Jun, the oncoprotein

NATURE.COM

NATURE NEWS

NATUREJOBS

NATUREEVENTS

ABOUT NPG

Help

Nature.com site index

my account

e-alerts

Journal home

Advance online publication

Contact editorial office

NPG Subject areas

Access material from all our publications in your subject area:

Molecular Cell Biology

Neuroscience

Pharmacology

Physics

30 April 2001, Volume 20, Number 19, Pages 2365-2377

Table of contents Previous Article Next [PDF]

Jun, the oncoprotein

Peter K Vogt

Department of Molecular and Experimental Medicine, The Scripps Reasearch Institute, 10550 North Torrey Pines Drive, La Jolla, California, CA, 9203, USA

Correspondence to: PK Vogt, Department of Molecular and Experimental Medicine, The Scripps Reasearch Institute, 10550 North Torrey Pines Drive, La Jolla, California, CA, 9203, USA

Abstract

Cellular Jun (c-Jun) and viral Jun (v-Jun) can induce oncogenic transformation. For this activity, c-Jun requires an upstream signal, delivered by the Jun N-terminal kinase (JNK). v-Jun does not interact with JNK; it is autonomous and constitutively active. v-Jun and c-Jun address overlapping but not identical sets of genes. Whether all genes essential for transformation reside within the overlap of the v-Jun and c-Jun target spectra remains to be determined. The search for transformation-relevant targets of Jun is moving into a new stage with the application of DNA microarrays technology. Genetic screens and functional tests remain a necessity for the identification of genes that control the oncogenic phenotype. Oncogene (2001) 20, 2365-2377.

Keywords

oncogenic transformation; downstream target; dimerization; leucine zipper; transactivation

Defining AP-1

The AP 1 transcription factor complex owes its initial definition to a fortuitous convergence of tumor virology and transcription research. AP-1 had been described as a 12-O-tetradecanoyl-phorbol-13-acetate (TPA) inducible transcription factor activity that addresses specific sequences in the enhancer of the metallothionein gene and in the 72-base pair repeat of the simian virus 40 enhancer region (Angel et al., 1987; Lee et al., 1987a,b). Jun was found as a novel oncoprotein encoded by a cellular insert in the genome of avian sarcoma virus 17 (ASV17), an acutely oncogenic retrovirus isolated from a spontaneous tumor in a chicken (Cavalieri et al., 1985; Maki et al., 1987). The two findings became linked by the discovery of homology between Jun and the yeast transcriptional regulator GCN4 (Vogt et al., 1987). GCN4 was known to bind to a sequence identical to that targeted by AP-1. Indeed, the DNA binding domains of Jun and of GCN4 were found to be functionally exchangeable (Struhl, 1987). Jun was thus quickly identified as a major component of AP-1 (Angel et al., 1988a; Bohmann et al., 1987; Bos et al., 1988). A second important step in the definition of AP-1 came with the discovery that the Fos-associated protein p39 is the product of the cellular jun gene (Rauscher et al., 1988; Sassone-Corsi et al., 1988a). This finding identified cellular Fos as a partner of Jun and as another component of the AP-1 complex. These initial discoveries catalyzed a wealth of inquiries into the structure, function, and regulation of Jun and the relationship of transcription to oncogenesis. Cloning of the cellular Jun gene showed it to be without introns and with an atypical TATA box (Hattori et al., 1988; Nishimura and Vogt, 1988). The human jun gene is located on chromosome 1 at region p31-32 (Haluska et al., 1988). Murine jun is on chromosome 4 subregion C5-C7 (Mattei et al., 1990). The viral Jun differs from its chicken cellular counterpart by a 27-amino acid deletion near the N-terminus (the delta deletion) and two or three (depending on the clone) amino acid substitutions in the C-terminal half (Nishimura and Vogt, 1988).

bZip proteins

The Jun protein can form homodimers, though heterodimers with Fos are more stable and have a higher affinity for the DNA target sequence (Allegretto et al., 1990; Halazonetis et al., 1988; Nakabeppu et al., 1988; Smeal et al., 1989). Dimerization requires the C-terminal region of Jun (Bos et al., 1989; Turner and Tjian, 1989). This domain contains five heptad repeats of leucines forming an amphipathic helix that upon dimerization takes up a coiled oil conformation with the leucines aligned along the contact surface (Rasmussen et al., 1991). This dimerization domain is referred to as a leucine zipper (Landschulz et al., 1988). Coiled coil-zipper structures can now be reliably predicted from sequence (Bornberg-Bauer et al., 1998; Hirst et al., 1996). A basic region, serving as DNA contact surface, is located immediately N-terminal to the leucine zipper. This characteristic arrangement of DNA binding and dimerization domains is found in numerous transcriptional regulators and has given them the family name 'bZIP proteins'. About 40 different bZIP proteins have already been identified in the human genome. The structure of the bZIP domain of the Jun/Fos heterodimer has been determined by X-ray crystallography (Glover and Harrison, 1995). A given leucine zipper pairs with a limited set of specific leucine zipper partners; the specificity is dictated by the non-leucine residues of the zipper (Alber, 1992). Thus, Jun can form homodimers and heterodimers with various bZIP proteins; Fos forms only heterodimers, and GCN4 only homodimers (Halazonetis et al., 1988; Kouzarides and Ziff, 1989; Ransone et al., 1989a,b; Sassone-Corsi et al., 1988b; Turner and Tjian, 1989).

Dimerization is a prerequisite for DNA binding (Halazonetis et al., 1988; Smeal et al., 1989). Dimerization of Jun and Fos also enhances their nuclear translocation (Chida et al., 1999). The consensus DNA binding sequence for AP-1 dimers is TGACTCA, it is referred to as TPA-response element, TRE (Lee et al., 1987a). The transactivation domain of Jun lacks defining structural features. It has been delineated by deletion analysis which discriminates between several domains contributing to transactivation (Angel et al., 1989). The N-terminal half of Jun contains the major transactivation domain. The dimerization, DNA binding, and transactivation domains of Jun are modular, exchangeable for functionally equivalent domains from other proteins. Such 'cut and paste' operations within the bZIP family usually preserve the basic activity of the protein. Even a fusion of the dimerization and DNA binding domains of Jun with the transactivation domain of the herpes simplex virus protein VP16 yields an active AP-1 protein (Schuur et al., 1993). The principal dimerization partners of Jun within the AP-1 complex are Fos and Fos-related antigens Fras (Cohen et al., 1989; Nishina et al., 1990). However, Jun can also form dimers with bZIP proteins from outside the AP-1 complex. These dimerization partners include some members of the CREB/ATF family of proteins and the oncogenic transcription factor Maf. The consensus sequence for CREB/ATF proteins is TGACGTCA, referred to as CRE. It bears close resemblance to the AP-1 target sequence. Heterodimers of Jun and certain ATF proteins are able to recognize the CRE sequence, whereas heterodimers of Jun and Maf bind equally well to the AP-1 and the CRE consensus. (Chatton et al., 1993; Hai and Curran, 1991; Kataoka et al., 1994; Nishizawa 1989). Thus dimerization with diverse partners can function as a modulator of target gene specificity. Importantly, v-Jun and c-Jun also differ in their target preferences. The spectra of genes regulated by the viral and the cellular form of the protein overlap but each of these two regulators also addresses specific genes not recognized by the other (Hadman et al., 1993). Such target genes that are differentially regulated by v-Jun and c-Jun have recently been identified (Fu et al., 2000). Jun has two close relatives in animal genomes, JunB and JunD (Hartl et al., 1991; Hirai et al., 1989; Ryder et al., 1989, 1988). The functions of JunB and JunD are less well understood than those of Jun but both proteins share with Jun the potential to induce oncogenic transformation (Castellazzi et al., 1991; Hartl and Vogt, 1992b; Kameda et al., 1993; Vandel et al., 1995; 1996).

Regulation of Jun activity

The DNA binding and transcriptional activities of Jun are regulated by several mechanisms. Historically the first was the induction of AP-1 activity by TPA tumor promoters which reflects a post-translational activating modification, stimulating the Jun protein to engage in an autoregulatory loop: the Jun enhancer contains two AP-1 binding sites, one of these has been reported to mediate positive autoregulation of the gene by Jun (Angel et al., 1988b; Lamph et al., 1988). Jun is the product of a paradigmatic immediate-early gene, highly responsive to external growth signals. But its regulation is also integrated into the cell cycle. Jun undergoes activating phosphorylation at the M-G1 transition. Since Jun can transactivate the promoter of cyclin D1, elevated Jun activity may be important in advancing the cell through G1 (Bakiri et al., 2000; Lallemand et al., 1997; Ryseck et al., 1988). The DNA binding activity of Jun is subject to redox regulation involving the DNA repair protein APE/REF-1 that targets Cys252 in the DNA binding domain of Jun (Abate et al., 1990; Frame et al., 1991; Fritz and Kaina, 1999b; Ng et al., 1993; Xanthoudakis et al., 1992). The mutations of Cys252 and of Ser226 found in v-Jun release the mutated Jun from redox control (Oehler et al., 1993) The Cys252 mutation also makes nuclear translocation of the mutated protein cell cycle dependent, controlled by phosphorylation of Ser252 (Chida and Vogt, 1992; Tagawa et al., 1995). The above-mentioned Ser226 reduces DNA binding when phosphorylated (Boyle et al., 1991; Oehler et al., 1993). However, the most important regulatory phosphorylation sites of Jun are Ser63 and Ser73 near the N terminus. The phosphorylation of these amino acids by JNK, a member of the stress-activated kinases, stimulates transcriptional activation by Jun (Derijard et al., 1994). JNK controls cellular but not viral Jun; this crucial difference between the two Juns will be discussed in greater detail below. Recently Jun has also been identified as a substrate of the Abl tyrosine protein kinase. The interaction is part of an interesting regulatory circuit in which phosphorylation of Jun by Abl leads to an activation of JNK (Barila et al., 2000).

Modulated and modulating Jun

Jun interacts with numerous other transcription factors, and this transcriptional crosstalk modulates the activities of Jun and its partners (Gottlicher et al., 1988). The following are some examples: The glucocorticoid receptor interferes with Jun activity, but under certain conditions can also exert an enhancing effect (Diamond et al., 1990; Schule et al., 1990; Teurich and Angel, 1995; Yang-Yen et al., 1990). The mechanism of this interaction is complex; it may involve the regulation of JNK (Gonzalez et al., 2000). The inhibition of Jun induced by the glucocorticoid receptor affects oncogenic transformation: it interferes with Jun induced transormation and secondarily with transformation by oncoproteins that depend on Jun, e.g. Src and Ras. However, inhibition by the glucocorticoid receptor does not affect oncoproteins that are independent of Jun action, e.g. Myc (Kameda and Iba, 1998). A potentially highly significant interaction takes place between Jun and various SMADs, components of the transforming growth factor beta signaling pathway. This interaction enhances transcription from AP-1 binding sites and interferes with transactivation from SMAD binding sites (Dennler et al., 2000; Liberati et al., 1999; Qing et al., 2000; Wong et al., 1999).

Transforming growth factor beta can also enhance c-Jun expression (Pertovaara et al., 1989). Expression of the transcriptional regulator called nuclear factor 1 (NF-1) strongly interferes with Jun-induced transformation in chicken embryo fibroblasts (CEF); it also inhibits transformation by other nuclear oncoproteins (Schuur et al., 1995). NF-1 does not directly interact with Jun. The mechanism of NF-1-induced inhibition is probably circuitous; NF-1 greatly increases the adherence of cells and may lower their susceptibility to transformation. A particularly interesting example of crosstalk is the interaction of Jun and Stat3 that involves both physical association of the proteins and binding to closely spaced sites on specific promoters. This interaction is cooperative; its possible effect on oncogenic transformation remains to be explored (Bromberg and Darnell, 2000; Schaefer et al., 1995; Zhang et al., 1999). Important interactions also take place between Jun and the transcription factors of the NFAT (nuclear factor of activated T-cells) family. NFAT proteins are critical in the regulation of cytokine and other immune response genes. Many of these genes contain composite NFAT-AP-1 binding sites in their regulatory regions to which Jun/Fos dimers and NFAT bind cooperatively (Rao, 1994; Rao et al., 1997). Dominant negative Jun inhibits NFAT transcriptional activation and interferes with the control of IL-2 expression (Petrak et al., 1994). Other modulating interactions with Jun include binding to human papilloma virus type 16 protein E7 (Antinore et al., 1996). This association enhances AP-1 activity and may be part of the mechanism by which human papilloma viruses transform cells. As part of the UV response, Jun interacts with p53, stimulating the activation of the mismatch repair enzyme MSH2 (Scherer et al., 2000). Jun also plays a part in the oncogenic transformation induced by the v-Rel oncoprotein (Kralova et al., 1998). Rel belongs to the NF- kappa B family of transcriptional regulators, it is the transforming protein expressed by the avian reticuloendotheliosis retrovirus. c-Jun is upregulated in Rel-transformed cells, and dominant negative Jun strongly interferes with transformation of avian lymphoid cells and of chicken embryo fibroblasts (CEF) by Rel. These observations suggest that Jun is an essential downstream effector for the oncogenicity of Rel. Negative modulation of Jun also occurs through binding to bZIP proteins, e.g. the Jun dimerizing protein JDP2 (Aronheim et al., 1997 ).

The effects of another Jun-binding protein, JIF-1, on Jun function are not clear (Imafuku et al., 1999; Monteclaro and Vogt, 1993). JIF-1 is a homolog of the highly conserved ribosomal protein QM and is expressed primarily in the cytoplasm (Loftus et al., 1997; Nika et al., 1997). The interaction of Jun/Fos dimers with promoter sequences is strongly influenced by the SWI/SNF chromatin remodeling complex (Ito et al., 2000). This observation may clarify aspects of specific promoter targeting by AP-1 that are still poorly understood.

Transformations

In discussing the oncogenicity of Jun, it is useful to make the distinction between transformations in which Jun is the only exogenous effector ('Solo-Transformations') and situations in which cooperation with another exogenous oncogenic effector is required ('Co-Transformations').

Solo-transformations

The avian retrovirus ASV17, expressing v-Jun fused to viral Gag sequences induces characteristic foci of transformed cells in cultures of primary CEF (Cavalieri et al., 1985; Maki et al., 1987). These foci consist of needle like cells growing in multiple layers and parallel arrays. Avian retroviral vectors expressing v-Jun without the Gag sequences are equally oncogenic (Bos et al., 1986, 1988). The transformed cells are capable of anchorage-independent growth.

ASV17 and avian retroviral vectors expressing v-Jun are highly tumorigenic in young chickens, inducing fibrosarcomas within a latent period of less than 2 weeks. Interestingly, c-Jun also transforms CEF as well as embryo fibroblasts of coturnyx quail, inducing foci and agar colony formation (Hartl et al., 1995; Castellazzi et al., 1990; Wong et al., 1992). Continuous viral passage of c-Jun-expressing vectors increases transforming activity, possibly due to an accumulation of mutations in the transduced cellular gene. CEF can also be tansformed by human, murine, and coturnyx quail c-Jun (Castellazzi et al., 1990, 1991; Metivier et al., 1993 Suzuki et al., 1991). However, the avian c-Jun is not tumorigenic in young chickens. The rare tumors that do arise after prolonged latent periods contain mutated and rearranged forms of the gene (Gao, 1996). v-Jun also transforms the continuous chicken embryo fibroblast line DF-1 (Himly et al., 1998). In this system, vectors expressing v-Jun without Gag are more effective than ASV17. In mammalian host systems, c-Jun, acting as a single oncoprotein, is able to transform the continuous line of rat fibroblasts known as Rat1a (Schutte et al., 1989). NIH3T3 cells can be transformed by c-Jun in conjunction with Fra-1 (Mechta et al., 1997). Formally, this is a co-transformation but Jun and Fra-1 probably form AP-1 dimers which qualify as a single oncogenic effector. Overexpression of c-Jun greatly enhances the tumorigenic properties of the MCF7 human mammary carcinoma cell line, inducing elevated motility, unresponsiveness to estrogen and tamoxifen, and increased tumor formation in nude mice (Doucas et al., 1991; Smith et al., 1999). v-Jun can transform the mouse fibroblast cell line C3H10T1/2 (Cohen et al., 2001). Repeated infections with a Jun-expressing retroviral vector are needed in this transformation; the system may depend on the integration of multiple proviruses into the same cell to achieve the requisite levels of v-Jun expression.

Co-transformations

The best-studied system of co-transformation uses primary rat embryo fibroblasts that are transfected with expression vectors of c-Jun and of mutated, constitutively active Ras (Schutte et al., 1989). Neither one of these oncoproteins alone induces transformation of primary fibroblast cultures, but in combination they are highly effective. In the oncogenic signal that is generated by the Ras-Jun combination, Ras appears to be located upstream of c-Jun, inducing JNK and constitutively activating Jun by phosphorylation of Ser63 and Ser73 (Behrens et al., 1999, 2000). A second system of co-transformation involving Jun targets avian hematopoietic cells. v-Jun alone fails to affect any of the hematopoietic lineages in chicken bone-marrow cultures. However, co-transfection with a v-ErbB expression vector leads to transformation of megakaryocytes (Garcia and Samarut, 1993). An enhancement of v-Jun transformation by v-ErbB is also seen in CEF cultures (Garcia and Samarut, 1990). V-ErbB is a constitutively active EGF receptor. One might therefore presume that it signals through the Ras-JNK pathway and that this signal is basically similar to the Ras-Jun co-transformation of primary rat fibroblasts. However, the partner in this co-transformation is not c-Jun but v-Jun, which cannot be phosphorylated by JNK (May et al., 1998a,b). The data suggests that v-Jun can respond to some upstream signal other than JNK, but the exact mechanism of the v-ErbB-v-Jun synergy remains to be clarified.

The Jun phenotype

The effectiveness of Jun as a transcriptional activator varies widely in different cell types, and this fact may in part explain the variations in oncogenicity (Imler et al., 1988). In avian fibroblast cultures, Jun can induce oncogenic transformation acting as a single oncoprotein; in cultures of mammalian cells, Jun usually, but not always, requires the cooperation of additional growth-stimulating factors. Formation of Jun tumors in young chickens is promoted by wounding (Marshall et al., 1992). Wounding is also a key requirement for the induction of sarcomas in jun-transgenic mice (Schuh et al., 1990). In mouse tumors induced by chemical carcinogens, the degree of malignancy is correlated with AP-1 activity, which in turn reflects the levels of phosphorylated c-Jun, ATf 2, and JunB (Zoumpourlis et al., 2000). Besides the standard indicators of oncogenicity, such as multilayered growth on solid substrates, growth in low serum concentrations, anchorage-independence, and tumorigenicity, Jun also induces more subtle changes. Jun-expressing cells show increased motility and invasiveness (Bos et al., 1999; Smith et al., 1999). In mammary epithelial cells, Jun induces the loss of polarity accompanied by a dissociation of beta-catenin from E-cadherin, redistribution of beta-catenin and a disruption of intercellular junctions (Fialka et al., 1996). Although it is clear that the growth promoting functions of Jun can be stimulated by a variety of exogenous and endogenous factors, it is not known whether these factors merely cause a gain of Jun function or whether they provide complementary and independent activities that are qualitatively distinct from that of Jun. In an illuminating set of experiments, different manifestations of the Jun phenotype have been dissected and correlated with preferential binding of Jun to specific bZIP proteins (van Dam et al., 1998). Mutated Jun that selectively dimerizes with members of the Fos family induces anchorage- but not growth factor-independence, while a mutated Jun that preferentially associates with ATF2 causes growth factor-independence but fails to support colony formation in nutrient agar. Combining the two mutants in the same cell leads to a fully transformed phenotype, indistinguishable from that induced by wild type Jun. These data probably reflect the different spectra of target genes and hence of genetic programs addressed by Jun-Fos and Jun-ATF2 dimers.

Constructs that allow regulation of Jun activity facilitate the analysis of the Jun phenotype. Fusions of Jun with the hormone-binding domain of the estrogen receptor can be tightly controlled by estrogen or tamoxifen (Fialka et al., 1996; Bader et al., 2000; Kruse et al., 1997). In one such construct, the transactivation domain of Jun is deleted, and that function is then provided by the transactivation domain present in the estrogen receptor portion of the molecule. Proteolysis of this Jun-estrogen receptor fusion generates only inactive fragments, preventing 'leakiness' of the construct (Kruse et al., 1997). Although in the presence of estrogen this construct transactivates a reporter carrying the AP-1 consensus sequence only weakly, it induces transformation in CEF cultures as efficiently as wild-type Jun. A similar mutant functions as a dominant negative in mammalian cells, inhibiting transcriptional activation from AP-1 sites in the presence of hormone and failing to co-transform with activated Ras in the presence or absence of ligand (Kim et al., 1996). In a second conditional system, based on continuous fibroblast lines derived from coturnix quail, Jun is regulated by a doxycycline-sensitive transactivator or a doxycycline-dependent reverse transactivator (Bader et al., 2000). In both the doxycycline- and the estrogen-regulated system, Jun is turned on within 2 h after the switch, yet phenotypic changes first appear after 24 h, and complete transformation may take as much as 72 h (Bader et al., 2000).

Similar to other highly transforming oncoproteins, like Src or Myc, Jun is a strong inhibitor of cellular differentiation. Myoblasts, prepared from the pectoral muscle of chicken embryos, can be induced to fuse into multinucleated, postmitotic myotubes and to differentiate into muscle tissue that shows spontaneous contraction in cell culture. Expression of v-Jun in such myoblasts keeps the cells cycling, interferes with cell fusion and prevents synthesis of muscle-specific proteins (Grossi et al., 1991; Su et al., 1991). Rare instances of fused myoblasts in Jun-transfected cultures are devoid of nuclear Jun. Jun expression and myogenesis appear to be mutually exclusive. Jun may be an antagonist of the helix-loop-helix transcription factor MyoD which plays a determining role in muscle differentiation. The two proteins have been reported to interact, and Jun inhibits transactivation of the MyoD promoter and transcription of genes controlled by MyoD (Bengal et al., 1992). More work on this putative antagonism would be required to explain the interference of Jun with myogenic differentiation. In avian chondrocytes, c-Jun is expressed during the early, immature phases of development and is downregulated when the cells mature and express differentiation-specific markers, such as alkaline phosphatase. Overexpression of c-Jun in this cell system significantly retards differentiation and blocks the retinoic acid induced expression of alkaline phosphatase (Kameda et al., 1997). Jun is also active in lens differentiation. Overexpression of Jun in lens epithelial cells retards the formation of lentoid bodies and reduces levels of beta A3/A1 crystallin mRNA. A transdominant negative Jun enhances differentiation and alpha A crystallin expression (Rinaudo et al., 1995).

Although Jun is known primarily for its growth-promoting activity, it also has the capacity to induce apoptosis, a property which Jun shares with other oncoproteins like Myc, E1A, or E2F. In NIH3T3 cells, overexpression of Jun triggers programmed cell death (Bossy-Wetzel et al., 1997). This occurs in the presence or absence of growth factors; it requires the bZIP and transactivation domains of Jun, suggesting apoptosis depends on transcriptional regulation. Bcl-2 or inhibitors of ICE/CED-3-type cysteine proteinases interfere with Jun-induced apoptosis. In CEF, v-Jun induces apoptosis only in the absence of growth factors (Clark and Gillespie, 1997). Normal CEF retain viability for some time without serum by staying in G0, whereas v-Jun-transfected CEF continue to cycle, expressing phosphorylated retinoblastoma protein, cyclin A, and CDK2 and undergo apoptosis. In cerebellar granule neurons, withdrawal of survival factors also induces apoptosis which is accompanied by higher levels of expressed Jun (Watson et al., 1998). A dominant negative mutant of Jun and a Jun mutant in which the JNK phosphorylation sites, serines 63 and 73, are mutated to alanines prevent this apoptosis. Overexpression of c-Jun also induces apoptosis in endothelial cells (Wang et al., 1999). The process is preventable by a dominant negative Jun mutant.

Conditions of oncogenicity

Mutational analysis

There is a hierarchical relationship between the three basic functions of Jun, dimerization, DNA binding, and transactivation. DNA binding depends on dimerization, and at the same time it is a prerequisite for transactivation. Deletion analysis of Jun in solo- and co-transformations has established a need for the bZIP and transactivation domains, leading to the conclusion that Jun transforms by interfering with transcriptional regulation and requiring dimerization, DNA binding, and transactivation (Alani et al., 1991; Morgan et al., 1992). Solo-transformation of avian cells in culture by v-Jun is additionally enhanced by point mutations. These replace cysteine 252, the target of redox regulation and serine 226, a site for negative regulatory phosphorylation (Morgan et al., 1993, 1994). The 27-amino acid deletion in the N-terminal region of v-Jun releases the protein from control by JNK (May et al., 1998b). These mutations in v-Jun act additively to confer tumorigenicity in the animal.

Proteins that must dimerize in order to function can be readily modified to act as dominant negatives. Mutations in the Jun DNA binding domain and deletions of all or part of the transactivation domain both serve this purpose (Brown et al., 1993; Ransone et al., 1990a). The latter type of mutant is the more potent inhibitor of Jun and has been used extensively to analyse Jun function (Brown et al., 1996; Petrak et al., 1994; Rinaudo et al., 1995; Watson et al., 1998; Young et al., 1999). Dominant negative Jun interferes with Jun-dependent transactivtion and oncogenic transformation (Brown et al., 1993; Domann et al., 1994). Since Jun functions at the nuclear end of certain mitotic signals, dominant negative Jun can interfere with upstream elements of such signals which include several oncoproteins: receptor tyrosine kinases, cytoplasmic tyrosine kinases, Ras, and Raf. Transformation by these oncoproteins is blocked by dominant negative Jun, and overexpression of the dominant negative mutant in cells already transformed by these oncoproteins forces reversion to the normal phenotype (Rapp et al., 1994; Suzuki et al., 1994). These data show that AP-1 activity is essential for transformation by these oncoproteins. Transformation by Myc was not changed by the expression of dominant negative Jun in CEF but was inhibited in mammalian cells probably because in these cells transformation by Myc is strongly affected by the Ras pathway. Dominant negative Jun also inhibits growth and tumorigenicity of several human ovarian carcinoma cell lines suggesting that Jun or an upstream signaling component that feeds into Jun plays an essential role in this type of cancer (Neyns et al., 1999).

Dimerization: the importance of the partner

Jun homodimerizes, and it also heterodimerizes with various bZIP proteins, raising the question as to whether oncogenic transformation can be accomplished by Jun homodimers or whether it depends on teaming up with another bZIP protein. The answer is provided by Jun constructs in which the native leucine zipper has been exchanged for one that only homodimerizes. The leucine zipper of the yeast transcription factor GCN4 and that of the Epstein-Barr virus transcription factor EB1 have this property. Jun-GCN4 and Jun-EB1 chimeras transform CEF in culture and Jun-EB1 is highly tumorigenic in the animal, indicating that Jun homodimers are fully oncogenic (Hartl and Vogt, 1992a; Hughes et al., 1992; Vandel et al., 1995, 1996). The c-Jun-EB1 construct is also capable of co-transformation of mammalian cells in conjunction with activated Ras. Jun-EB1 homodimers show even increased oncogenicity for avian cells as compared to v-Jun (Jurdic et al., 1995). This gain of function is qualitative as well as quantitative, including a distinct morphology of the Jun-EB1-transformed CEF and expansion of the tumor spectrum in vivo. These data show that the Jun dimerization domain has a strong, unanticipated effect on tumorigenicity. An even more striking illustration of this fact can be seen with Jun mutants that are engineered to dimerize preferentially with ATF2 (Huguier et al., 1998). These mutants show enhanced oncogenicity, producing tumors in chickens much more rapidly than v-Jun. ATF2-Jun heterodimers bind preferentially to the CRE sequence, and v-Jun, but not c-Jun, also shows a predilection for that sequence. One might therefore speculate that addressing target genes that are controlled by CRE-like elements is an important aspect of Jun-induced oncogenic transformation.

DNA binding: the importance of an address

The DNA binding domain of a transcription factor determines the spectrum of genes that are controlled by that protein. The TRE consensus sequence of the AP-1 family is a minimal shared target of the various AP-1 dimers, but not all combinations bind to the TRE with the same affinity (Claret et al., 1996). Experiments with defined dimers suggest that variant target sequences will be seen by some AP-1 combinations but not by others (van Dam et al., 1998). Comparisons between v-Jun and c-Jun reveal differences in target preference that may be important in transformation (Fu et al., 1999, 2000; Gao et al., 1996; M Gao, 1996, unpublished observations). Despite this variation in targeting, there appears to be a common core of genes that need to be addressed to effect oncogenic transformation. Mutations in the Jun DNA binding domain that eliminate interaction with the TRE lead to a complete loss of oncogenicity; mutations that preserve TRE recognition may cause a reduction in oncogenicity (Basso et al., 2000). Although the basic information of DNA recognition is encoded in the structure of the DNA binding domain, it is important to remember that this is a dimeric structure and that different dimerization partners can significantly modulate the specificity of DNA targeting (Ito et al., 2000; van Dam et al., 1998). DNA binding of Jun is also affected by interactions of the transactivation domain with regulatory proteins such as JAB-1 (Claret et al., 1996). Surprisingly, nuclear extracts of CEF transformed by v-Jun do not show increased binding to the TRE target sequence despite the overexpression of v-Jun in the cells (Hawker et al., 1993). However it would be premature to conclude that oncogenic transformation does not involve changes in the levels of DNA binding, because interactions with variant TRE or CRE sequences would have to be considered as well. In contrast to v-Jun in avian cells, co-transformation of mammalian cells by Jun and Ras increases AP-1 DNA binding activity (Mechta et al., 1997).

Transactivation: the importance of delta

The delta domain (amino acids 34-60 of c-Jun) is critical for the regulation of transcriptional activation by c-Jun. The deletion of this domain in v-Jun accounts for distinct and perhaps fundamental differences between the two versions of the protein. Early studies showed that c-Jun needs to be phosphorylated at serines 63 and 73 to become an active transcriptional regulator (Adler et al., 1992a; Mechta et al., 1997; Smeal et al., 1991, 1992). Initial purification of the responsible kinase activity pointed to the family of stress-activated protein kinases (SAPKs) (Adler et al., 1992b; Dai et al., 1995; Hibi et al., 1993; Pulverer et al., 1993). The corresponding kinase gene was cloned and its product termed Jun N-terminal kinase (JNK1) (Derijard et al., 1994). Cloning of a second Jun kinase, JNK2, followed soon after (Kallunki et al., 1994). The Jun kinases use the delta domain as an obligatory docking surface in their interaction with c-Jun (Dai et al., 1995; Kallunki et al., 1996; May et al., 1998a). The minimal essential sequence for docking extends from amino acids 31-42. One of the first reported functions of delta was its engagement of a cell-specific inhibitor of c-Jun (Baichwal et al., 1991; Baichwal and Tjian, 1990; Bohmann and Tjian, 1989). Although this inhibitor has not been identified with certainty, it probably is catalytically inactive JNK which can bind to c-Jun (Dai et al., 1995; Ljungdahl et al., 1997; May et al., 1998a). A polypeptide representing the delta domain inhibits JNK and interferes with Jun N-terminal phosphorylation (Adler et al., 1994).

It is generally assumed that JNK docks on the molecule that it phosphorylates, but there are some data that suggest intermolecular phosphorylation occurring in AP-1 dimers in which at least one of the partners is docking-competent (De Graeve et al., 1999; Kallunki et al., 1996). JNK is itself activated by phosphorylation triggered by a signal that originates in Ras. The signal follows a pathway distinct from that leading to the Erk subgroup of mitogen-activated kinases (Adler et al., 1992b; Minden et al., 1994; Westwick et al., 1994). (For review see Minden and Karin, 1997). Components of that signal and their sequential interactions have not been fully worked out. Two interesting participants in the signal to JNK are the Ras-like GTPase Ral and its guanine nucleotide exchange factor Rlf (de Ruiter et al., 2000). Two JNK kinases, JNKK1 and JNKK2, have been indentified (Lin et al., 1995; Xia et al., 1998). JNKK2 forms a tripartite complex with MEKK2 and JNK that mediates highly efficient activation of JNK (Cheng et al., 2000). Another candidate for a JNK kinase, designated DPK, has recently been described; its place among the MEK kinases remains to be determined (Zhang et al., 2000). One JNK-activating kinase, termed mixed lineage kinase 3, is able to transform NIH3T3 cells (Hartkamp et al., 1999).

Interestingly, Wortmannin, an inhibitor of PI3-kinase, interferes with the UV-induced activation of JNK, possibly by blocking a signaling reaction that lies upstream of Ras (Fritz and Kaina, 1999a). JNK connects c-Jun to incoming signals, responding to mitogens, stress signals, and genotoxic substances (Adler et al., 1995a,b; Derijard et al., 1994; Franklin et al., 1993; Smeal et al., 1991, 1992). JNK-mediated phosphorylation of c-Jun at serines 63 and 73 is required for two principal functions of the protein: transcriptional activation and co-transformation of mammalian cells in conjunction with Ras (Smeal et al., 1991; Xiao and Lang, 2000). N-terminal phosphorylation is also essential for transformation by Fos and for c-Jun-dependent apoptosis (Behrens et al., 1999; Watson et al., 1998). c-Jun with serines 63 and 73 mutated to alanine fails to transactivate or to co-transform, and a JNK1 dominant negative mutant also interferes with c-Jun dependent transformation.

The situation with v-Jun contrasts sharply to that of c-Jun. v-Jun, with its deletion of amino acids 34 to 60 does not interact with JNK, and consequently it is not phosphorylated on serines 63 and 73 by JNK (Adler et al., 1992a; Black et al., 1991; Dai et al., 1995; May et al., 1998a,b). Although dimerization of v-Jun with a partner that is competent for JNK docking could in theory allow for intermolecular phosphorylation by JNK, such cooperative combinations apparently do not occur (De Graeve et al., 1999; Kallunki et al., 1996). Despite this lack of positive regulation, the transactivation domain of v-Jun shows higher inherent activity than that of c-Jun when both are tested as Gal4 fusions (Black et al., 1994; Bohmann and Tjian, 1989). v-Jun is disconnected from regulatory upstream signals of the SAPK/JNK pathway. Substitutions of serines 63 and 73 with alanines do not affect transformation of CEF or transcriptional activation by v-Jun. However, part of this difference between c-Jun and v-Jun in their dependence on JNK is attributable to the cell system. c-Jun transforms avian cells as a single oncoprotein, and that transformation also does not require N-terminal phosphorylation; mutation of the JNK phosphorylation sites does not affect the transforming ability of c-Jun for avian cells (Metivier et al., 1993; Ui et al., 1998).

In Jun-transformed CEF, the expression of the endogenous c-Jun is severely downregulated, and v-Jun replaces c-Jun in the AP-1 dimers (Kilbey et al., 1996). Activation of reporter constructs driven by a TRE is also reduced in v-Jun-transformed CEF. v-Jun interferes with c-Jun-mediated transactivation from such constructs (Hussain et al., 1998). Two mechanisms have been proposed for the downregulation of transcription from the proximal TRE of the c-Jun promoter and of other TRE-dependent promoters: (a) binding of v-Jun containing, non-transactivating heterodimers to consensus TRE sites and (b) binding of dominant negative v-Jun, generated by proteolytic cleavage, to these sites (Gao et al., 1996; Hussain et al., 1998). Reduced transactivation from TRE sites and enhanced inherent transactivation potential by v-Jun are not incompatible, because the latter has been established with Gal4 fusions and the former probably reflects the fact that the preferred binding sequence of v-Jun is closer to the CRE than to the TRE consensus (Gao et al., 1996).

For c-Jun, transactivation potential and oncogenic activity in cooperation with Ras appear to be roughly correlated (Alani et al., 1991; Smeal et al., 1991). But there is a notable exception: the yeast AP-1-related protein, GCN4, transactivates from AP-1 sites in rat cells, yet it does not co-transform with Ras (Oliviero et al., 1992). In avian cells, transactivation of TRE-driven reporters by various v-Jun and c-Jun constructs shows no correlation with transformation; there are potent transformers that transactivate poorly and there are efficient transactivators that transform only marginally (Havarstein et al., 1992). Yet in this system too, deletion analysis shows that the transactivation domain is required for transformation in accord with the hypothesis that transformation results from the transcriptional deregulation of specific genes. The solution to these apparent inconsistencies may lie with the affinity of v-Jun to a variant target sequence. Promoters controlled by this as yet to be determined sequence may show a perfect correlation between transactivation and transforming potential.

Targets

Transcription factors are the ultimate recipients of incoming signals and convert the signals into patterns of gene expression. oncogenic transformation of mammalian cells by c-Jun is driven by a signal from activated Ras; v-Jun, on the other hand, is autonomous and constitutively active. Both induce transformation by corrupting the transcriptional regulation of specific target genes that are then expressed at higher or lower levels in transformed cells as compared to normal progenitor cells. Among these genes that are differentially expressed in Jun-transformed cells, there is a group whose deregulation is essential for inducing and maintaining the neoplastic cellular phenotype, the oncogenic effector genes and a presumably larger group whose differential expression is of no consequence to the growth behavior of the cell, the innocent bystanders. The identification of the oncogenic effector genes and their functions is essential for an understanding of transformation and has become the central preoccupation of the field. Several techniques have been introduced for the study of differential gene expression. Initial subtractive schemes including differential display, cDNA-based RDA, directional tag PCR, and SABRE have been superseded by more sophisticated and comprehensive approaches such as TOGA, SAGE and DNA microarrays (De Risi et al., 1996; Hubank and Schatz, 1999; Lavery et al., 1997; Liang and Pardee, 1995; Sutcliffe et al., 2000; Usui et al., 1994; Velculescu et al., 1995). Since transformation by v-Jun is restricted to avian cells for which chip technology is not yet available, and SAGE, based on human genome information, is not applicable, the subtractive techniques are still being applied and are yielding some valuable results (Fu et al., 1999, 2000; Goller et al., 1998; Hadman et al., 1996, 1998; Hartl and Bister, 1995, 1998; Oberst et al., 1997). In addition, mouse and human transformation systems are being developed for v-Jun and for c-Jun, and these are subject to analysis by microarrays (Cohen et al., 2001; Rinehart-Kim et al., 2000; Smith et al., 1999).

Table 1 summarizes information on genes that are differentially expressed in Jun-transformed cells. So far only three of these genes induce a change of cellular phenotype when they are up- or downregulated by themselves. Overexpression of heparin-binding epidermal growth factor, a protein that is upregulated in Jun-transformed CEF, induces a partially transformed phenotype in the same cell system (Fu et al., 1999). It is not clear whether the upregulation of this target is essential for Jun-induced transformation or whether it is ancillary. SSeCKS, a scaffolding protein that is a PKC substrate and can also inhibit G1 to S transition, is downregulated in rodent cells transformed by Jun, Src, Ras, Fos or Myc but shows no change in cells expressing Raf, Mos or Neu. Re-expression of SSeCKS results in a partial reversion of the Jun-transformed and the Src-transformed phenotype (Cohen et al., 2001; Erlichman et al., 1999; Lin and Gelman, 1997; Lin et al., 1996, 2000). SSeCKS is a candidate tumor suppressor. Lowered levels of SSeCKS may be a property of cells transformed by diverse oncogenic mechanisms. The third differentially expressed gene capable of inducing a phenotype code for SPARC, an extracellular matrix protein that is downregulated in CEF as a result of v-Jun-induced transformation. The re-expression of SPARC does not affect the phenotype of Jun-transformed cells in culture but significantly lowers their oncogenicity in the animal (Mettouchi et al., 1994; Vial and Castellazzi, 2000; Vial et al., 2000). In MCF-7 cells overexpressing c-Jun, SPARC is upregulated (Rinehart-Kim et al., 2000). This inconsistency could result from the different target spectra of v-Jun and c-Jun or it could be related to the difference in cell type.

It is likely that the phenotype of the transformed cell results from the combined action of several positively and several negatively misregulated targets. Each individual target may be necessary but its differential expression may not be sufficient for transformation. In this situation, verification of upregulated targets could be accomplished by turning the target off, e.g. with an antisense nucleotide. Essentiality of the target would then result in reversion to the normal phenotype. Individual downregulated targets could be verified in an analogous manner by re-expressing them.

In the near future, microarray techniques will rapidly and copiously expand the list of genes that are differentially expressed in Jun-transformed cells. This information will generate a major challenge, the discrimination between oncogenic effectors and innocent bystanders. The first steps in this necessary weeding process will probably consist of genetic screens. For instance, there are v-Jun mutants that transform effectively but transactivate TRE-driven reporters poorly (Havarstein et al., 1992). Presumably such mutants address genes that are essential for transformation and leave at least some of the bystanders untouched. Another genetic screen could use naturally-occurring oncogenic AP-1 proteins and artificially constructed ones (using available modular domains) and look for genes that are differentially expressed in cells transformed by all these proteins. The underlying assumption here is that the essential oncogenic effectors would be shared by various AP-1-mediated mechanisms. The use of transforming constructs that only homodimerize could also shrink the list of candidate targets (Vandel et al., 1996). Constructs that induce either growth factor- but not anchorage-indepence and those that induce the reverse could help with functional assignments of targets (van Dam et al., 1998). The time course of target activation or repression can be determined with regulated Jun constructs such as the Jun-estrogen receptor fusions and the tetracyclin-controlled quail cell lines (Fialka et al., 1996; Kim et al., 1996; Bader et al., 2000; Kruse et al., 1997). Genes responding early after Jun is switched on are candidates for direct targets, controlled by an interaction of Jun with the promoter-enhancer sequence of that gene. With the estrogen receptor constructs, the differential regulation of direct targets should be insensitive to inhibitors of protein synthesis, because the activation of Jun in this system consists in the translocation of pre-existing proteins into the nucleus (Kruse et al., 1997; Goller et al., 1998). On the other hand, regulation of indirect targets by the estrogen receptor fusions should in most cases be sensitive to inhibition of protein synthesis, because it would depend on the synthesis of subordinate transcriptional regulators. In the end though, none of these screens will be able to substitute for complete promoter analyses or for direct functional tests which remain a daunting task.

Coda

The mechanism of Jun-induced oncogenic transformation is known only in general terms; its most important aspect is misregulation of specific target genes. However, it would be naïve to consider transformation merely a consequence of a gain of Jun function. The dimerization partners of c-Jun prevailing in transformed cells are different from those in normal cells, and that difference probably entails a qualitative change in target spectrum (Mechta et al., 1997; Zoumpourlis et al., 2000). The highly oncogenic v-Jun differs further from its cellular counterpart in response to signals, stability and target spectrum (Black et al., 1994; Dai et al., 1995; Frame et al., 1991; Hussain et al., 1998; Kilbey et al., 1996; Treier et al., 1994).

A puzzling aspect of v-Jun is the antagonism to c-Jun. The suggestion that this antagonism is essential for transformation would cast c-Jun in the role of a tumor suppressor but is difficult to reconcile with the fact that overexpression of c-Jun transforms avian and mammalian cells and that the c-Jun-Ras co-transformation depends on elevated c-Jun activity. It is also in apparent contradiction to the transforming activity of a JNK-activating kinase (Hartkamp et al., 1999).

Jun is not overtly involved in human tumors. It has not been identified as a partner in a cancer-specific chromosomal translocation, is not amplified in tumors and usually not overexpressed. Yet Jun is located at the end of signal cascades that include important oncogenes active in human tumors. This position in cellular signaling makes Jun ipso facto a participant in numerous and diverse mechanisms of oncogenesis. Not surprisingly, dominant negative mutants of Jun attenuate the growth behavior of various human tumor cell lines as they also interfere with transformation by oncogenes linked to the Ras pathway (Neyns et al., 1999; Rapp et al., 1994; Suzuki et al., 1994; Ui et al., 2000). Jun activity may therefore be a determinant factor in many tumors. Regulating that activity with small molecules appears a worthy and promising goal.

Acknowledgements

Research of the author is supported by grants from the National Cancer Institute, National Institutes of Health. Masa Aoki, Thorsten Berg, Weixing Chen, Jason Iacovoni, Barbara Kempf, Daniel Maslyar, Makoto Nishizawa, Corinna Sonderegger, Anke Waha and Mark Ware made valuable comments on the manuscript and generously tolerated the preoccupation of their mentor with this review. Special thanks are due to Anke Waha who compiled Table 1. This is manuscript number 13798-MEM from the Scripps Research Institute. This paper is dedicated to two charmingly independent and affectionate tabbies, Tampopo and Iris, who helped shorten and re-shorten the manuscript and the reference list.

References

Abate C, Patel L, Rauscher FJd, Curran T. (1990). Science 249, 1157-1161. MEDLINE

Adler V, Franklin CC, Kraft AS. (1992a). Proc. Natl. Acad. Sci. USA 89, 5341-5345. MEDLINE

Adler V, Fuchs SY, Kim J, Kraft A, King MP, Pelling J, Ronai Z. (1995a). Cell Growth Differ. 6, 1437-1446.

Adler V. Polotskaya A, Wagner F, Kraft AS. (1992c). J. Biol. Chem. 267, 17001-17005. MEDLINE

Adler V, Schaffer A, Kim J, Dolan L, Ronai Z. (1995b). J. Biol. Chem. 270, 26071-26077. MEDLINE

Adler V, Unlap T, Kraft AS. (1994). J. Biol. Chem. 269, 11186-11191. MEDLINE

Ainbinder E, Bergelson S, Pinkus R, Daniel V. (1997). Eur. J. Biochem. 243, 49-57. MEDLINE

Alani R, Brown P, Binetruy B, Dosaka H, Rosenberg RK, Angel P, Karin M, Birrer MJ. (1991). Mol. Cell. Biol. 11, 6286-6295. MEDLINE

Alber T. (1992). Curr. Opin. Genet. Dev. 2, 205-210. MEDLINE

Allegretto EA, Smeal T, Angel P, Spiegelman BM, Karin M. (1990). J. Cell. Biochem. 42, 193-206. MEDLINE

Angel P, Allegretto EA, Okino ST, Hattori K, Boyle WJ, Hunter T, Karin M. (1988a). Nature 332, 166-171. MEDLINE