ATM Engages Autodegradation of the E3 Ubiquitin Ligase COP1 After DNA Damage

See allHide authors and affiliations

Science  25 Aug 2006:
Vol. 313, Issue 5790, pp. 1122-1126
DOI: 10.1126/science.1127335


The ataxia telangiectasia mutated (ATM) protein kinase is a critical component of a DNA-damage response network configured to maintain genomic integrity. The abundance of an essential downstream effecter of this pathway, the tumor suppressor protein p53, is tightly regulated by controlled degradation through COP1 and other E3 ubiquitin ligases, such as MDM2 and Pirh2; however, the signal transduction pathway that regulates the COP1-p53 axis following DNA damage remains enigmatic. We observed that in response to DNA damage, ATM phosphorylated COP1 on Ser387 and stimulated a rapid autodegradation mechanism. Ionizing radiation triggered an ATM-dependent movement of COP1 from the nucleus to the cytoplasm, and ATM-dependent phosphorylation of COP1 on Ser387 was both necessary and sufficient to disrupt the COP1-p53 complex and subsequently to abrogate the ubiquitination and degradation of p53. Furthermore, phosphorylation of COP1 on Ser387 was required to permit p53 to become stabilized and to exert its tumor suppressor properties in response to DNA damage.

Genomic integrity is a prerequisite in mammals to maintaining cellular and tissue homeostasis. Given that the majority of tumors display genetic instability (1), the malfunction of the DNA-damage sensor system contributes to the selection of transformed somatic cells. ATM acts as a sentinel for DNA damage and phosphorylates key effecter substrates including the tumor suppressor p53, breast cancer–associated gene–1 (BRCA1), and Checkpoint kinase 2 (Chk2) (2).

The p53 protein is a critical component of the DNA damage response that induces the expression of a plethora of genes whose products are implicated in cell cycle arrest, senescence, apoptosis, and DNA repair (3). Furthermore, mice null for p53 typically develop lymphomas and sarcomas displaying aneuploidy (4). In response to DNA damage, p53 is stabilized (5, 6), and ATM directly phosphorylates p53 at the N terminus (7) on Ser15 and resultantly increases recruitment of the transcriptional coactivator p300 (8). ATM also phosphorylates the E3 ubiquitin ligase MDM2 and the related MDMX protein and dampens their ability to negatively regulate p53 (911). The E3 ubiquitin ligase COP1 is also a critical negative regulator of p53 that is overexpressed in breast and ovarian cancers (12, 13); however, the signal transduction pathway that regulates the COP1-p53 axis following DNA damage remains elusive. We therefore examined the steady-state levels of COP1 in fibroblasts after DNA damage induced by 10 Gy of ionizing radiation (IR). Steady-state levels of COP1 started to decline within 30 min and decreased to an almost undetectable level by 1 hour after exposure to IR (Fig. 1A). This correlated closely with an increase in steady-state levels of p53 and activation of the downstream target gene p21 (Fig. 1A and fig. S1). The decreased abundance of COP1 protein was not attributed to a decrease in COP1 mRNA levels (fig. S1) and occurred with as low as 2 Gy of IR (fig. S2). In addition, the half-life of COP1 was less than 10 min after exposure of cells to 10 Gy IR compared with 30 min for untreated cells (Fig. 1B).

Fig. 1.

Dependence of COP1 abundance after DNA damage on phosphorylation at Ser387 by ATM. (A) Steady-state amounts of COP1 are reduced after IR. GM03490 fibroblasts were irradiated with 10 Gy IR, harvested, and immunoblotted as indicated. (B) COP1 abundance after treatment of cells with IR and cycloheximide (CHX). GM0637 (ATM+/+) fibroblasts were exposed to 10 Gy IR and, after 2 hours, treated with CHX and sampled at the indicated time points. Bands were quantified by densitometry and represented as a line graph. (C) Delayed loss of COP1 in A-T cells. GM02052 fibroblasts were irradiated with 10 Gy IR, harvested, and immunoblotted as indicated. (D) Resistance of COP1-S387A to degradation on exposure to IR. U2-OS cells were transfected with COP1-WT or COP1-S387A for 24 hours and subsequently treated with 10 Gy IR for 2 hours before harvesting and immunoblotting. GFP- and pS1981-specific antibodies were used as a transfection control and DNA damage marker, respectively. (E) Phosphorylation of COP1 GST peptides containing Ser387 in vitro with immunoprecipitated ATM. ATM antibodies were used to immunoprecipitate proteins from lysates of GM03490 and GM02052 fibroblasts exposed to 10 Gy IR for 30 min. Resultant immunoprecipitates were incubated with GST peptides derived from COP1 or p53 for an in vitro kinase assay. (F) Phosphorylation of COP1 on Ser387 after DNA damage in an ATM-dependent manner. GM0637 and GM9607 fibroblasts were treated with 5 μg/ml bleomycin for the indicated times, and lysates were prepared for immunoprecipitation with antibodies to COP1 or direct immunoblotting.

Because ATM is the primary responder in the DNA damage–signaling pathway (14), we next determined whether the reduced abundance of COP1 after exposure of cells to IR was dependent on the presence of ATM. We exposed human fibroblasts derived from a patient with ataxia telangiectasia (A-T), which lack functional ATM, to 10 Gy IR. The reduction in steady-state levels of COP1 protein was absent in A-T fibroblasts (Fig. 1C) at early time points. However, COP1 decrease in abundance was evident at later time points, suggesting an alternative pathway may be at play. These observations also occurred after as low as 2 Gy of IR (fig. S3). Thinking that ATM might be directly phosphorylating COP1, we searched for potential phosphorylation sites that conformed to the ATM consensus on COP1 (15). Five serine-glutamine (SQ) motifs (fig. S4) are present within COP1; however, the third one, SQ3 (Ser387) was of primary interest, because it is highly conserved between COP1 orthologs (fig. S5) and because mutation of this serine residue to alanine prevented the reduction in steady-state abundance (Fig. 1D), as well as the reduction in the half-life of COP1 in cells exposed to IR (fig. S6).

We used glutathione S-transferase (GST) fusion peptides to test whether Ser387 was a bona fide target for ATM-mediated phosphorylation in vitro. ATM-specific immunoprecipitates from wild-type cells, but not those from ATM null fibroblasts (Fig. 1E), catalyzed phosphorylation of a peptide harboring Ser387 (GST-COP1-WT) and a peptide from p53 having Ser15 (GST-p53) (a classical ATM substrate) (16). The peptide GST-COP1-A containing S387A, the mutation in which Ser387 is replaced by alanine (A), was not phosphorylated. Similar results were obtained with incubation of full-length COP1 with recombinant ATM, but not ATM-kinase dead (ATM-KD) mutant (fig. S7). We used a polyclonal phospho-specific antibody to phosphorylated Ser387 (figs. S8, S9, and S10) to determine whether Ser387 is indeed phosphorylated in vivo following DNA damage. COP1 phosphorylated at Ser387 was readily detected in ATM-WT fibroblasts, whereas this was absent in A-T cells (Fig. 1F).

Together, these data suggested that ATM could be directly responsible for phosphorylation and degradation of COP1 levels. Overexpression of ATM caused a decrease in the abundance of COP1 protein, whereas ATM-KD had no effect (Fig. 2A), and the COP1-S387A mutant expressed in place of COP1-WT was resistant to such ATM-mediated degradation (Fig. 2B).

Fig. 2.

Auto-E3 degradation of COP1 initiated by ATM. (A) Exogenous ATM promotes a reduction in COP1 steady-state levels. Saos-2 cells were transiently transfected with COP1, with or without a titration of ATM or ATM-KD, for 24 hours. (B) Resistance of the COP1-S387A mutant to an ATM-mediated reduction in steady-state abundance. Saos-2 cells were transiently transfected with COP1 or COP1-S387A, with or without a titration of ATM, for 24 hours. (C) Requirement of the 26S proteasome for ATM-induced reduction of COP1. Saos-2 cells were transfected with COP1, with or without ATM, for 24 hours followed by 6 hours' incubation with proteasome inhibitor or vehicle control. (D and E) Ubiquitination of COP1 promoted by ATM in a kinase-dependent manner and dependence on phosphorylation at Ser387. H1299 cells were transfected for 24 hours with the plasmids, as indicated, and were treated with proteasome inhibitor for 4 hours before immunoprecipitation and immunoblotting. (F) Resistance of COP1-C136/139S RING mutant to ATM-induced degradation. Saos-2 cells were transfected with COP1 or COP1-C136/139S, with or without ATM or ATM-KD, for 24 hours before lysates were harvested and immunoblotted. (G) Phosphate-mimetic at Ser387 promotes autoubiquitination of COP1 in vitro. E. coli–derived COP1, COP1-S387A, or COP1-S387D was subject to an autoubiquitination in vitro assay, with or without components of the ubiquitin system as indicated.

To find out whether COP1 turnover required the 26S proteasome, we treated cells with a proteasome inhibitor after transfection with hemagglutinin-tagged COP1 (HA-COP1) and FLAG-tagged ATM (FLAG-ATM) (Fig. 2C). ATM-induced loss of COP1 was completely abrogated when cells were treated with a proteasome inhibitor. Because recognition of substrates by the 19S cap of the proteasome is facilitated by addition of a polyubiquitin chain (17), we investigated COP1 to find out if it was ubiquitinated in response to activation of ATM. H1299 cells transfected with HA-tagged ubiquitin, ATM or ATM-KD, and COP1 were treated with proteasome inhibitor to allow accumulation of ubiquitinated substrates. Transfection of COP1 alone revealed a low abundance of ubiquitin products (Fig. 2D). When ATM was transfected with COP1, however, there was an increase in the signal representing ubiquitinated COP1. ATM-KD did not show these effects. This ubiquitination event appeared to be entirely dependent on phosphorylation at Ser387, because no ubiquitination of the COP1-S387A mutant was detected (Fig. 2E). Furthermore, exposure of cells to 10 Gy IR resulted in the enhanced ubiquitination of COP1-WT, whereas this was diminished with the COP1-S387A mutant (fig. S11). If COP1 were autoubiquitinating, then ATM would be unable to promote degradation of a COP1-RING mutant in which both cysteine residues in the RING domain were converted to serine (C136S/C139S). Therefore, we transfected cells with either FLAG-COP1 or FLAG-C136S/C139S and ATM or ATM-KD and assessed steady-state amounts of COP1 protein by immunoblotting (Fig. 2F). The RING mutant was refractory to ATM-induced degradation compared with COP1-WT. To explore the possibility that ATM phosphorylation promotes autoubiquitination of COP1, we generated a COP1-S387D mutant (in which Ser387 is replaced by aspartic acid) that was expressed in and purified from Escherichia coli, to mimic phosphorylation on this serine residue, for an in vitro autoubiquitination assay (Fig. 2G). The COP1-WT and COP1-S387A proteins had detectable auto-E3 degradation activity, whereas the S387D COP1 mutant showed a substantial increase in auto-E3 ligase activity. Similar results were also obtained from COP1-S387D derived from mammalian cells (figs. S12 and S13).

When we examined the localization of exogenous COP1 in cells by immunofluorescence, COP1 was primarily nuclear; however, after exposure of cells to 10 Gy IR for 2 hours, COP1 was predominantly cytosolic (fig. S14). Endogenous COP1 detected by biochemical fractionation within primary fibroblasts was also exclusively localized to the nuclear fraction in wild-type and A-T fibroblasts (Fig. 3A). Treatment of the wild-type fibroblasts with etoposide resulted in a substantial portion of COP1 redistributed to the cytosolic fraction, but not within the A-T fibroblasts. In addition, within cells expressing the COP1-S387D mutant, COP1 was exclusively cytoplasmic, whereas the COP1-S387A mutant was confined entirely to the nucleus in the presence or absence of DNA damage (Fig. 3B). Biochemical fractionation of wild-type fibroblasts confirmed that the pS387 COP1 species was predominantly localized to the cytosolic compartment (fig. S15).

Fig. 3.

COP1 localization to the cytoplasm promoted by ATM. (A) Cellular fractionation following DNA damage. GM03490 and GM02052 fibroblasts were treated with or without 10 μM etoposide for 30 min; cytoplasmic and nuclear extracts were prepared. (B) Requirement of phosphorylation at Ser387 for localization to the cytoplasm following DNA damage. H1299 cells were transfected with COP1-WT, COP1-S387A, or COP1-S387D for 24 hours and then irradiated with 10 Gy IR. Cytoplasmic and nuclear extracts were prepared 2 hours after irradiation.

We wondered whether ATM-mediated regulation of COP1 abundance might contribute to the increase in p53 stability following DNA damage (5, 6). We therefore transfected H1299 cells (p53–/–) with constructs encoding FLAG-COP1 with or without HA-p53, treated the cells with bleomycin, and monitored the interaction between COP1 and p53 by immunoprecipitation (Fig. 4A). Interaction between COP1 and p53 was attenuated after DNA damage. In cells expressing the COP1-S387A mutant, DNA damage had no effect on the p53/COP1-S387A complex, which implies that phosphorylation of Ser387 is required to disrupt the COP1-p53 complex (Fig. 4B). In addition, modification of Ser387 appeared to be sufficient to disrupt the p53-COP1 because binding was reduced between p53 and COP1-S387D mutant (Fig. 4C). Similar results were observed in vitro with purified COP1-S387D and E. coli–derived and purified GST-p53 (fig. S16). COP1-S387D derived from E. coli also showed a decrease in ability to ubiquitinate p53 in vitro (Fig. 4D). These data suggest that phosphorylation of COP1 on Ser387 by ATM might compromise COP1's ability to degrade p53 and to inhibit its tumor suppressor function. Therefore, we transfected Saos-2 cells with p53 and various amounts of COP1-S387D, COP1-S387A, or COP1-WT (Fig. 4E). Proteins from lysates were Western blotted with p53-specific antibody, and a decrease in abundance of p53 was observed after transfection with COP1-S387A or COP1-WT, whereas only a small decrease occurred after cotransfection of COP1-S387D. Furthermore, the COP1-S387D mutant was inefficient at ubiquitinating p53 in vivo (fig. S17).

Fig. 4.

Effects of ATM-dependent phosphorylation of COP1 on negative regulation of p53. (A and B) Reduced interaction of COP1 with p53 and dependence on phosphorylation at Ser387. H1299 cells were transfected with constructs as indicated, and cells were treated with bleomycin (5 μg/ml) for 3 hours, followed by incubation with proteasome inhibitor for 3 hours. (C) Phosphorylation at Ser387 is sufficient to disrupt binding to p53. H1299 cells were transfected as indicated and treated with proteasome inhibitor. (D) Reduced ubiquitination of p53 by COP1 in vitro. HA-tagged p53 derived from in vitro translated material was incubated with COP1-WT or COP1-S387D with components of the ubiquitin system as indicated. (E) Decreased efficiency of degrading p53. Saos-2 cells were transfected for 24 hours with the constructs indicated, and lysates were harvested for immunoblotting. Remaining p53 was determined by densitometry. (F) Requirement of phosphorylation of COP1 on Ser387 for p53 stabilization after DNA damage. H1299 cells were transfected for 24 hours as indicated, followed by treatment of cells with 10 μM etoposide for 6 hours. (G) COP1-S387A mutant inhibits a p53-dependent reduction in cell viability following DNA damage. Transfected cells from (I) were treated with 10 μM etoposide for 24 hours, and cellular viability was determined. Data are the means ± SEM of three independent experiments. (H) Down-regulation of COP1 is required for full p53 activation after DNA damage. A549 cells were transfected with siRNA as indicated for 48 hours before treatment with 10 μM etoposide for 4 hours. (I) Model of the DNA damage response pathway between ATM and p53. (I) Model of the DNA damage response pathway between ATM and p53.

We used a colony-formation assay to monitor the long-term effects of COP1-WT and COP1-S387D in inhibiting p53-dependent function (fig. S18). In H1299 cells, transfection of p53 resulted in very few surviving colonies (6 × 103 CFU), whereas cotransfection of COP1 with p53 resulted in formation of a greater number of colonies (25 × 103 CFU). However, the COP1-S387D mutant failed to support colony formation (6 × 103 CFU) when p53 was transfected.

To determine whether phosphorylation of COP1 on Ser387 was required for p53 stabilization after DNA damage, we transfected H1299 cells with p53 with or without COP1-WT or COP1-S387A, and treated them with or without etoposide for 6 hours (Fig. 4F). Cotransfection of COP1-S387A or COP1-WT with p53 resulted in a reduction in steady-state amounts of p53. However, this reduction in abundance of p53 diminished in cells treated with etoposide. In contrast, the p53 was still reduced in cells expressing the COP1-S387A mutant after addition of etoposide. Cellular viability was also assessed 24 hours after etoposide treatment (Fig. 4G). Expression of COP1-WT diminished the effect of transfected p53 to reduce cell viability; however, this effect of COP1-WT was diminished in cells that had been treated with etoposide. On the contrary, the COP1-S387A mutant retained the ability to inhibit a p53-dependent reduction in cellular viability in cells that had been treated with etoposide. In addition, COP1 ablation by small interfering RNA (siRNA) in A549 cells blunted the fold activation of p53 after DNA damage, which indicated that degradation of COP1 is required for full p53 activation (Fig. 4H). Collectively, these data indicate that phosphorylation at Ser387 is required for COP1 degradation and to permit p53 stabilization after genotoxic stress, and these observations may help explain the delayed response to genomic abnormalities in A-T patients (Fig. 4I).

Supporting Online Material

Materials and Methods

Figs. S1 to S19


References and Notes

Stay Connected to Science

Navigate This Article