Mammalian COP9 signalosome




The COP9 signalosome (CSN) complex is highly conserved from yeast to human. Although the plant CSN was first identified as a negative regulator of photomorphogenesis, the mammalian CSN is linked to different biological responses such as checkpoint control, signal transduction, development and the cell cycle. Frequent over-expression of the CSN subunit in a variety of human cancers suggests its involvement in cell transformation and tumorigenesis. The best-known biochemical function associated with the CSN is the control of protein stability via the ubiquitin–proteasome system through regulation of cullin-RING-E3 ubiquitin ligase activity by deneddylation, by controlling the activity of COP1 E3 ligase, or by counteracting ubiquitin-mediated degradation through a CSN-associated deubiquitinating enzyme. In addition to affecting the stability of transcription factors, the CSN may regulate gene transcription by directly associating with chromatin. This review summarizes recent findings and discusses the physiological role and the cellular function of the mammalian CSN in terms of the regulation of cell proliferation.


In 1996, a novel protein complex consisting of eight subunits was identified as a negative regulator of photomorphogenesis in Arabidopsis (designated as the COP9 complex) (Chamovitz et al. 1996). Photomorphogenesis is the developmental process that occurs in response to the dark/light environment. The COP9 complex was proven to modulate gene expression dependent on light (Wei & Deng 1992; Chamovitz 2009). Later in 1998, the same protein complex was rediscovered during the course of the purification of proteasomes from the lysate of mammalian red blood cells (designated as the Jab1-containing signalosome) (Seeger et al. 1998) as well as purified from pig spleen (Wei et al. 1998), indicating that its function is beyond the control of light-mediated plant signal transduction. The nomenclature of this protein complex was eventually unified as ‘the COP9 signalosome (CSN) complex, and each subunit was designated as CSN1 to CSN8 depending on its size (from the largest, CSN1, to the smallest, CSN8) (Deng et al. 2000). Since its discovery in plants and mammals, the CSN has been identified in a variety of different eukaryotic organisms [Saccaromyces cerevisiae (Wee et al. 2002; Maytal-Kivity et al. 2003), Schizosaccharomyces pombe (Mundt et al. 1999), Aspergillus nidulans (Busch et al. 2003), Neurospora crassa (He et al. 2005), Caenorhabditis elegans (Luke-Glaser et al. 2007), Drosophila melanogaster (Freilich et al. 1999)]. The range of biological responses in which the CSN is involved is extensive, including embryonic development, the cell cycle, checkpoint control, T-cell development, signal transduction, oocyte maturation, autophagy and circadian rhythm, reflecting the multifunctionality of the complex (see review Wei & Deng 2003).

Each mammalian CSN subunit was independently identified as an interactor of several signaling molecules (see Table 1). Notably, the fifth mammalian component, CSN5, was first identified as Jun-activation-domain-binding protein (Jab) 1 (Claret et al. 1996), independent of the CSN complex. Later, Jab1/CSN5 was repeatedly found to be an interactor of various intracellular regulators, such as the Cdk inhibitor p27, the leucocyte antigen receptor, various transcription factors and the tumor suppressor p53 (see Table 1), placing CSN5 in a unique position among the 8 CSN subunits. Besides the holo-CSN complex, each subunit of the CSN may have unique functions. Certain CSN subunits exist as a monomer or different small subcomplexes (Kwok et al. 1998; Freilich et al. 1999; Oron et al. 2002; Tomoda et al. 2002; Fukumoto et al. 2005; Sharon et al. 2009) (Fig. 1). The molecular identity and functions of each subcomplex remain to be clarified but may contribute to further multifunctionality.

Table 1.   Interactors for the mammalian CSN subunits
CSN subunitInteractorFunctionReference
CSN15/6 kinaseProtein/lipid kinaseSun et al. 2002
CSN2Interferon consensus sequence binding protein (ICSBP)Transcription factorCohen et al. 2000
Cullin 1, 2Core component of CRLLyapina et al. 2001; Schwechheimer et al. 2001; Yang et al. 2002
TRThyroid hormone receptorLee et al. 1995
DAX-1, COUP-TF1, ecdysone receptorNuclear receptorDressel et al. 1999; Altincicek et al. 2000
CSN3MLF1 (myeloid leukemia factor 1)Shuttling proteinYoneda-Kato et al. 2005; Yoneda-Kato & Kato 2008
IKKγIκB kinaseHong et al. 2001
Int6/eIF3eTranslation initiation factor 3Hoareau Alves et al. 2002
CK2Protein kinaseUhle et al. 2003
PKDProtein kinase DUhle et al. 2003
CSN5c-Jun, JunDAP-1 transcription activatorClaret et al. 1996
P27Kip1CDK inhibitorTomoda et al. 1999
Bcl3IκB familyDechend et al. 1999
LFA-1Integrin receptor (leukocyte functional antigen 1)Bianchi et al. 2000; Perez et al. 2003
rLHRLutropin/choriogonadotropin receptorLi et al. 2000
Macrophage migration inhibitor factor (MIF)CytokineKleemann et al. 2000
PRProgesterone receptorChauchereau et al. 2000
SRC-1Steroid receptor coactivatorChauchereau et al. 2000
P53Tumor suppressive transcriptionz factorBech-Otschir et al. 2001; Oh et al. 2006a; Zhang et al. 2008
ERαEstrogen receptor αCallige et al. 2005
P8 Malicet et al. 2006
RUNX3Runt-related transcription factor 3Kim et al. 2009
WNVCpWest Nile Virus Capsid proteinOh et al. 2006b
HIF-1αHypoxia inducible transcription factorBae et al. 2002; Bemis et al. 2004
Smad7TGF-β signal transducerKim et al. 2004
Smad4TGF-β signal transducerWan et al. 2002
PGP9.5Ubiquitin carboxy terminal hydrolaseCaballero et al. 2002
9-1-1 complexDNA-damage sensor (Rad1-Rad9-Hus1 complex)Huang et al. 2007
EB1Microtubule end-binding protein 1Peth et al. 2007
DNA topoisomerase II α Yun et al. 2004
HPOHepatopoietinLu et al. 2002
Hepatitis B virus X protein Tanaka et al. 2006
Thioredoxin Hwang et al. 2004
VDUP1 Jeon et al. 2005
HAND2Transcription factorDai et al. 2004
Psoriasin Emberley et al. 2003
L-type Ca2+ channels Kameda et al. 2006
Pretease-activated receptor-2 Luo et al. 2006
IRE1α Oono et al. 2004
53BP1Mitotic checkpoint activationKwak et al. 2005
E2F1Transcription factorHallstrom & Nevins 2006
Brn-2POU transcription factorHuang et al. 2005
Id3 Bounpheng et al. 1999
SMYD3Histone methyl transferaseMori et al. 2008
CSN6VprHIV-1 proteinMahalingam et al. 1998
Int6/eIF3eTranslation initiation factor 3Hoareau Alves et al. 2002
Roc1SCF componentLyapina et al. 2001; Schwechheimer et al. 2001
CSN7Int6/eIF3eTranslation initiation factor 3Hoareau Alves et al. 2002
CK2Protein kinaseUhle et al. 2003
PMF-1Polyamine modulated factorWang et al. 2001
Figure 1.

 The model of the COP9 signalosome (CSN) structure. The models of the holo-CSN and small subcomplexes are drawn (Sharon et al. 2009). The subcomplex consisting of CSN4-7 components is designated as mini-CSN (Tomoda et al. 2002). Oncogenic CSN5 is shown in red, and putative tumor suppressors CSN2, 3 and 8 are shown in blue.

Each of the eight components of the CSN contains unique domains, MPN (Mpr1-Pad1-N-terminal, for CSN5 and 6) and PCI (proteasome, COP9, initiation factor 3, for CSN1, 2, 3, 4, 7 and 8) (Hofmann & Bucher 1998). MPN and PCI domains are also found in basically all subunits of the lid complex of the proteasome and some of the eIF3 complex (Hofmann & Bucher 1998), which show a remarkable resemblance to the CSN in terms of subunit composition and the amino acid sequence of each subunit (Glickman et al. 1998) (Table 2). Although MPN/PCI domain-containing proteins were originally identified in these three complexes, recent reports indicate that cells contain a variety of MPN/PCI proteins that are not incorporated into these complexes (designated as an orphan MPN/PCI protein) (Kikuchi et al. 2003; McCullough et al. 2004; Pena et al. 2007; Cooper et al. 2009; Pick et al. 2009). However, the precise functions of these proteins have not been fully investigated. The similarity in the composition and sequence of subunits suggests that these three complexes originated from a common ancestor and share similar molecular properties. Although the precise molecular mechanisms remain to be clarified, it seems likely that at least one of the physiological roles of the CSN lies in the protein life cycle; from synthesis to degradation.

Table 2.   Subunit comparison among CSN, proteasome lid and eIF3 complexes
CSNProteasome LideIF3Domain
  1. *They all contains the PCI domain but are unable to be corresponded to the exact CSN subunits.

  2. †JAMM domain.

  3. ‡Mammalian cells contain two highly homologous genes.

  4. §Some eIF3 subunits contain neither PCI nor MPN domain.


Molecular action of the mammalian CSN

The CSN is a multifunctional protein complex and is associated with many different activities, such as the regulation of protein stability, transcription, protein phosphorylation and intracellular distribution (Figs 2 and 3) (Table 3).

Figure 2.

 Regulation of the ubiquitin ligases by the COP9 signalosome (CSN).

Figure 3.

 Other functions of the COP9 signalosome (CSN).

Table 3.   Molecular functions of mammalian CSN
Regulation of protein stability
 Regulation of CRL by deneddylationLyapina et al. 2001; Schwechheimer et al. 2001; Cope et al. 2002; Groisman et al. 2003; Pintard et al. 2003
 Regulation of the ubiquitin ligase COP1Dornan et al. 2004, 2006; Yoneda-Kato et al. 2005
 Regulation of deubiquitinationSchweitzer et al. 2007
Transcriptional regulation/specificity factorClaret et al. 1996; Hallstrom & Nevins 2006
Protein phosphorylationSeeger et al. 1998; Naumann et al. 1999; Bech-Otschir et al. 2001; Sun et al. 2002; Uhle et al. 2003
Regulation of subcellular distributionTomoda et al. 1999; Oh et al. 2006a,b; Huang et al. 2007; Zhang et al. 2008; Kim et al. 2009

Regulation of protein stability

The most studied function of the CSN is the regulation of protein degradation (Schwechheimer & Deng 2001; Bech-Otschir et al. 2002; von Arnim 2003; Cope & Deshaies 2003; Wolf et al. 2003). This is carried out through the ubiquitin–proteasome system in three ways: (i) regulation of the cullin-RING-E3 ligases (CRL) by deneddylation; (ii) regulation of the ubiquitin ligase COP1; and (iii) regulation of deubiquitination.

Deneddylation of cullins

Among the many different ubiquitin ligases, CRL form the largest family, which includes SCF, VHL and so on, depending on the types of cullin subunits (Cul1, 2, 3, 4A, 4B, 5 and 7) (Petroski & Deshaies 2005). The cullin in CRLs is post-translationally modified by covalent conjugation with the ubiquitin-like modifier Nedd8/Rub1 (Chiba & Tanaka 2004). This modification is critical for the enzymatic activity of CRLs and the CSN catalyzes the reaction to remove Nedd8 from cullin (Lyapina et al. 2001; Schwechheimer et al. 2001; Cope et al. 2002; Groisman et al. 2003; Pintard et al. 2003), thereby regulating the activity of CRLs (Fig. 2). This process is catalyzed by the metaloprotease enzyme and the reaction is called deneddylation (Cope et al. 2002).

Because the modification of Nedd8 is absolutely required for the activation of CRLs (Ohh et al. 2002), it was originally postulated that down-regulation of the CSN up-regulates CRL activity by increasing levels of neddylated cullins. However, experimental results showed that the CSN is actually required for the activation of CRLs (Schwechheimer et al. 2001; Cope et al. 2002; Doronkin et al. 2003; Feng et al. 2003; Groisman et al. 2003; Pintard et al. 2003; Wang et al. 2003). This seems paradoxical, but it is now thought that the active turnover of the Nedd8 linkage is critical for the CRLs to exert their enzymatic activity (e.g. turnover of the substrate or adaptor protein) (for details see reviews: Schwechheimer & Deng 2001; Bech-Otschir et al. 2002; Cope & Deshaies 2003; von Arnim 2003; Wolf et al. 2003).

The fifth component (CSN5/Jab1) of the CSN complex plays a central role in neddylation (Cope et al. 2002). Among the 8 subunits of the CSN, CSN5 and 6 contain the MPN domain, but CSN5 harbors a very specific type of MPN domain designated the JAb1/Mpn domain metalloenzyme (JAMM) domain (Cope et al. 2002), which is conserved and plays a critical role among the different metaloproteases (in the case of RPN11, this activity is used to remove ubiquitin from the substrate before the degradation by the proteasome (Verma et al. 2002). However, the single CSN5 polypeptide does not exhibit metaloprotease activity (Cope et al. 2002), suggesting that another CSN component, most probably the entire holo-CSN complex, is required for the deneddylase activity (Sharon et al. 2009). As the CSN’s components form different combinations (Kwok et al. 1998; Freilich et al. 1999; Oron et al. 2002; Tomoda et al. 2002; Fukumoto et al. 2005), it is important to determine which (sub)complexes possess this enzymatic activity and how the metaloprotease activity is regulated.

Although the modification of Nedd8 is critical for the activation of CRLs, down-regulation of the CSN complex does not necessarily affect the expression of all specific CRL targets equally (Wu et al. 2003; Tomoda et al. 2004; Cope & Deshaies 2006; Fukumoto et al. 2006; Harari-SteinBerg et al. 2007; Menon et al. 2007; Panattoni et al. 2008; Knowles et al. 2009), suggesting that the stability of a particular protein is regulated in multiple ways and a deficiency in one of these pathways does not lead to a total catastrophe. Alternatively, most mammalian cells may contain more than one deneddylating enzyme other than Jab1/CSN. In fact, it was showed that DEN1 exhibits deneddylating activity (Gan-Erdene et al. 2003; Mendoza et al. 2003; Wu et al. 2003; Chan et al. 2008), suggesting multiple regulators of cullin deneddylation. Furthermore, several proteins besides cullins are reported to be modified by Nedd8/Rub1, including the tumor suppressor p53 (Xirodimas et al. 2004) and ribosomal proteins (Xirodimas et al. 2008), indicating the need to think about the substrate specificity of different deneddylases. Two specific questions remain: (i) whether cullins are deneddylated by multiple deneddylases; and (ii) whether different neddylated proteins are deneddylated by different deneddylases?

The ubiquitin ligase COP1

In plants, a series of COP mutants that regulate photomorphogenesis have been identified and their genes cloned, some of which encode the components of the CSN, indicating that the CSN is the regulator of photomorphogenesis (Wei & Deng 1999). COP1 was also identified as a regulator of photomorphogenesis but its product is not contained in the CSN subunits, instead COP1 seems to be a ubiquitin ligase functioning downstream of the CSN (Fig. 2), regulating the stability of Hy5, HYH, HFR1 and LAF1 (von Arnim & Deng 1994; Osterlund et al. 1999; Yi & Deng 2005), key transcription factors controlling light-mediated gene expression. In contrast to the CSN, which exists from yeast to human/higher plants, COP1 is not found in yeast, but is highly conserved in mammals (Wang et al. 1999a). Although the overall physiological role of the mammalian COP1 is yet to be clarified, several functions have been reported. (1) COP1 seems to play a role in regulating the stability of p53 (Dornan et al. 2004) in response to DNA damage (Yoneda-Kato et al. 2005; Dornan et al. 2006). Currently, it is not entirely clear whether COP1 acts directly on p53, but it is feasible to place COP1 in the signaling pathway in response to genotoxic stress. (2) COP1 is also reported to regulate the stability of transcription factors such as c-Jun and FOXO1 (Bianchi et al. 2003; Wertz et al. 2004; Kato et al. 2008). In the case of c-Jun, COP1 forms a complex with the DET1 (de-etiolated-1)-DDB1 (DNA damage binding protein-1)-CUL4A (cullin 4A)-ROC1 (regulator of Cullins-1) complex, thereby serving as an adaptor for the specific substrate (Wertz et al. 2004). (3) Furthermore, COP1 also plays a role in the regulation of lipid metabolism by regulating acetyl-coenzyme A carboxylase through pseudokinase Tribbles 3 (Qi et al. 2006), and in insulin-modulated gluconeogenesis via the cAMP-responsive coactivator TORC2 (Dentin et al. 2007). Although a strong genetic link between the CSN and COP1 has been observed in plants, there is little evidence yet to be found in the mammalian system. However, the third component of the CSN (CSN3) seems to act upstream of COP1 in mammalian cells upon stimulation by UV (Yoneda-Kato et al. 2005). In this case, myeloid leukemia factor 1 (MLF1) interacts with CSN3 and modulates the expression and the activity of COP1 to control p53 (Fig. 2). MLF1 contains both nuclear localization signal and nuclear export signal (NES) sequences and shuttles between the nucleus and the cytoplasm (Yoneda-Kato & Kato 2008), suggesting that MLF1 regulates the intracellular distribution of COP1 (and the CSN).


The CSN was originally thought to regulate CRL activity in a negative fashion, but is actually a positive regulator of CRLs. It is true that a cycle of neddylation and deneddylation is required for the proper activation of CRLs, but the regulation is also governed by the deubiquitinating enzyme associated with the CSN. In yeast, Ubp12 is bound to the CSN, and deubiquitinates the adaptor subunits of CRL thereby stabilizing the functional CRL complexes (Wee et al. 2005). In mammals, USP15 was found to be associated with the CSN and play a role in the NF-κB pathway (Fig. 4C) (Schweitzer et al. 2007). NF-κB is a transcription factor playing an important part in many biological responses such as inflammation. NF-κB stays in the cytoplasm in a complex with its negative regulator IκBα in the absence of extracellular stimuli. IκBα is ubiquitinated by CRL SCFβ-TrCP and subsequently degraded by the proteasome in stimulated cells, which eventually leads to the nuclear translocation and activation of NF-κB (Karin & Ben-Neriah 2000). In this signaling pathway, the CSN binds to and controls the turnover of IκBα by promoting its deubiquitination through USP15. It is desired to address whether USP15 in association with the CSN functions outside of NF-κB signaling in a different biological setting.

Figure 4.

 (A) Regulation of the CSN by Ras and Bcr-Abl through MAPK and PI3-K. (B) Control of Ras-mediated transformation by the CSN. (C) Modulation of NF-κB signaling by the CSN and USP15.

Transcriptional regulation and a specificity factor

Among proteins that are reported to be associated with the CSN or its components, many are transcriptional regulators; therefore, it is important to investigate the function of the CSN in terms of transcriptional regulation (Chamovitz 2009). The fifth subunit (CSN5) of the mammalian CSN was originally identified as a transcriptional co-activator for c-Jun, and thereby designated Jun-activation-domain-binding protein 1, Jab1 (Claret et al. 1996). Since then, a number of transcription factors including E2F-1 have been reported to interact with Jab1/CSN5 (Table 1). Although it is not entirely clear how Jab1/CSN5 acts on transcription factors, the protein seems to function to select a specific factor among the members of a family (Fig. 3). For example, the Jun family contains c-Jun, JunB, JunD, etc. and interacts with the Fos family of transcription factors to form the AP1 dimeric transcription factor. Jab1 interacts with and activates c-Jun and JunD, but not JunB, thereby selecting a subset of factors among the family to be activated (Claret et al. 1996). Another example is E2F-1. The E2F family of transcription factors comprises eight members and forms a hetero-dimer with the double-positive (DP) family of transcription factors (DP-1, -2 and -3) generating the E2F transcription factor, which acts positively and negatively on cell cycle progression depending on the composition of subunits. Jab1 selectively binds to and activates E2F-1, thereby determining the specificity of the different E2F family proteins (Hallstrom & Nevins 2006). It is said that this selective binding may confer the exclusive activity of E2F-1, in that E2F-1, but not the other E2Fs (e.g. E2F-2, 3, 4, 5, 6, 7 and 8), can induce apoptosis. It is not clear how Jab1 regulates transcriptional activity. In the case of c-Jun, Jab1 seems to stimulate binding to the AP-1 site (Claret et al. 1996). But this may not apply to other Jab1-binding transcription factors. Recently, it was suggested that Jab1/CSN5 or the holo CSN complex is able to associate with chromatin (Menon et al. 2007; Mori et al. 2008). The precise mechanism involved remains to be investigated, but in the case of Jab1/CSN5, the protein interacted with the histone methyl transferase SMYD3, thereby regulating the transcription (Mori et al. 2008).

Protein phosphorylation

In the original study, during the purification of the mammalian CSN from red blood cells, Seeger et al. found that kinase activity is associated with the CSN complex, which targets c-Jun, NF-κB, IκBα and p53 as potential substrates (Fig. 3) (Seeger et al. 1998; Bech-Otschir et al. 2001). In the case of p53, a CSN-associated kinase phosphorylates threonine 155 in the middle portion of the protein, which destabilizes p53 (Bech-Otschir et al. 2001). Curcumin, one of the main ingredients of the spice turmeric, is the most effective inhibitor of the CSN-associated kinase (Henke et al. 1999), and, interestingly, is known to be antitumorigenic (Huang et al. 1995) and antiangiogenic (Arbiser et al. 1998), suggesting the CSN-associated kinase activity to be a potential target in cancer therapy (Braumann et al. 2008).

To date, several kinases have been found to be associated with the CSN complex, including casein kinase II (CK2) (Uhle et al. 2003), protein kinase D (PKD) (Uhle et al. 2003) and inositol 1,3,4-trisphosphate 5/6-kinase (5/6-kinase) (Wilson et al. 2001; Sun et al. 2002). These kinases can phosphorylate CSN-kinase substrates such as c-Jun, ATF-2 and p53, and sensitive to curcumin, but differ in their interaction with the surface of the CSN complex (the binding subunit of the CSN). It remains unclear whether these kinases play a distinctive role in responses to different stimuli and particularly, whether the CSN-associated kinases are regulated, and if so, how and when.

Subcellular distribution

The CSN has been implicated in the control of the subcellular localization of target proteins (Fig. 3). The plant CSN functions upstream of COP1 to regulate its nuclear/cytoplasmic localization (von Arnim & Deng 1994). In mammals, the fifth subunit (CSN5/Jab1) controls the nuclear export of the Cdk inhibitor p27 (Tomoda et al. 1999). Besides p27, p53 (Oh et al. 2006a; Zhang et al. 2008), Runx3 (Kim et al. 2009), 9-1-1 complex (Huang et al. 2007) and WNVCp (Oh et al. 2006b) have been reported as potential targets, whose subcellular distribution is regulated by the CSN or its subunit (mostly Jab1/CSN5). In some cases, but not necessarily all, cytoplasmic exportation is connected with the subsequent degradation of the target protein (Tomoda et al. 1999; Oh et al. 2006a,b; Huang et al. 2007; Zhang et al. 2008; Kim et al. 2009). Although Jab1/CSN5 contains a NES sequence (Tomoda et al. 2002), the physiological role of this sequence in the subcellular localization of targets remains to be clarified, and, therefore, precisely how the CSN regulates subcellular localization remains ill defined.

Biological function of the CSN in mammals

The CSN is involved in many biological responses (Wei & Deng 1999, 2003). In contrast to the plant CSN, however, the mammalian CSN seems to be more closely involved in cell proliferation and maintenance because the knockout of any CSN subunit genes (CSN2, 3, 5 and 8) in mice leads to death at a very early stage during embryonic development mostly because of defects in cell proliferation and survival (Lykke-Andersen et al. 2003; Yan et al. 2003; Tomoda et al. 2004; Menon et al. 2007). This phenotype mirrors the one found in mice devoid of the Nedd8-activating enzyme Uba3 (Tateishi et al. 2001) and cullins (Dealy et al. 1999; Wang et al. 1999b; Li et al. 2002). Together with the finding that the expression pattern of Nedd8 resembles that of the CSN (Jab1/CSN5) (Carrabino et al. 2004), these results support the notion that the CSN is the regulator of CRLs via deneddylation in an in vivo setting. However, the findings differ slightly for each knockout mouse (e.g. induction of p53), suggesting that, in addition to deneddylation governed by the holo-complex, other unique functions most probably carried out by each subunit play a role in animals.

Because the mammalian CSN is involved in the control of cell proliferation, disruption of its actions may result in proliferative disorders. In fact, aberrant expression of the CSN subunit, mostly the fifth component Jab1/CSN5, is frequently observed in a variety of cancers. Although how disregulation of the CSN leads to the development of tumors is not clear, the mammalian CSN is concerned with several important biological functions such as cell cycle regulation (see below), signal transduction (see below), checkpoint control (see below), apoptosis (da Silva Correia et al. 2007; Hetfeld et al. 2008) and autophagy (Pearce et al. 2009). Because the immune system is vital to an animal’s development and survival, the functional significance of the CSN is often investigated in the T-cell developmental system using conditional knockout mice.

CSN (Jab1/CSN5) and cancer

In a variety of human cancers, the fifth component CSN5 or Jab1 protein has been found to be over-expressed (Shen et al. 2000; Sui et al. 2001, 2006; Korbonits et al. 2002; Tsuchida et al. 2002; Esteva et al. 2003; Ito et al. 2003a,b, 2005; Kouvaraki et al. 2003; Rassidakis et al. 2003; Shintani et al. 2003; Fukumoto et al. 2004; Goto et al. 2004; Ivan et al. 2004; Osoegawa et al. 2004; Dong et al. 2005; Patil et al. 2005; Tomoda et al. 2005; Adler et al. 2006; Harada et al. 2006; Berg et al. 2007; Hsu et al. 2008a,b; Wang et al. 2008). The Jab1 gene is rarely found mutated except for some instances of gene amplification (Patil et al. 2005; Adler et al. 2006), suggesting that a mutation in the upstream regulators of Jab1 is responsible for the up-regulation of Jab1 expression and development of cancer. A high level of Jab1 is often correlated with a poor prognosis, and sometimes related to a low level of the Cdk inhibitor p27. In an experiment using various human cancer cell lines, knockdown of Jab1 inhibited the proliferation of tumor cells (Adler et al. 2006; Fukumoto et al. 2006), suggesting that over-expression of Jab1 not only serves as a marker of malignant transformation, but also actually contributes to tumorigenesis. However, how Jab1 over-expression contributes to the development of cancer remains to be investigated. The targets of Jab1 known to be involved in tumorigenesis include several tumor suppressors such as p53, p27, Smad7 and Runx3 (Table 1), and the Cdk inhibitor p16INK4a gene (Mori et al. 2008). However, the spectrum of cancers, in which Jab1 is over-expressed, is not well correlated with that of tumor suppressors, indicating other novel targets of Jab1 to be responsible for tumorigenesis.

In addition to CSN5, CSN2, CSN3 and CSN8 are also suggested to be involved in tumorigenesis, but in a different way. Based on the screening of genes able to overcome senescence, Leal et al. (2008) cloned the CSN2 gene. As senescence is part of the program to eliminate cells that contained oncogenic mutations, CSN2 is defined as a putative tumor suppressor. As another example, CSN3 functions, together with myeloid leukemia factor 1, to activate the p53 checkpoint pathway (Yoneda-Kato et al. 2005). In fact, knockdown of CSN3 facilitates the proliferation of mouse fibroblasts (Yoneda-Kato et al. 2005). A similar phonotype is observed in CSN8-knockdown cells. These results indicate CSN3 and CSN8 to be a putative tumor suppressor. However, as the CSN3 gene was amplified in osteosarcoma cells (Yan et al. 2007), CSN3 may have multiple functions. Structural analysis of the CSN shows that the CSN complex can be separated into two subdomains (Fig. 1), one containing CSN4, 5, 6, 7 and the other containing CSN1, 2, 3, 8 (Sharon et al. 2009). It is worth noting that CSN5 and CSN2/3/8 are situated in different subdomains. Therefore, the CSN may govern dual functions by manipulating two different subdomains in terms of cell transformation and proliferation.

A recent report using a genome-wide RNAi screen in Ras-transformed cells showed that a reduction in several CSN subunits (CSN3, CSN4 and CSN8) and Nedd8 leads to growth defects in tumor cells harboring Ras mutations (Luo et al. 2009), indicating that Nedd8/CSN is required for Ras-mediated transformation (Fig. 4B). Therefore, the CSN-associated activity may be a potential target in cancer therapy. In fact, Nedd8-activating enzyme was proposed as a new target for cancer treatment (Soucy et al. 2009) and curcumin derivatives (effective inhibitor of the CSN-associated kinase) were suggested as potential antitumor drugs (Li et al. 2009).

Using a more direct approach to investigate how Jab1 contributes to tumorigenesis, a stable form of Jab1/CSN5 was ectopically expressed in mice (Mori et al. 2008). The transgenic mice developed myeloproliferative disorders with expansion of the stem cell population, but the phenotype was less severe than expected, suggesting that Jab1/CSN5 collaborates with other oncogenic/tumor suppressive proteins. Actually, Jab1/CSN5 is suggested to cooperate with c-myc oncoprotein in breast cancer cells (Adler et al. 2006).

CSN in signal transduction; link to the Ras and NF-κB pathways

Because one important issue is whether and if so how the mammalian CSN is regulated by extracellular signals, it is necessary to investigate the CSN’s function in accordance with signal transduction. It is suggested that the signaling molecule Ras acts upstream of the CSN (Fig. 4A). First, as mentioned above, the CSN’s integrity is critical for growth of the Ras-transformed cells (Luo et al. 2009). Second, the Jab1/CSN5-CSN connection is regulated in an anchorage-dependent manner, which is abrogated by ras transformation (Fukumoto et al. 2005). Finally, in CML (chronic myeloid leukemia) cells, the CSN functions downstream of Bcr-Abl kinase via MAP kinase and PI-3 kinase (Fig. 4A) (Tomoda et al. 2005), both of which are mediators of ras transformation. However, precisely how (and even whether) the CSN is regulated by the ras pathway is not clear. In contrast, the NF-κB pathway seems to be more closely connected with the CSN’s actions (Fig. 4C). As mentioned above in the Deubiquitination section, the CSN targets IκBα and promotes its deubiquitination through USP15 (Schweitzer et al. 2007). As is the case with this pathway, too, it is questioned whether the CSN’s activity is regulated by TNFα.

CSN and T-cell development

Because the immune system is indispensable to development and embryogenesis, the role of the mammalian CSN was investigated using the conditional knockout procedure in T cells. Originally, Jab1/CSN5 was implicated in the functions of cytokine macrophage migration inhibitor factor (Kleemann et al. 2000) and leucocyte functional antigen 1 (LFA-1) (Bianchi et al. 2000; Perez et al. 2003). So far, two CSN subunits [CSN5 (Panattoni et al. 2008) and CSN8 (Menon et al. 2007)] have been conditionally knockedout in T cells. In both cases, conditionally knockedout animals suffered reduced proliferation, survival and ligand-induced responses of T cells. Cell cycle progression is also abrogated in CSN-deficient T cells. Defects were observed in the entry into the cell cycle from the G0 quiescent state and in the S phase progression. Cells underwent massive apoptosis at the double-negative to DP transition stage. In most cases, impaired gene expression was detected. Therefore, the CSN controls various aspects of T-cell physiology, including antigen-receptor-activated cell cycle entry and cytokine production as well as T-cell homeostasis.

CSN and DNA-damage checkpoint control

The CSN and its downstream effector COP1 regulate the checkpoint protein p53 (Dornan et al. 2004, 2006; Yoneda-Kato et al. 2005). Therefore, the CSN plays an important role in checkpoint control. However, besides acting directly on p53, the CSN is involved in multiple steps in the DNA-damage-triggered checkpoint control pathway. Ataxia-telangiectasia mutated (ATM) protein directly interacted with the CSN and phosphorylated CSN3 upon the formation of double strand breaks (Shiloh 2006; Matsuoka et al. 2007). Nucleotide excision repair proteins, such as DDB1/2 and CSA, form a complex with cullin 4A and Roc1, whose ubiquitin ligase activity in response to UV irradiation is regulated by the CSN (Groisman et al. 2003). Cdt1, a licensing factor of the prereplication complex, is rapidly proteolysed after UV- and γ-irradiation and this process is regulated by the Cul4-Roc1 ubiquitin E3 ligase and the CSN complexes (Higa et al. 2003). COP1, a downstream effector of the CSN, is regulated by phosphorylation mediated by ATM kinase (Dornan et al. 2006). Furthermore, Jab1/CSN5 mediates degradation of the Rad1-Rad9-Hus1 complex, serving as a DNA-damage sensor in checkpoint signaling and a mediator in the DNA repair pathway (Huang et al. 2007). Therefore, the CSN complex affects the DNA-damage checkpoint control pathway at multiple points.

CSN and cell cycle regulation

The CSN and its subunits are involved in cell cycle progression at many points. The first link between the CSN and the cell cycle machinery is that the CDK inhibitor p27 is regulated by Jab1/CSN5 (Tomoda et al. 1999), implying a role for the CSN in G1 progression. Discovery of the CSN as the deneddylase implied a second mechanism of p27 regulation, in which the CSN acts upstream of the ubiquitin ligase SCFSkp2 (Yang et al. 2002), that controls G1/S cell cycle regulators including p27. The subsequent findings that Skp2 forms a complex with Cul4A and DDB1 to target p27 (Bondar et al. 2006) and the CSN directly regulates Skp2 levels (Denti et al. 2006) showed that the regulatory mechanism for G1 progression is more complicated than originally imagined.

Knockout and knockdown experiments showed that reduction of the CSN causes multiple defects in the cell cycle in a tissue-specific and subunit-dependent manner. Conditional knockout of Jab1/CSN5 in T cells inhibited progression through S phase without significant effects on the G1 phase (Panattoni et al. 2008). CSN8-deficient T cells showed defects in reentry into the cell cycle from the quiescent stage (Menon et al. 2007). In this case, the CSN directly regulates the transcription of the genes encoding G1 cyclins E, D2, and D3, and CDKs 2 and 4, thereby controlling the phosphorylation of the Rb (retinoblastoma) protein. It is interesting to note that, in Drosophila, the CSN physically interacts with Rb protein to control gene expression (Ullah et al. 2007). It may be important to examine this hypothesis in a mammalian system. It is also noteworthy that, in Arabidopsis, CSN mutants (defects in CSN3, 4 or 5ab) are delayed in G2 phase progression, which accompanied induction of the DNA damage-response pathway (Dohmann et al. 2008). Recent papers reported that the CSN (CSN2 subunit) physically interacts with the components of the anaphase-promoting complex/cyclosome (APC/C, APC1, 4 and 6) (Kob et al. 2009), presenting a direct connection between the CSN and G2/M phase progression. Because the APC/C component is known to control the endocycle (Zielke et al. 2008), the CSN may regulate endoreplication as well. Thus, the CSN controls multiple steps in the cell cycle progression, dependent on tissue-specificity and subunit-specificity. It appears important to segregate the CSN subunits and the subcomplexes and analyze each’s action in a cell cycle phase-dependent manner.

Conclusion and future prospects

During the last several years, the concept that the CSN is a deneddylase has become well established. Remaining questions to be answered include is the mammalian CSN the main deneddylase? If not, how many deneddylases function in the cell? Except for cullins, how many different neddylated proteins are in the cell, and what is the substrate specificity of different deneddylases? With a better understanding of the deneddylase, we should like to know the relationship between the deneddylase and the other functions of the CSN such as transcription, phosphorylation and translocation. After the deneddylase, we may be able to ask the next three questions: (i) Whether or not the CSN is regulated, and if so, how? (ii) What is the molecular basis supporting the multifunctionality of the CSN? (iii) What is the precise mechanism regulating cell cycle progression and cell death? Answering these questions will also provide clues as to the role of the CSN in tumor development and other diseases.


This work was partly supported by Grants-in-Aid for Scientific Research and for Cancer Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan.