the up-regulation of cJun and Fra1 can one another. Nevertheless, these clones remained well confer these two properties to the cells. The experiments attached to the culture dishes, a phenotype shared with presented in Figure 7a and b demonstrated that these normal ®broblasts. By contrast, clones overexpressing two parameters (attenuation of cFos serum inducibility both cJun and Fra1 (CF1, CF2, CF3) displayed and enhanced AP1 DNA binding activity) were observed enhanced transformed properties: they were much in cells overexpressing both cJun and Fra1 proteins smaller than NIH3T3, presented a highly refractile (CF1, CF2 and CF3 clones). The constitutive expression phase-contrast image, and grew with increased cell of cJun or Fra1 alone (C1, F1, F2 clones) slightly overlap. They were similar to the Ras-transformed in¯uences these two properties (results not shown) but ®broblasts (R2) with fuciform shape and compact less markedly than overexpression of both proteins. cFos nuclei, two characteristics of loss of contact inhibition. protein synthesis was still inducible in response to serum Nevertheless, the CF clones were slightly ¯atter than in C1, F1, or F2 clones, though less strongly than in R2 cells and remained better attached to the culture NIH3T3. AP1 DNA binding activity was also slightly dishes. enhanced in these cell lines. Nevertheless, only co- F1, F2 and C1 clones grew slowly in low serum and overexpression of both cJun and Fra1 led to complete formed small colonies in soft agar. Fra1-overexpressing repression of c-fos induction and a strong increase in ®broblasts grew slower than the cJun clone in serum AP1 activity. Finally, as in Ras-transformed ®broblasts, free medium and resulted in fewer and smaller colonies we observed also that Fra2 protein levels increased in in agar. Overexpression of both cJun and Fra1 in 3T3 CFs cells relative to NIH3T3 (data not shown). ®broblasts synergized in conferring an almost fully transformed phenotype, though still less ecient than mutated Ras. CF clones gave rise to less colonies in a soft agar (19 to 32% of cells seeded) than Ras- cJun+Fra1 Ras transformed ®broblasts (74%) and the size of the colonies was smaller (Figure 5b). This observation suggests that anchorage-independent growth was NIH CF1 CF2 CF3 R2 Serum – +– + – + – + – + slower in CF clones than in Ras-expressing fibro- L blasts. This hypothesis was con®rmed by following the cFos L growth rate of these clones in serum-free medium (Figure 6). CF clones grew slower in serum-deprived b medium than Ras-transformed ®broblasts but they still cJun+Fra1 Ras proliferated faster under these conditions than normal ®broblasts (NIH3T3), cJun (C1), or Fra1 (F1, F2) CF1 CF2 CF3 NIH overexpressing clones. R2 L

AP1 L

c cJun+Fra1 Ras CF1 CF2 CF3 NIH R2 L

Cyclin D1 L

Figure 7 Overexpression of cJun and Fra1 mimics transformation by Ras. (a) The cFos serum response was measured in normal NIH3T3 ®broblasts (NIH), cJun+Fra1 (CF1, CF2, CF3) Figure 6 Growth in low serum of the Fra1 overexpressing and Ras (R2) overexpressing clones as described in Figure 3. (b) clones. Cells were seeded in 0.25% FCS at day 0 at a density of The AP1 DNA binding activity was performed as depicted in 105 cells per 5 cm culture dish and counted the subsequent days Figure 4. (c) The level of Cyclin D1 protein was measured in (day 1, day 3 and day 5). Plotted numbers and resulting growth equal amount of nuclear cell extracts by immunoblot experiment curve represent the average of two representative experiments with a speci®c mouse monoclonal antibody API composition and activity in Ras-transformed fibroblasts FMechtaet al 844 Increased levels of cJun and Fra1 will result also in Both c-jun and junB mRNA levels were shown enhanced transcriptional activation of AP1 target genes. previously to be induced by transient expression of Since cJun-Fra1 heterodimers are strong activators of a Ras (Sistonen et al., 1989). We con®rm that the canonical TRE element (Suzuki et al., 1991; L Bakiri, accumulation of cJun in stable Ras-transformed personal communication). ®broblasts results from an increase in c-jun mRNA Enhanced AP1 activity in Ras-transformed fibro- level as well as from the stabilization of its translation blasts could be responsible for increasing and product. Ras leads to phosphorylation of cJun at two maintaining the expression of proliferation promoting sites in its transactivation domain which increases its genes. Ras transformation results in an elevated level transcriptional activation activity (BineÂ truy et al., of Cyclin D1 protein (Filmus et al., 1994; Liu et al., 1991). Elevated c-jun gene expression partially results 1995). The human Cyclin D1 gene is known to be from positive autoregulation via two AP1 binding sites regulated by AP1 (Herber et al., 1994) suggesting that present in its promoter, as the AP1 binding activity

the accumulation of this important G1/S regulator targeting these two sequences is enhanced in Ras might be correlated with the enhanced AP1 activity transformed cells (data not shown). The enhanced observed in our stable Ras clones. The AP1-mediated phosphorylation of cJun may also increase its induction of Cyclin D1 expression by activated Ras is stability. By contrast with cJun, an AP1-independent also consistent with data showing that cAMP (which mechanism of regulation has been described for junB blocks Ras signal transduction) inhibits Cyclin D1 gene induction by Ras. The enhanced junB expression synthesis (Sewing et al., 1993) in addition to serum- or in Ras-transformed cells is, at least partially, mediated TPA-induction of c-jun (Mechta et al., 1989). We by c-Ets1 and c-Ets2 transcription factors through therefore checked whether overexpression of cJun and several ETS-binding sites in the murine junB promoter Fra1 would result in increased Cyclin-D1 protein (Coer et al., 1994). levels. Immunoblot experiments using a speci®c anti- The synthesis of Fra1 also becomes constitutive in Ras Cyclin-D1 antibody demonstrated that overexpression transformed ®broblasts. Similar to c-jun, fra1 is known of both cJun and Fra1 led to Cyclin-D1 accumulation to be regulated by AP1 itself. Transcriptional activation to a level comparable to that in the R2 clone (Figure of the fra1 gene by AP1 is mediated by regulatory 7c). These reults con®rm that accumulation of cJun- sequences in the ®rst intron (Bergers et al., 1995). and Fra1-containing AP1 dimers confer several proper- Elevated levels of fra1 mRNA were observed in ties of Ras-overexpressing cells and contribute to the ®broblasts overexpressing cJun, FosB, or Fra1 itself, transformed phenotype of the cells. indicating that the fra1 gene is up-regulated by itself or by other members of the AP1 family (Busslinger and Bergers, 1994; Miao and Curran, 1994). Consistent with Discussion this idea, we observed that the level of Fra1 protein is increased in a cJun-overexpressing clone (C1) (Figure Mitogenic signal transduction initiated by activation of 5a). Thus, in Ras transformed ®broblasts, the Fra1 growth factor receptors at the plasma membrane leads accumulation may also result from an AP-1 mediated to conversion of Ras into its GTP bound active form positive auto-regulation loop. Fra1 up-regulation might followed by a cascade of kinase activation. This is be one consequence of the enhanced levels of cJun, JunB followed rapidly by phosphorylation of transcription and Fra2. factors and induction of immediate early gene As stated above, we never detected cFos, FosB, or expression. One of the transcription factors that is DFosB proteins in the established Ras-overexpressing rapidly induced during this transition is AP1, clones. By contrast, microinjection of Ras oncoprotein composed of members of the Jun and Fos families. into quiescent NIH3T3 induces transient c-fos expression In the present study, we investigated the long-term (Stacey et al., 1987). We cannot exclude the possibility eects of constitutive stimulation of the Ras-signaling that there is a transient c-fos induction in our Ras clones pathway on the accumulation of Jun and Fos proteins. at the beginning of Ras expression, but cFos protein Previous studies have focused on short term eects and remains completely undetectable in the stable Ras analysis at the mRNA level. transformed cells. These results suggest that c-fos We show that the members of the AP1 family are induction is required for initiation but not for dierentially regulated by constitutive oncogenic Ras maintenance of transformation. Moreover, c-fos is expression, suggesting that they are implicated in refractory to serum stimulation in most transformed distinct ways in the Ras signal transduction pathway. clones suggesting that c-fos regulation is altered in Overexpression of the Ha-Ras oncoprotein in mouse transformed cells. Down-regulation of c-fos does not 3T3 ®broblasts increased the levels of cJun, JunB, result from deletion of the c-fos gene due to cellular Fra1, and to a lesser extent Fra2 proteins. JunD, clonal selection as c-fos expression can be weakly previously identi®ed as an antagonist of Ras transfor- stimulated by treatment of Ras-transformed ®broblasts mation was found to decrease in most cases. with a high combination of high serum and ultraviolet Surprisingly, no trace of cFos or FosB proteins was irradiation (unpublished observation). Serum induction found in these transformed cells. The jun and fos genes of c-fos has already been shown to be markedly also responded dierently to serum stimulation of these attenuated in cells transformed by v-src, v-sis, v-ras or cells. The inducibility of JunB and Fra2 genes is v-raf (Yu et al., 1993). The v-src-mediated repression of markedly attenuated or even abolished in Ras- c-fos induction occurs at the transcriptional level transformed ®broblasts whereas elevated cJun and suggesting that prolonged oncogenic Ras expression Fra1 protein levels remain steady under these could also inhibit c-fos transcription. Expression of c-fos conditions. Finally, cFos and FosB are only partially following growth factor treatment of normal ®broblasts or not inducible at all by such serum treatment. is transient and is stringently turned o (Lin et al., 1988; API composition and activity in Ras-transformed fibroblasts FMechtaet al 845 SchoÈ nthal et al., 1988). The mechanism of c-fos down- Materials and methods regulation is complex and involves dierent transcription factors including the cFos product itself (Sassone-Corsi Cell culture et al., 1988b; SchoÈ nthal et al., 1989). This auto- NIH3T3 cells and Ras-derived clones were maintained in repression requires the C-terminal domain of cFos. Dulbecco's Modi®ed Eagle's minimal essential Medium Similar sequences which also contain repressing activity (DMEM) containing 7% Fetal Calf Serum (FCS) and exist in the other Fos family proteins, particularly Fra1 antibiotics. (O®r et al., 1990; Wilson and Treisman, 1988). The accumulation of Fra1 in Ras-transformed cells might be Preparation of stable cell lines important for mediating c-fos repression. This hypothesis is consistent with the fact that the ectopic expression The Ha-ras cell lines were established from a plasmid (pEJ6.6) containing a genomic activated Ha-ras gene of cJun and Fra1 also blocks the serum-response of the c- isolated from a human bladder carcinoma. An expression fos gene. cJun and Fra1 might play a direct role in the vector (pSV-neor) encoding a neomycin-resistance gene was transcriptional silencing of c-fos, perhaps by inhibition used as a selection marker and was cotransfected with the of the serum responsive elements (SRE) in the c-fos pEJ6.6 plasmid in NIH3T3 cells in 1:10 proportion. Fra1- promoter. Indeed, the binding activity to the c-fos SRE overexpressing cells were obtained by transfecting a sequence was decreased in Ras-transformed mouse plasmid carrying the mouse fral cDNA driven by the osteoblast cell lines (Nose et al., 1989). SRa promoter along with a hygromycin B selectable The attenuation of c-fos serum inducibility may also marker contained in the pcDEBD vector (Nakabeppu et involve down-regulation of growth factor receptors. al., 1993). NIH3T3 cells or C1 clone (Pfarr et al., 1994) 5 The expression of the b-chain of the PDGF receptor in were seeded at a density of 5610 cells per 10 cm diameter culture dish in DMEM complemented with 7% FCS. After mouse ®broblasts was shown to be repressed by the v- 24 h the cells were transfected using the calcium phosphate ras or v-src oncogenes (Vaziri and Faller, 1995). The coprecipitation method (Wigler et al., 1977). Two days reduced number of cell surface PDGF receptors post-transfection, the cells were trypsinized, diluted 10- or present in transformed ®broblasts may desensitize the 20-fold and selected in 100 mg/ml of G418 (geneticin cells to subsequent stimulation by PDGF. antibiotic from Gibco BRL) or 200 mg/ml of hygromycin One critical event in Ras-mediated transformation B (Calbiochem). Individual resistant clones were subcloned seems to be the constitutive activation of AP1 after 1 week of selection. transcription factors (Wasylyk et al., 1988). AP1 activity is enhanced by expression of cellular or viral Soft-agar assays transforming oncogenes (such as activated Ras, Src, For assay of cell growth in soft agar, three dilutions of PymT) but not by expression of immortalizing cells were analysed. 102,103,104 cells were plated in 5 cm oncogenes (including SV40T, PyLT, myc). Moreover, culture dishes onto a 3 ml basal layer (0.3% agar in our results show a correlation between Ras transforma- DMEM plus 7% FCS) (Difco) over a solidi®ed cushion of tion eciency and AP1 activity. The requirement for 0.6% agar. Individual macroscopic colonies were counted strong AP1 activity in Ras-transformed ®broblasts has after 2 weeks of growth. been demonstrated in previous studies. Expression of transdominant Jun or Fos mutants is capable of Immuno¯uorescence microscopy reverting transformed cells to a normal phenotype (Lloyd et al., 1991; Suzuki et al., 1994). Overexpression The dierent clones were seeded at the same initial density of wild-type Fra1 protein in NIH3T3 ®broblasts is by onto 20 mm glass coverslips, previously coated with a poly- L-lysine solution (Sigma). Cells were grown on these itself weakly transforming, though it is a more potent treated coverslips in 7% FCS DMEM. All the following transforming gene in Rat1A cells (Bergers et al., 1995). steps were performed at room temperature in phosphate Overexpression of cJun alone also leads to partial buered saline solutions (PBS). Cells were ®xed in 0.5% transformation of 3T3 ®broblasts. However, co-over- paraformaldehyde for 20 min, permeabilized in 0.1% expression of both cJun and Fra1 strongly increases Triton X-100 for 5 min, rinsed twice and blocked for the severity of the transformed phenotype. These 30 min in 0.05% Tween 20. The coverslips were incubated observations suggest that the intrinsic transforming 2 h with the primary antibody (mouse monoclonal pan-Ras capacity of cJun and Fra1 is crucial in the establish- antibody from Oncogene Science #OP40), in PBS plus ment of Ras-mediated carcinogenesis. The enhanced 0.05% Tween, rinsed three times and incubated 1 h with AP1 activity in Ras-transformed ®broblasts suggests a the secondary antibody (Fluorescein-coupled anti-mouse IgG, Amersham). The coverslips were washed in PBS and model for signal ampli®cation and conversion of distilled water, mounted with an antibleaching glycerol mix transient early events into longer term eects on gene (Citi¯uor±AF1) and viewed on a Zeiss Axiphot epi- expression. Ras transformation enhances the Cyclin D1 ¯uorescence microscope. Cells were photographed with protein level, one of the major G1/S transition control identical exposure times on TMAX 400 ®lms (Kodak). factors. A recent report suggests that increased expression of Cyclin D1 in Ras-transformants is Cell extracts and immunoblot analysis responsible for the shortening of the G1 phase of these cells (Liu et al., 1995). Overexpression of Cyclin Whole cell extracts and Western blotting were performed as described in Pfarr et al. (1994). The Jun and Fos speci®c D1 in NIH3T3 has been shown to accelerate G1 progression and to decrease serum requirement for antibodies were prepared as described in Pfarr et al. (1994; Lallemand, et al., 1997). The cyclin D1 antibody was a kind growth (Quelle et al., 1993). Our results suggest that gift of Jiri Bartek. The blots were incubated with a transformation by Ras leads to accumulation of the horseradish peroxidase-conjugated secondary antibody (1/ cJun and Fra1 transcription factors and enhances AP1 8000 dilution) followed by detection with enhanced activity which in turn results in increased Cyclin D1 chemiluminescence (ECL detection system from Amer- protein level which may accelerate G1 progression. sham) and exposed to autoradiography ®lms. API composition and activity in Ras-transformed fibroblasts FMechtaet al 846 Nuclear extracts and gel retardation assays 10 mM HEPES (pH=7), 100 mg/ml BSA, 17.5% glycerol, 2mM DTT, 0.5 mM Spermidine, 2 mM MgCl and 1 mgof All solutions used for preparation of cell extracts and 2 poly(dI-dC).poly(dI-dC). Next, 0.5 ng of a double electrophoretic mobility shift assays contained 5 mg/ml of stranded 32P-labelled AP1 binding site (from the collage- each of the following protease inhibitors: Aprotinin, nase promoter 5'-AGCTAGCTGACTCAGATGTCCT-3') Pepstatin A, Leupeptin, Antipain, and Chymostatin (all was incubated for an additional 10 min at 48C. DNA- from Sigma). Cultured cells were washed with cold PBS protein complexes were resolved on 6% native polyacry- and rapidly harvested by centrifugation. Cell pellets were lamide gels (Protogel from National Diagnostic), ®xed in resuspended in HNB buer (approximately 200 mlfor107 10% Methanol+10% acetic acid for 10 min, dried and cells): 0.5 M sucrose, 15 mM Tris, 60 mM KCl, 0.25 mM visusalized by autoradiography. EDTA 0.125 mM EGTA, 0.5 mM Spermidine, 0.15 mM Spermine. One half volume of HNB plus 1% NP40 was added to the solution and the cell lysates were immediately cleared in a microcentrifuge at 6000 r.p.m. Acknowledgements for 3 min. The pellets were resuspended in NEB1 solution: We are grateful to Jean de Guntzburg for the human Ras

10 mM HEPES (pH=8.0), 25% glycerol, 1.5 mM MgCl2, cDNA, Makoto Noda and Bodan Wasylyk for DT cells, 0.25 mM EDTA and a volume of NEB2 (NEB1 plus 0.7 M Bruno Tocque for advice in the detection of Ras oncoprotein NaCl) was added. The mixture ws agitated (on a roller and Jiri Bartek for the kind gift of Cyclin-D1 antibody and Y system)for30minat48C and centrifuged 30 min at Nakabeppu for the gift of Fra1 expression vector. We 13 000 r.p.m. Protein concentration was determined by the warmly thank Brigitte Bourachot for generous assistance in method of Bradford (Bio-rad protein assay). Extracts of cell culture, David Sourdive for precious help in computer dierent cell samples were stored in aliquots at 7808C. sorcery and Latifa Bakiri for valuable discussions. This work Equivalent amounts (2.5 mg) of nuclear extracts of the was supported by grants from the ARC, the Ligue Nationale dierent cell samples were preincubated with puri®ed Francaise Contre le Cancer and the European Economic antisera (2 ml) during 10 min at 48C in binding buer: Community Biomedicine and Health Program.

References

Alani R, Brown P, Binetruy B, Dosaka H, Rosenberg RK, Halazonetis TD, Georgopoulos K, Greenberg ME and Leder Angel P, Karin M and Birrer MJ. (1991). Mol. Cell. Biol., P. (1988). Cell, 55, 917 ± 924. 11, 6286 ± 6295. Herber B, Truss M, Beato M and Muller R. (1994). AlmendralJM,SommerD,BravoMC,BurckhardtH, Oncogene, 9, 2105 ± 2107. Perera J and Bravo R. (1988). Mol. Cell. Biol., 8, 2140 ± Herrera RE, Shaw PE and Nordheim A. (1989). Nature, 340, 2152. 68 ± 70. Barbacid M. (1987). Ann. Rev. Biochem., 56, 779 ± 827. Herschman HR. (1991). Annu. Rev. Bioch., 60, 281 ± 319. Bergers G, Graninger P, Braselmann S, Wrighton C and Hibi M, Lin A, Smeal T, Minden A and Karin M. (1993). Busslinger M. (1995). Mol. Cell. Biol., 15, 3748 ± 3758. Genes Dev., 7, 2135 ± 2148. BineÂ truy B, Smeal T and Karin M. (1991). Nature, 351, 122 ± Hipskind RA, Buscher D. Nordheim A and Baccarini M. 127. (1994). Genes Dev., 8, 1803 ± 1816. Bohmann D, Bos TJ, Admon A, Nishimura T, Vogt PK and Hirai SI, Bourachot B and Yaniv M. (1990). Oncogene, 5, Tijan R. (1987). Science, 238, 1368 ± 1392. 39 ± 46. Bollag G and McCormick F. (1991). Annu. Rev. Cell. Biol., 7, Hirai SI, Ryseck RP, Mechta F, Bravo R and Yaniv M. 601 ± 632. (1989). EMBO J., 8, 1433 ± 1439. Boulton TG, Nye SH, Robbins DJ, Ip NY, Radziejewska E, Howe LR, Leevers SJ, Gomez N, Nakielny S, Cohen P and Morgenbesser SD, DePinho RA, Payanotatos N, Cobb Marshall CJ. (1992). Cell, 71, 335 ± 342. MH and Yancopoulos GD. (1991). Cell, 65, 663 ± 675. KoÈ nig H. (1991). Nucl. Acid. Res., 19, 3607 ± 3611. Busslinger M and Bergers G. (1994). The Fos and Jun families Kyriakis JM, App H, Zhang X, Banerjee P, Brautigan DL, of transcription factors. CRC Press, INC 10, 133 ± 150. Rapp UR and Avruch J. (1992). Nature, 358, 417 ± 421. Chiu R, Angel P and Karin M. (1989). Cell, 59, 979 ± 986. Lallemand D, Spyrou G, Yaniv M and Pfarr C. (1997). Coer P, de JM, Mettouchi A, Binetruy B, Ghysdael J and Oncogene, 14, 819 ± 830. Kruijer W. (1994). Oncogene, 9, 911 ± 921. Landschulz WH, Johnson PF and McKnight SL. (1988). Cohen DR and Curran T. (1988). Mol. Cell. Biol., 8, 2063 ± Science, 240, 1759 ± 1764. 2067. Lau LF and Nathans D. (1987). Proc. Natl. Acad. Sci., 84, Curran T and Franza BJ. (1988). Cell, 55, 395 ± 397. 1182 ± 1186. de Groot R, Schoorlemmer J, van GS and Kruijer W. (1990). Lin AH, Groppi VE and Gorman RR. (1988). Mol. Cell. Nucleic Acids Res., 18, 3195 ± 3302. Biol., 8, 5052 ± 5055. de Vries-Smit, A.M. Burgering BM, Leevers SJ, Marshall CJ Liu JJ, Chao JR, Jiang MC, Ng SY, Yen J and Yang-Yen H. and Bos JL. (1992). Nature, 357, 602 ± 604. (1995). Mol. Cell. Biol., 15, 3654 ± 3663. Dent P, Haser W, Haystead TAJ, Vincent LA, Roberts TM Lloyd A, Yancheva N and Wasylyk B. (1991). Nature, 352, and Sturgill TW. (1992). Science, 257, 1404 ± 1407. 635 ± 638. DeÂ rijard B, Hibi M, Wu IH, Barrett T, Su B, Deng T, Karin Lowenstein EJ, Daly RJ, Batzer AG, Li W, Margolis B, M and Davies RJ. (1994). Cell, 76, 1025 ± 1037. Lammers R, Ullrich A, Skolnik EY, Bar-Sagi D and Devary Y, Gottlieb RA, Lau EF and Karin M. (1991). Mol. Schlessinger J. (1992). Cell, 70, 431 ± 442. and Cell. Biol., 11, 2804 ± 2811. Maki Y, Bos TJ, Davis C, Starbuck M and Vogt PK. (1987). Dony C and Gruss P. (1987). Nature, 328, 711 ± 714. Proc.Natl.Acad.Sci.USA,84, 2848 ± 2852. Feig LA and Cooper GM. (1988). Mol. Cell. Biol., 8, 3235 ± Mechta F, Piette J, Hirai SI and Yaniv M. (1989). New 3243. Biologist, 1, 297 ± 304. Filmus J, Robles AI, Shi W, Wong MJ, Colombo LL and Miao GG and Curran T. (1994). Mol. Cell. Biol., 14, 4295 ± Conti CJ. (1994). Oncogene, 9, 3627 ± 3633. 4310. Garner MM and Revzin A. (1981). Nucl. Acid. Res., 9, Mulcahy LS, Smith MR and Stacey DW. (1985). Nature, 3047 ± 3060. 241, 241 ± 243. API composition and activity in Ras-transformed fibroblasts FMechtaet al 847 Nakabeppu Y, Oda S and Sekiguchi M. (1993). Mol. Cell Sewing A, Burger C, Brusselbach S, Schalk C, Lucibello FC Biol., 13, 4157 ± 4166. and Muller R. (1993). J. Cell. Sci., 104, 545 ± 554. Nakabeppu Y, Ryder K and Nathans D. (1988). Cell, 55, Shaw PE, SchroÈ ter H and Nordheim A. (1989). Cell, 56, 907 ± 915. 563 ± 572. Nishina H, Sato H, Suzuki T, Sato M and Iba H. (1990). Shih C and Weinberg RA. (1982). Cell, 29, 161 ± 169. Proc. Natl. Acad. Sci., 87, 3619 ± 3621. Sistonen L, HoÈ lttaÈ E, MaÈ kelaÈ TP, Keski-Oja J and Atilato K. Noda M, Selinger Z, Scolnick EM and Bassin RH. (1983). (1989). EMBO J., 8, 815 ± 822. Proc. Natl. Acad. Sci., 80, 5602 ± 5606. Smeal T, Binetruy B, Mercola DA and Karin M. (1991). Norman C, Runswick M, Pollock R and Teisman R. (1988). Nature, 354, 494 ± 496. Cell, 55, 989 ± 1003. Smith MR, De Gudicibus SJ and Stacey DW. (1989). Nature, NoseK,ItamiM,SatakeM,ItoYandKurokiT.(1989). 320, 540 ± 543. Mol. Carcinog., 2, 208 ± 216. Stacey DW, Watson T, Kung HF and Curran T. (1987). Mol. O®r R, Dwarki VJ, Rashid D and Verma IM. (1990). Nature, Cell. Biol., 7, 523 ± 527. 348, 80 ± 82. Suzuki T, Murakami M, Onai N, Fukuda E, Hashimoto Y, Pfarr CM, Mechta F, Spyrou G, Lallemand D, Carillo S and Sonobe MH, Kameda T, Ichinose M, Miki K and Iba H. Yaniv M. (1994). Cell, 76, 747 ± 760. (1994). J. Virol., 68, 3527 ± 3535. Piette J, Hirai S and Yaniv M. (1988). Proc. Natl. Acad. Sci., Suzuki T, Okuno H, Yoshida T, Endo T, Nishina H and Iba 85, 3401 ± 3405. H. (1991). Nucl. Acid Res., 19, 5537 ± 5542. Prywes R, Dutta A, Cromlish JA and Roeder RG. (1988). Treisman R. (1987). EMBO J., 6, 2711 ± 2717. Proc. Natl. Acad. Sci., 85, 7206 ± 7210. van Dam H, Duyndam M, Rottier R, Bosch A, Herrlich P, Prywes R and Roeder RG. (1987). Mol. Cell. Biol., 7, 3482 ± Zantema A, Angel P and Van der Erb AJ. (1993). EMBO. 3489. J., 12, 479 ± 487. Quelle DE, Ashmun RA, Shurtle SA, Kato JY, Bar-Sagi D, Vaziri C and Faller DV. (1995). Mol. Cell Biol., 15, 1244 ± Roussel MF and Sherr CJ. (1993). Genes and Dev., 7, 1253. 1559 ± 1571. Wasylyk C, Imler JL and Wasylyk B. (1988). EMBO J., 7, Ryder K, Lanahan A, Perez AE and Nathans D. (1989). 2475 ± 2483. Proc. Natl. Acad. Sci. USA, 86, 1500 ± 1503. Westwick JK, Cox AD, Der CJ, Cobb MH Hibi M and Ryder K, Lau LF and Nathans D. (1988). Proc. Natl Acad Brenner DA. (1994). Proc. Natl. Acad. Sci USA., 91, Sci USA, 85, 1487 ± 1491. 6030 ± 6034. Ryseck RP and Bravo R. (1991). Oncogene, 6, 533 ± 542. Wigler M, Silverstein S, Lee LS, Pellicer A, Cheng YC and Ryseck RP, Hirai SI, Yaniv M and Bravo R. (1988). Nature, Axel R. (1977). Cell, 11, 223 ± 232. 384, 535 ± 537. Wilkinson DG, Bhatt S, Ryseck RP and Bravo R. (1989). Sassone-Corsi P, Ransone LJ, Lamph WW and Verma IM. Development, 106, 465 ± 471. (1988a). Nature, 336, 692 ± 695. Wilson T and Treisman R. (1988). Nature, 336, 396 ± 399. Sassone-Corsi P, Sisson JC and Verma I. (1988b). Nature, Wood KW, Sarnecki C, Roberts TM and Blenis J. (1992). 344, 314 ± 319. Cell, 89, 1041 ± 1050. SchoÈ nthal A, Buscher M, Angel P, Rahmsdorf HJ, Ponta H, Yu CL, Prochownik EV, Imperiale MJ and Jove R. (1993). Hattori K, Chiu R, Karin M and Herrlich P. (1989). Mol. Cell. Biol., 13, 2011 ± 2019. Oncogene, 4, 629 ± 636. Zerial M, Toshi L, Ryseck RP, Schuerman M, MuÈ ller R and SchoÈ nthalA,HerrlichP,RahmsdorfHJandPontaH. Bravo R. (1989). EMBO J., 8, 805 ± 813. (1988). Nature, 354, 325 ± 334 Zinck R, Hipskind RA, Pingoud V and Nordheim A. (1993). SchroÈ ter H, Mueller CGF, Meese K and Nordheim A. EMBO J., 12, 2377 ± 2387. (1990). EMBO J., 9, 1123 ± 1130.