Molecular Biology Reports 26: 29–34, 1999. © 1999 Kluwer Academic Publishers. Printed in the Netherlands.
29
Comparison of human COP9 signalosome and 26S proteasome ‘lid’ Wolfgang Henke1 , Katherine Ferrell1 , Dawadschargal Bech-Otschir1 , Michael Seeger1 , Rüdiger Schade2 , Peter Jungblut3, Michael Naumann3 & Wolfgang Dubiel1,∗ 1 Institute of Biochemistry and 2 Institute of Pharmacology and Toxicology, Medical Faculty Charit´ e, Humboldt Uni-
versity; 3 Max-Planck-Institut für Infektionsbiologie, Abt. Molekulare Biologie, 10117 Berlin, Germany; ∗ Author for correspondence (E-mail:
[email protected])
Key words: COP9 signalosome, 26S proteasome, 19S regulator, JNK, curcumin
Abstract The human core COP9 signalosome consists of eight subunits which have been identified, cloned and sequenced. The components of COP9 signalosome possess homologies with eight non-ATPase regulatory subunits of the 26S proteasome. These polypeptides of the 19S regulator form a reversibly binding subcomplex called the ‘lid’. We isolated the ‘lid’ from human red blood cells and compared it with the COP9 signalosome complex. In addition to the non-ATPase regulatory polypeptides, we found a high molecular mass ATPase copurifying with the human ‘lid’. The COP9 signalosome-associated kinase activity is either not at all or only weakly affected by common kinase inhibitors such as 1-(5-Isoquinolinesulfonyl)-2-methyl-piperazine (H7), 5,6-dichloro-1-β-D-ribofuranosylbenzimidazole (DRB) or Wortmannin. Curcumin, a tumor suppressor and effector of AP-1 activation, is a potent inhibitor of the COP9 signalosome kinase activity with a Ki of about 10 µM. Since curcumin is known as an inhibitor of the c-Jun N-terminal kinase (JNK) signaling pathway acting upstream of the MAP kinase kinase kinase level, one site of action of the COP9 signalosome might be proximal to regulators on that level.
Introduction Recently we identified and characterized a new protein complex from human red blood cells and HL-60 cells and called it signalosome [1]. The complex consists of 8 polypeptides, most of which have been identified as components of signal transduction pathways. Recent structural data on the COP9 complex demonstrate that the signalosome is identical to the COP9 complex [2], originally discovered in A. thaliana and implicated to be involved in light-mediated signaling [3]. Therefore, in agreement with Wei et al. (personal communication), the complex should be renamed the COP9 signalosome. The connection with signal transduction has been supported by the fact that our COP9 signalosome preparations possess kinase activity. The COP9 signalosome-associated kinase activity specifically phosphorylates a number of factors involved in transcriptional regulation including c-Jun, IκBα and p105 [1]. There is another complex, called cytokine-
responsive IκB kinase (IKK) complex, which specifically phosphorylates IκBα on Ser32 and 36 which leads to the ubiquitinylation and subsequent degradation of the polypeptide (for rev. see [4]). Unfortunately the IKK complex has been called ‘signalsome’ [5] which might lead to confusion with the COP9 signalosome. Since the specificity of IκBα phosphorylation (W.D. & M.N., publication in prep.) and the subunit composition [1, 6] are different for COP9 signalosome and the IKK complex, we conclude that the two particles are not identical, which does not exclude that they might share subunits. In addition to its relationship to signal transduction, we showed by sequence homologies and subcellular localization that the COP9 signalosome is also related to the proteasomal system, in particular to the 19S regulator of the 26S proteasome. In fact, proteasome-COP9-initiation factor 3 (PCI) domains and Mpr1p-Pad1p (MPN) domains have been identified in subunits of the 19S regulator, the COP9 signalosome and the initiation factor 3 complex [7].
30 There is increasing evidence that multimeric protein complexes are important regulatory entities in signal transduction and in the Ub pathway as shown for the cyclosome/APC complex [8, 9] and the SCFcomplexes [10] during cell cycle and for the IKK complex [4] in IκB degradation. These complexes represent relays which interpret data from signaling via phosphorylation-dephosphorylation and switch to signaling via ubiquitin. We have isolated a subcomplex of the 19S regulator from human erythrocytes which is, with high probability, identical to the 26S proteasome ‘lid’ first described in yeast [11]. It copurifies with an ∼80 kDa protein, a putative member of the AAA (ATPaseAssociated-Activity) family of ATPases. Here we show structural features of the COP9 signalosome and the 26S proteasome ‘lid’ and further characteristics of the COP9 signalosome-associated kinase activity.
Structural aspects of human COP9 signalosome and 26S proteasome ‘lid’ All eight subunits of the human core COP9 signalosome have been identified, cloned and sequenced and are summarized in Table 1. We recently cloned and sequenced the first complete human Sgn2 = Trip15 cDNA, available under Genbank # AF084260. Human Sgn4 and Sgn7 cDNAs were assembled from ESTs (Kay Hofmann, pers. comm.). The corresponding homologs from mice were published by Wei et al. [2]. Figure 1 shows a silver-stained two-dimensional (2D) gel of human COP9 signalosome. The spots in the framed areas were excised and analyzed by massspectrometry. The data show that all identified spots are modifications of the 8 known core subunits. Although the exact nature of the modifications is unclear at the moment, different grades of phosphorylation are very likely as indicated by the existence of canonical phosphorylation sites in all subunits. In addition, two forms of COP9 signalosome complex with identical molecular masses were separated by a Mono Q ion exchange column (data not shown) suggesting posttranslational modification by phosphorylation. We do not know the identity of all spots seen on the 2D gel. There are a number of unidentified spots larger than Sgn1 and between Sgn5 and Sgn7. Some of these polypeptides might be degradation products. In addition to Sgn7 = COPS7a there might be an alternative form, COPS7b, as shown for mouse
Figure 1. Two-dimensional electrophoresis of human COP9 signalosome. COP9 signalosome was purified as described [1]. One µg of gel-eluted complex was loaded for the silver-stained gel shown. In order to identify spots, 10 µg of protein were loaded and the gel was stained with coomassie. The spots were excised, subjected to tryptic digestion and analyzed by matrix assisted laser-desorption ionization (MALDI) mass spectrometry.
COP9 signalosome [2]. Using antibody screenings during signalosome preparation, we noticed a complex containing the non-ATPase 19S regulator subunit S10a. The S10a-containing complex comigrates with signalosome in non-denaturing electrophoresis (Figure 2a) and in glycerol gradient centrifugations, since it has a molecular mass similar to that of the COP9 signalosome. However, the two complexes can be well separated by ion exchange chromatography (Mono Q) where the S10-containing complex elutes at about 0.25 M KCl and the COP9 signalosome at 0.35 M KCl. Fractions containing the new complex were pooled and tested by immunoblotting. As shown in Figure 2b, in addition to the non-ATPase S10 = Rpn7, it contains the non-ATPase 19S regulator subunits S3 = Rpn3 (as seen with the anti 19S regulator antibody), S12 = Rpn8 and S13 = Rpn11. It does not exhibit typical regulatory ATPases such as S4 = Rpt2 (see Figure 2b), S60 = Rpt5, S7 = Rpt1 and S8 = Rpt6 and it does not contain the Ub conjugate binding protein S5a = Rpn10 (data not shown). The silver stained gel demonstrates no subunits of an appropriate mass to be the non-ATPases S1 = Rpn2 and S2 = Rpn1. Although we do not have antibodies against all subunits of the 19S regulator, we assume that the S10a-containing
31 Table 1. Subunits of human COP9 signalosome and their homologs in other species and in the 26S proteasome ‘lid’ Subunit
Organism
Sgn1 Gps1 COPS1 COP11/FUS6 Sgn2 Trip15 Trip15 COPS2 Sgn3 COPS3 LC15 Sgn4 COPS4 COS41.8 Sgn5 JAB1 COPS5 COPS5 Sgn6 34kDaMov34 COPS6 Sgn7 ACOB COPS7a Sgn8 COPS8 COPS8
human human human A. thaliana human human Drosophila mouse human mouse L. chilense human mouse C. intestinalis human human mouse A. thaliana human human mouse human E. nidulans mouse human human A. thaliana
Acc. # subunits of the 26S proteasome ‘lid’
Homology to
Domain
S10a, p44, Rpn7
PCI
S9, p44.5, Rpn6
PCI
S3, p58, Rpn3
PCI
p55, Rpn5
PCI
S13, Poh1, Rpn11
MPN
S12, p40, Rpn8
MPN
S11, p40.5, Rpn9
PCI
S14, p31, Rpn12
PCI
U20285 L26498 AF084260 L40388 U57758 AF071312 AF031647 AF071313 U19099 AF071314 1764018 U65928 U70736
U70735 AF071315 U18265 AF071316 U51205 L32874
complex is very similar to, if not identical with, the 26S proteasome ‘lid’ recently isolated from S5a = Rpn10 deletion mutants of S. cerevisiae [11]. As pointed out before [11], the ‘lid’ reversibly binds to the base consisting of the 20S proteasome, the six 19S regulator ATPases, S1 = Rpn2 and S2 = Rpn1 and presumably S5a = Rpn10 as an anchor for the ‘lid’. Interestingly, the ‘lid’ contains those 19S regulator subunits which are the most homologous to components of the COP9 signalosome (see Table 1). According to Table 1, each of the two complexes consists of six polypeptides with PCI and two with MPN domains, the function of which are unknown. In many cases the homology goes far beyond these domains. In addition, when anti ATPase antibodies against S4 or S60 = TBP1 were used at low dilutions, we found
cross reactivity with an ∼80 kDa polypeptide copurifying with the ‘lid’ (see Figure 2b, anti S4). The nature and function of that protein is currently under investigation. At the moment we do not know whether the 26S proteasome ‘lid’ from human red blood cells occurs in free form or whether the 26S proteasome dissociates during isolation procedures. Because the COP9 signalosome copurifies with the 26S proteasome throughout most of our preparation and because of the high homologies to the ‘lid’, we speculate that the ‘lid’ might be substituted by COP9 signalosome. Whether this speculation is true and whether the ‘lid’, like COP9 signalosome, possesses an associated kinase activity are questions currently under investigation.
32 Table 2. Effects of kinase inhibitors on COP9 signalosome-associated kinase activity Inhibitor None H7 DRB Staurosporine Wortmannin
% activity
(50 µM) (50 µM) (50 µM) (100 µM)
100 84 52 26 100
Kinase assays were performed as described in Figure 3. H7 – 1-(5-Isoquinolinesulfonyl)-2methyl-piperazine; DRB – 5,6-dichloro-1-βD-ribofuranosyl-benzimidazole.
The COP9 signalosome-associated kinase activity is inhibited by curcumin
Figure 2. Comparison of human COP9 signalosome and 26S proteasome ‘lid’. The signalosome was prepared as described [1] with an additional Mono Q column on the pooled complexes. Elution was performed with a 0.1 to 1 M KCl gradient and signalosome eluted at about 0.35 M KCl as determined by Western blots using the anti Sgn3 antibody. Fractions 21 and 22 were pooled and concentrated to 1 mg/ml by Microcon 30 (Amicon). The ‘lid’ was obtained by pooling fractions 6–8 of the first 10–40% glycerol gradient containing mostly subcomplexes of the 26S proteasome and signalosome. This pool was treated the same way as the signalosome [1]. The ‘lid’ was detected by anti S10a antibody. From the last Mono Q column it eluted at about 0.25 M KCl. Mono Q fraction 9 and 10 were concentrated by Microcon 30 to 1 mg/ml. a) Non-denaturing electrophoresis of COP9 signalosome (SGN), ‘lid’ and 26S proteasome (26S). Four µg of each complex were loaded onto a 4–15% Phast gel. The 26S proteasome was prepared as previously described [1]. b) Immunoblotting of the human 26S proteasome ‘lid’. Anti S4 and anti S19 antibodies are gifts from Martin Rechsteiner and the anti S13 = Poh1 antibody was obtained from Chris Norbury. The anti S10a antibody was produced in rabbits by standard methods. For comparison, SDS PAGE patterns visualized by coomassie of COP9 signalosome and of the 26S proteasome and immunoblots of the 26S proteasome with anti S4 and anti S19 antibodies are shown. Relevant subunits of the 26S proteasome are indicated at the right.
Although the eight subunits of the core COP9 signalosome do not exhibit typical kinase domains, the isolated complex possesses a serine/threonine kinase activity [1]. Whether the kinase is one of the unidentified spots seen in 2D electrophoresis (Figure 1) is not yet clear. To characterize the COP9 signalosomeassociated kinase activity we tested a number of typical kinase inhibitors using recombinant 6xHis-c-Jun which, in addition to other sites, is phosphorylated by COP9 signalosome on serines 63 and 73 in the activation domain of the transcription factor [1]. The results are summarized in Table 2. Although H7 is known as a potent inhibitor of serine/threonine kinases, its effect on COP9 signalosome-associated kinase activity is marginal, as is staurosporine which has no effect in the nM concentration region, the normal inhibitory region for many kinases (data not shown). Taken together the data would exclude kinases such as PKC and PKA. The small effect of DRB, an inhibitor of casein kinase II with a Ki of 6 µM [12], makes the presence of such a kinase in our preparations unlikely. Wortmannin was used to exclude the presence of the double-stranded DNA-dependent protein kinase, DNA-PK, which is highly sensitive to the inhibitor [13]. Finally our interest turned to a substance which is a dietary pigment in curry, called curcumin, which has been described as a supressor of tumor initiation and promotion [14, 15]. Curcumin is also known as a strong inhibitor of AP-1 activation [16]. Since COP9 signalosome might be involved in the AP-1 activation pathway via c-Jun phosphorylation, we tested the effect of curcumin on the signalosome-mediated c-Jun phosphorylation. Figure 3 shows the concentration de-
33 The identification of the responsible kinase and functions of the COP9 signalosome in different cellular compartments are currently under investigation.
Conclusions
Figure 3. Inhibition of COP9 signalosome-associated kinase activity by curcumin. The kinase assay was performed as described [1] with 1 µg of 6xHis-c-Jun as substrate. Phosphoimages were quantified by area integration using the Image Quant program (Molecular Dynamics). The data represent means ± S.E.M. obtained in three independent experiments with duplicates. A Ki of 11 µM was estimated by fitting all data with a non-linear regression procedure.
pendency of the COP9 signalosome-associated kinase activity inhibition by curcumin. There is an increase of activity at very low curcumin concentrations which cannot be explained at the moment. By non-linear regression a Ki of 11 µM was calculated. Recently it has been shown that the c-Jun N-terminal kinase (JNK) signaling pathway induced by various agonists is inhibited by curcumin with an 50% inhibition concentration of 5 to 10 µM [17]. However, neither kinases of the cascade such as JNK (MAP kinase), MKK4 (MAP kinase kinase) nor MEKK1 (MAP kinase kinase kinase) are direct targets of curcumin. Therefore, the authors concluded that curcumin acts on regulators proximally upstream of the MAP kinase kinase kinase level. Since MEKK1 has been described as an activator of the IκBα kinase complex, which induces ubiquitinylation and subsequent degradation of the phosphorylated NF-κB inhibitor [18], it might explain the supression of both AP-1 and NF-κB signaling by curcumin [16, 19]. These inhibitory effects may account for the anti-inflammatory and anti-oxidant properties of curcumin [19, 20]. In that context, the curcumin effect on the COP9 signalosome-associated kinase activity might mean that the complex acts upstream of the MAP kinase kinase kinase level. On the other hand, the COP9 signalosome has been localized to the nucleus [1].
(1) 2D electrophoresis reveals that subunits of the COP9 signalosome occur in multiple forms which are probably due to different grades of phosphorylation. This observation supports the idea of COP9 signalosome as an mediator of signal transduction. (2) There is a subcomplex of the 19S regulator called ‘lid’ which dissociates from 26S proteasomes in preparations from human red blood cells having no deletion in S5a. Whether the lid exists in free form in human cells is unknown. (3) The human ‘lid’ copurifies with an ∼80 kDa putative AAA-ATPase of unknown function. (4) High homologies between subunits of the COP9 signalosome and the ‘lid’ led us to assume that the two complexes might be interchangeable in the 26S proteasome. (5) Common inhibitors of serine/threonine, casein and other kinases had little or no effects on the COP9 signalosome-associated kinase activity. (6) Curcumin, an anti-inflammatory and anti-carcinogenic agent, is a potent inhibitor of the COP9 signalosome-associated kinase activity with a Ki of about 10 µM.
Acknowledgements These studies were supported by grants from the EC (Biomed II, PL9540447) and from the Deutsche Forschungsgemeinschaft (DU229/5-1) to W.D. and (NA 292/5-1) to M.N.
References 1.
2. 3. 4. 5.
Seeger M, Kraft R, Ferrell K, Bech-Otschir D, Dumdey R, Schade R, Gordon C, Naumann M & Dubiel W (1998) FASEB J. 12: 469–478 Wei N, Tsuge T, Serino G, Dohmae N, Takio K, Matsui M & Deng X-W (1998) Curr. Biol. 8: 919–922 Chamovitz DA, Wei N, Osterlund MT, von Arnim AG, Staub JM, Matsui M & Deng X-W (1996) Cell 86: 115–121 Maniatis T (1997) Science 278: 818–819 Mercurio F, Zhu H, Murray BW, Shevchenko A, Bennett BL, Li JW, Young DB, Barbosa M, Mann M, Manning A & Rao A (1997) Science 278: 860–866
34 6. Scheidereit C (1998) Nature 395: 225–226 7. Hofmann K & Bucher P (1998) Trends Biochem. Sci. 23: 204– 205 8. Hershko A (1997) Curr. Opin. Cell Biol. 9: 788–799 9. Peters J-M, King RW, Höög Ch & Kirschner MW (1996) Science 274: 1199–1201 10. Krek W (1998) Curr. Opin. Genet. Dev. 8: 36–42 11. Glickman MH, Rubin DM, Coux O, Wefes I, Pfeifer G, Cjeka Z, Baumeister W, Fried VA & Finley D (1998) Cell 94: 615– 623 12. Zandomeni R, Zandomeni MC, Shugar D & Weinmann R (1986) J. Biol. Chem. 261: 3414–3419 13. Shieh S-Y, Ikeda M, Taya Y & Prives C (1997) Cell 91: 325– 334
14.
15. 16. 17. 18. 19. 20.
Huang MT, Ma W, Lu YP, Chang RL, Fisher C, Manchand PS, Newmark HL & Conney AH (1995) Carcinogenesis 16: 2493–2497 Rao CV, Simi B & Reddy BS (1993) Carcinogenesis 14: 2219–2225 Huang TS, Lee SC & Lin JK (1991) Proc. Natl. Acad. Sci. USA 88: 5292–5296 Chen Y-R & Tan T-H (1998) Oncogene 17: 173–178 Lee FS, Hagler J, Chen ZJ & Maniatis T (1997) Cell 88: 213– 222 Singh S & Aggarwal BB (1995) J. Biol. Chem. 270: 24995– 25000 Ruby AJ, Kuttan G, Babu KD, Rajasekharan KN & Kuttan R (1995) Cancer Lett. 94: 97–83
