Fra-1 controls motility of bladder cancer cells via transcriptional upregulation of the receptor tyrosine kinase AXL.

Fos-related antigen 1 (Fra-1) is a Fos family member overexpressed in several types of human cancers. Here, we report that Fra-1 is highly expressed in the muscle-invasive form of the carcinoma of the bladder (80%) and to a lesser extent in superficial bladder cancer (42%). We demonstrate that in this type of cancer Fra-1 is regulated via a C-terminal instability signal and C-terminal phosphorylation. We show that manipulation of Fra-1 expression levels in bladder cancer cell lines affects cell morphology, motility and proliferation. The gene coding for AXL tyrosine kinase is directly upregulated by Fra-1 in bladder cancer and in other cell lines. Importantly, our data demonstrate that AXL mediates the effect of Fra-1 on tumour cell motility but not on cell proliferation. We suggest that AXL may represent an attractive therapeutic target in cancers expressing high Fra-1 levels.

Heterodimerization with Fra-1 cooperates with the ERK pathway to stabilize c-Jun in response to the RAS oncoprotein.

Multiple tumorigenic pathways converge on the activating protein-1 (AP-1) family of dimeric transcription complexes by affecting transcription, mRNA decay, posttranslational modifications, as well as stability of its JUN and FOS components. Several mechanisms have been implicated in the phosphorylation- and ubiquitylation-dependent control of c-Jun protein stability. Although its dimer composition has a major role in the regulation of AP-1, little is known about the influence of heterodimerization partners on the half-life of c-Jun. The FOS family member Fra-1 is overexpressed in various tumors and cancer cell lines wherein it controls motility, invasiveness, cell survival and cell division. Oncogene-induced accumulation of Fra-1 results from both increased transcription and phosphorylation-dependent stabilization of the protein. In this report, we describe a novel role of Fra-1 as a posttranslational regulator of c-Jun. By using both constitutively and inducible transformed rat thyroid cell lines, we found that c-Jun is stabilized in response to RAS oncoprotein expression. This stabilization requires the activity of the extracellular signal-related kinase (ERK) pathway, along with c-Jun heterodimerization with Fra-1. In particular, heterodimerization with Fra-1 inhibits c-Jun breakdown by a mechanism dependent on the phosphorylation of the Fra-1 C-terminal domain that positively controls the stability of the protein in response to ERK signaling. Therefore, Fra-1 modulates AP-1 dimer composition by promoting the accumulation of c-Jun in response to oncogenic RAS signaling.

Identification of a C-terminal tripeptide motif involved in the control of rapid proteasomal degradation of c-Fos proto-oncoprotein during the G(0)-to-S phase transition.

c-Fos proto-oncoprotein is rapidly and transiently expressed in cells undergoing the G(0)-to-S phase transition in response to stimulation for growth by serum. Under these conditions, the rapid decay of the protein occurring after induction is accounted for by efficient recognition and degradation by the proteasome. PEST motifs are sequences rich in Pro, Glu, Asp, Ser and Thr which have been proposed to constitute protein instability determinants. c-Fos contains three such motifs, one of which comprises the C-terminal 20 amino acids and has already been proposed to be the major determinant of c-Fos instability. Using site-directed mutagenesis and an expression system reproducing c-fos gene transient expression in transfected cells, we have analysed the turnover of c-Fos mutants deleted of the various PEST sequences in synchronized mouse embryo fibroblasts. Our data showed no role for the two internal PEST motifs in c-Fos instability. However, deletion of the C-terminal PEST region led to only a twofold stabilization of the protein. Taken together, these data indicate that c-Fos instability during the G0-to-S phase transition is governed by a major non-PEST destabilizer and a C-terminal degradation-accelerating element. Further dissection of c-Fos C-terminal region showed that the degradation-accelerating effect is not contributed by the whole PEST sequence but by a short PTL tripeptide which cannot be considered as a PEST motif and which can act in the absence of any PEST environment. Interestingly, the PTL motif is conserved in other members of the fos multigene family. Nevertheless, its contribution to protein instability is restricted to c-Fos suggesting that the mechanisms whereby the various Fos proteins are broken down are, at least partially, different. MAP kinases-mediated phosphorylation of two serines close to PTL, which are both phosphorylated all over the G(0)-to-S phase transition, have been proposed by others to stabilize c-Fos protein significantly. We, however, showed that the PTL motif does not exert its effect by counteracting a stabilizing effect of these phosphorylations under our experimental conditions.

Degradation of cellular and viral Fos proteins.

c-Fos proto-oncoprotein is a short-lived transcription factor with oncogenic potential. We have shown that it is massively degraded by the proteasome in vivo under various experimental conditions. Other proteolytic systems including lysosomes and calpains, might, however, also marginally operate on it. Although there is evidence that c-Fos can be ubiquitinylated in vitro, the unambiguous demonstration that ubiquitinylation is necessary for its addressing to the proteasome in vivo is still lacking. c-Jun, one of the main dimerization partners of c-Fos within the AP-1 transcription complex, is also an unstable protein. Its degradation is clearly proteasome- and ubiquitin-dependent in vivo. Interestingly, several lines of evidence indicate that the addressing of c-Fos and c-Jun to the proteasome is, at least in part, governed by different mechanisms. c-Fos has been transduced by two murine osteosarcomatogenic retroviruses under mutated forms which are more stable and more oncogenic. The stabilization is not simply accounted for by simple deletion of c-Fos main destabilizer but, rather, by a complex balance between opposing destabilizing and stabilizing mutations. Though mutations in viral Fos proteins confer full resistance to proteasomal degradation, stabilization is limited because mutations also entail sensitivity to an unidentified proteolytic system. This observation is consistent with the idea that Fos-expressing viruses have evolved to ensure control protein levels to avoid high protein accumulation-linked apoptosis. In conclusion, the unveiling of the complex mechanism network responsible for the degradation of AP-1 family members is still at its beginning and a number of issues regarding the regulation of this process and the addressing to the proteasome are still unresolved.

Human cyclin C protein is stabilized by its associated kinase cdk8, independently of its catalytic activity.

Cyclin C belongs to the cyclin family of proteins that control cell cycle transitions through activation of specific catalytic subunits, the cyclin-dependent kinases (CDKs). However, there is as yet no evidence for any role of cyclin C and its partner, cdk8, in cell cycle regulation. Rather, the cyclin C-cdk8 complex was found associated with the RNA polymerase II transcription machinery. The periodic degradation of bona fide cyclins is crucial for cell-cycle progression and depends on the catalytic activity of the associated CDK. Here we show that endogenous cyclin C protein is quite stable with a half-life of 4 h. In contrast, exogenously expressed cyclin C is very unstable (half-life 15 min) and degraded by the ubiquitin-proteasome pathway. Co-expression with its associated cdk, however, strongly stabilizes cyclin C and results in a protein half-life near that of endogenous cyclin C. In stark contrast to data reported for other members of the cyclin family, both catalytically active and inactive cdk8 induce cyclin C stabilization. Moreover, this stabilization is accompanied in both cases by phosphorylation of the cyclin, which is not detectable when unstable. Our results indicate that cyclin C has apparently diverged from other cyclins in the regulation of its stability by its CDK partner.

Cellular and viral Fos proteins are degraded by different proteolytic systems.

c-Fos proto-oncoprotein is a short-lived transcription factor degraded by the proteasome in vivo. Its mutated forms expressed by the mouse osteosarcomatogenic retroviruses, FBJ-MSV and FBR-MSV, are stabilized two- and threefold, respectively. To elucidate the mechanisms underlying v-Fos(FBJ) and v-Fos(FBR) protein stabilization, we conducted a genetic analysis in which the half-lives and the sensitivities to various cell-permeable protease inhibitors of a variety of cellular and viral protein mutants were measured. Our data showed that the decreased degradation of v-Fos(FBJ) and v-Fos(FBR) is not simply explained by the deletion of a c-Fos destabilizing C-terminal domain. Rather, it involves a complex balance between opposing destabilizing and stabilizing mutations which are distinct and which include virally-introduced peptide motifs in both cases. The mutations in viral Fos proteins conferred both total insensitivity to proteasomal degradation and sensitivity to another proteolytic system not naturally operating on c-Fos, explaining the limited stabilization of the two proteins. This observation is consistent with the idea that FBR-MSV and FBJ-MSV expression machineries have evolved to ensure controlled protein levels. Importantly, our data illustrate that the degradation of unstable proteins does not necessarily involve the proteasome and provide support to the notion that highly related proteins can be broken down by different proteolytic systems in living cells.

Molecular characterization of the thermosensitive E1 ubiquitin-activating enzyme cell mutant A31N-ts20. Requirements upon different levels of E1 for the ubiquitination/degradation of the various protein substrates in vivo.

According to our current knowledge, protein ubiquitination involves three steps: activation of ubiquitin through formation of an energy-rich bond with an E1 ubiquitin-activating enzyme; and transfer of activated ubiquitin onto E2 ubiquitin-conjugating enzymes, which, in turn, alone, or in combination with E3 ubiquitin-protein ligase enzymes, transfer ubiquitin onto target proteins. A31N-ts20 cells are mouse embryo fibroblasts, thermosensitive for E1. We show here that: (a) the enzymatic activity of the enzyme is heat-inactivatable in vitro; and (b) a major mechanism responsible for E1 inactivation in vivo consists of accelerated destruction. Surprisingly, a >90% reduction in E1 abundance little alters the formation of the bulk of protein-ubiquitin conjugates when A31N-ts20 cells are grown at the nonpermissive temperature, indicating that cautious interpretation of results is required when studying ubiquitination of specific substrates using this cell line. Surprisingly, our data also indicate that, in vivo, ubiquitination of the various protein substrates in A31N-ts20 cells requires different amounts of E1, indicating that this mutant cell line can be used for unveiling the existence of differences in the intimate mechanisms responsible for the ubiquitination of the various cell proteins in vivo, and for providing criteria of reliability when developing in vitro ubiquitination assays for specific proteins.

The sensitivity of c-Jun and c-Fos proteins to calpains depends on conformational determinants of the monomers and not on formation of dimers.

Milli- and micro-calpains are ubiquitous cytoplasmic cysteine proteases activated by calcium. They display a relatively strict specificity for their substrates which they usually cleave at only a limited number of sites. Motifs responsible for recognition by calpains have not been characterized yet, and recently a role for PEST motifs in this process has been ruled out. c-Fos and c-Jun transcription factors are highly sensitive to calpains in vitro. They thus provide favourable protein contexts for studying the structural requirements for recognition and degradation by these proteases. Using in vitro degradation assays and site-directed mutagenesis, we report here that susceptibility to calpains is primarily determined by conformational determinants of the monomers and not by the quaternary structure of c-Fos and c-Jun proteins. The multiple cleavage sites borne by both proteins can be divided into at least two classes of sensitivity, the most sensitive ones being easily visualized in the presence of rate-limiting amounts of calpains. One site located at position 90-91 in c-Fos protein is extremely sensitive. However, efficient proteolysis did not have any strict dependence on the nature of the amino acids on either side of the scissile bond in the region extending from P2 to P'2. The structural integrity of the monomers is not crucial for recognition by calpains. Rather, sensitive sites can be recognized independently and their recognition is dependent on the local conformation of peptide regions that may span several tens of amino acids and maybe more in the case of the identified c-Fos hypersensitive site.

Are there multiple proteolytic pathways contributing to c-Fos, c-Jun and p53 protein degradation in vivo?

The c-Fos and c-Jun oncoproteins and the p53 tumor suppressor protein are short-lived transcription factors. Several catabolic pathways contribute to their degradation in vivo. c-Fos and c-Jun are thus mostly degraded by the proteasome, but there is indirect evidence that, under certain experimental/physiological conditions, calpains participate in their destruction, at least to a limited extent. Lysosomes have also been reported to participate in the destruction of c-Fos. Along the same lines, p53 is mostly degraded following the ubiquitin/proteasome pathway and calpains also seem to participate in its degradation. Moreover, c-Fos, c-Jun and p53 turnovers are regulated upon activation of intracellular signalling cascades. All taken together, these observations underline the complexity of the mechanisms responsible for the selective destruction of proteins within cells.

Differential directing of c-Fos and c-Jun proteins to the proteasome in serum-stimulated mouse embryo fibroblasts.

c-Fos and c-Jun proteins are highly unstable transcription factors that heterodimerize within the AP-1 transcription complex. Their accumulation is transiently induced at the beginning of the G0-to-S phase transition in quiescent cells stimulated for growth. To address the mechanisms responsible for rapid clearance of c-Fos and c-Jun proteins under these experimental conditions, we have used the ts20 mouse embryo fibroblasts which express a thermosensitive mutant of the E1 enzyme of the ubiquitin pathway. The use of cell-permeant protease inhibitors indicates that both proteins are degraded by the proteasome and excludes any major contribution for calpains and lysosomes during the G0-to-S phase transition. Synchronisation of ts20 cells at the non permissive temperature blocks the degradation of c-Jun, indicating that this process is E1-dependent. In contrast, c-Fos is broken down according to an apparently E1-independent pathway in ts20 cells, although a role for ubiquitinylation in this process cannot be formally ruled out. Interestingly, c-Jun is highly unstable in c-Fos-null mouse embryo fibroblasts stimulated for growth. Taken together, these observations show that in vivo during a G0-to-S phase transition (i) the precise mechanisms triggering c-Fos and c-Jun directing to the proteasome are not identical, (ii) the presence of c-Fos is not an absolute prerequisite for the degradation of c-Jun and (iii) the degradation of c-Jun is not required for that of c-Fos.

Complex mechanisms for c-fos and c-jun degradation.

c-fos and c-jun proto-oncogenes have originally been found in mutated forms in murine and avian oncogenic retroviruses. They both define multigenic families of transcription factors. Both c-jun and c-fos proteins are metabolically unstable. In vivo and in vitro work by various groups suggests that multiple proteolytic machineries, including the lysosomes, the proteasome and the ubiquitous calpains, may participate in the destruction of c-fos and c-jun. The relative contribution of each pathway is far from being known and it cannot be excluded that it varies according to the cell context and/or the physiological conditions. It has been demonstrated that, in certain occurrences, the degradation of both c-fos and c-jun by the proteasome in vivo involves the ubiquitin pathway. However, the possibility that proteasomal degradation can also occur in a manner independent of the E1 enzyme of the ubiquitin cycle remains an open issue.

PEST motifs are not required for rapid calpain-mediated proteolysis of c-fos protein.

Cytoplasmic degradation of c-fos protein is extremely rapid. Under certain conditions, it is a multi-step process initiated by calcium-dependent and ATP-independent proteases called calpains. PEST motifs are peptide regions rich in proline, glutamic acid/aspartic acid and serine/threonine residues, commonly assumed to constitute built-in signals for rapid recognition by intracellular proteases and particularly by calpains. Using a cell-free degradation assay and site-directed mutagenesis, we report here that the three PEST motifs of c-fos are not required for rapid cleavage by calpains. Testing the susceptibility of PEST motif-bearing and non-bearing transcription factors including GATA1, GATA3, Myo D, c-erbA, Tal-1 and Sry, demonstrates that PEST sequences are neither necessary nor sufficient for specifying degradation of other proteins by calpains. This conclusion is strengthened by the observation that certain proteins, reportedly known to be cleavable by calpains, are devoid of PEST motifs.

Ubiquitinylation is not an absolute requirement for degradation of c-Jun protein by the 26 S proteasome.

Degradation of rapidly turned over cellular proteins is commonly thought to be energy dependent, to require tagging of protein substrates by multi-ubiquitin chains, and to involve the 26 S proteasome, which is the major neutral proteolytic activity in both the cytosol and the nucleus. The c-Jun oncoprotein is very unstable in vivo. Using cell-free degradation assays, we show that ubiquitinylation, along with other types of tagging, is not an absolute prerequisite for ATP-dependent degradation of c-Jun by the 26 S proteasome. This indicates that a protein may bear intrinsic structural determinants allowing its selective recognition and breakdown by the 26 S proteasome. Moreover, taken together with observations by different groups, our data point to the notion of the existence of multiple degradation pathways operating on c-Jun.

Evidence indicating proximity in the nucleosome between the histone H4 N termini and the globular domain of histone H1.

Proteolysis of rat liver chromatin by the Arg-C peptidase, clostripain, is characterized by a progressive fragmentation of the N-terminal segments of the four core histones H2A, H2B, H3 and H4, until a well-defined limit digest is reached. This work addresses the case of histone H4. Two intermediate proteolytic sites are identified for this histone, i.e. Arg3 and Arg17, before the limit digest is achieved through cleavage of the polypeptide chain after Arg19. The accessibility of these intermediate sites depends strongly on the presence or absence of histone H1. When H1 is absent, both intermediate sites of histone H4 are similarly accessible, whereas one of them, Arg3, becomes totally inaccessible in the presence of histone H1. Di- and trinucleosomes were used with the aim of avoiding any interference with superstructural effects which can occur with longer polynucleosomes in the presence of H1. We also investigated the accessibility of the Arg sites of H1 that are located primarily in the central globular domain of this histone. In free histone H1, all the centrally located Arg sites are accessible to clostripain. In contrast, in the chromatin-bound state none of these sites is accessible. Besides the arginyl sites in the central globular domain of H1, two Arg residues are observed with the most abundant H1d variant in rat chromatin, one in the N-terminal region and the other in the C-terminal region. The restricted number of proteolytic fragments observed with chromatin-bound H1 is accounted for by the cleavage of H1 after these Arg residues located on the outside of the globular domain. Our results suggest that mutual steric effects are at play between histones H1 and H4 and indicate that the N termini of both histones H4 in the nucleosome lie in close proximity to the globular domain of H1. Based on these observations and taking into account the known structural features of the nucleosome, we propose a model for positioning the N-terminal segments of both histones H4 at the periphery of the nucleoprotein structure. In this model both H4 segments are located within the expanded DNA minor grooves, at periods +/- 1, symmetrically disposed relatively to the nucleosome dyad axis. This arrangement brings the amino ends of both H4 molecules in close contact with the H1 globular domain thus accounting for the observed inaccessibility of the Arg3 site of H4 in the presence of H1.

A 1H-NMR study of the transcription factor 1 from Bacillus subtilis phage SPO1 by selective 2H-labeling. Complete assignment and structural analysis of the aromatic resonances for a 22-kDa homodimer.

1H-NMR experiments have been performed on transcription factor 1 (TF1) encoded by Bacillus subtilis phage SPO1. To study this 22-kDa homodimeric DNA-binding protein, a selective 2H-labeling strategy has been employed. Complete sequence-specific assignments of all the resonances from the five aromatic residues were determined by a modified standard sequential-assignment procedure. The reduced contribution of spin diffusion upon the long-mixing-time nuclear-Overhauser-enhancement spectroscopy for the selectively 2H-labeled variants, as opposed to the fully 1H-containing protein, has allowed for the identification of the spin systems and of the long-range dipolar contacts between Phe28 and Phe47 protons in the protein core and between Phe61 and Phe97 protons. The latter suggests an interaction between the proposed beta-ribbon DNA-binding arm and the carboxy terminus of the paired monomer. A previously proposed TF1 structural model [Geiduschek, E. P., Schneider, G. J. & Sayre, M. H. (1990) J. Struct. Biol. 104, 84-90)] has been modified using constrained-energy-minimization calculations incorporating the experimentally determined set of aromatic-to-aromatic contacts. This new model has been analyzed with regard to the relative mobility and the relative solvent accessibility of the aromatic residues which have been measured by the nonselective T1 relaxation times of the aromatic resonances for the fully 1H-containing protein and the relaxation time enhancements upon selective 2H-labeling, respectively.