Oncogene (2000) 19, 1752 ± 1763 ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc B-ATF functions as a negative regulator of AP-1 mediated transcription and blocks cellular transformation by Ras and Fos

Deborah R Echlin2,4, Hyi-Jeong Tae3,4, Natalia Mitin1 and Elizabeth J Taparowsky*,1

1Department of Biological Sciences, Purdue University, West Lafayette, Indiana, IN 47907-1392, USA

B-ATF is a nuclear basic leucine zipper protein that amino acids (b) that are contiguous with the ZIP belongs to the AP-1/ATF superfamily of transcription domains of the molecules (Kerppola and Curran, 1991; factors. Northern blot analysis reveals that the human B- Glover and Harrison 1995). Homodimers or hetero- ATF gene is expressed most highly in hematopoietic dimers comprised of members of this extended family tissues. Interaction studies in vitro and in vivo show that of bZIP proteins bind closely related DNA sites the leucine zipper of B-ATF mediates dimerization with referred to as the TRE (for 12-O-tetradecanolyphor- members of the Jun family of proteins. Chimeric proteins bol-13-acetate-response element) or the CRE (cyclic consisting of portions of B-ATF and the DNA binding AMP-response element). Jun/Fos dimers preferentially domain of the yeast activator GAL4 do not stimulate bind to TREs while complexes containing ATFs bind reporter gene expression in mammalian cells, indicating preferentially to CREs (Hai and Curran, 1991). that B-ATF does not contain a conventional transcription In recent years, a number of new bZIP proteins have activation domain. Jun/B-ATF dimers display similar been identi®ed that associate speci®cally with Jun and/ DNA binding pro®les as Jun/Fos dimers, with a bias or Fos. The bZIP protein ATF3/LRF-1 forms dimers toward binding TRE (12-O-tetradecanolyphorbol-13- with Jun, not Fos, to generate a complex that binds acetate-response element) over CRE (cyclic AMP- and activates expression of the tissue-speci®c proenke- response element) DNA sites. B-ATF inhibits transcrip- phalon gene through CRE sites (Hsu et al., 1991). tional activation of a reporter gene containing TRE sites Interestingly, as a homodimer, ATF3/LRF-1 also binds in a dose-dependent manner, presumably by competing CRE sites, yet functions as a transcriptional repressor with Fos for Jun and forming transcriptionally inert Jun/ (Chen et al., 1994). c-Maf and the neural retina bZIP B-ATF heterodimers. Stable expression of B-ATF in protein NRL form heterodimers with either Fos or Jun C3H10T1/2 cells does not reduce cell viability, but does (Kerppola and Curran, 1994) and activate transcription result in a reduced cellular growth rate when compared by binding MAREs ± extended AP-1 sites that contain to controls. This e€ect is dominant in the presence of the a TRE or a CRE (Kataoka et al., 1994). Fos-speci®c growth promoting e€ects of the H-Ras or the v-Fos dimerization is a feature of the small Maf proteins oncoproteins, since expression of B-ATF restricts the (MafF, MafG and MafK) which, unlike c-Maf, do not eciency of focus formation by these transforming encode a transcription activation domain (Fujiwara et agents. These ®ndings demonstrate that B-ATF is a al., 1993; Kataoka et al., 1994). Most recently, two tissue-speci®c transcription factor with the potential to novel proteins called JDP1 and JDP2 were identi®ed as function as a dominant-negative to AP-1. Oncogene speci®c dimerization partners for Jun (Aronheim et al., (2000) 19, 1752 ± 1763. 1997). The role of these proteins in cellular events has yet to be determined. Keywords: AP-1; transcription regulation; protein: The existence of so many AP-1 components DNA interactions; transformation presumbably re¯ects a mechanism by which gene targets under AP-1 transcriptional control are regu- lated in a growth-speci®c and/or a tissue-speci®c Introduction manner. Indeed, while AP-1 activity was de®ned originally as the transcription complex playing a The DNA binding activity described as AP-1 (activat- crucial role in up-regulating gene expression following ing protein-1) represents dimers of the Jun (c-Jun, the stimulation of quiescent cells with growth factors JunB and JunD), Fos (c-Fos, FosB, Fra1 and Fra2), or (reviewed in Angel and Karin, 1991), it is now CREB/ATF (CREB, ATF2, ATF3/LRF1, CREB, recognized that AP-1 DNA binding activity is main- Mafs and others) superfamily of proteins (reviewed in tained in quiescent cells (Pfarr et al., 1994), directs Vogt et al., 1990; Angel and Karin, 1991; Karin et al., certain cells to di€erentiate (Grigoriadis et al., 1993; 1997). AP-1 components form dimers through leucine Pfarr et al., 1994) and plays a role in programmed cell zipper (ZIP) motifs and bind DNA through basic death (Colotta et al., 1992; Smeyne et al., 1993; Ham et al., 1995; Bossy-Wetzel et al., 1997; Wisdom et al., 1999). Clearly, it is the participation of speci®c *Correspondence: EJ Taparowsky subunits in AP-1, combined with the in¯uence of Current addresses: 2Department of Medicine, Section of Hematology/ post-translational protein modi®cations on the activ- Oncology, University of Chicago Medical Center, Chicago, IL 60637, ities of these subunits (reviewed in Karin et al., 1997), USA; 3Biotechnology Research Institute, LG Chemical Ltd., 104-1 that determines whether AP-1 functions as a transcrip- Munji-dong, Yusung, Taejon 305-380, South Korea tional activator or repressor toward a speci®c target 4These two authors to be treated as co-®rst authors Received 15 October 1999; revised 13 January 2000; accepted 25 gene. In agreement with this, the small Maf proteins, January 2000 when dimerized with Fos, are transcriptionally inactive B-ATF functions as a dominant-negative to AP-1 DR Echlin et al 1753 and function to block erythroid di€erentiation by was probed with a 32P-labeled, B-ATF cDNA fragment competing for the DNA binding sites that become (Figure 1, top panel). B-ATF mRNA expression is occupied by transcriptionally active Maf/NF-E2 com- moderate in the spleen, lymph nodes, appendix and plexes in erythrocytes (Kataoka et al., 1995). Similarly, peripheral blood, is low in the thymus and bone as NIH3T3 ®broblasts undergo growth arrest, or as marrow samples, and is not detected in fetal liver. This 3T3T adipocytes di€erentiate, the level of c-Jun restricted pattern of B-ATF gene expression indicates decreases concomitant with an increase in JunD, that B-ATF mRNA is associated with the later stages suggesting opposing cellular roles for AP-1 complexes of myeloid and lymphoid lineage development and is containing c-Jun or JunD (Pfarr et al., 1994; Wang et not a feature of developing erythrocytes. To investigate al., 1996). how the pattern of junB mRNA expression compares In 1995, our laboratory isolated a cDNA from a to B-ATF in these tissues, the ®lter was stripped and human B cell library that encodes a 125 amino acid reprobed with a junB cDNA fragment. As shown in nuclear-localized protein with a central bZIP domain Figure 1 (middle panel), the level of junB mRNA displaying a high degree of sequence similarity to the expression is more widespread and generally higher AP-1/ATF superfamily of bZIP transcription factors than B-ATF. JunB mRNA is abundant in peripheral (Dorsey et al., 1995). We called this protein B-ATF blood and, like B-ATF, is expressed moderately in and have shown that it does not form homodimers, spleen, lymph nodes and appendix. Lower, yet does not dimerize with Fos, but does form hetero- detectable, amounts of junB mRNA also are observed dimers with the Jun family of proteins (c-Jun, JunD, in the thymus and in fetal liver. These expression data and JunB) to bind AP-1 (TRE) consensus DNA sites were interpreted using control hybridization signals (Dorsey et al., 1995). B-ATF mRNA is expressed at generated from a labeled DNA probe speci®c for the detectable levels in very few human tissues, with the constitutively expressed G3PDH mRNA (Figure 1, highest levels of expression restricted to hematopoietic bottom panel). tissues. In this paper, we present data showing that B- ATF does not contain a functional transcription B-ATF does not contain a transcription activation domain activation domain and that Jun/B-ATF heterodimers are transcriptionally inert. The ability of B-ATF to The B-ATF protein is one of the smallest nuclear bZIP function as a dominant-negative to AP-1 is revealed by proteins identi®ed to data and consists of a basic standard transcription activation assays and by the leucine zipper domain ¯anked by 29 amino terminal ability of forced expression of B-ATF to slow the residues and 50 carboxyl terminal amino acids. The growth of ®broblasts and to dramatically reduce transformation of these cells by the v-fos or H-ras oncogenes. Expression of a fusion protein in which B- ATF is linked to the potent transcription activation domain of the herpes simplex virus VP16 protein augments reporter gene expression by Fos and Jun and enhances focus formation by H-ras. This con®rms the functional participation of B-ATF in cellular AP-1 complexes and suggests that, in hematopoietic cells where B-ATF is normally expressed, the novel proper- ties of this bZIP protein contribute to the modulation of AP-1 activity that is critical to cell lineage establishment and expansion (reviewed in Foletta et al., 1998).

Results

B-ATF mRNA is expressed predominantly in hematopoietic tissues and cell lines The B-ATF cDNA was isolated from an EBV- immortalized human B cell cDNA library. Using a yeast two-hybrid screen with B-ATF as the bait protein and the same B cell cDNA library as the source of interacting protein partners, we found that JunB served Figure 1 B-ATF is expressed in human hematopoietic tissues. as a speci®c dimerization partner of B-ATF (Dorsey et Northern blot hybridization of poly A+ mRNA isolated from al., 1995). Follow-up studies by our lab and another human hematopoietic tissues was performed as described in revealed that B-ATF mRNA expression is restricted to Materials and methods. The top panel shows the results of hybridizing with a radiolabeled fragment of the B-ATF cDNA EBV-positive B cell lines such as Raji and 721 and to and reveals the presence of B-ATF transcripts in all samples human T lymphocyte cell lines such as Jurkat and Sup- except fetal liver. The blot was stripped and subsequently T1 (Hasewaga et al., 1996; Johansen and Taparowsky, hybridized with a human junB cDNA probe to assess expression manuscript in preparation). In an e€ort to survey the of this B-ATF dimerization partner in these same tissues (middle level of B-ATF gene expression in various human panel). The blot was stripped a second time and hybridized with a G3PDH probe to check the amount and integrity of the mRNA blood forming tissues, a Northern blot of poly A+ loaded in each lane. RNA size markers are indicated to the left of mRNA isolated from one fetal and six adult tissues the exposures showing B-ATF and junB mRNA expression

Oncogene B-ATF functions as a dominant-negative to AP-1 DR Echlin et al 1754 ability of B-ATF to form heterodimers with JunB, c- amino acids 1 ± 347 of JunB as a GAL4 DB fusion Jun and JunD and bind AP-1 DNA sites (Dorsey et protein and ®ve additional plasmids expressing over- al., 1995) suggests that B-ATF plays a role in the lapping portions of B-ATF as fusions with the cellular activities of AP-1 perhaps by altering tran- transactivation domain (TAD) of herpes simplex virus scription activation by Jun. To determine if B-ATF VP16 (Figure 3). C3H10T1/2 cells were transiently possesses a functional transcription activation domain transfected with (GAL4)5 CAT, GAL4 DB-JunB and (TAD), various portions of the B-ATF protein were each TAD-B-ATF protein as described in Materials expressed as fusions with the DNA binding domain and methods. For these assays, CAT expression (DB) of the yeast GAL4 protein (Figure 2). C3H10T1/2 stimulated by the potent chimeric activator GAL4 cells were transiently transfected as described in DB-VP16 was set to 100% and co-expression of GAL4 Materials and methods with each GAL4 DB activator DB-JunB and the VP16 TAD, or of GAL4 DB and B- and a chloramphenicol acetyl transferase (CAT) ATF (1 ± 125) TAD, served as the negative controls reporter plasmid containing ®ve tandem GAL4 protein (0.5 and 0.7% of GAL4 DB-VP16, respectively). binding sites. Basal transcription activation by GAL4 Signi®cant levels of CAT activity were observed only DB alone was set to 1.00 and the level of CAT activity when TAD fusions representing full-length B-ATF generated by the well-characterized chimeric activator (amino acids 1 ± 125) or an amino terminal truncation GAL4 DB-E1A served as the positive control (Figure of B-ATF (amino acids 49 ± 125) containing the leucine 2). Neither the full-length B-ATF protein (1 ± 125), nor zipper domain (amino acids 54 ± 81) were co-expressed fusion proteins containing the amino- or carboxyl- with GAL4 DB-JunB. The strength of the mammalian terminal portion of B-ATF, were able to stimulate two-hybrid interaction between JunB and B-ATF is signi®cant (GAL4)5 CAT reporter gene expression in relatively low when compared to the CAT activity C3H10T1/2 cells. Western blot analysis of the cell generated by a monomeric activator (e.g. GAL4 DB- extracts demonstrated that all of the activators used in VP16). However, in control experiments, we observed this experiment were expressed to comparable levels that the Jun/B-ATF interaction results in a level of (data not shown). These data support our prior activation that is at least twofold higher than that observation that GAL4 DB-B-ATF(1 ± 125) does not produced from the interaction of two well-character- activate transcription in Saccharomyces cerevisae ized dimerization partners, Myc and Max (data not (Dorsey et al., 1995) and suggest that the participation shown). Therefore, we conclude that in mammalian of B-ATF in AP-1 complexes is unlikely to enhance cells expressing both JunB and B-ATF, direct protein- transcription. Instead, B-ATF may function as a protein interaction is possible and is mediated by the dominant-negative to block AP-1-mediated gene ex- leucine zipper region of B-ATF. pression in cells. Jun/B-ATF heterodimers show a preference for binding B-ATF interacts with JunB through the B-ATF leucine AP-1 consensus DNA zipper motif Members of the AP-1/ATF superfamily of transcrip- The ability of JunB and B-ATF to interact is assumed tion factors function as dimers that bind two closely to be mediated by the amphipathic a-helical leucine related target DNA motifs called the AP-1 (or TRE) zipper motifs present in each protein. To test this site (-TGA[C/G]TCA-) and the CRE site (-TGACGT- hypothesis and to demonstrate a two-hybrid, protein- CA-). In many instances, there is preferred binding by protein interaction between JunB and B-ATF in dimers to one site over the other, and this is thought to mammalian cells, we generated a vector expressing provide a mechanism by which small changes in the

Figure 2 GAL4 DB-B-ATF fusion proteins and their transactivation potential. The full-length B-ATF protein and two subdomains of B-ATF were expressed as fusions with the DNA binding domain (DB) of the yeast activator GAL4. C3H10T1/2 cells were transiently transfected as described in Materials and methods with 5 mg of the (GAL4)5 CAT reporter and 5 mg of the indicated GAL4 activator. The CAT activity in each cell extract is expressed as the fold-increase over the activity obtained from the GAL4 DB plasmid alone, which is set at 1.00. Expression of a GAL4 fusion protein containing the transcription activation domain of the adenovirus E1A transactivator was used as the positive control. Each value represents the average of at least three independent transfection experiments. Error bars indicate the standard deviation (s.d.) which is enumerated to the right of the bar graph

Oncogene B-ATF functions as a dominant-negative to AP-1 DR Echlin et al 1755

Figure 3 Two-hybrid interaction of JunB and B-ATF in C3H10T1/2 cells. C3H10T1/2 cells were transiently transfected with 5 mgof the (GAL4)5 CAT reporter, 5 mg of the indicated GAL4 DB plasmids and 5 mg of each plasmid encoding various subregions of B- ATF fused to the transcription activation domain (TAD) of VP16. The CAT activity in each cell extract is expressed relative to the level of activity measured for the monomeric GAL4 DB-VP16 which is set at 100%. Each value represents the average of at least three independent determinations and the error bars indicate standard error (s.e.) which is enumerated to the right of the bar graph level of particular bZIP proteins generate signi®cant at position 6 of the core. These nucleotide changes changes in target promoter activation (Hai and Curran have been implicated in the di€erential binding of AP-1 1991). B-ATF is a tissue-speci®c member of the AP-1/ complexes to DNA in vitro (Bohmann et al., 1987). As ATF superfamily and forms dimers with the Jun shown in Figure 5, the competition pro®le of the Jun/ proteins to bind AP-1 DNA (Dorsey et al., 1995). To B-ATF shifted complex with the mut1 and mut2 investigate the DNA binding preference of Jun/B-ATF oligonucleotides re¯ects the behavior of the Jun/Fos complexes, electrophoretic mobility shift assays complex assayed in parallel with the same competitors. (EMSA) were performed using 32P-labeled AP-1 (TRE) or CRE oligonucleotides. In vitro transcribed B-ATF inhibits transcription activation by AP-1 and translated B-ATF and JunB proteins were used to generate protein/DNA complexes that were resolved by The shared DNA binding preference of Jun/B-ATF electrophoresis through native polyacrylamide gels and and Jun/Fos complexes, coupled with the results of the visualized by autoradiography. Previously, we demon- GAL4 transcription assays (Figure 2), suggests that B- strated that Jun/B-ATF heterodimers display strong ATF competes with other AP-1 family members (e.g., binding to AP-1 DNA (Dorsey et al., 1995). In this Fos) for the intracellular pool of Jun and generates study, we have extended this observation and now DNA binding complexes that display a reduced show that this binding is competed eciently by adding transactivation potential. To test this hypothesis, excess unlabeled AP-1 DNA, is reduced modestly by 3TRE-Lux, a luciferase reporter gene controlled by CRE DNA and is not a€ected at all by competing with three tandem AP-1 consensus sites, was used to the unrelated consensus target DNA for the muscle measure AP-1 activity in mouse F9 teratocarcinoma regulatory factors (MRF) (Figure 4A). JunB/B-ATF cells (Figure 6) as described in Materials and methods. complexes also bind labeled CRE DNA (Figure 4A), F9 cells display a low level of endogenous AP-1 although the binding is weaker and is competed activity and thus have been used extensively to eciently by both the homologous CRE and the investigate transcription activation by AP-1 compo- heterologous AP-1 oligonucleotides. In Figure 4B, the nents (Angel and Karin, 1991). As shown in Figure 6A, relative anity of JunB/B-ATF complexes for AP-1 the 3TRE-Lux reporter shows measurable basal versus CRE DNA is compared using competition expression in F9 cells which is increased twofold EMSA. While a 50% reduction in AP-1 binding is following the addition of c-Fos. In the presence of B- obvious when a tenfold molar excess of unlabeled AP-1 ATF, c-Fos activation was reduced by 50%, support- DNA is added, there is no signi®cant decrease in AP-1 ing the hypothesis that B-ATF interferes with the binding until 50- or 100-fold excess of CRE DNA is formation of transcriptionally active (i.e., Jun/Fos) AP- added. These data demonstrate that JunB/B-ATF 1 complexes. Expression of B-ATF in the absence of complexes behave similarly to Jun/Jun or Jun/Fos co-transfected c-Fos results in a consistent 30% complexes and show a preferred binding to AP-1 decrease in the basal activity of the luciferase reporter DNA. gene, which again suggests that B-ATF disrupts the Additional in vitro evidence for the shared preference basal activity of AP-1 complexes in F9 cells. The of Jun/B-ATF and Jun/Fos complexes for AP-1 target ability of B-ATF to inhibit AP-1 mediated transactiva- DNA was obtained by competition EMSA using the tion was tested further by transfecting F9 cells with a wildtype AP-1 oligonucleotide and two non-consensus set amount of c-Fos expression vector (1 mg) and AP-1 oligonucleotides. AP-1 mut1 contains an A to T increasing amounts of pEMscribe B-ATF (0.4 ± 10 mg) transversion in the last position of the AP-1 core (Figure 6B). As predicted, the levels of 3TRE-Lux consensus and AP-1 mut2 contains a C to A transition reporter gene expression decreased as the relative input

Oncogene B-ATF functions as a dominant-negative to AP-1 DR Echlin et al 1756

Figure 4 JunB/B-ATF heterodimers preferentially bind AP-1 DNA. (A) EMSA using 32P-labeled, double-stranded DNA containing a single AP-1 or CRE site was performed with the indicated in vitro translated proteins. Control lanes represent the addition of unprogrammed rabbit reticulocyte lysate to the probe DNA. Complexes representing JunB/B-ATF bound to AP-1 DNA (a) or CRE DNA (b) are indicated by arrows. 100-fold excess of the indicated unlabeled competitor DNA was added to the binding reactions. JunB/B-ATF complexes bind both the AP-1 and CRE DNA, yet are competed more eciently by AP-1 DNA. The unrelated MRF DNA does not compete for JunB/B-ATF. (B) In vitro translated proteins were incubated with a 32P-labeled AP-1 probe and increasing amounts of unlabeled AP-1 (left lanes) or CRE DNA (right lanes) competitor DNA. Results support the conclusion from (A) which indicates preferred binding by JunB/B-ATF complexes to AP-1 DNA

of B-ATF to c-Fos DNA is increased. Control VP16 TAD, B-ATF or B-ATF TAD (see Materials and experiments demonstrate that increasing levels of the methods for details on the construction of this basic helix ± loop ± helix transcription factor Mist1 does plasmid). Parallel groups were examined for 3TRE- not reduce AP-1-mediated transactivation by c-Fos Lux reporter gene activation in the presence or absence (Figure 6C) and that B-ATF does not inhibit the of co-transfected c-Fos and c-Jun. As shown in Figure expression of non-AP-1 regulated reporter genes under 7, without prior activation by Fos and Jun (black similar experimental conditions (Wang et al., 1999). bars), the expression of B-ATF results in a 50% If B-ATF functions as a non-activating competitor reduction in AP-1 transactivation (compare lanes 1 and of c-Fos in the AP-1 dimer, it follows that a chimeric 3), while the expression of B-ATF TAD results in protein consisting of B-ATF and the TAD from VP16 almost a tenfold activation over basal levels (compare should augment the transactivation of AP-1 reporter lanes 1 and 5). This e€ect also is observed in the genes. For these experiments, C3H10T1/2 cells were presence of c-Fos and c-Jun where the expression of B- transfected with 3TRE-Lux and expression vectors for ATF TAD increases AP-1 reporter gene activation

Oncogene B-ATF functions as a dominant-negative to AP-1 DR Echlin et al 1757

Figure 5 Jun/B-ATF and Jun/Fos complexes display the same binding preferences for non-consensus AP-1 sites. In vitro translated proteins were incubated with a 32P-labeled AP-1 DNA probe in the presence of a 100-fold excess of the indicated competitor DNA. The control lane represents the addition of unprogrammed rabbit reticulocyte lysate to the probe DNA. The migration of the complexes representing c-Jun/B-ATF (a) or c-Jun/c-Fos (b) bound to AP-1 DNA is indicated by arrows. The nucleotide sequence of the wildtype (wt), mut1 and mut2 AP-1 DNA is presented below the EMSA. The similar binding pro®les of the c-Jun/B-ATF and c- Jun/c-Fos complexes suggest a shared preference for the same non-consensus target sites

®vefold over control (compare lanes 2 and 6). Taken C3H10T1/2 culture reached saturation density on day together with the data in Figure 6, these observations 6 after plating with a total cell count of approximately strongly support the hypothesis that wildtype B-ATF 16106 cells/60 mm dish. Ras neo 11A cells, which functions in mammalian cells as a component of AP-1 display similar growth kinetics as C3H10T1/2 cells, are and as a negative modulator of the transcription not contacted-inhibited and grow to a saturation potential of the complex. density that is threefold higher than the wildtype cells. The B-ATF-4 cell line displays di€erent growth kinetics, yet reaches the same saturation density as Biological activities of B-ATF confirm its role as a control cells by day 10 of the experiment. Two negative regulator of AP-1 additional B-ATF expressing C3H10T1/2 cell lines The ability of B-ATF to block endogenous AP-1 generated in this study also display a similar slow transactivation can be tested by examining the e€ects growth phenotype (data not shown). of B-ATF in cell growth and transformation assays The ability to establish several B-ATF expressing cell where the importance of AP-1 activity is well- lines for long term study suggests that forced documented (Lloyd et al., 1991; Smeal et al., 1991; expression of B-ATF is not toxic to C3H10T1/2 cells. Vandel et al., 1996; Johnson et al., 1996). We generated To demonstrate this directly, we stably transfected a C3H10T1/2 cell line, B-ATF-4, stably expressing C3H10T1/2 cells with pDCR, a vector carrying a moderate levels of a hemaglutinin-antigen (HA)-tagged neomycin resistance gene cassette or with the pDCR human B-ATF protein and compared the growth of vector also expressing HA-tagged B-ATF (pDCR-B- this cell line to wildtype C3H10T1/2 cells and to a ATF). After 14 days of growth in selective medium, the previously characterized C3H10T1/2 cell line expres- cultures were ®xed, stained with Giemsa and the sing oncogenic Ras p21 (see Materials and methods for resulting colonies were counted. For both groups, the details) (Figure 8A). The linear growth of the total number of G-418 resistant colonies was equiva-

Oncogene B-ATF functions as a dominant-negative to AP-1 DR Echlin et al 1758 lent (data not shown), yet the appearance of the control colonies was dramatically di€erent when compared to the B-ATF colonies (Figure 8B). C3H10T1/2-B-ATF colonies are more sparsely popu- lated and less densely packed than the control colonies. These results support the growth curve studies using, in this case, a mixed population of cell clones expressing B-ATF. We conclude that while B-ATF expressing C3H10T1/2 cells remain viable, the cell cycle is altered such that their generation time is lengthened signi®- cantly. To further examine the role of B-ATF as a negative regulator of AP-1-mediated activities, cellular transfor- mation by activated human H-Ras p21 in the presence or absence of B-ATF was analysed. AP-1 function is required for Ras-induced cellular transformation (Lloyd et al., 1991; Smeal et al., 1991; Vandel et al., 1996; Johnson et al., 1996) and we wanted to test the potential of B-ATF to counteract Ras-stimulated cell growth and transformation. Having at our disposal the

Figure 7 A B-ATF TAD fusion protein activates AP-1 reporter gene activity. C3H10T1/2 cells were transiently transfected with 1 mg of the 3TRE-Lux reporter, 0.5 mg of RSV-lacZ and 2.5 mgof pNLVP (VP16 TAD), pEMscribe B-ATF (B-ATF) or B-ATF TAD in the presence or absence of pEMscribe c-fos and pEMscribe c-jun (0.25 mg of each). The total amount of DNA per transfection group was kept constant by the addition of pEMSVscribea-2 vector DNA. Forty-eight hours after transfec- tion, cells extracts were prepared and assayed for luciferase and b- galactosidase activity as described previously. Results are expressed as the fold-increase over the luciferase activity measured for 3TRE-Lux in the presence of pNLVP alone which is set at 1.0. Each value represents the average of six independent experiments. Error bars indicate the standard error of the mean

Figure 6 B-ATF inhibits transactivation of an AP-1 reporter gene in vivo.(A) F9 teratocarcinoma cells were transiently pEMscribe c-fos, 0.5 mg of RSV-lacZ, the indicated amount of transfected with 1 mg of the 3TRE-Lux reporter gene, 2.5 mgof pEMscribe B-ATF and the appropriate amount of pEMSVscri- pEMscribe c-fos and/or pEMscribe B-ATF, and 0.5 mg of RSV- bea-2 vector DNA to adjust plasmid input to 12.5 mg per group. lacZ to control for transfection eciency. The total input of DNA Results are expressed relative to the luciferase activity in the c-Fos in all groups was kept constant (6.5 mg) by adding the appropriate control group which is set at 100. Each value represents the amount of pEMSVscribea-2 vector DNA. Forty-eight hours post- average of at least three independent experiments. Error bars transfection, cell extracts were prepared. Luciferase activity and b- represent the standard error of the mean. (C) Control experiment galactosidase activity were determined as described in Materials performed using the basic ± helix ± loop ± helix transcription factor and methods. Results are expressed relative to the luciferase Mist1. The amounts of reporter, RSV-lacZ, pEMscribe c-fos, activity in the c-Fos group which is set at 100. Each value pcDNA3-Mist1 and pcDNA3 vector DNA are as indicated in (B). represents the average of six independent experiments. Error bars Each value represents the average of three independent represent the standard error of the mean. (B) F9 cells were determinations. Error bars represent the standard error of the transfected as described above with 1 mg of 3TRE-Lux, 1 mgof mean

Oncogene B-ATF functions as a dominant-negative to AP-1 DR Echlin et al 1759

Figure 8 B-ATF alters the growth properties of C3H10T1/2 ®broblasts. (A) Growth analysis of C3H10T1/2 ®broblasts, C3H10T1/2 cells transformed by the human H-ras oncogene (ras neo 11A) and B-ATF-4, a cell line that stably expresses exogenous B-ATF. Cells were plated in 60-mm dishes on day 0 at a density of 16104 cells per dish. Growth medium was changed on days 1, 4, and 7; duplicate plates for each group were counted daily starting on day 2. The plotted results represent the average of counts from three independent experiments. Error bars indicate the standard error of the mean. (B) Phase contrast view (406) of a representative G-418 resistant colony generated from the stable transfection of C3H10T1/2 cells with pDCR (neo control) (left panel) or pDCR-B-ATF (right panel). Note the reduced cell density of the B-ATF expressing colony, despite growth for an equivalent length of time (14 days)

Table 1 E€ects of B-ATF on focus formation in C3H10T1/2 cells Number of focia Efficiency of focus formation Input DNA Exp. # 1 2 3 4 5 (%+s.e.m.)b

H-Ras 148 132 98 184 141 100% H-Ras+B-ATF 53 85 29 119 84 51% (+8%) B-ATF ± ± ± 0 ± 0% H-Ras+B-ATF TAD ± 401 ± 286 372 241% (+44%) B-ATF TAD ± 0 ± 0 ± 0% H-Ras+VP16 TAD ± 146 ± 202 157 110% (+1%) VP16 TAD ± 0 ± ± ± 0% no DNA ± 0 ± 0 ± 0% aThe total number of foci counted for the six plates in each experimental group. bPer cent focus formation is calculated within each experimental group by setting the number of foci obtained with H-Ras alone to 100%. The eciency of focus formation for each group represents the average of per cents calculated from individual experiments. The standard error of the mean (s.e.m.) for each group is noted in parentheses

Oncogene B-ATF functions as a dominant-negative to AP-1 DR Echlin et al 1760 indicating that AP-1 activation is necessary, but not sucient, to produce this biological phenomenon. As a more direct test of the ability of B-ATF to inhibit cellular transformation events relying on AP-1 activity, C3H10T1/2 cells were stably transfected with FBJ/R v-fos, pCDR-B-ATF, or FBJ/R v-fos and pDCR-B-ATF as described in Materials and methods. Figure 9 shows representative plates from this experiment. v-Fos alone induces a high level of focus formation in C3H10T1/2 cells, while a ®vefold excess of B-ATF reduces the amount of foci obtained with v- Fos by approximately 90%. A Western blot of cell extracts prepared from control cultures transiently transfected with the v-Fos and/or B-ATF expression vectors demonstrated that both proteins accumulate to similar levels in the cells. These results directly address the potential of B-ATF to inhibit Fos biological activity and support the hypothesis that B-ATF competes with Fos for the available pool of Jun dimerization partners and, in doing so, inhibits AP-1- mediated gene expression and the biological phenom- ena that rely on this expression.

Discussion

In this paper, we have characterized the activities of B- ATF, a relatively new member of the AP-1 transcrip- tion factor family. Analyses of the mRNA encoding B- ATF reveals that the 0.9 bp transcript is expressed Figure 9 B-ATF inhibits cellular transformation by v-Fos. Standard C3H10T1/2 focus assays were performed as described predominantly in hematopoietic tissues and represents in Materials and methods. In this experiment, cultures were the full-length transcript of a single gene (Dorsey et al., scored for focus formation 21 days following transfection due to 1995; Meyer et al., 1998). In humans, the highest level the small, compact phenotype of v-Fos foci. A representative of B-ATF mRNA is detected in circulating T cells and plate from each experimental group is pictured. Note the in EBV-infected B lymphocytes (Johansen and Tapar- dramatic reduction in the number of foci when v-Fos is co- expressed with B-ATF owsky, manuscript in preparation). In the mouse, a more detailed analysis of B-ATF mRNA and protein expression has been performed and reveals a di€er- B-ATF TAD construct (Figure 7), we also were ential expression of B-ATF during discrete stages of T interested in examining if this arti®cial AP-1 `activator' cell development in the thymus and constitutive would augment Ras transformation or perhaps func- expression of B-ATF in B lymphocytes (Williams and tion as a transforming protein on its own. C3H10T1/2 Taparowsky, unpublished observations). The role of cells were stably transfected with pT24H-ras and either AP-1 complexes in the immune system is well- pDCR vector, pDCR-B-ATF, or B-ATF TAD. Cells documented (reviewed in Foletta et al., 1998), in¯uen- were maintained in reduced serum for a minimum of cing a myriad of growth and di€erentiation decisions 14 days after which time the cultures were ®xed, relating to the development and function of hemato- stained and scored for focus formation. Table 1 lists poietic cells. As a result, it was of paramount the number of transformation events scored in importance to explore the biochemical properties of individual experiments. No spontaneous foci formed B-ATF in order to assess how its presence or absence in the control plates and B-ATF expression alone did may in¯uence cellular AP-1 activity. not transform C3H10T1/2 cells. H-Ras expression We have shown previously that B-ATF does not induces a moderate level of focus formation (set to form homodimers and preferentially forms heterodi- 100%) which is reduced by approximately 50% in the mers with all three members of the Jun family of bZIP presence of pDCR-B-ATF. We believe that this proteins. In this study, we extended this observation to reduced level of focus formation is due to the show that heterodimer formation relies on the leucine inhibition of the Ras transformed phenotype, rather zipper of B-ATF and that the DNA binding site than to the general suppression of cell growth as preference for Jun/B-ATF dimers is strikingly similar described above, since two plates from the H-Ras plus to that of Jun/Fos dimers. This strongly suggests that B-ATF group were maintained for an additional 7 days in cellular situations where B-ATF, Fos and Jun are following transfection (21 days total) and showed no co-expressed, B-ATF competes with Fos for the evidence of slow-growing, microscopic foci that would available pool of Jun and forms dimers that will be able to account for the reduced number of foci occupy the same target DNA. However, in contrast to overall. Interestingly, transfection of C3H10T1/2 cells Fos, B-ATF does not possess a conventional transcrip- with the B-ATF TAD expression vector did not cause tion activation domain and thus suggests that the focus formation on its own, yet increased Ras- competition between B-ATF and Fos for the tran- mediated focus formation by approximately 2.5-fold, scriptionally active members of the Jun protein family

Oncogene B-ATF functions as a dominant-negative to AP-1 DR Echlin et al 1761 (primarily JunB and c-Jun) will have a dramatic e€ect replacing the basic region of Fos with an acidic on AP-1 target gene expression. Indeed, while Jun/Fos extension (Olive et al., 1997). These arti®cial molecules complexes activate transcription from the 3TRE-Lux function as predicted and in a direct comparison, reporter gene, we have observed that the expression of duplicate the e€ects of B-ATF on AP-1 reporter gene exogenous B-ATF inhibits AP-1 reporter gene activity activity and in C3H10T1/2 focus assays (Olive et al., in cells. Furthermore, we generated a B-ATF protein 1997; Echlin and Taparowsky, unpublished observa- fused to the potent transcription activation domain of tions). VP16 and demonstrated that this chimeric molecule In 1997, Aronheim et al. identi®ed a protein in rat exerts a positive impact on 3TRE-Lux reporter gene pituitary cells called JDP-1 whose sequence similarity expression. The transcriptional e€ects of B-ATF on to B-ATF suggests that it too will function as an this arti®cial AP-1 reporter gene have been observed in inhibitor of AP-1 (Aronheim et al., 1997). The two di€erent cell lines and have been duplicated in existence of at least two distinct bZIP molecules with separate studies (Burg and Taparowsky, manuscript in this property reveals a new mechanism for controlling preparation) using human T lymphocytes and the AP-1 activity that is similar to the role of dominant- tissue-speci®c GM-CSF reporter gene which is regu- negative ICER proteins in regulating members of the lated by the cooperative interaction between AP-1 and CREB/ATF family of proteins (reviewed in Sassone- Ets transcription factors (Wang et al., 1994). Corsi et al., 1995) or the DFosB (FosB2) protein, the Our simple interpretation of the observed e€ects of product of an alternatively spliced mRNA from the B-ATF on reporter gene activity is that B-ATF mouse fosB gene (Nakabeppu and Nathans, 1991; functions as a dominant-negative to AP-1. In an Carrozza et al., 1997), in regulating AP-1 (Nakabeppu attempt to utilize B-ATF as a biological reagent to and Nathans, 1991; Carrozza et al., 1997). In addition, block cellular processes that have been shown to rely the existence of B-ATF and JDP-1 has provided us on the positive activation of AP-1 (Lloyd et al., 1991; with the opportunity to compare the bZIP regions of Smeal et al., 1991; Vandel et al., 1996; Johnson et al., these inhibitors with the bZIP regions of other AP-1 1996), we used C3H10T1/2 focus assays to test if B- family members. Interestingly, the basic regions of both ATF would inhibit transformation induced by the B-ATF and JDP-1 contain a serine in place of a expression of oncogenic H-Ras of v-Fos. In both cases, conserved cysteine residue that is responsible for the B-ATF expression resulted in a reduced number of foci redox-regulated DNA binding properties of AP-1 compared to controls. This suggests that the in¯uence (reviewed in Xanthoudakis and Curran, 1996). The of H-Ras and v-Fos on the growth and morphology of post-translational modi®cation of B-ATF in vivo and cells can be counteracted by B-ATF, presumably the potential impact of various modi®cations on through its negative e€ect on Jun-mediated transacti- dimerization and DNA binding is the subject of future vation. Support for this conclusion was obtained from studies and will help to de®ne the cellular conditions the analysis of C3H10T1/2 cells overexpressing B-ATF, required for B-ATF to eciently block the activity of which remain viable yet display a slow growth Jun containing AP-1 protein complexes. phenotype, again suggesting that B-ATF impairs the function of AP-1. Interestingly, while the transcription- ally active B-ATF-TAD construct was able to augment Materials and methods transformation by activated H-Ras, it did not perform equivalently to v-Fos and was unable to cause focus Plasmid constructions formation on its own. This may be a consequence of The GAL4 (1 ± 147) vector, which contains the DNA binding di€erences in the protein-protein interactions speci®ed domain (DB) of the yeast activator protein GAL4, the by the TADs of VP16 and Fos, or to additional (GAL4)5 E1B TATA CAT reporter and the chimeric mutations in v-Fos that augment its activity. In this activators, GAL4 DB-E1A and GAL4 DB-VP16, have been regard, we have not observed focus formation described previously (Min et al., 1994). GAL4 DB B-ATF following the overexpression of the c-Fos protein in (1 ± 125) was constructed by inserting the EcoRI/XbaI C3H10T1/2 cells (Tae and Taparowsky, unpublished fragment from pBS-B-ATF (Dorsey et al., 1995) into observations). In analysing the results of these culture EcoRI/XbaI-linearized GAL4 (1 ± 147). Portions of the B- experiments, we are encouraged by the slow growth ATF cDNA spanning residues 1 ± 32 and 81 ± 125, ¯anked by phenotype of the B-ATF-4 cell line since it is similar to 5' EcoR1 and 3' XbaI restriction sites, were generated by PCR ampli®cation and cloned into EcoRI/XbaI-restricted that described for c-jun7/7 null ®broblasts (Johnson GAL4 (1 ± 147) to create GAL4 DB-B-ATF (1 ± 32) and et al., 1993; Schreiber et al., 1999). Additional GAL4 DB-B-ATF (81 ± 125). GAL4 DB-JunB was con- experiments to characterize the cell cycle of B-ATF-4 structed by excising the junB cDNA from pBS-junB (Dorsey cells and to examine potential changes in the et al., 1995) with EcoRI and XbaI and cloning the fragment expression of AP-1 target genes such as p53 (Schreiber into GAL4 (1 ± 147). Regions of the B-ATF cDNA were et al., 1999) and cyclin D1 (Wisdom et al., 1999) are ampli®ed by PCR with primers designed to provide 5' EcoRI underway. and 3' XbaI restriction sites and subcloned into pBS-KS+ The B-ATF protein represents the ®rst bZIP protein, (Stratagene, La Jolla, CA, USA) to generate pBS-B-ATF (1 ± encoded by the full-length transcript of a distinct gene 125), pBS-B-ATF (1 ± 31), pBS-B-ATF (1 ± 50) and pBS-B- (Meyer et al., 1998), that has been shown to function ATF (49 ± 125). The pBS-B-ATF plasmids were restricted with XhoI and XbaI and the inserts subcloned into pNLVP as a dominant-negative to AP-1 by forming transcrip- (Dang et al., 1991) to generate TAD-B-ATF (1 ± 125), TAD- tionally inert dimers with the Jun family of proteins. B-ATF (1 ± 31), TAD-B-ATF (1 ± 50) and TAD-B-ATF (49 ± Earlier studies by a number of labs sought to design 125). pEMscribe B-ATF contains the human B-ATF coding synthetic dominant-negatives to AP-1 by removing the region modi®ed with EcoRI linkers prior to insertion into TAD of Fos (Wick et al., 1992; Brown et al., 1994) or EcoRI-linearized pEMSVscribea-2 (Davis et al., 1987). Jun (Lloyd et al., 1991; Brown et al., 1993), or by pEMscribe c-jun and pEMscribe c-fos were constructed

Oncogene B-ATF functions as a dominant-negative to AP-1 DR Echlin et al 1762 similarly using the rat pSP65 c-jun and c-fos cDNAs as the phosphate-DNA precipitates containing 30 mg of C3H10T1/2 starting material (gift of S Konieczny, Purdue University). carrier DNA and either 50 ng of pDCR or pDCR-B-ATF The 3TRE-Lux reporter (gift of J Massague, Howard Hughes were added to each of two plates. Twenty-four hours after Medical Institute Research Laboratories, Memorial Sloan- transfection, the cultures were passaged 1 : 6 and maintained Kettering Cancer Center, New York, NY, USA) contains in medium supplemented with G-418. After 14 days, the three tandem TRE elements inserted upstream of a minimal cultures were ®xed with 100% methanol for 10 min and TATA box promoter which drives expression of the ®re¯y stained with Giemsa (EM Diagnostic Systems, Gibbstown, luciferase gene. RSV-lacZ, the b-galactosidase expression NJ) to establish the eciency of colony formation. Individual plasmid used to normalize transfection eciency, has been colonies from live cultures were photographed and then described previously (Min et al., 1994), as have the pDCR selected using cloning cylinders, expanded in culture and and pDCR-B-ATF DNAs (Dorsey et al., 1995) and the analysed for the expression of exogenous B-ATF by Western pcDNA3 and pcDNA3-Mist1 plasmids (Lemercier et al., blot hybridization with HA antiserum (Boehringer Man- 1997). B-ATFTAD was generated from a three-part ligation nheim, Indianapolis, IN, USA) as described previously using a PCR ampli®ed B-ATF cDNA fragment, tailed with (Ramocki et al., 1997). EcoRI and ClaI restriction sites, a PCR ampli®ed VP16 TAD Focus assays were performed as described previously (Min fragment tailed with ClaI and XbaI restriction sites, and et al., 1994). Individual precipitates containing 30 mgof EcoRI/XbaI-cleaved pcDNA3 (Invitrogen; Carlsbad, CA, carrier DNA and the indicated combinations of 100 ng pT24 USA). In B-ATFTAD, the VP16 TAD is expressed as an H-ras and 300 ng of each B-ATF plasmid DNA, or 250 ng of in-frame fusion at the carboxyl terminus of B-ATF. The FBJ/R v-fos DNA and 1.25 mgofB-ATF DNA were added to each v-fos and pT24 H-ras plasmids used for the C3H10T1/2 of two plates. The increased amount of DNA used for the focus assays have been described previously (Miller et al., Fos transfections was necessary in order to provide a more 1985; Taparowsky et al., 1982). dramatic contrast in focus formation, given that the small, compact phenotype of v-fos-induced C3H10T1/2 foci are dicult to count (see Figure 9). Twenty-four hours after Northern blot hybridization transfection, the cultures were passaged 1 : 3 and maintained A human immune system Multiple Tissue Northern (MTN) in reduced serum medium. Fourteen to 21 days after blot (Clontech, Palo Alto, CA, USA) was hybridized with a passaging, the plates were ®xed and stained with Giemsa as 32P-labeled B-ATF cDNA probe using the hybridization and described above and the number of foci per experimental wash conditions described in Dorsey et al. (1995). The group tabulated. hybridized ®lter was stripped as described in Davenport and For the growth curve, C3H10T1/2 cells, ras neo 11A cells Taparowsky (1992) and rehybridized as above using a 32P- and B-ATF-4 cells were plated in growth medium at a density labeled human junB cDNA fragment obtained from the of 16104 cells per 60 mm culture dish on day 0. The growth 32 SmaI/XbaI digestion of pBS-junB and a P-labeled G3PDH medium was changed on day 1, day 4 and day 7. Cell counts DNA probe (Clonetech). for duplicate plates for each cell line were determined daily starting on day 2. Cell culture and transfection C3H10T1/2 mouse ®broblasts (American Type Tissue Electrophoretic mobility shift assays (EMSA) Culture Collection, CCL226, Rockville, MD, USA) were maintained in basal modi®ed Eagle medium (BME) (GIBCO/ Proteins used for EMSA were generated by in vitro BRL, Grand Island, NY, USA) supplemented with 10% fetal transcription and translation. The human junB,c-fos, and bovine serum (Bio-Whittaker, Walkersville, MD, USA), B-ATF cDNAs in pBS-KS+ and the human c-jun cDNA in penicillin (100 U/ml) and streptomycin (100 mg/ml). For the pEMSVscribea-2 were linearized and 5 mg of each DNA used focus assays, the BME was supplemented to 5% with fetal with the MAXIscript in vitro Transcription Kit (Ambion, bovine serum; for the colony assays, 400 mg/ml G-418 Austin, TX, USA) to produce m7GpppG-capped RNA. (GIBCO/BRL) was added to complete growth medium. The Proteins were produced by in vitro translation in rabbit ras neo 11A (Teegarden et al., 1990) and B-ATF-4 cell lines reticulocyte lysates (Promega, Madison, WI, USA) and were derived from C3H10T1/2 cells and maintained in protein quanti®ed in parallel reactions supplemented with growth medium supplemented with 400 mg/ml G-418. F9 L-[35S]methionine (4800 Ci/mmol; Amersham Pharmacia teratocarcinoma cells (American Type Tissue Culture Collec- Biotech, Piscataway, NJ, USA). Using sequence information tion, CRL1720) were maintained in Dulbecco's high glucose provided by a number of commercial sources, double- modi®ed Eagle medium (HG-DMEM) (GIBCO/BRL) con- stranded DNA containing consensus AP-1 (5'- taining 10% fetal bovine serum, penicillin (100 U/ml) and AGCTTCGCTTGATGAGTCAGCCG-3'), mutant AP-1 streptomycin (100 mg/ml). (Figure 5) or CRE (5'-GATCCAGAGATTGCCTGACGT- Transient cell transfections were carried out using calcium CAGAGAGCTA-3') motifs were synthesized and end-labeled phosphate-DNA precipitation as described previously (Min et using T4 polynucleotide kinase and [g-32P]dATP (6000 Ci/ al., 1992) with the amounts of plasmid DNA indicated in the mmol; Amersham Pharmacia Biotech). All DNA probes were Figure legends. DNA precipitates were incubated with cells separated from unincorporated nucleotides using the spin for 5 h, after which the cells were shocked osmotically with column procedure (Sambrook et al., 1989). Approximately 20% glycerol in unsupplemented BME or HG-DMEM for 26105 c.p.m. of each DNA probe was incubated at RT for 2 min and then maintained in complete growth medium for 20 min with 4 ml total volume of each in vitro translated an additional 48 h. Cell extracts were prepared, normalized protein in a reaction mixture containing 25 mM HEPES (pH for b-galactosidase activity, and assayed for chloramphenicol 7.9), 50 mM potassium chloride, 5 mM dithiothreitol, 5 mM acetyl transferase (CAT) activity or for luciferase activity as magnesium chloride, 0.5 mM EDTA and 10% glycerol. For previously described (Min and Taparowsky 1992; Ramocki et competition EMSA, unlabeled AP-1, mutant AP-1, CRE or al., 1997). double-stranded MRF DNA (5'-GATCCGTCTGAGGAGA- C3H10T1/2 colony assays were performed as described CAGCTGCAGCTCCA-3') was added prior to the addition previously (Min et al., 1994). Brie¯y, C3H10T1/2 cells were of the labeled probe. Binding reactions were electrophoresed seeded 24 h prior to transfection at 56105 C3H10T1/2 cells through a 4% nondenaturing polyacrylamide gel at 100 V for per 100 mm culture dish and treated 2 h prior to transfection 3 h in 12.5 mM Tris-HCl (pH 8.5), 95 mM glycine and with complete growth medium containing 10 mg/ml chlor- 0.5 mM EDTA. The gels were dried and exposed to X-ray oquine (Sigma, St. Louis, MO, USA). Individual calcium ®lm overnight at 7808C with an intensifying screen.

Oncogene B-ATF functions as a dominant-negative to AP-1 DR Echlin et al 1763 Acknowledgments was supported by a grant awarded to EJ Taparowsky from The authors thank all of the members of the EJ the National Cancer Institute (CA-78264) and a predoctor- Taparowsky and S Konieczny laboratories at Purdue for al fellowship awarded to DR Echlin from the US Army their assistance during the course of this study. This work Breast Cancer Research Program.

References

Angel P and Karin M. (1991). Biochim. Biophys. Acta, 1072, Kerppola TK and Curran T. (1994). Oncogene, 9, 675 ± 684. 129 ± 157. Lemercier C, To RQ, Swanson BJ, Lyons GE and Konieczny Aronheim A, Zandi E, Hennemann HS, Elledge SJ and SF. (1997). Dev. Biol., 182, 101 ± 113. Karin M. (1997). Mol. Cell. Biol., 17, 3094 ± 3102. Lloyd A, Yancheva N and Wasylyk B. (1991). Nature, 352, Bohmann D, Bos TJ, Admon A, Nishimura T, Vogt PK and 635 ± 638. Tjian R. (1987). Science, 238, 1386 ± 1392. Meyer NP, Johansen LM, Tae H-J, Budde PP, Williams KL Bossy-Wetzel E, Bakiri L and Yaniv M. (1997). EMBO J., and Taparowsky EJ. (1998). Mammal. Genome, 9, 849 ± 16, 1695 ± 1709. 852. Brown PH, Chen TK and Birrer MJ. (1994). Oncogene, 9, Miller AD, Verma IM and Curran T. (1985). J. Virol., 55, 791 ± 799. 521 ± 526. Brown PH, Alani R, Preis LH, Szabo E and Birrer MJ. Min S, Crider-Miller SJ and Taparowsky EJ. (1994). Cell (1993). Oncogene, 8, 877 ± 886. Growth Di€er., 5, 563 ± 573. CarrozzaML,JacobsH,ActonD,VermaIandBernsA. Min S and Taparowsky EJ. (1992). Oncogene, 7, 1531 ± 1540. (1997). Oncogene, 14, 1083 ± 1091. Nakabeppu Y and Nathans D. (1991). Cell, 64, 751 ± 759. Chen BPC, Liang G, Whelan J and Hai T. (1994). J. Biol. Olive M, Krylov D, Echlin DR, Gardner K, Taparowsky EJ Chem., 269, 15819 ± 15826. and Vinson C. (1997). J. Biol. Chem., 272, 18586 ± 18594. Colotta F, Polentarutti N, Sironi M and Mantovani A. Pfarr CM, Mechta F, Spyrou G, Lallemand D, Carillo S and (1992). J. Biol. Chem., 267, 18278 ± 18283. Yaniv M. (1994). Cell, 76, 747 ± 760. Dang CV, Barrett J, Villa-Garcia M, Resar MS, Kato GJ Ramocki MB, Johnson SE, White MA, Ashendel CL, and Fearon ER. (1991). Mol. Cell. Biol., 11, 954 ± 962. Konieczny SF and Taparowsky EJ. (1997). Mol. Cell. Davenport EA and Taparowsky EJ. (1992). Exp. Cell Res., Biol., 17, 3547 ± 3555. 202, 532 ± 540. Sambrook J, Fritsch EF and Mainiatis T. (1989). Molecular Davis RL, Weintraub H and Lassar AB. (1987). Cell, 51, Cloning: A Laboratory Manual. Cold Spring Harbor 987 ± 1000. Laboratory Press: Cold Spring Harbor, NY. Dorsey MJ, Tae H-J, Sollenberger KG, Mascarenhas NT, Sassone-Corsi P. (1995). Ann. Rev. Cell Dev. Biol., 11, 355 ± Johansen LM and Taparowsky EJ. (1995). Oncogene, 11, 377. 2255 ± 2265. Schreiber M, Kolbus A, Piu F, Szabowski A, MoÈ hle- Foletta VC, Segal DH and Cohen DR. (1998). J. Leuk. Biol., Steinlein U, Tian J, Karin M, Angel P and Wagner EF. 63, 139 ± 152. (1999). Genes Dev., 13, 607 ± 619. Fujiwara KT, Kataoka K and Nishizawa M. (1993). Smeal T, Binetruy B, Mercola DA, Birrer M and Karin M. Oncogene, 8, 2371 ± 2380. (1991). Nature, 354, 494 ± 496. Glover JNM and Harrison SC. (1995). Nature, 373, 257 ± SmeyneRJ,VendrellM,HaywardM,BakerSJ,MiaoGG, 261. Schilling LM, Robertson LM, Curran T and Morgan J. Grigoriadis AE, Schellander K, Wang Z-Q and Wagner EF. (1993). Nature, 363, 166 ± 169. (1993). J. Cell Biol., 122, 685 ± 701. Taparowsky EJ, Suard Y, Fasano O, Shimizu K, Goldfarb Hai T and Curran T. (1991). Proc. Natl. Acad. Sci. USA, 88, M and Wigler M. (1982). Nature, 300, 762 ± 764. 3720 ± 3724. Teegarden D, Taparowsky EJ and Kent C. (1990). J. Biol. Ham J, Babij C, Whit®eld J, Pfarr CM, Lallemand D, Yaniv Chem., 265, 6042 ± 6047. M and Rubin LL. (1995). Neuron, 14, 927 ± 939. Vandel L, Montreau N, Vial E, Pfarr CM, Binetruy B and Hasegawa H, Utsunomiya Y, Kishimoto K, Tange Y, Castellazzi M. (1996). Mol. Cell. Biol., 16, 1881 ± 1888. Yasukawa M and Fujita S. (1996). Biochem. Biophys. Vogt PK and Bos TJ. (1990). Adv. Cancer Res., 55, 1 ± 35. Res. Comm., 222, 164 ± 170. Wang C-Y, Bassuk AG, Boise LH, Thompson CB, Bravo R Hsu J-C, Laz T, Mohn KL and Taub R. (1991). Proc. Natl. and Leiden JM. (1994). Mol. Cell. Biol., 14, 1153 ± 1159. Acad. Sci. USA, 88, 3511 ± 3515. Wang H, Zirong X and Scott RE. (1996). J. Cell Biol., 135, Johnson R, Spiegelman B, Hanahan D and Wisdom R. 1151 ± 1162. (1996). Mol. Cell. Biol., 16, 4504 ± 4511. Wang X, Peters MA, Utama FE, Wang Y and Taparowsky Johnson RS, van Lingen B, Papaioannou VE and Spiegel- EJ. (1999). Mol. Endocrinol., 13, 254 ± 267. man BM. (1993). Genes Dev., 7, 1309 ± 1317. Wick M, Lucibello FC and MuÈ ller R. (1992). Oncogene, 7, KarinM,LiuZ-GandZandiE.(1997).Curr. Opin. Cell 859 ± 867. Biol., 9, 240 ± 246. Wisdom R, Johnson RS and Moore C. (1999). EMBO J., 18, Kataoka K, Igarashi K, Itoh K, Fujiwara KT, Noda M, 188 ± 197. Yamamoto M and Nishizawa M. (1995). Mol. Cell. Biol., Xanthoudakis S and Curan T. (1996). Biological Reactive 15, 2180 ± 2190. Intermediates V. Snyder R et al. (eds), Plenum Press: New Kataoka K, Noda M and Nishizawa M. (1994). Mol. Cell. York. Biol., 14, 700 ± 712. Kerppola TK and Curran T. (1991). Curr. Opin. Struct. Biol., 1, 71 ± 79.

Oncogene