COP1D, an alternatively spliced constitutive ... - Nature

Oncogene (2008) 27, 2401–2411

& 2008 Nature Publishing Group All rights reserved 0950-9232/08 $30.00 www.nature.com/onc

ORIGINAL ARTICLE

COP1D, an alternatively spliced constitutive photomorphogenic-1 (COP1) product, stabilizes UV stress-induced c-Jun through inhibition of full-length COP1 MG Savio1,3, G Rotondo1,3, S Maglie1, G Rossetti1, JR Bender2 and R Pardi1 1

Unit of Leukocyte Biology, Vita-Salute San Raffaele University School of Medicine, DIBIT-Scientific Institute San Raffaele, Milano, Italy and 2Raymond and Beverly Sackler Foundation Cardiovascular Laboratory, Section of Cardiovascular Medicine, Yale University School of Medicine, New Haven, CT, USA

COP1 is an evolutionarily conserved RING-finger ubiquitin ligase acting within a Cullin-RING ligase (CRL) complex that promotes polyubiquitination of c-Jun and p53. Stability of the above substrates is affected by posttranslational changes priming the proteins for polyubiquitination and proteasome-dependent degradation. However, degradation of both substrates is controlled indirectly by signaling pathways affecting the E3 ligases involved in their polyubiquitination. Here, we report the identification of COP1D, a ubiquitously expressed splice variant of COP1 lacking a portion of a coiled-coil region involved in intermolecular associations. While being unable to associate with other components of the CRL complex, COP1D exerts a dominant-negative function over the full-length protein, due to its ability to heterodimerize with COP1 and sequester it from the enzymatically active complex. Ectopic expression of COP1D antagonizes the function of COP1, while its selective downregulation by RNA interference promotes more efficient degradation of c-Jun and p53 by the full-length protein. The COP1/COP1D mRNA ratio is modulated by UV stress and a decreased COP1/COP1D ratio correlates with elevated c-Jun, but not p53 protein levels in invasive ductal breast cancer. Thus, dynamic changes of the COP1/COP1D ratio provide an additional level of regulation of the half-life of the substrates of this E3 ligase under homeostatic or pathological conditions. Oncogene (2008) 27, 2401–2411; doi:10.1038/sj.onc.1210892; published online 29 October 2007 Keywords: AP-1; transcription; ubiquitin ligase; alternative splicing; genotoxic stress

Correspondence: Dr R Pardi, Unit of Leukocyte Biology, Vita-Salute San Raffaele University School of Medicine, DIBIT-Scientific Institute San Raffaele, Milano 20132, Italy. E-mail: [email protected] 3 These authors contributed equally to this work. Received 21 May 2007; revised 19 September 2007; accepted 28 September 2007; published online 29 October 2007

Introduction The E3 ubiquitin ligase COP1 has recently been shown to recognize selected Jun family members as well as p53 as specific substrates in mammalian cells (Bianchi et al., 2003; Wertz et al., 2004; Dornan et al., 2004b), although the physiological context in which it operates in higher organisms is still unclear. It was established early on that Arabidopsis thaliana COP1 (AtCOP1) functions as an essential negative regulator of light-mediated plant development (Yi and Deng, 2005). The COP1 protein comprises three signature domains: a RING-finger motif, followed by a coiled-coil domain and seven WD40 repeats, all of which have been implicated in mediating the interaction of COP1 with other proteins and/or its self-dimerization (Torii et al., 1998), as originally shown in Arabidopsis. Notably, HY5, the preferred substrate of AtCOP1, is a basic region leucine zipper transcription factor that shares with selected Jun family members the structural motifs involved in COP1 binding (Yi and Deng, 2005). Although COP1 is capable of self-ubiquitinating efficiently in vitro (Bianchi et al., 2003), it appears to require the cooperation of other factors in mediating ubiquitination of c-Jun in vivo. It has been suggested that mammalian COP1 functions as an adaptor protein recruiting c-Jun to an E3 complex containing DET1, DDB1, cullin4A and Roc1, through direct interaction with DET1 (Wertz et al., 2004; Yanagawa et al., 2004). Mammalian COP1 directly associates with c-Jun and represses AP-1-mediated transcription (Bianchi et al., 2003; Yi et al., 2005). The structural requirements for polyubiquitination of p53 by COP1 are less well defined but may differ from those involved in c-Jun degradation, as it has been shown that integrity of the COP1 RING domain, which is dispensable for the degradation of c-Jun, is instead required for p53 ubiquitination and degradation (Wertz et al., 2004). A recent study (Dornan et al., 2006) has shown that Serine 387 of COP1 is a direct substrate of the ATM protein kinase in response to g-irradiation. This results in functional inactivation of the ligase, contributing to stabilization of p53. This study did not address the consequences of ATM-induced

COP1 dominant interfering splice variant MG Savio et al

2402

phosphorylation on the ability of COP1 to affect the polyubiquitination of c-Jun. Although it is well established that the short half-life of the c-Jun and p53 proteins is linked to their polyubiquitination and proteasome-dependent degradation, their dynamics and the sequential post-translational changes involved are incompletely understood and appear to be increasingly complex. The activity of cJun and JunB is enhanced by phosphorylation of their transcriptional activation domain by Jun N-terminal kinases (JNKs) (Karin, 1995). Early mutagenesis studies indicated that JNK-dependent phosphorylation on Ser residues 63 and 73 in the transactivation domain can also stabilize c-Jun (Fuchs et al., 1997; Musti et al., 1997). Recently, however, JNK-mediated phosphorylation at the same residues was shown to accelerate c-Jun degradation by allowing its recognition by the E3 ligase Fbw7-containing Skp/Cullin/F-box protein complex (SCFFbw7) (Nateri et al., 2004). The role of Fbw7 in c-Jun polyubiquitination has been recently defined further by Wei et al., who reported that phosphorylation of c-Jun by GSK3 creates a high-affinity binding site for the E3 ligase Fbw7, which targets c-Jun for polyubiquitination and proteasomal degradation (Wei et al., 2005). Similar to c-Jun, p53 is also tightly regulated. Mdm2 was shown to downregulate p53 protein level and activities (Momand et al., 1992; Oliner et al., 1992). Mdm2 itself is a highly regulated protein under DNA damage conditions and was thought for quite some time to be the sole E3 ligase responsible for p53 degradation under normal physiologic conditions. Amplifications of the Mdm2 gene in 7% of human tumors account for one mechanism of overexpression. Recent data suggest that Mdm2-mediated ubiquitination is not the only important factor for p53 regulation, as in vivo knock-in experiments show that a p53 mutant protein, lacking the major ubiquitination sites for Mdm2, has a normal halflife and is stabilized and activated in response to stress (Feng et al., 2005; Krummel et al., 2005). In addition to Mdm2, other E3 ligases have been shown to impart specificity toward p53 and promote its proteasomemediated degradation, such as Pirh2 (Leng et al., 2003), ARF-BP1 (Chen et al., 2005) and COP1 (Dornan et al., 2004b). COP1 is also a p53-inducible gene and COP1 depletion by siRNA enhances p53-mediated G1 arrest and can sensitize cells to ionizing radiation (Dornan et al., 2004a). We have been investigating how the function of COP1 is regulated by extracellular stimuli in mammalian cells, which lack the photomorphogenic response involved in COP1 regulation in plants. We describe here an alternatively spliced variant of COP1, COP1D, which lacks exon 7 in the coiled-coil domain and exerts a dominant-interfering function over the full-length protein at physiological concentrations. Altered ratios between COP1 and its splice variant are induced during UV-stress responses and can be detected in invasive breast cancer, suggesting the existence of a further level of regulation of c-Jun homeostasis in DNA-damaged or transformed cells. Oncogene

Results Alternatively spliced COP1 transcripts are produced at variable levels in different cell lines and tissues To achieve a deeper understanding of structure–function relationships in COP1, we performed an extensive in silico analysis. We determined that up to 11 different predicted transcripts can be generated by alternative splicing of the HsCOP1 pre-mRNA (Figure 1a). We validated the existence of the above splice variants in several cell lines by nested, semiquantitative reverse transcription (RT)–PCR analysis, and determined that in addition to the predominant full-length (FL) COP1 isoform, the COP1D variant, lacking exon 7, is the most abundantly expressed (not shown). Wertz et al. reported the occurrence of COP1D24, a COP1 splice variant that differs from COP1D as, in addition to exon 7, it lacks 12 nucleotides (resulting in an in-frame deletion of 4 amino acids) in exon 4, likely due to the existence of competing 50 splice sites in this exon. To differentially assess the presence of the two isoforms, we took advantage of the existence of a unique DraIII restriction site within the 12 nt sequence that is removed in the COP1D24 variant, combined to the observation that the 12 nt deletion in this isoform causes the loss of DraIII restriction site and creates a new unique BstXI site. Figure 1b shows that both variants are expressed at similar levels in U2OS and BJ1 cell lines, and in PBMC. This was confirmed by sequencing the RT–PCR products, which revealed the coexistence of the two isoforms (not shown). Further analyses were therefore carried out with primers that amplify both COP1D and COP1D24. Figures 1c and d show that FL COP1 and COP1D mRNA levels are highly variable depending on the cell line and tissue analysed. COP1D is unable to promote proteasome-dependent degradation of c-Jun We cloned the COP1D variant by PCR amplification of total cDNA obtained from HeLa cells. To address whether COP1D is functionally competent to promote degradation of c-Jun, we expressed COP1D in the presence or absence of DET1 and comparatively assessed the ability of FL COP1 and its splice variant to affect the intracellular levels of exogenously expressed c-Jun. As previously shown (Wertz et al., 2004), we confirmed that while FL COP1 affects a decrease of cJun levels, COP1D is unable to do so independently of the presence of DET1 (Figure 2). In separate experiments, using the proteasome inhibitor MG132, we confirmed that the reduced intracellular levels of c-Jun observed upon overexpression of FL COP1 are a consequence of proteasome-mediated degradation of the protooncogene product (not shown). COP1D antagonizes the function of FL-COP1 at physiologically relevant ratios The above findings show that COP1D is unable to affect the degradation of c-Jun in spite of the presence of a conserved substrate-binding domain (comprising the


2403

Figure 1 Relative abundance of the COP1 and COP1D isoforms in different cell lines and tissues. (a) Intron–exon organization of the human COP1 gene and its predicted splice variants. Black squares represent ubiquitously spliced-out exons. COP1c (full length) isoform includes all other exons while COP1D lacks exon 7 (dark grey squares). (b) Comparative assessment of COP1D and COP1D24 by selective restriction digestion of RT–PCR products from the indicated cell lines. Shown are amplicons from the various cell lines, either undigested or digested with DraIII or BstXI. (c and d) COP1 and COP1D steady-state mRNA expression, as assessed by realtime PCR (Taqman) in different human cell lines (c) and fetal tissues (d). Values were normalized to COP1 levels as detected in the control cDNA provided by the manufacturer. Values are the mean7s.d. of triplicate samples.

WD40 repeats). We therefore tested the hypothesis that COP1D exerts a dominant-interfering function over the full-length protein. To this aim, we assessed the degradation of exogenous c-Jun by FL COP1 and DET1 in the presence of varying FL COP1/COP1D ratios. Figure 3a shows that COP1D efficiently antagonizes the activity of FL COP1. Such antagonism is already apparent at FL COP1/COP1D ratios of 10:1 and results in the complete inhibition of FL COP1 function at a 5:1 ratio, which is physiologically relevant in most tissues analysed (Figures 1c and d). A further increase in the intracellular concentration of COP1D, results in an elevation of c-Jun levels above those observed in mocktransfected cells. This suggests that exogenous COP1D protects c-Jun from the activity of endogenous factors affecting its stability under steady-state conditions. To further address this hypothesis, we assessed the intracellular levels of endogenous c-Jun in cells expressing increasing concentrations of COP1D, either under steady-state conditions or at various time points following UV irradiation. Figure 3b shows the kinetics of endogenous c-Jun induction by UV stress in subconfluent HeLa cells. The contribution of proteasomedependent degradation to the expression profile of the

c-Jun protein was revealed by exposing cells to MG132 (not shown). Expression of exogenous COP1D results in a dose-dependent increase in c-Jun, which appears to be particularly relevant at intermediate to late time points post-UV irradiation. These findings confirm that COP1D antagonizes an endogenous process affecting the stability of the c-Jun proto-oncogene product. COP1D forms heterodimers with full-length COP1, thus preventing its association with DET1 Next, we addressed the mechanism(s) underlying the dominant-interfering function of COP1D over the fulllength protein. AtCOP1 has been shown to be able to homodimerize through a yet undefined region located in the coiled-coil domain (Torii et al., 1998). Thus, in principle COP1D could titer down the full-length protein by forming heterodimers that bind less efficiently to the other components of the DET1 containing E3 complex. Alternatively, COP1D, whose substrate-binding region is intact, could sequester the substrate from the FL COP1based ubiquitin ligase complex. To this aim, FL COP1 and COP1D were modified by the addition of different tags (Flag and T7). The constructs were cotransfected Oncogene


2404

Figure 2 COP1D is unable to promote c-Jun degradation. HeLa cells were transfected with c-Jun and co-transfected with equal amounts of T7-COP1 or T7-COP1D, in the presence or absence of HSV-DET1, followed by SDS–PAGE of cell lysates. Total protein levels of c-Jun were detected by immunoblotting with anti-c-Jun antibody. Anti-T7 and anti-HSV antibodies were used to control the relative protein levels of the transfected protein products. Antib-tubulin antibody was used as a loading control. Shown is a representative experiment out of the six independent ones performed. The arrowhead denotes an unspecific band revealed by the anti-T7 antibody.

and assessed by immunofluorescence and co-immunoprecipitation analysis. The subcellular localization of the two isoforms is largely overlapping and predominantly nuclear, as expected based on the conservation of both nuclear localization signals operating in COP1 (Supplementary Figure S1A). Reciprocal co-immunoprecipitations were carried out with antibodies specific for one tag followed by immunoblotting of the eluates with antibodies specific for the other tag. Supplementary Figure S1B shows that FL COP1 and COP1D form both homoand heterodimers at stoichiometric ratios. Notably, the expression of stoichiometric levels of COP1D completely eliminates the association of FL COP1 with both DDB1 and DET1 (Figure 4), suggesting that the splice variant exerts a dominant-interfering function by sequestering the full-length protein from the enzymatically active ubiquitin ligase complex. Differential expression of FL COP1 and COP1D in UV irradiated cells To explore the physiological relevance of the COP1D isoform, we initially assessed by quantitative PCR analysis the steady-state levels and kinetics of COP1D and FL COP1 transcripts in cells exposed to UV stress, as the currently known substrates of COP1 in mammalian cells (c-Jun and p53) are both involved in the effector phase of such response. Upon UV stress, we observed an initial decrease in COP1/COP1D ratio, followed by a progressive increase at later time points, mainly due to variations in the levels of the FL COP1 transcript (Figure 5a). As a result, the overall FL COP1/ COP1D ratio ranged from 1.4370.2 at the time of UV exposure to 2.3170.4 at 16 h post-UV stress. We assessed whether the aforementioned variation in FL Oncogene

Figure 3 COP1D effect on COP1 activity under steady-state or stress conditions. (a) HeLa cells were transfected with constant amounts of c-Jun, Flag-COP1, HSV-DET1 and increasing concentrations of T7COP1D (arrowhead; shown are the relative COP1D/COP1 cDNA ratios used), followed by SDS–PAGE of cell lysates and immunoblotting with anti-c-Jun antibody. Anti-tag and anti-b-actin antibodies were used to control for transfection efficiency and as loading controls, respectively. Numbers in parenthesis indicate densitometric ratios of c-Jun over COP1untransfected cells. (b) HeLa cells were transfected with empty vector or with the indicated amounts of T7COP1D. Forty-eight hours after transfection, cells were left either untreated (no UV) or exposed to 30 J m2 of UV, followed by incubation for the indicated time points. Endogenous c-Jun levels were assessed by immunoblotting and quantitated by densitometric analysis (bottom panel). b-Actin was used as a normalization control. Experiments shown in (a) and (b) are representative of the six independent ones performed.

COP1/COP1D ratio was the result of an UV-induced alteration in transcript stability, selectively affecting one of the two mRNAs. To this aim, cells were exposed to a


2405

Figure 4 COP1D prevents the association of COP1 with the DET1/DDB1 complex. HeLa cells were transfected with FlagCOP1, in the presence or absence of HSV-DET1, and cotransfected with increasing amounts of T7COP1D (arrowhead). Cell lysates were subjected to immunoprecipitation with the indicated antibodies. Eluates were resolved on SDS–PAGE followed by immunoblotting with the indicated antibodies. Total cell lysates were loaded in parallel gels to assess the expression level of the various proteins. b-Actin was used as loading control.

dose of 30 J m2 of UV-C. Four hours later, 5,6-dichloro1-b-D-ribobenzimidazole (DRB) was added to the cells to block de novo mRNA synthesis and cells were chased for various time points to assess mRNA decay by quantitative PCR. Figure 5b shows that the two transcripts have overlapping decay profiles both in untreated and in UV-irradiated cells, suggesting that the observed variations are likely due to a regulated splice site selection process, which is known to be controlled by extracellular signaling pathways, including those associated with genotoxic stress (van der Houven van Oordt et al., 2000; Shin and Manley, 2004). Selective downregulation of FL COP1 or COP1D has opposite effects on UV stress-induced c-Jun levels and kinetics The altered FL COP1/COP1D ratio observed in UV stressed cells led us to hypothesize that COP1D may dynamically interfere with the function of FL COP1, when its levels are proportionally increased over the FL isoforms above a critical threshold. To test such hypothesis, we developed an RNAi strategy. We successfully generated siRNA oligoduplexes that affect selectively and efficiently the levels of either the FL or the spliced variant’s mRNA, as judged by the downregulation of the exogenously

Figure 5 COP1 and COP1D mRNA induction and stability in UV-stressed cells. (a) Real-time PCR analysis of COP1 and COP1D mRNA, under steady-state conditions or at various time points following UV exposure (30 J m2). Values are the mean7s.d. of four independent experiments (*Po0.05; **Po0.01). (b) Total RNAs were extracted at different time points after 5,6-dichloro-1b-D-ribobenzimidazole (DRB) addition either under steady-state conditions (top panel) or 4 h after UV irradiation (30 J m2) (bottom panel). COP1 and COP1D mRNA relative amounts were quantified by real-time PCR. Values are expressed as % variations over COP1 levels in DRB-untreated cells. Shown is a representative experiment out of the three performed.

expressed constructs coding for either isoform (Figure 6a), as well as by the selective modulation of the endogenous transcripts (Supplementary Figures S2A and B). We also generated nonisoform-specific siRNA that effectively Oncogene


2406

knocked down both isoforms (Figure 6a). Reconstitution experiments (Supplementary Figure S2C) showed that the designed siRNA were exerting a target-specific effect. In both HeLa and U2OS cells we could detect a UV stressinduced increase in the endogenous COP1 protein using both commercially available (not shown) and customdesigned anti-COP1 antibodies (Figures 6b and 7d). As we noticed that the increase in the COP1 protein level induced by UV stress frequently exceeded the corresponding increase at the mRNA level across several cell lines (compare Figure 5a to 7d), we performed cyclohexamideinhibition experiments to assess the protein’s half-life under steady-state conditions or upon UV stress. Results (Supplementary Figure S3) show that both FL COP1 and COP1D are relatively short-lived proteins and that UV irradiation promotes their stabilization via a post-translational mechanism. Both steady-state and UV-induced COP1 levels were markedly affected by siRNA’s targeting the endogenous COP1 transcript (Figure 6f). Using such siRNA and their appropriate controls, we treated cells prior to UV exposure and monitored the intracellular levels and kinetics of the endogenous c-Jun protein at various time points post-UV stress. c-Jun accumulation

could be obtained by downregulating selectively the FL COP1 isoform (Figure 6c). Conversely, selective downregulation of the COP1D splice variant results in decreased intracellular levels of c-Jun at intermediate time points post-UV irradiation (Figure 6d). To confirm that COP1D exerts its function exclusively by interfering with the FL isoform, we assessed the interfering function of exogenously expressed COP1D on intracellular c-Jun levels in cells whose endogenous FL COP1 levels had been downregulated by selective RNAi. Figure 6e shows that exogenous COP1D no longer affects the intracellular levels of c-Jun in cells lacking endogenous FL COP1, thus confirming that COP1D specifically interferes with the function of its FL isoform. Role of COP1 and COP1D in the control of p53 levels To assess whether degradation of p53 by COP1 was also affected by its splice variant, we utilized the p53competent cell line U2OS, as endogenous ligase activities in HeLa cells are blunted by the presence of the human papillomavirus E6 protein, which greatly destabilizes p53 (Koivusalo et al., 2006). We utilized this

Figure 6 Selective downregulation of COP1 or COP1D has opposite effects on endogenous c-Jun levels in UV-stressed cells. (a) HeLa cells were transfected with identical amounts of T7COP1 and T7COP1D and co-transfected with the reported siRNA. A scrambled siRNA was used as negative control. Immunoblotting was performed with anti-T7 and anti-b-tubulin antibodies to control for COP1 isoform expression and as loading control, respectively. (b) HeLa cells were transfected with the indicated siRNA, followed by UV irradiation and incubation for the indicated time points. Cell lysates were subjected to immunoprecipitation with anti-COP1 and antib-tubulin (for normalization) antibodies. Eluates were loaded on an SDS–PAGE and assessed by immunoblotting for endogenous COP1 levels. Densitometric ratios of COP1 over control siRNA-transfected cells are reported in parenthesis. (c and d) HeLa cells were transfected with the indicated siRNA, followed by UV irradiation. Endogenous c-Jun levels were assessed by immunoblotting. Variations of c-Jun levels over siRNA control-transfected cells, at each corresponding time point, are indicated in parenthesis. (e) Cells were transfected with either empty vector or T7COP1D and co-transfected with siRNA targeting endogenous FL COP1 followed by UV irradiation. At the indicated time points post-irradiation cell lysates were subjected to SDS–PAGE and assessed by immunoblotting for endogenous c-Jun levels. Anti-T7 and anti-b-tubulin antibodies were used to control for transfection efficiency and protein loading, respectively. Numbers in parenthesis represent densitometric ratios of c-Jun over the siRNA control-transfected cells at each time point. (f) Normalized densitometric values of endogenous COP1 protein levels, relative to the experiment shown in panel b. Throughout the experiments, controls (C) indicate non UV-treated cells. Oncogene


2407

Figure 7 Modulation of COP1 and COP1D levels in p53-competent cells. (a and b) U2OS cells were transfected with the indicated siRNA, followed by UV irradiation (30 J m2). Cells were harvested at the indicated time points after UV exposure. c-Jun and p53 levels were assessed by immunoblotting; b-tubulin was used as loading control. (c) U2OS cells were transfected with empty vector or with increasing concentrations of FlagCOP1D. Forty-eight hours after transfection cells were either left untreated (no UV) or exposed to 30 J m2 of UV. Endogenous c-Jun, p53 levels were assessed by immunoblotting. b-Tubulin was used as a loading control. (d) U2OS cells were transfected with the indicated siRNA, followed by UV irradiation. Cell lysates were subjected to immunoprecipitation with anti-COP1 and anti-tubulin antibodies, and assessed by immunoblotting for endogenous COP1 levels. Densitometric ratios of COP1 in treated samples over control siRNA-transfected cells are reported in parenthesis.

cell line to perform both FL COP1 and COP1D downregulation and COP1D overexpression experiments and assessed endogenous p53 levels in cells exposed to UV stress. Figure 7 shows that both steady-state and UV-induced p53 levels are markedly increased by either treatment. Variable COP1/COP1D ratios in breast cancer samples correlate with endogenous c-Jun levels Overexpression of COP1 has been linked to downregulation of p53 levels in selected cancer histotypes (Dornan et al., 2004a). To assess whether variations of the COP1/COP1D ratio paralleled the altered levels of c-Jun and p53 frequently observed in neoplastic tissues, we screened a panel of human breast cancer specimens (stage II and IIIA invasive ductal carcinomas) by comparing the expression profile of COP1 and COP1D mRNA (as detected by real-time PCR) with the protein levels of p53 and c-Jun (Figure 8 and Supplementary Figure S4). As a reference for the assessment of steadystate COP1 transcript levels, we pooled total mRNA extracted from 10 independent samples of normal breast tissue (Figure 8). Notably, steady-state expression of COP1 mRNA was particularly elevated in normal breast tissue compared to the internal control utilized to screen the array of normal tissues (Figure 1). COP1 mRNA levels were highly heterogeneous and markedly reduced in 8 out of 14 tumor samples analysed when compared to normal breast tissue, while the levels of COP1D showed minor variations (Figure 8a). This resulted in an elevation of the COP1D/COP1 ratio in a large fraction of samples (Figure 8b). Interestingly, we found that

increased COP1D/COP1 ratios correlated with higher c-Jun, but not p53 protein levels in 14 independent samples (Figure 8c). Notably, the COP1D/COP1 ratio showed a better correlation with c-Jun protein levels compared with COP1 mRNA levels alone (Supplementary Figure S5), providing indirect evidence that this splice variant may exert a modulating function over the full-length protein in vivo.

Discussion This work provides evidence for the existence of a novel regulatory mechanism underlying the function of the mammalian COP1 ubiquitin ligase. Such mechanism involves the expression of a splice variant of COP1 that exerts a specific dominant-interfering function over the full-length protein at physiologically attainable ratios. Several lines of evidence in this work suggest that COP1D plays a physiologically relevant role in the functional regulation of the full-length isoform. First, ectopic expression studies show that COP1D, which is functionally incompetent to promote substrate degradation, exerts a dominant-negative function over the full-length isoform at sub-stoichiometric ratios. Co-immunoprecipitation experiments show that the most likely explanation for the observed functional downregulation of FL COP1 by COP1D is that the latter heterodimerizes with the former and the resulting heterodimers are unable to stably associate with the other component elements of the COP1-containing E3 ligase complex, as previously suggested for AtCOP1 (Torii et al., 1998; Subramanian Oncogene


2408

Figure 8 Altered COP1/COP1D ratio in invasive breast cancer parallels variations in c-Jun protein levels. (a) COP1 and COP1D mRNA expression, and COP1D/COP1 mRNA ratio (b) as assessed by real-time PCR (Taqman) in a panel of human breast cancer samples. Values were normalized to steady-state COP1 levels as detected in a pool of 10 independent normal breast tissue samples (normal) and compared to the internal control (control) provided by the manufacturer of the tissue mRNA array shown in Figure 1. Values in (a) are the mean7s.d. of triplicate samples. (c) Correlation analysis between the COP1D/COP1 mRNA ratio and either c-Jun (left panel) or p53 (right panel) protein levels as detected by immunoblotting. Each dot represents an individual cancer sample.

et al., 2004). COP1D antagonizes efficiently the function of FL COP1, as judged by co-expression experiments. The functional effect of COP1D at sub-stoichiometric concentrations can be explained if under the conditions used COP1 levels are limiting to achieve significant degradation of the substrate. This could be due to the extremely short half-life of the COP1 protein (Torii et al., 1998; Bianchi et al., 2003), which can be barely detected in unstimulated cells. The increased steady-state level of endogenous c-Jun and p53 in cells ectopically expressing COP1D provides a further line of evidence supporting a functional role for COP1D. Finally, selective downOncogene

regulation of COP1D by RNA interference results in decreased levels of c-Jun at intermediate time points postUV irradiation, when the COP1/COP1D ratio reaches its lowest observable values. Both of the above effects are lost in cells in which FL COP1 has been selectively knocked down. Taken together, these findings strongly argue in favor of a negative regulation of COP1 by COP1D at physiological levels of the splice variant. We observed that COP1D, as previously reported for COP1D24 (Wertz et al., 2004), is unable to associate to DDB1 and DET1, two of the components of the previously identified COP1-containing E3 ligase complex.


2409

Similar to COP1D, COP1D24 promotes a slight increase in endogenous c-Jun levels upon ectopic expression in U2OS cells (Wertz et al., 2004). This effect has been ascribed to sequestration of the substrate by the functionally incompetent ligase. In view of our findings that COP1D efficiently heterodimerizes with the FL isoform, we favor the hypothesis that the dominantinterfering function of the splice variant is due to displacement of the active E3 ligase from the other components of the COP1-containing E3 ligase complex. Using p53 competent cell lines, we have shown that p53 is an additional substrate of COP1 whose stability can be modulated by the mechanism we have unveiled in this work. As stated above, we observed highly variable steadystate expression levels of the COP1 transcript in various cell lines and tissues. This finding, coupled with the observation that the decay rate of the COP1 transcript is marginally affected by extracellular stimuli such as UV irradiation, suggests that COP1 mRNA levels are controlled at the level of transcription both under homeostatic conditions and upon UV stress. Precedents exist in the plant COP1 orthologue indicating that lightinduced transcriptional regulation is a relevant aspect of COP1 expression and function (Tsuge et al., 2001). By analogy with plant COP1, we found that UV stress modulates the expression levels of the COP1 mRNA, yielding a biphasic response that is characterized by a downregulation at early time points followed by a roughly two-fold upregulation later in the response. The late increase in COP1 transcript levels post-UV stress could be interpreted as a p53-dependent event (Dornan et al., 2004b). However, since UV irradiation promotes an increase of COP1 transcription in HeLa cells, which reportedly harbor a functionally inactive p53 gene (Matlashewski et al., 1986), additional mechanisms likely exist which contribute to the observed modulation of COP1 mRNA levels in UV-stressed cells. The observed variations in the COP1/COP1D ratio can possibly be explained by the control of splice site selection and the efficiency of splicing in UV exposed cells, as previously reported (Shin and Manley, 2004). Further, our data point to the existence of a posttranslational mechanism leading to stabilization of the COP1 protein in UV-exposed cells. What could be the functional relevance of the coordinate regulation of COP1 function by its splice variant in UV stressed cells? The c-Jun proto-oncogene is strongly and selectively induced by UV stress and it appears to have a dual role in the UV response. At early time points post-UV irradiation, c-Jun sensitizes cells to apoptosis in a p53-dependent fashion, at least in part by skewing the transcriptional activity of p53 toward the expression of proapoptotic, as opposed to cell cycle arresting genes (Shin and Manley, 2004). At later time points, however, c-Jun appears to be essential for cell cycle re-entry of cells exposed to UV stress (Shin and Manley, 2004). Within this context, the function of COP1 could be to transiently lower the intracellular levels of c-Jun at critical time points post-UV stress, thus favoring cell cycle arrest and repair of DNA prior to

resumption of the cell cycle. In this respect, dynamic changes in the COP1/COP1D ratio provide a fine-tuning of this ubiquitin ligase’s function at defined time points following its induction by DNA damaging stimuli. Whether misregulation of COP1 expression or function contributes to cancer progression remains to be established. Dornan et al. (2004a) reported overexpression of COP1 protein in a high percentage of ovarian and breast cancer samples, which paralleled a downregulation of p53 levels in a fraction of samples. Our profiling of COP1 and its splice variant by quantitative mRNA analysis indicates that COP1 mRNA levels are most commonly decreased in invasive breast cancer and suggests that dysregulation of COP1/COP1D ratios correlates with altered expression of c-Jun rather than p53. Notably, overexpression of AP-1 transcription factors is a hallmark of breast cancer (Langer et al., 2006) and previous studies have shown that hCdc4, a E3 ligase targeting c-Jun for proteasome degradation, is selectively lost in a fraction of invasive breast tumors (Ekholm-Reed et al., 2004). Clearly, a more extensive assessment of the role of COP1, both in clinical settings and in preclinical models, is needed to unequivocally establish if expression or regulation of the COP1 gene and its protein product are a target of malignant transformation and cancer progression.

Materials and methods Cloning of hCOP1 and hCOP1D in a Tet-off vector system An hCOP1 cDNA (IMAGE clone 1707833, Human Genome Mapping Project Resource Center, Cambridge, UK) was cloned in the pBI4 Tet-inducible vector (Clontech, BD Bioscience, Palo Alto, CA, USA). Either Flag or T7 tags sequences were inserted by PCR in between the ATG and the second codon. To clone the hCOP1D, HeLa total RNA was extracted with the RNeasy mini-kit (Qiagen, Hilden, Germany) following the manufacturer’s instructions. cDNA second strand synthesis was performed with PfuUltra (Stratagene, La Jolla, CA, USA), with primers 50 -TAGGAGGCAGCAGCTC TAGC-30 and 50 -ACTGTCATCTGAGATACGAGACATC C-30 , which amplify a fragment between exon 1 and exon 11 of COP1. The faster migrating band was eluted from 1% agarose gel with Gel-extraction Kit (Qiagen). The eluted fragment was subcloned in the full-length sequence of hCOP1, with Flag or T7 tags. Selective COP1D24 and COP1D identification COP1D transcript was amplified with primers EX4-fwd (50 ACAGAAGCAAAGATTTG AGGAAAAGAGG-30 ) and EX6-rev (50 -CCATAAGAATCTGTAGTTGGGCTGC-30 ). Two bands, the previously described COP1D24 (GenBank Accession Number AY509921) and the novel isoform COP1D, were separated on a precast TBE gel, 4–12% agarose (Invitrogen, Carlsbad, CA, USA). The two bands were selectively digested with restriction enzymes DraIII and BstXI. Real-time PCR Total RNA was extracted from different cell lines (BJ1, HeLa, PBMC, HEK293T, U2OS) either in steady-state condition or after UV exposure (30 J m2) with the RNeasy Mini Kit (Qiagen) and reverse-transcribed. Total RNA was extracted Oncogene


2410 from a panel of surgically removed stage II/IIIA invasive ductal breast carcinomas and from surrounding normal breast tissue with the RNeasy Lipid Tissue Mini Kit (Qiagen), followed by reverse transcription. cDNAs were analysed for COP1 and COP1D mRNA levels by real-time PCR, following the manufacturer’s provided protocol (PE Applied Biosystem, Foster City, CA, USA). COP1 and COP1D mRNA stability, under steady-state conditions or 4 h after UV stress (30 J m2) was assessed by treating cells with 100 mM DRB, (5,6-dichloro-1-b-D-ribofuranosylbenzimidazole) for 1, 2 and 6 h prior to total RNA extraction and Taqman analysis as described in the ABI Prism 7700 Sequence Detection System User’s Manual. Western blotting and co-immunoprecipitation HeLa and U2OS cells were transfected using Lipofectamine 2000 (Invitrogen). Forty-eight hours after transfection the cells were harvested and processed according to Bianchi et al. (2003). Immunoblotting was performed with anti-Flag (M5, Sigma-Aldrich, St Louis, MO, USA), anti-c-Jun (BD Transduction Laboratories, San Diego, CA, USA), anti-HSV-tag (Novagen, Merck Darmstadt, Germany), anti-T7-tag (Novagen) antibodies and anti-actin (Sigma-Aldrich) for normalization. Detection was performed with the ECL-Plus reagent (Amersham, Piscataway, NJ, USA). Breast cancer samples and normal breast tissue were lysed with radioimmunoprecipitation buffer and followed by SDS–PAGE and western blotting with anti-c-Jun, anti-p53 (1C12, Cell Signaling Technology, Beverly, MA, USA) and anti-actin antibodies for normalization. Three different anti-Cop1 antibodies were tested: K16 (Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA), PW9725 (Biomol Research Laboratories, Plymouth Meeting, PA, USA) and a rabbit anti-Cop1 custom designed from Invitrogen (amino acids 1–452). Western blots have been performed with anti-Cop1 antibody from Invitrogen.

For co-immunoprecipitation experiments, HeLa cells were transfected with Lipofectamine 2000. Cells were harvested after 48 h in lysis buffer (50 mM Tris–HCl, pH8, 50 mM NaCl, 1% NP-40, protease inhibitors, MG132 20 mM, N-ethyl maleimide 10 mM). Immunoprecipitation was carried out with anti-Flag (M5), anti-HSV, anti-DDB1 (PC718, Oncogene Research Products, San Diego, CA, USA) antibodies. HeLa and U2OS cells were lysed at different time points after UV exposure, with lysis buffer; 500 mg of total lysates were subjected to immunoprecipitation with anti-COP1 (see above) and anti-b-tubulin (for normalization) antibodies. SiRNA transfection Synthetic siRNA were from Invitrogen; the siRNA sense sequences used in this study are as follows: nonspecific control, 50 -UAAUGUAUUGGAACGCAUAUU-30 ; COP1 FL, 50 -G GAUAUUAAGAGAGUGGAA-30 ; COP1D, 50 -AGAGG AAAUGAGUGGCUU-30 . A totoal of 30% confluent HeLa cells were transfected with Lipofectamine2000. Forty-eight hours after the transfection, cells were harvested and total RNA and proteins were extracted, as described above. A final concentration 10 nM for all siRNA was used. Acknowledgements We thank S Putignano and B Clissi for technical help, L Cremonesi for assistance in designing the RT–PCR strategy, C Doglioni and S Grassi for providing breast cancer samples. This work was supported in part by grants from AIRC, FIRB and MIUR (PRIN projects) to RP and from National Institutes of Health Grant R01 HL43331 and the Raymond and Beverly Sackler Foundation to JRB. This study was carried out in the framework of the Italian MUIR Center of Excellence in Physiopathology of Cell Differentiation.

References Bianchi E, Denti S, Catena R, Rossetti G, Polo S, Gasparian S et al. (2003). Characterization of human constitutive photomorphogenesis protein 1, a RING finger ubiquitin ligase that interacts with Jun transcription factors and modulates their transcriptional activity. J Biol Chem 278: 19682–19690. Chen D, Kon N, Li M, Zhang W, Qin J, Gu W. (2005). ARFBP1/Mule is a critical mediator of the ARF tumor suppressor. Cell 121: 1071–1083. Dornan D, Bheddah S, Newton K, Ince W, Frantz GD, Dowd P et al. (2004a). COP1, the negative regulator of p53, is overexpressed in breast and ovarian adenocarcinomas. Cancer Res 64: 7226–7230. Dornan D, Shimizu H, Mah A, Dudhela T, Eby M, O’rourke K et al. (2006). ATM engages autodegradation of the E3 ubiquitin ligase COP1 after DNA damage. Science 313: 1122–1126. Dornan D, Wertz I, Shimizu H, Arnott D, Frantz GD, Dowd P et al. (2004b). The ubiquitin ligase COP1 is a critical negative regulator of p53. Nature 429: 86–92. Ekholm-Reed S, Spruck CH, Sangfelt O, van Drogen F, Mueller-Holzner E, Widschwendter M et al. (2004). Mutation of hCDC4 leads to cell cycle deregulation of cyclin E in cancer. Cancer Res 64: 795–800. Feng L, Lin T, Uranishi H, Gu W, Xu Y. (2005). Functional analysis of the roles of posttranslational modifications at the p53 C terminus in regulating p53 stability and activity. Mol Cell Biol 25: 5389–5395. Oncogene

Fuchs SY, Xie B, Adler V, Fried VA, Davis RJ, Ronai Z. (1997). c-Jun NH2-terminal kinases target the ubiquitination of their associated transcription factors. J Biol Chem 272: 32163–32168. Karin M. (1995). The regulation of AP-1 activity by mitogenactivated protein kinases. J Biol Chem 270: 16483–16486. Koivusalo R, Mialon A, Pitka¨nen H, Westermarck J, Hietanen S. (2006). Activation of p53 in cervical cancer cells by human papillomavirus E6 RNA interference is transient but can be sustained by inhibiting endogenous nuclear export-dependent p53 antagonists. Cancer Res 66: 11817–11824. Krummel KA, Lee CJ, Toledo F, Wahl GM. (2005). The C-terminal lysines fine-tune P53 stress responses in a mouse model but are not required for stability control or transactivation. Proc Natl Acad Sci USA 102: 10188–10193. Langer S, Singer CF, Hudelist G, Dampier B, Kaserer K, Vinatzer U et al. (2006). Jun and Fos family protein expression in human breast cancer: correlation of protein expression and clinicopathological parameters. Eur J Gynaecol Oncol 27: 345–352. Leng RP, Lin Y, Ma W, Wu H, Lemmers B, Chung S et al. (2003). Pirh2, a p53-induced ubiquitin-protein ligase, promotes p53 degradation. Cell 112: 779–791. Matlashewski G, Banks L, Pim D, Crawford L. (1986). Analysis of human p53 proteins and mRNA levels in normal and transformed cells. Eur J Biochem 154: 665–672. Momand J, Zambetti GP, Olson DC, George D, Levine AJ. (1992). The mdm-2 oncogene product forms a complex with


2411 the p53 protein and inhibits p53-mediated transactivation. Cell 69: 1237–1245. Musti AM, Treier M, Bohmann D. (1997). Reduced ubiquitindependent degradation of c-Jun after phosphorylation by MAP kinases. Science 275: 400–402. Nateri AS, Riera-Sans L, Da Costa C, Behrens A. (2004). The ubiquitin ligase SCFFbw7 antagonizes apoptotic JNK signaling. Science 303: 1374–1378. Oliner JD, Kinzler KW, Meltzer PS, George DL, Vogelstein B. (1992). Amplification of a gene encoding a p53-associated protein in human sarcomas. Nature 358: 80–83. Shin C, Manley JL. (2004). Cell signaling and the control of pre-mRNA splicing. Nat Rev Mol Cell Biol 5: 727–738. Subramanian C, Kim BH, Lyssenko NN, Xu X, Johnson CH, von Arnim AG. (2004). The Arabidopsis repressor of light signaling, COP1, is regulated by nuclear exclusion: mutational analysis by bioluminescence resonance energy transfer. Proc Natl Acad Sci USA 101: 6798–6802. Torii KU, McNellis TW, Deng XW. (1998). Functional dissection of Arabidopsis COP1 reveals specific roles of its three structural modules in light control of seedling development. EMBO J 17: 5577–5587. Tsuge T, Inagaki N, Yoshizumi T, Shimada H, Kawamoto T, Matsuki R et al. (2001). Phytochrome-mediated control of COP1 gene expression in rice plants. Mol Genet Genomics 265: 43–50.

van der Houven van Oordt W, Diaz-Meco MT, Lozano J, Krainer AR, Moscat J, Ca`ceres JF. (2000). The MKK(3/6)p38-signaling cascade alters the subcellular distribution of hnRNP A1 and modulates alternative splicing regulation. J Cell Biol 149: 307–316. Wei W, Jin J, Schlisio S, Harper JW, Kaelin Jr WG. (2005). The v-Jun point mutation allows c-Jun to escape GSK3dependent recognition and destruction by the Fbw7 ubiquitin ligase. Cancer Cell 8: 25–33. Wertz IE, O’Rourke KM, Zhang Z, Dornan D, Arnott D, Deshaies RJ et al. (2004). Human De-etiolated-1 regulates c-Jun by assembling a CUL4A ubiquitin ligase. Science 303: 1371–1374. Yanagawa Y, Sullivan JA, Komatsu S, Gusmaroli G, Suzuki G, Yin J et al. (2004). Arabidopsis COP10 forms a complex with DDB1 and DET1 in vivo and enhances the activity of ubiquitin conjugating enzymes. Genes Dev 18: 2172–2181. Yi C, Deng XW. (2005). COP1—from plant photomorphogenesis to mammalian tumorigenesis. Trends Cell Biol 15: 618–625. Yi C, Li S, Chen X, Wiemer EA, Wang J, Wei N et al. (2005). Major vault protein, in concert with constitutively photomorphogenic 1, negatively regulates c-Jun-mediated activator protein 1 transcription in mammalian cells. Cancer Res 65: 5835–5840.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc).

Oncogene