Roles of COP9 signalosome in cancer

Abstract

The constitutive photomorphogenesis 9 signalosome (COP9 or CSN) is an evolutionarily conserved multiprotein complex found in plants and animals. Because of the homology between the COP9 signalosome and the 19S lid complex of the proteosome, COP9 has been postulated to play a role in regulating the degradation of polyubiquitinated proteins. Many tumor suppressor and oncogene products are regulated by ubiquitination- and proteosome-mediated protein degradation. Therefore, it is conceivable that COP9 plays a significant role in cancer, regulating processes relevant to carcinogenesis and cancer progression (e.g., cell cycle control, signal transduction and apoptosis). In mammalian cells, it consists of eight subunits (CSN1 to CSN8). The relevance and importance of some subunits of COP9 to cancer are emerging. However, the mechanistic regulation of each subunit in cancer remains unclear. Among the CSN subunits, CSN5 and CSN6 are the only two that each contain an MPN (Mpr1p and Pad1p N-terminal) domain. The deneddylation activity of an MPN domain toward cullin-RING ubiquitin ligases (CRL) may coordinate CRL-mediated ubiquitination activity. More recent evidence shows that CSN5 and CSN6 are implicated in ubiquitin-mediated proteolysis of important mediators in carcinogenesis and cancer progression. Here, we discuss the mechanisms by which some CSN subunits are involved in cancer to provide a much needed perspective regarding COP9 in cancer research, hoping that these insights will lay the groundwork for cancer intervention.

Functions of the COP9 Signalosome (CSN): Ubiquitin-Mediated Protein Degradation

The COP9 signalosome (CSN) is an evolutionarily conserved multiprotein complex found in plants and animals. This protein complex, which consists of eight subunits (CSN1 to CSN8), is first characterized as a repressor of plant photomorphogenesis.1–4 Photomorphogenesis is regulated by expression of light-activated genes, resulting in inhibited hypocotyl elongation and the absence of an apical hook during seedling process. COP9 employs its associated deneddylation activity toward cullin-RING ubiquitin ligases (CRL), thereby coordinating CRL-mediated ubiquitination activity. While mammalian cells have CSN, mammals don't do photomorphogenesis. The role of mammalian CSN remains largely elusive, although CSN is implicated in a wide variety of regulatory processes, including cell cycle control, transcriptional activation4,5 and tumorigenesis.6,7 It has an emerging role in cancer. However, the mechanistic regulation of each subunit in cancer is still not well-characterized.

Mammalian CSN 1, 2, 3, 4, 7, 8 each contains a PCI (proteasome, COP9 signalosome, translation initiation factor) domain, which may serve as a scaffold in the assembly of the COP9 signalosome. CSN6 and CSN5 are the two subunits that each contain an MPN (Mpr1p and Pad1p N-terminal) domain.8–11 Mpr1 and Pad1 are yeast proteins of yeast proteasome subunits. The MPN domain is shared by subunits from three protein complexes: the proteasome (Rpn 11 and 8), COP9 signalosome (CSN5 and Can6), and translation initiation factors (eIF3 p47 and p40).4,12,13 The MPN domain of CSN5 is involved in regulating cullin deneddylation.11 The MPN domain of CSN6 is homologous to that of CSN5, but its function remains unknown. On the basis of size-exclusion chromatography and immunolocalization data, it has been suggested that the different subunits or subcomplexes of COP9 signalosome can possess distinct functions.14 In yeast genetic studies, neither csn4- nor csn5-null mutants have phenotypes similar to those of csn1- and csn2-null mutants, suggesting that different subunits of the COP9 signalosome mediate distinct functions.14

The well-characterized functions of the CSN are their roles in ubiquitin-mediated protein degradation.15–18 And these activities are manifested through regulating cullin-RING ligases, RING-containing ubiquitin ligases and deubiquitination. SCF (Skp1-cullin-F-box) ubiquitin ligases target many proteins for ubiquitin-mediated degradation.19 The C terminus of cullin is required for interaction with ROC1 (a RING protein) and facilitates the assembly CRLs, while the N terminus of cullin is required for associating with Skp1, which can recruit a substrate adaptor F-box protein. Importantly, cullin is covalently modified by an ubiquitin-like protein Nedd8 (neural precursor cell expressed developmentally downregulated gene 8).20 Modification of cullins with Nedd8 (neddylation) is important for the activity of CRLs.21 The removal of Nedd8 from cullin (deneddylation)22 is catalyzed by CSN5 due to its very important metalloprotease activity.11 As mentioned above, CSN6 and CSN5 are the two subunits that each contain an MPN domain. CSN5 has the MPN domain containing a JAMM (JAB1/MPN/Mov34) motif linked to the metalloprotease motif (EXnHXHX10D) that causes cullin deneddylation, a process important for regulating CRL activity.15,18,21,22 Sequence alignment between CSN5 and CSN6 reveals that CSN6 has the MPN domain but without a JAMM motif.23 A recent mass spectrometry study shows that CSN6 is the one subunit that directly interacts with CSN5.24 It raises the question of whether both CSN6 and CSN5 are involved in cullin deneddylation. In addition to CRLs, CSN can regulate RING-containing ubiquitin ligases, such as MDM26,25 and constitutively photomorphogenic 1 (COP1),26 functioning as an important platform for mediating functional interactions between proteasome and various ubiquitin ligases. This aspect will be further discussed in this review. As for deubiquitination, CSN is known to associate with deubiquitination enzyme and thus protect certain protein stability. For example, CSN-associated deubiquitinylase USP15 can deubiquitinylate IκBα, thereby regulating NFκb signaling.27 Thus CSN is a platform to regulate the protein stability of certain proteins depending on the context of intracellular signaling.

CSN is Overexpressed in Cancer: Potential Oncogene

Notably, the eight subunits of the COP9 signalosome are each paralogous to one of the eight subunits that form the lid complex (19S) of the 26S proteasome.28,29 The lid complex can recognize ubiquitinated substrates and then funnel them into the proteolytic core complex for degradation. Given this characteristic, the COP9 signalosome does play a role in ubiquitin-mediated protein degradation, including many tumor suppressor or oncogene products.7 For examples, COP9 signalosome is implicated in ubiquitin-mediated proteolysis of p27,30,31 p53,6,32 Mdm2,6,25 Smad7,33 Runx3,34 Id1,35 Skp236 and HIF1.7,37 Analyzing human cancer patient transcriptomic data sets from Oncomine38–40 and Gene Expression Omnibus revealed that many types of cancer have CSN6 or CSN5 overexpression (Fig. 1). It remains to be determined whether other subunits have a role in cancer. Only few examples are documented and could sometimes be important for tumor suppression. For example, CSN3 is involved in myeloid leukemia factor 1-meditaed growth arrest,41 and CSN3 deficiency impairs p53 activation, facilitates the cell proliferation and affects COP1-mediated p53 degradation.41 However, this seems to be controversial because Csn3-knockout mouse embryos have increased cell death (Table 1), and lower expression of CSN3 correlates with better survival in patients bearing tumors with a Ras signature.42 In another interesting case, CSN2 overexpression can lead to VEGF production43 and prolong the stability of cyclin A by affecting anaphase-promoting complex. CSN2 overexpression leads to chromosome instability,44 but Csn2 gene is found lost in several types of cancer.45 Csn4 knockdown leads to down-regulation of Skp2, which is involved in p27 degradation.36,46,47 Thus, it is obvious that the COP9 signalosome subunits have emerging roles in cancer. Here, we will focus on some of the recent discoveries.

CSN5 overexpression.

CSN is known to play an essential role in the signaling processes that control many aspects of plant and Drosophila development.3,48 Only recently has the biological function of each subunit in mammalian cells begun to be characterized. CSN5 is overexpressed in many types of cancer (Fig. 1 and reviewed in ref. 49). In breast cancer, CSN5 overexpression correlates with downregulation of p27 in invasive breast cancers.50,51 CSN5 is responsible for deneddylation of cullin-based ubiquitin ligases, and this activity is important for promoting cell transformation.52 However, the Csn5 gene is rarely mutated, and it remains to be determined whether an upstream regulator is responsible for the overexpression of CSN5. For instance, CSN5 can collaborate with Myc in promoting cell invasion,53 but it is not clear if Myc can regulate CSN5 expression. Also, it remains to be determined whether a myriad of other CSN5 targeted substrates can play roles in carcinogenesis and/or tumor progression.

CSN6 overexpression.

CSN6's role in cancer was poorly documented. Interestingly, CSN6 interacts with human immunodeficiency virus type-1 (HIV-1) viral protein R (Vpr), which, in turn, regulates CSN6's subcellular localization.54 Expression of Vpr or antagonizing the expressing of CSN6 results in the blockade of cell proliferation and accumulation of cells in the G₂/M phase of the cell cycle.54,55 Consequently, Vpr prevents cell proliferation of infected cells and collaborates with the matrix protein to enable HIV-1 to enter the nucleus of the nondividing cells. Vpr can interact with p5356 and modulate the transcriptional activity of p53;55 however, it is not clear whether CSN6 is involved in such a process. It is noteworthy that cells treated with antisense CSN6 were growth inhibited,54 suggesting that CSN6 plays an important role in cell cycle and has growth promoting activity. Based on the observations discussed above, the following questions are raised: Can CSN6 antagonize cell cycle regulators such as p53 to promote cell growth? Does CSN6 promote degradation of p53? If CSN6 is linked to p53 activity, what is the correlation between CSN6 and p53 during tumorigenesis? These questions are addressed by Zhao et al.6 (see discussion below).

Our recent study has employed the System for Integrative Genomic Microarray Analysis (SIGMA),57 to evaluate the genetic loss or gain of CSN6 (located at 7q22.1) using the data from the BCCRC SMRT arrays.6 High-resolution comparative genomic hybridization analysis of a different set of breast cancer cell lines and tumor samples reveals a high percentage of samples that have amplification of the genomic region where CSN6 resides.6 The gene amplification of CSN6 in breast cancer samples is also experimentally confirmed using quantitative PCR.6 Significantly, amplification of CSN6 gene is detected in a high percentage of breast cancer samples, and there is a positive correlation between CSN6 gene copy numbers with the breast tumor size. Also, when malignant follicular thyroid carcinomas were compared with benign thyroid lesions (follicular adenomas, adenomatous nodules and multinodular goiters) and normal thyroid tissue, the malignant follicular carcinomas expressed higher levels of CSN6 than benign thyroid lesions/tissues.6 Again, analyzing human cancer patient transcriptomic data sets from Oncomine demonstrates that in addition to breast cancer, other types of cancer have CSN6 overexpression (Fig. 1). These results indicate that overexpression of CSN6 in cancer is a common phenomenon and is not restricted to a few specific types or cases of cancer.

Physiological Significance of CSN Overexpression in Cancer: Affecting p27 and p53

The consequence of CSN overexpression in cancer could lead to dysregulation of several important products of oncogene or tumor suppressor gene. They are discussed below.

CSN5 causes p27 degradation.

It is shown that CSN is involved in degrading proteins such as CDK inhibitor p27.30 p27 is frequently downregulated in human cancers.58 Its function is to protect normal cells from undergoing abnormal cell proliferation. Stabilization of p27 is essential for the maintenance of its tumor suppressor function. Therefore, deregulation of proteins that affect p27 protein stability is expected to have an impact on human tumorigenesis. Several studies suggest a link between CSN5 overexpression and p27 protein downregulation in cancers.50,59 It is known that p27 is a nuclear protein with a short half-life,60 and its stability is important for its biological function.58 For degradation, p27 is ubiquitinated by SCF^Skp2 ubiquitin ligase consisting of Skp1, cullin, ROC1 and a distinct F-box protein, Skp2, that acts as substrate-recognition component to ubiquitinate phosphorylated p27.47,61,62 It is shown that mammalian COP9 signalosome collaborates with SCF^Skp2 to regulate p27 degradation through ubiquitination.36 CSN5 was the first CSN subunit characterized to be involved in regulating the exportation of p27 to the cytoplasm.30 Modified p27 proteins, which are deleted at the CSN5 binding domain, are resistant to CSN5-mediated degradation and can inhibit HER2-activated cell growth, CDK2 activity, cell proliferation and transformation efficiently.63 In our previous studies, HER2 signaling affects CSN5 subcellular localization to mediate p27 degradation.64,65 Precisely how the HER2 regulates subcellular localization of CSN5 remains to be defined. So far, whether other COP9 subunits regulate p27 remains unknown.

CSN causes dysregulation of Mdm2-p53 axis.

Cells respond to DNA damage by activating a complex DNA-damage response pathway that induces cell cycle arrest and transcriptional and post-transcriptional activation of a set of genes. This process will lead to activation and stabilization of the p53 tumor suppressor. Stabilized p53 will stop the cell cycle, allow time for repair and prevent mutations from being propagated. It is also very important for maintaining the integrity of the genome. Given that p53's function is lost in almost half of all human cancers66 and that stabilization of p53 is important for its tumor suppressor function, there is always an urgent need to know how p53 levels are regulated.

Ubiquitin-mediated protein degradation is one well-known pathway for p53 regulation. Ubiquitination of p53 regulates p53 concentrations in the cell to maintain its effects on tumorigenesis and normal cell growth. Mdm2, a RING domain-containing p53 ubiquitin ligase, is known to bind and regulate p53 ubiquitination.67 Ubiquitinated p53 is degraded by 26S proteasome. In response to DNA damage, activated ATM (ataxia telangiectasia mutated) can phosphorylate p53 or Mdm2.68 ATM-mediated phosphorylation of p53 interferes with its association with Mdm2,69 and phosphorylation of Mdm2 inhibits Mdm2 RING domain oligomerization required for p53 ubiquitination,68,70 thereby stabilizing the p53 protein and activating its transcriptional activity. Stabilized p53 can activate downstream genes, such as p2171 and 14-3-3σ,72,73 which, in turn, inhibit cell cycle progression. Recently, the in vivo biochemical pathways of two MPN-containing CSN subunits involved in regulating p53 levels have been characterized. These two CSN proteins have been shown to regulate p53 ubiquitination, adding compound layers of complexity involved in tight regulation of p53 levels.

CSN5 leads to Mdm2 stabilization and p53 degradation. The finding that embryos of CSN5-null mice have an accumulation of p53 suggests that this protein is a crucial regulator of p53 stability.74 Another observation from an experiment of CSN-associated kinase inhibitor cur-cumin also suggests the CSN5-p53 link.25 Curcumin, a yellow plant pigment with antiproliferative and apoptotic activities, is a potent suppressor of tumor formation. Although curcumin is used in phase I clinical trials for the treatment of cancer, the mechanism by which curcumin mediates tumor suppression remains not well-understood. It has been shown that curcumin can mediate downregulation of both CSN5 and MDM2, which leads to subsequent p53 stabilization, offering new clues as to how curcumin may have tumor suppressive activities.25 The curcumin studies suggest that CSN5 is a critical regulator of both p53 and MDM2. Further biochemical studies indicate that CSN5 associates with p53, and that CSN5 expression leads to p53 degradation, facilitating MDM2-mediated p53 ubiquitination and promoting p53 nuclear export.25 A CSN5 fragment (1–222 aa) containing MPN domain is sufficient to downregulate p53,25 but it remains to be determined how the MPN is involved in this process. CSN5-mediated p53 degradation is MDM2-dependent, because CSN5 has no impact on p53 degradation in an MDM2-null background.25 Furthermore, CSN5 expression results in stabilization of MDM2 through reducing MDM2 self-ubiquitination, thereby decelerating turnover rate of MDM2. It remains unclear how CSN5 reduces MDM2 self-ubiquitination. It is possible that CSN5 has impact on some of the regulators of MDM2, such as ARF,75 PML,76 which will result in MDM2 stabilization. Interestingly, CSN5 can mediate the nuclear export of p53,25,77 which, in turn, will facilitate the degradation of p53. Further, a CSN5 expression construct that contains just MPN I-VI domain (1–222 aa), can mediate p53 cytoplasmic location. In contrast, a construct encoding MPN I-III region (1–120 aa) is inefficient in affecting the subcellular localization of p53.25 Given that CSN5 can stabilize MDM2, a known protein involved in p53 nuclear export, it is conceivable that CSN5 expression leads to the translocation of p53 from the nucleus to the cytoplasm due to its impact on MDM2. Alternatively, CSN5 (120–222 aa) may be directly involved in p53 nuclear export, which requires further investigation. The biological significance indicates that CSN5-mediated MDM2 stabilization leads to antagonizing the transcriptional activity of p53.25 These results demonstrate that CSN5 is a pivotal regulator for both p53 and MDM2.

CSN6 impacts on negative regulator of p53 to degrade p53. MDM2 is frequently amplified/overexpressed in many different types of malignancies.78 Overexpression of MDM2 leads to the accelerated degradation of p53 protein in cancers, thereby dampening the tumor suppressor function of p53. The amplification/overexpression of MDM2 is obviously important in tumorigenesis,78 but it remains unclear why some malignancies have high MDM2 expression at protein levels that are not due to gene amplification, increased transcription or enhanced translation.78 Recent studies show that CSN6 overexpression positively correlates with MDM2 protein overexpression in human breast cancer sample studies.6 The data also show that CSN6 gene amplification was found without coexisting MDM2 gene amplification and p53 mutation/deletion,6 indicating that overexpression of CSN6 leads to increased stability of MDM2.

Biochemical studies indicate that CSN6 interacts with MDM2 directly. Interestingly, CSN6 is mapped to bind to MDM2 at a region (294–384 aa) containing a conserved C4 zinc finger where ARF75 and ribosomal proteins L5,79 L1180 and L2381,82 bind to inhibit the E3 ligase activity of MDM2.6 Whether CSN6 antagonizes these negative regulators through binding at the zinc finger region remains to be investigated. However, it is shown that Lys364 is the MDM2 autoubiquitination site through a mass spectrometry analysis.6 It happens that Lys364 is located within the interaction region (294–384 aa) between CSN6 and MDM2. Further evidence shows that MDM2 K364R mutant has dramatic reduction of autoubiquitination levels in both in vitro and in vivo ubiquitination assays.6 Also, the K364R MDM2 mutant has a longer half-life than wild-type MDM2.6 This observation strongly suggests that the interaction between CSN6 and MDM2 blocks the autoubiquitination and degradation of MDM2 (Fig. 2). It is important to point out that CSN6-mediated decreased MDM2 ubiquitination is not through enhancing the deubiquitinating enzyme activities of COP9-associated deubiquitinase USP1583 and MDM2 deubiquitinase USP2,84 because CSN6 still has the capability to stabilize MDM2 regardless of the depletion of either USP2 or USP15.6 This result strongly suggests that CSN6-mediated stabilization of MDM2 is independent of USP2 or USP15.

Given that CSN6 interacts with and positively regulates MDM2, this important regulation has been translated into negative impact of CSN6 on p53. Increase of CSN6 reduces steady-state levels of p53 in a dose-dependent manner. CSN6 accelerates MDM2-mediated p53 ubiquitination in a dose-dependent manner. CSN6's positive impact on MDM2 has been translated into enhancing p53 degradation and further antagonizing p53 transcriptional activity. CSN6 impairs the transcriptional activity of p53.6 Accordingly, the gene expression of p53 transcriptional targets, including p21, 14-3-3σ, BAX and PUMA, is downregulated in cells overexpressing CSN6 gene.6 Together, CSN6's direct binding on MDM2 and subsequent blocking of MDM2 ubiquitination contribute to the function of the MDM2-p53 axis.

As another E3 ligase for p53, COP1 drives the ubiquitination and proteasomal degradation of p53, thereby maintaining low steady-state levels of p53 in unstressed cells.85 Recently, CSN6 was found to associate with COP1 and is involved in 14-3-3σ ubiquitin-mediated degradation.26 CSN6 associates with COP1 endogenously, and an in vitro binding assay confirms that CSN6 directly binds to COP1.26 CSN6 expression actually leads to stabilization of COP1 through reducing COP1 self-ubiquitination and decelerating COP1's turnover rate.26 Mechanism studies indicate that CSN6 increases COP1 stability through inhibition of ubiquitin-mediated COP1 proteasomal degradation. How CSN6 prevents COP1 self-ubiquitination remains to be studied. It is possible that the CSN6-COP1 axis is driving p53 ubiquitination by (1) the CSN6-COP1 axis causing 14-3 3σ degradation, thereby blocking 14-3-3σ's positive effect on p53 stability; (2) CSN6 regulating COP1 in a positive way to directly enhance COP1-mediated p53 ubiquitination and degradation.

Other CSN-associated kinases and p53 degradation. It is important to point out that several kinases, including inositol-1,3,4-trisphosphate 5/6-kinase (5/6-kinase),86 protein kinase D (PKD) and CK2,87 can be copurified with the COP9 signalosome and are implicated in phosphorylating p53 and subsequent p53 degradation. Interestingly, the protein kinase activity of these CSN-associated kinases can be inhibited by curcumin.32 Under this inhibition, p53 protein levels were elevated in the presence of curcumin.32 Nonetheless, the mechanism by which these different phosphorylations on p53 can lead to p53 destabilization remains to be elucidated. It is not clear whether the known p53 ligases, such as MDM2, COP1 and Pirh2,88 are involved in this phosphorylation-dependent p53 degradation. Recently, a new p53 ubiquitin ligase (Kelch domain-containing F-box protein) JFK is involved in this phosphorylationdependent p53 degradation process.89,90 CSN-associated kinase can cause p53 phosphorylation and is important in JFK-mediated p53 ubiquitination and degradation. Importantly, CSN5 knockdown also impairs JFK-mediated p53 degradation. It appears that CSN5 is involved in this regulation, but it raises more questions regarding the cross-talk between CSN5, JFK and CSN-associated kinases, and how this crosstalk is regulated.

Oncogenic Activity of CSN Subunits: In Vivo Evidence

Although several subunits of CSN are overexpressed in cancer, it is important to demonstrate their oncogenic activity in mouse models to illustrate their roles in cancer. It is shown that CSN5 isopeptidase activity (located in MPN domain) is critical for transformation and breast cancer formation.52 A CSN5 D151N mutant that alters one of the conserved D residues coordinating the JAMM⁺ domain activity and that has dominant-negative activity causes the reduction of tumor cell proliferation and lower tumor grade.52 Thus, CSN5's isopeptidase activity is required for CSN5-mediated tumorigenicity. Another transgenic mice study shows that CSN5 overexpression could lead to myeloproliferative disorder and reduce the expression of p16 ink4a,91 suggesting its role in tumor development. CSN5 also associates with Myc and potentiates Myc transcriptional activity in breast cancer.53 Paradoxically, it is important to point out that, in this case, CSN5 is negatively affecting the stability of Myc.53 The detailed regulation between Myc and CSN5 warrants further investigation.

The role of CSN6 in tumorigenesis was also investigated by using the xenograft cancer model. Given the important role of CSN6 in MDM2-p53 axis, a wild-type p53-containing cancer cell line was used to address whether overexpression or downregulation of CSN6 could affect the tumor formation process.6 CSN6 overexpression in cells leads to increased cell foci formation (transformation), tumor growth rate and tumor weight.6 As expected, knockdown of CSN6 using shRNA decreased the cell foci formation, tumor growth rate and tumor weight compared with control shRNA.6 Significantly, immunohistochemistry staining results indicate that there are more MDM2-positive staining signals in the CSN6 overexpression tumors than the vector controls, while less p53 signals among the CSN6 overexpression tumors.6 In summary, CSN6's functions as an oncoprotein in tumorigenesis through regulating MDM2 and p53 protein levels can be recapitulated in mouse xenograft models. It is not known whether the MPN domain of CSN6 is critical in the tumor promoting process. Obviously, the biological function of the MPN domain in CSN6 remains to be further determined, as its polar residues that resemble the active site residues of metalloprotease in other MPN domain92 can be involved in proteasome-associated deneddylation activity22 or other types of uncharacterized activity. It remains to be elucidated whether roles of other PCI-containing CSN subunits have impacts on cancer formation using animal models.

CSN Knockout: Developmental Defects

Genetic studies are an important means for understanding the molecule's biological function. Recently, several studies have demonstrated that several mammalian CSN subunits are involved in developmental process. For example, targeted disruption of CSN2, CSN3, CSN5, CSN8 and CSN6 resulted in defected embryo development6,74,93–95 (see Table 1). Disruption of these genes all leads to early embryonic lethality. In most of the cases, the abnormal apoptosis effect is observed in these embryos (Table 1). The apoptosis is accompanied with elevated p53 and/or p21. But the detailed mechanistic regulation remains largely unknown. We recently performed targeted disruption of Csn6 gene in mice and have characterized the phenotypes of these mice.6 In the case of csn6 target gene disruption, the deregulated MDM2-p53 axis is characterized. This animal model is valuable for studying its biological function further. The embryonic lethality of Csn6-null mice was addressed further by crossing with p53-null mice, and loss of p53 rescued the embryonic lethality of Csn6-null embryos from E7.5 to E10.5. This degree of partial rescue still underscores the importance of CSN6 in the MDM2-p53 axis. It is possible that CSN6 could have other important targets required for the mouse embryogenesis other than MDM2 and p53. Thus, the embryonic lethality due to loss of csn6 may not be compensated by depletion of p53 only. This scenario deserves further study. Although both CSN5 and CSN6 are the only two MPN domain-containing subunits in the COP9 signalosome, it appears that they cannot compensate for each other in early embryo development. Because knocking out either csn5 or csn6 gene still causes lethality, and the embryo dies almost at the same embryonic stage (E7.5), this suggests that these two genes are not functionally redundant.

It is important to point out that partial loss-of-function of CSN6 created by anti-sense strategy results in diverse developmental defects in plants, such as homeotic organ transformation and symmetric body organization change,8 attesting that the importance of Csn6 gene in development is evolutionally conserved. Given that embryonic lethality hinders further investigation of the in vivo physiological significance of CSN subunits, it is important to address the problems with conditional gene targeting strategy. CSN8 and CSN5 are two successful examples that address the roles of CSN in hepatocytes and T-cell development.95,96

CSN and DNA Damage Response: Important Mediator

It was revealed that CSN has a role in regulating DNA damage repair. For example, Csn5 deficiency during meiosis can lead to activation of a DNA damage checkpoint in D. melanogaster.48 Also, COP9 signalosome can regulate ubiquitin ligase activity of the damaged DNA binding protein 1&2 (DDB1&2) and Cockayne syndrome group A (CSA) complexes in response to UV irradiation.97 CSN forms complexes with Cul4a, Roc1, ddb2 or CSA to regulate their associated ubiquitin ligase activity.97 In such a regulation, CSN is required for global genome repair and transcription couple repair. In addition, Cdc10-dependent transcript 1 (CDT1), a licensing factor of the pre-replication complex (preRC), is regulated by CSN.98 CDC6 and CDT1 associate with Origin Recognition Complex and promote loading of the MCM2–7 proteins onto chromatin to assemble the preRC. CTD1 is degraded after DNA damage, and CSN1 or CSN5 deficiency leads to accumulation of CDT1. These results indicate the critical role of CSN in DNA damage response. CSN can also use its associated deubiquitination enzyme to do the DNA damage repair.83,99 An elaborated review in this aspect has been documented.100

Csn6 Haploinsufficiency Attenuates p53-Mediated DNA Damage Response

Although it is indicated that DNA damage can relay its signal to regulate COP9 signalosome and targets, it remains not well-characterized how this process can regulate the activity of p53, one of the most important regulators in response to DNA damage. As for CSN6, given its negative role in p53 stability, csn6 haploinsufficiency causes increased DNA damage-mediated apoptosis. p53 has an indispensable role in γ-IR-induced apoptosis in mouse thymocytes.101,102 Remarkably, Csn6^+/− thymocytes demonstrate more susceptibility to γ-IR-induced apoptosis,6 indicating that reduced expression of CSN6 sensitized these cells to p53-dependent apoptosis after DNA damage. Accordingly, the protein levels of p53, PUMA and specifically cleaved PARP (p85) are dramatically increased in Csn6^+/− thymocytes.6 Taken together, these data suggest that CSN6 plays an important role in regulating p53, thereby contributing to p53-dependent apoptosis in response to DNA damage. p53-mediated apoptosis is involved in lethality of high-dose γ-IR.103 Results also show that Csn6^+/− mice are more sensitive to high-dose γ-IR than Csn6^+/+ mice in terms of survival. This observation strongly supports that accumulation of p53 in Csn6^+/− mice sensitized cells to γ-IR-induced apoptosis, which, in turn, increases the lethality caused by high-dose γ-IR.

Loss of CSN6 and Tumorigenicity

It is known that p53 plays an important role in suppressing carcinogenesis after low-dose γ-IR exposure.104 Haploinsufficiency of Csn6, which increases p53, can attenuate tumorigenesis in vivo. Three cohorts of Csn6^+/−, Csn6^+/+ and p53^+/− mice were treated with a single sub-lethal dose (4.5 Gy) of γ-IR. Both irradiated Csn6^+/− and Csn6^+/+ mice took longer to develop malignancies than p53^+/− mice. Analysis of tumors from litter-mates of two genotypes (Csn6^+/−; Csn6^+/+) revealed that there were fewer neoplasms per mouse in Csn6^+/− mice than in Csn6^+/+ mice.6 Atypical lymphoid proliferation was observed in various tissues of Csn6^+/− mice, while high-grade lymphomas were typically observed in Csn6^+/+ mice. Csn6^+/− mice showed no sarcomas, malignancies that positively correlated with the overexpression of MDM2,67 in contrast to a high percentage incidence of sarcomas (angiosarcoma and osteosarcoma) in Csn6^+/+ mice. Kaplan-Meier analysis showed that both Csn6^+/+ and Csn6^+/− mice had significantly better survival than p53^+/− mice, and the survival of Csn6^+/− mice was better than that of Csn6^+/+ mice, consistent with a possible dosage effect of p53 level (i.e., relative p53 levels: Csn6^+/− > Csn6^+/+ > p53^+/−) on tumor-specific survival.6 These results indicate that Csn6 haploinsufficiency inhibits γ-IR-induced tumorigenesis. In summary, loss of one copy of Csn6 can confer tumor suppressive activity by increasing the level of p53. These intriguing observations suggest that CSN6 may be a potential drug target for cancer prevention or therapy.

Conclusion

In this review, we have discussed roles of COP9 in promoting cancer, particularly focusing on recent findings about CSN5 and CSN6. The potential involvement of COP9 in cancer could be linked to its role in ubiquitin-meditated protein degradation activity. Nonetheless, CSN's roles in regulating cullin-based or RING domain-containing ubiquitin ligases remain to be fully elucidated.105 The roles of PCI-containing COP9 subunits in cancer remain enigmatic.

Although it is not fully investigated, some of the discussed regulatory pathways could be shared by some COP9 subunits. For example, CSN5 and CSN6 may be able to potentiate the activity of MDM2 in a similar way, including blocking MDM2 self-ubiquitination. However, it is important to point out that, although the MPN domain of CSN6 is 70% homologous to that of CSN5, which is involved in regulating cullin deneddylation,11 the function of the CSN6 MPN domain has yet to be investigated. Is the CSN6 MPN domain also involved in regulating cullin neddylation? The deneddylation activity of CSN5 seems important for its transformation activity. However, can this deneddylation activity have an impact on MDM2 stability? Does CSN6 also regulate p27? If so, how can this be achieved? On the basis of size-exclusion chromatography and immunolocalization data, it has been suggested that the different subunits or subcomplexes of COP9 signalosome can possess distinct functions.14 At issue is whether we can dissect the role of each CSN subunit without considering the fact that they are part of a holocomplex or sub-complexes? It is possible that functions may be assigned to certain COP9 subcomplexes with different compositions.

Here, we have focused on the COP9 subunits that participate in p53 ubiquitination, including CSN5 and CSN6. Importantly, p53 mutation or deletion is found in half of all human cancers,66 and the defects in the other half remain to be characterized. For cancers with wild-type p53 genotype, it is possible that different genes involved in p53 degradation are deregulated. Since the CSN-p53 axis is required in response to DNA damage,6 dysregulation of COP9 subunits (e.g., CSN6 amplification/overexpression) that cause p53 degradation may contribute to some of the cancers that do not have p53 mutations or deletions.

p53 activity is highly regulated by modification, including phosphorylation, ubiquitination, sumoylation and neddylation. Little is known regarding how these modifications are regulated by COP9. As a subunit with a JAMM motif to deneddylate cullin, can CSN5 affect p53 neddylation directly? As for the phosphorylation event, the COP9-associated kinases seem to add another layer of regulation on p53 stability. However, why are several kinases involved? Are these kinases also regulated by DNA damage? Could these kinases regulate p53 ubiquitin ligases, such as JFK, COP1, Mdmx70,106 or MDM2, to participate in p53 degradation? Given the fact that COP9 negatively regulates p53 activity in response to DNA damage, these regulations warrant future investigation. We propose that COP9 subunits, such as CSN5 and CSN6, are potential targets for designing drugs to restore proper p53 activities in cancers with a wild-type p53 genotype. In the future, we expect that more targeted gene disruption studies of other COP9 subunits or transgenic mice constructions will facilitate understanding of the function of COP9 in cancer. A comprehensive understanding of the role of CSN in cancer will speed up the discovery of strategies for cancer therapy.

Figures and Tables

Figure 1 Transcriptomic analyses of CSN5 and CSN6 overexpression in human cancer patients. Human cancer patient data sets were obtained from Oncomine and Gene Expression Omnibus. Data were analyzed using Oncomine analysis tools and Nexus Expression 2.0. Only patients with more than 40% increase in CSN5 (A) or CSN6 (B) mRNA levels compared with normal tissues were scored as “CSN5 overexpression” or “CSN6 overexpression,” respectively. N represented the total number of patients analyzed in each type of cancer.

Figure 2 A model for the role of CSN6 in regulating MDM2-p53 axis. MDM2 is self-ubiquitinated at K346, which leads to fast turnover. CSN6 binds to MDM2 at a region 294–384 aa, potentially blocking MDM2-mediated self-ubiquitination at K346 site. Thus, MDM2 is more stable and can enhance p53 ubiquitination, which, in turn, will block p53-mediated biological functions.

Table 1 Targeted gene disruption studies of COP9 signalosome gene

Gene name	Phenotypes	Functional studies/references
CSN2	embryonic lethal; embryos die at E3.5; heterozygous mice survive well.	Abnormal elevation of cyclin E in csn2^−/− embryos; csn2^−/− embryos have elevated levels of p53 and p21.93
CSN3	embryonic lethal; embryos die at E8.5; heterozygous mice survive well.	csn3^−/− embryos have increased cell death but no cell proliferation.94
CSN5	embryonic lethal; embryos die at E8.5; heterozygous mice survive well; act as a regulator of p27, p53 and cyclin E.	csn5^+/− MEF demonstrated impaired cell growth; transgenic expression of csn5 rescues the csn5^−/− embryonic lethality; csn5 transgenic mice develop myeloproliferative disorders.74,91
CSN6	embryonic lethal; embryos die at E7.5; heterozygous mice survive well; act as a negative regulator of p53 and positive regulator of MDM2.	csn6-deficiency enhances p53-mediated apoptosis; loss of csn6 attenuates IR-induced tumorigenesis; embryonic lethality is partially rescued in p53 null background.6
CSN8	embryonic lethal; embryos die at E7.5; heterozygous mice survive well; E7.5 embryo demonstrated retarded growth and differentiation.	Loss of csn8 impairs peripheral T cell homeostasis; csn8-deficient T cells demonstrate compromised TCR-dependent response.95

Acknowledgments

We would like to thank the Susan Komen Breast Cancer Foundation and NIH grant (RO1CA 089266 to M.H. Lee) for research support. We apologize to our many colleagues whose work that we were not able to cite due to space constraints.