COP9 Signalosome-directed c-Jun Activation/Stabilization Is Independent of JNK*

The basic region-leucine zipper transcription factor c-Jun regulates gene expression and cell function. It participates in the formation of homo- or heterodimeric complexes that specifically bind to DNA sequences called activating protein 1 (AP-1) sites. The stability and activity of c-Jun is regulated by phosphorylation within the N-terminal activation domain. Mitogen- and stress-activated c-Jun N-terminal kinases (JNKs) were previously the only described enzymes phosphorylating c-Jun at the N terminus in vivo. In this report we demonstrate a JNK-independent activation of c-Jun in vivo directed by the constitutive photomorphogenesis (COP9) signalosome. The overexpression of signalosome subunit 2 (Sgn2), a subunit of the COP9 signalosome, leads to de novo assembly of the complex with a partial incorporation of the overexpressed subunit. The de novoformation of COP9 signalosome parallels an increase of c-Jun protein resulting in elevated AP-1 transcriptional activity. The c-Jun activation caused by Sgn2 overexpression is independent of JNK and mitogen-activated protein kinase kinase 4. The data demonstrate the existence of a novel COP9 signalosome-directed c-Jun activation pathway.

The transcriptional activity of c-Jun is dependent on its cellular concentration, which is regulated by induction of c-Jun gene expression and by its ubiquitin-and 26 S proteasome-dependent degradation (1). In addition, posttranslational modifications such as phosphorylation and dephosphorylation in response to many stimuli (for review see Ref. 2) modulate both activity and stability of the transcription factor. In most cells, the constitutive c-Jun levels are very low because of its short half-life. Phosphorylation at the N-terminal activation domain including Ser-63 and Ser-73 leads to a reduced ubiquitin-dependent degradation of c-Jun (3,4). Increased c-Jun levels contribute to elevated AP-1 1 transcriptional activity by the formation of homodimers or heterodimers with other transcription factors such as Fos. The AP-1 factor c-Jun is involved in important functions such as cell proliferation, differentiation, and survival (2).
Prior to this communication mitogen-and stress-activated c-Jun N-terminal kinases (JNKs) have been described as the only enzymes phosphorylating c-Jun on Ser-63 and Ser-73 in vivo. The JNKs are constitutively inactive. They are components of the MAP kinase signaling pathway. In response to many stimuli such as PMA, JNKs are activated by phosphorylation via MAP kinase kinases, MKK4 and MKK7, which are in turn phosphorylated by numerous MAP kinase kinase kinases (for review see Ref. 5). Although there are a few reports on JNK-independent c-Jun/AP-1 activation (e.g. Ref. 6), no alternative pathway was identified. Here we show the existence of a COP9 signalosome-directed AP-1 activation pathway.
The COP9 signalosome complex, originally identified in plant cells (7), consists of 8 subunits, which are conserved between plant and human (8,9). The significant sequence homologies between components of the COP9 signalosome and the 26 S proteasome lid (8 -12) and the colocalization of the two complexes led to the speculation that COP9 signalosome and 26 S proteasome cooperate in the regulation of signaling pathways (9). Whereas the 26 S proteasome lid components are essential for the degradation of many transcriptional factors (10) (for review see Ref. 13), COP9 signalosome might stabilize those proteins. The COP9 signalosome is involved in light signaling in plants (7) and isolated human COP9 signalosome is associated with kinase activity that phosphorylates regulators of transcription (9). Recently it has been demonstrated that the purified human COP9 signalosome complex phosphorylates c-Jun at the N-terminal activation domain including Ser-63 and Ser-73. In contrast to JNK, the isolated COP9 signalosome modifies only full-length c-Jun, whereas JNK phosphorylates N-terminal c-Jun fragments such as the ⌬c-Jun-(1-79) and ⌬c-Jun-(1-226) (9). The COP9 signalosome kinase has not yet been identified. Because none of the COP9 signalosome subunits contain a recognizable kinase domain, we refer to it as an associated kinase activity. Data presented here demonstrate that COP9 signalosome-directed phosphorylation of c-Jun results in the stabilization of the transcription factor in vivo accompanied by an elevated AP-1 activity.

EXPERIMENTAL PROCEDURES
In Vitro Kinase Assays-COP9 signalosome was isolated from human red blood cells as described previously (9). Kinase assays were performed with His-tagged full-length c-Jun as substrate (9). His-tagged full-length c-Jun, His-tagged ⌬c-Jun-(1-226), and His-tagged signalosome subunit 5 (Sgn5) were produced in Escherichia coli from pQE expression vectors and isolated using the Ni-nitrilotriacetic acid purification kit (Qiagen). The complete Sgn2 cDNA has been deposited in the GenBank TM data base under GenBank TM Accession number AF084260.
Transient Transfections and Reporter Assays-HeLa cells were grown in RPMI 1640 containing 4 mM glutamine (Life Technologies, Inc.), 100 units/ml penicillin, 100 g/ml streptomycin, and 10% fetal calf serum (Life Technologies, Inc.) in a humidified 5% CO 2 atmosphere. The cells were seeded in tissue culture plates for 48 h prior to infection. * This work was supported by grants DU 229/5-1 (to W. D.) and Na 292/5-1 (to M. N.) from the Deutsche Forschungsgemeinschaft. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF084260.
16 h before infection, the medium was replaced by fresh RPMI 1640 medium supplemented with 5% fetal calf serum. Transactivating activity of AP-1 was measured at 50 -70% confluence by cotransfection of a luciferase expression plasmid containing three repeats of the AP-1 binding site and other expression constructs using cationic liposomes (DAC-30, Eurogentec, Sart Tilman, Belgium). Expression constructs Sgn5, Sgn2, and ⌬c-Jun-(1-226) were cloned into the pcDNA3 vector (Invitrogen) expressing a N-terminal Flag-tagged sequence. 16 h after transfection cells were either treated with 40 nM PMA (Sigma) or left untreated. Luciferase assays were performed 3-4 h after treatment as recommended by the manufacturer's instructions (Promega). The results were recorded on a Wallac 409 counter (Berthold-Wallac). Representative results from more than three independent experiments are shown as fold induction or percentage induction compared with the control. Activities varied Ͻ10% between transfection experiments.
Immunoprecipitation and Protein Kinase Assays-To analyze the kinase activities of JNK and p38, cells were lysed in RIPA buffer containing 20 mM Tris, pH 8.0, 150 mM NaCl, 0.5% Nonidet P-40, 0.05% SDS, 5% glycerol, 1 mM EGTA, 10 mM NaF, 10 mM K 2 HPO 4 , 1 mM Na 3 VO 4 , 100 M phenylmethylsulfonyl fluoride, 10 M pepstatin A, and 4 M aprotinin. For immunoprecipitation, RIPA buffer-lysed cells were disrupted and incubated with anti-JNK1 (sc-474, Santa Cruz Biotechnology) or anti-p38 (sc-535, Santa Cruz Biotechnology) antibodies. Immunocomplexes were recovered and washed, and immunoprecipitates were used for in vitro kinase reactions using 1 g of GST-c-Jun (Santa Cruz Biotechnology) as substrate for JNK and 1 g of GST-ATF2 (Santa Cruz Biotechnology) for p38. The samples were separated in SDS-PAGE and dried, and substrate phosphorylation was visualized by autoradiography. Equal amounts of each sample were used for immunoblot analysis to indicate equivalent protein amounts in all lanes as described previously (14).
Density Gradient Centrifugation-For separation in 10 -30% glycerol gradients 16 h after transfection, cells were lysed with RIPA buffer as described above. Lysate from approximately 1 ϫ 10 7 cells was subjected to density gradient centrifugation (9). Trichloroacetic acid-precipitated proteins from fractions 6 -15 were separated by 12.5% SDS-PAGE, and Western blots using anti-Flag antibody (Stratagen) were developed according to the ECL protocol (Amersham Pharmacia Biotech). The same blots were stripped and reprobed with anti-c-Jun (Santa Cruz Biotechnology) and anti-Sgn3 (9) antibodies. Estimation of the amounts of Flag-tagged Sgn2 incorporated into cellular COP9 signalosome and of endogeneous COP9 signalosome was performed using the ImageQuant program (Molecular Dynamics).

RESULTS AND DISCUSSION
It has been shown that c-Jun interacts via amino acids 31-57 with Jun activation domain-binding protein 1 (Jab1) (15), recently identified as Sgn5 of the COP9 signalosome (9). To test whether COP9 signalosome-associated kinase phosphorylation of full-length c-Jun is dependent on Sgn5-c-Jun interaction, we performed in vitro competition assays with ⌬c-Jun-(1-226). As mentioned above, ⌬c-Jun-(1-226) is not phosphorylated by COP9 signalosome but should bind to Sgn5. Fig. 1 shows that increasing amounts of ⌬c-Jun-(1-226) added to constant amounts of full-length c-Jun inhibit c-Jun phosphorylation by purified COP9 signalosome. 2.4 g of ⌬c-Jun-(1-226) corre-sponding to a molar excess of approximately 3:1 are almost sufficient for complete inhibition of full-length c-Jun modification. Moreover, Fig. 1 also demonstrates that phosphorylation of full-length c-Jun by COP9 signalosome can be inhibited by recombinant Sgn5, which competes with the complex for the substrate. These data show that COP9 signalosome-directed c-Jun phosphorylation is dependent on binding to the integrated Sgn5 subunit. It has been suggested that COP9 signalosome might preferentially phosphorylate c-Jun dimers (9) formed by the interaction of C termini. Therefore, one can imagine a scenario in which one molecule of a c-Jun homodimer binds to Sgn5 in the COP9 signalosome complex, whereas its partner is phosphorylated by the associated kinase.
To verify these in vitro data under cellular conditions we had to find a way to increase intracellular COP9 signalosome activity that could be measured by elevated AP-1 activity. Therefore, the impact of COP9 signalosome subunit 2 (Sgn2) and Sgn5 overexpression in HeLa cells on AP-1 activity was tested by transient transfections in a luciferase reporter assay. As seen in Fig. 2, Sgn2 overexpression stimulates AP-1 activity significantly more than PMA, a well known stimulator of the JNK pathway. Sgn2 overexpression exerts a dose-dependent effect on AP-1 activity, reducing it at large amounts of transfected cDNA. In contrast, Sgn5 overexpression does not affect AP-1 activity. Although this is not in agreement with former findings (15), it does confirm recent data (16) and corresponds to our in vitro results shown in Fig. 1. Accordingly, free Sgn5 might trap cellular c-Jun and prevent its phosphorylation.
Why does Sgn2 overexpression lead to an increased AP-1 activity, whereas Sgn5 does not? One part of the explanation might be that in HeLa cells Sgn2 concentration limits COP9 signalosome assembly. The overexpressed protein could cause de novo complex formation with the consequence of an elevated COP9 signalosome activity resulting in an effective c-Jun stabilization. On the other hand, overexpressed Sgn5 may not incorporate into the complex. To prove this hypothesis we expressed Flag-tagged Sgn2 or Sgn5 in HeLa cells. Lysates of 10 7 cells each were separated by glycerol density gradients. As shown in Fig. 3, COP9 signalosome was mostly localized in fractions 7-10 according to its molecular mass of 450 kDa (9). Approximately 40% of the total Flag-Sgn2 protein sedimented with the COP9 signalosome, indicating that it was incorporated into de novo assembled COP9 signalosome. Compared with control cells, de novo assembly of COP9 signalosome in Sgn2-transfected cells led to an approximately 2-fold increase of the complex amount as estimated from immunoblots with anti-Sgn3 antibody. In contrast, less than 1% of Flag-Sgn5 incorporation into the COP9 signalosome was observed (see Fig. 3). To see whether the transfections had an impact on c-Jun stabilization, the same glycerol gradient fractions were analyzed for c-Jun amounts using an anti-c-Jun antibody. Whereas endogeneous c-Jun can barely be detected in Sgn5transfected cells and is very low in the controls, increased cellular c-Jun amounts were found in Sgn2-transfected HeLa cells (Fig. 3). This increase of c-Jun concentration is most likely because of stabilization of the protein and is responsible for the increase of AP-1 activity.
As mentioned above, c-Jun is stabilized by phosphorylation including serines 63 and 73, and JNKs have been described to be the responsible kinases (2). To study the possibility whether Sgn2 overexpression leads to JNK activation resulting in elevated AP-1 activity, JNK was immunoprecipitated from Sgn2, transfected Sgn5, or PMA-stimulated cells. The precipitate was assayed for immunocomplex kinase activity with full-length c-Jun as a substrate. As shown in Fig. 4A, PMA treatment led to JNK activation as expected. In contrast, there was no JNKdependent c-Jun phosphorylation as a consequence of Sgn2 or Sgn5 overexpression. Similar data were obtained in experiments in which another regulator of AP-1 activity, p38 MAP kinase (5), was analyzed (data not shown). These data demonstrate that increased AP-1 activity induced by Sgn2 overexpression is independent of JNK or p38 kinase activities. Thus, AP-1 activity can be stimulated via c-Jun phosphorylation by two different pathways COP9 signalosome-dependent and JNKdependent signaling. To further discriminate between the two signaling pathways dominant inhibitory MKK4(K116R) (DNMKK4) was transfected into HeLa cells. MKK4 is a physiological activator of JNK at the MAP kinase kinase level and also functions as an activator of p38 MAP kinase (5). As illustrated in Fig. 4B, transfection of DNMKK4 into PMA stimu-lated cells led to a dose-dependent inhibition of AP-1 activity. This c-Jun activation is dependent on JNK activation by MKK4. On the other hand, the COP9 signalosome-directed c-Jun activation is not affected by DNMKK4, again demonstrating JNK-independent signaling.
⌬c-Jun-(1-226) and recombinant Sgn5 inhibit the phosphorylation of full-length c-Jun by the COP9 signalosome-associated kinase in vitro (see Fig. 1). Whether similar effects could be obtained under cellular conditions was tested in HeLa cells cotransfected with Sgn2 and ⌬c-Jun-(1-226) or Sgn5 cDNAs (Fig. 5). Consistent with in vitro results, increasing amounts of both ⌬c-Jun-(1-226), which cannot form dimers and is unable to stimulate AP-1 activity, as well as free Sgn5 led to a complete inhibition of AP-1 activity induced by Sgn2. These data show that Sgn2-stimulated AP-1 activity is dependent on COP9 signalosome-directed phosphorylation of full-length c-Jun. Similar data were obtained with PMA-stimulated cells using ⌬c-Jun-(1-226) as competitor for JNK.
The presented data demonstrate that ectopically expressed Sgn2 incorporates into the cellular COP9 signalosome complex accompanied by a significant de novo complex formation. Increased amounts of COP9 signalosome lead to a stabilization of endogeneous c-Jun and increased AP-1 transactivation activity. Thus, transcriptional regulation of Sgn2 might represent a mechanism for controlling COP9 signalosome amounts and cellular activity, e.g. c-Jun activation/stabilization. The resulting stimulation of AP-1 activity is independent of JNK and MKK4 activities but depends on COP9 signalosome. Therefore, stabilization of c-Jun is because of phosphorylation of the transcription factor at its N-terminal activation domain by the COP9 signalosome-associated kinase as demonstrated with the isolated complex (9). Additional evidence for the existence of a JNK-independent COP9 signalosome-directed c-Jun signaling comes from the fact that the activity of the purified complex is inhibited by curcumin (11), a known inhibitor of AP-1 activity (17). Interestingly, there seems to be a cross-talk between the COP9 signalosome-directed c-Jun activation and the JNK pathway. The G-protein suppressor 1 (Gps1), identical to signalosome subunit 1, has been shown to act as a suppressor of JNK (18). It is perhaps advantageous for the cell to block the stress-activated protein kinases, although the COP9 signalosome is active.
In addition to c-Jun stabilization, COP9 signalosomedependent phosphorylation might also affect the transport of the transcription factor into the nucleus. It has been shown in Arabidopsis that a functional COP9 signalosome complex is essential for the nuclear accumulation of COP1, a transcriptional regulator, in dark adapted plants (19). In addition, the relocalization into the cytoplasm of another protein which binds Sgn5, p27 Kip1 , might be regulated by COP9 signalosome (16). However, because the p27 Kip1 relocalization was induced by Sgn5 overexpression, one should be cautious with the interpretation in light of the effects of Sgn5 overexpression presented in this paper. If COP9 signalosome is involved in the regulation of p27 Kip1 , large amounts of free Sgn5 might trap the cell cycle regulator and prevent its interaction with the COP9 signalosome.
The high homologies of COP9 signalosome subunits with components of the 26 S proteasome lid (8 -12) could be because of a common ancestor and perhaps a functional divergence of the two complexes during evolution. In the case of c-Jun, interaction with the COP9 signalosome leads to stabilization of the transcription factor, whereas the 26 S proteasome lid is involved in its ubiquitin-dependent degradation. The balance of the two processes, stabilization and degradation of c-Jun, is crucial for the decision whether cells proliferate, differentiate, or go into apoptosis.