Rapid Destruction of Human Cdc25A in Response to DNA Damage

See allHide authors and affiliations

Science  26 May 2000:
Vol. 288, Issue 5470, pp. 1425-1429
DOI: 10.1126/science.288.5470.1425


To protect genome integrity and ensure survival, eukaryotic cells exposed to genotoxic stress cease proliferating to provide time for DNA repair. Human cells responded to ultraviolet light or ionizing radiation by rapid, ubiquitin- and proteasome-dependent protein degradation of Cdc25A, a phosphatase that is required for progression from G1 to S phase of the cell cycle. This response involved activated Chk1 protein kinase but not the p53 pathway, and the persisting inhibitory tyrosine phosphorylation of Cdk2 blocked entry into S phase and DNA replication. Overexpression of Cdc25A bypassed this mechanism, leading to enhanced DNA damage and decreased cell survival. These results identify specific degradation of Cdc25A as part of the DNA damage checkpoint mechanism and suggest how Cdc25A overexpression in human cancers might contribute to tumorigenesis.

High-fidelity maintenance of genomic integrity in eukaryotes is ensured by DNA repair and cell cycle checkpoints, surveillance pathways that respond to DNA damage by inhibiting critical cell cycle events (1). The key G1-phase checkpoint in mammalian cells relies on the ability of the p53 tumor suppressor to induce transcription of p21CIP1/WAF1 (2), an inhibitor of cyclin-dependent kinases (Cdks). However, transient inhibition of Cdk2 in response to DNA damage occurs even in cells lacking p53 (3) or p21 (4). Given the importance of checkpoints for prevention of genetic diseases including cancer, we wanted to explore these alternative mechanisms of G1arrest. Because inhibitory tyrosine phosphorylation of the mitotic kinase Cdk1 due to inactivation of the Cdc25C phosphatase underlies the G2-M checkpoint (5), we examined the role of Cdc25 family members, of which at least Cdc25A is essential for entry into S phase (6), in checkpoint control of the G1-S transition.

Exposure of proliferating human U-2-OS osteosarcoma cells (wild type for p53) to ultraviolet light (15 J/m2) resulted in a rapid decline in phosphatase activity of Cdc25A but not the activities of Cdc25B or Cdc25C (Fig. 1A). This was accompanied by a decline of the Cdc25A but not Cdc25B or Cdc25C proteins (Fig. 1B). Steady-state amounts of Cdc25A mRNA remained unaffected by ultraviolet light (UV) (7), whereas treatment with the proteasome inhibitor LLnL immediately after UV irradiation prevented reduction in the amount of both the endogenous (Fig. 1C) and the ectopic Cdc25A protein expressed from a heterologous promoter (Fig. 1D). In an in vivo protein ubiquitination assay (8), we detected abundant species of polyubiquitinated Cdc25A (Fig. 1E), and protein turnover measurements demonstrated shortening of the Cdc25A half-life in U-2-OS cells exposed to UV (Fig. 1F). These experiments showed that Cdc25A is targeted by the ubiquitin-proteasome pathway and that its turnover becomes accelerated in response to UV light.

Figure 1

Proteasome-dependent degradation of Cdc25A in response to DNA damage. (A) Inhibition of the activity of Cdc25A but not that of Cdc25B or Cdc25C in cells exposed to UV. Asynchronous U-2-OS cells were harvested 2 hours after UV treatment (15 J/m2) (19) and assayed for phosphatase activities as described (20). (B) UV-induced decrease in abundance of the Cdc25A protein, revealed by immunoblotting (21) of cells treated as in (A). (C) Prevention of UV-induced decrease in abundance and activity of Cdc25A by the proteasome inhibitor LLnL. LLnL (25 μg/ml) or dimethyl sulfoxide (DMS0) was added to the culture medium of irradiated (1 hour) or control U-2-OS cells immediately after treatment. Samples were immunoblotted or assayed for phosphatase activity. (D) Degradation of ectopic Cdc25A in a proteasome-dependent manner in UV-irradiated U-2-OS/B3C4 cells (22). Expression of the transgene was induced for 12 hours, and cells were treated with LLnL as in (C). Immunoblotting with Cdc25A-specific antibody DCS-122 (21). (E) Ubiquitination of Cdc25A in vivo. U-2-OS cells were transfected with expression plasmids for hemagglutinin (HA)-tagged Cdc25A or p21 (positive control) and six times histidine-tagged ubiquitin (His-Ub). After denaturing cell lysis, the polyubiquitinated species were purified on a Ni2+ column, electrophoretically separated, and visualized by immunoblotting with antibody to HA tag (8). (F) Half-life of Cdc25A protein in U-2-OS/B3C4 cells treated as in (D). Cycloheximide (25 μg/ml) was added to the medium 1 hour after irradiation and cells were collected at the indicated times. HA-Cdc25A was detected by immunoblotting with antibody DCS-122. (G) Effect of UV on Cdc25A in human diploid fibroblasts (IMR-90) exposed to UV (15 J/m2) and assayed by immunoblotting and phosphatase assays 2 hours after irradiation. (H) Down-regulation of Cdc25A in response to γ-irradiation. Irradiated (8 Gy) U-2-OS cells (19) were harvested after 2 hours and subjected to immunoblotting and Cdc25A phosphatase activity analysis.

To explore the possibility that Cdc25A protein stability represents a previously unrecognized target of a checkpoint response to genotoxic stress, we first reproduced the experiments in normal diploid cells. IMR-90 human fibroblasts irradiated with UV showed decreased phosphatase activity of Cdc25A and decreased abundance of the Cdc25A protein (Fig. 1G). We detected an analogous decline of Cdc25A protein abundance and activity in both U-2-OS and IMR-90 cells exposed to ionizing radiation (Fig. 1H) (7). Thus, in human cells, DNA damage triggers specific destruction of Cdc25A.

We next identified the cyclin E–Cdk2 kinase as the downstream target of the inactivation of Cdc25A. In time-course experiments with U-2-OS cells, Cdc25A protein and its phosphatase activity declined within 30 min after exposure to UV (Fig. 2A). Cyclin D–associated kinase activity and the amounts of cyclins D and E remained unchanged (Fig. 2A) (7), whereas cyclin E–Cdk2 activity declined with dynamics matching those of the UV effects on Cdc25A (Fig. 2A). This inhibition was not attributable to recruitment of p21 into cyclin E–Cdk2 complexes or to decreased interaction of cyclin E with Cdk2 (Fig. 2A). These data indicated that the inhibition of cyclin E–Cdk2 in response to UV could reflect persistent inhibitory phosphorylation on Tyr15within the ATP binding lobe of Cdk2 (9).

Figure 2

Effect of UV-induced degradation of Cdc25A on the activity of cyclin E–Cdk2. (A) Total amounts of cyclin E and Cdc25A, their associated kinase/phosphatase activities, and association of p21 and Cdk2 with cyclin E in extracts of U-2-OS cells (20, 21) harvested at the indicated times after UV irradiation (WCE, whole-cell extracts). (B) Rates of tyrosine dephosphorylation and kinase activation of cyclin E–Cdk2 (10) are decreased by UV treatment. U-2-OS cells were exposed to UV or left untreated and harvested 1 hour later. Endogenous kinases were inhibited by addition of 10 mM EDTA. Dephosphorylation of Cdk2 was initiated by transferring the lysates from ice to 30°C. At the indicated times, lysates were prepared for immunoprecipitation of cyclin E by immunoblotting with specific antibodies to Tyr15-phosphorylated Cdk2 or total Cdk2 (left). Kinase activity of cyclin E–Cdk2 was measured in an in vitro histone H1 kinase assay (right) (results presented as mean values and standard deviations from three independent experiments). (C) Effect of expression of ectopic Cdc25A in U-2-OS/B3C4 (22) cells. Tetracycline was removed from the medium 3 hours before UV treatment. Dephosphorylation (left) and kinase activation (right) of Cdk2 were assayed in vitro and data are presented as in (B). Equivalent amounts of Cdk2 were immunoprecipitated from each sample.

To test whether the inhibition of Cdk2 was causally linked to the UV-induced destruction of Cdc25A, we measured cyclin E–Cdk2 activity as a function of the extent of the inhibitory tyrosine phosphorylation of Cdk2 (10). Whereas extracts of mock-treated cells supported tyrosine dephosphorylation and corresponding activation of cyclin E–Cdk2, such phosphatase activity was largely lost in lysates from cells exposed to UV (Fig. 2B). Expression of Cdc25A in amounts that saturated the cellular capacity to efficiently degrade it (7) led to an increase of cyclin E–Cdk2 activity in nonirradiated cells, and UV treatment did not inhibit dephosphorylation of Cdk2 or cyclin E–Cdk2 activity in such cells (Fig. 2C). A similar, albeit less pronounced, effect was observed when cyclin A–associated kinase activity was assayed (7). Collectively, these data directly implicate the reduced activity of Cdc25A in UV-irradiated cells in the enhanced tyrosine phosphorylation and consequent inhibition of Cdk2, a kinase essential for DNA replication.

The p53 protein is a regulator of the G1 checkpoint (2), which raises the possibility that it was also required for DNA damage–induced degradation of Cdc25A. We established a U-2-OS subline conditionally expressing a dominant-negative allele of p53, p53DD (11). Expression of p53DD abolished the p53-dependent transcription of a reporter gene (Fig. 3A) and expression of endogenous p53 targets such as p21 (7). After UV treatment, both the p53 wild-type and p53DD-expressing cells had decreased amounts and activity of Cdc25A and ceased to replicate DNA to a very similar extent (Fig. 3A) (7), consistent with similar dynamics of Cdc25A degradation and cyclin E–Cdk2 inhibition in UV-exposed Saos-2 osteosarcoma cells, which lack p53 (Fig. 3B). Thus, the destruction of Cdc25A in response to UV appears to occur independently of p53-mediated transcription.

Figure 3

UV-induced degradation of Cdc25A is p53 independent but requires Chk1. (A) Dominant-negative p53 (p53DD) was expressed after removal of tetracycline for 24 hours. p53-dependent transcription in U-2-OS/C6 cells was revealed by immunostaining for luciferase after microinjection of cells with a p53-dependent reporter plasmid (22) (left). Cdc25A abundance and activity after 2 hours of UV treatment were determined by immunoblotting and phosphatase activity analyses in cells in which p53DD was not (+Tet) or was (−Tet) induced 24 hours before irradiation (right). (B) The Cdc25A response and inhibition of cyclin E–Cdk2 in Saos-2 cells that lack p53. Cells were exposed to UV (15 J/m2) or left untreated, harvested 2 hours later, and analyzed by immunoblotting and phosphatase/kinase activity measurements (20, 21). (C) UV-stimulated association of Cdc25A and Chk1. U-2-OS cells were transiently transfected with HA-tagged Cdc25A expression plasmid and either mock-treated, UV-irradiated (15 J/m2), or UV-irradiated after a 1-hour incubation with caffeine (10 mM). After an additional 2 hours in the constant presence of caffeine, cells were analyzed for levels and complexes of HA-Cdc25A and Chk1 as indicated. (D) Effect of caffeine on UV-induced down-regulation of Cdc25A. U-2-OS cells were treated with UV and caffeine as in (C), and Cdc25A protein level and phosphatase activity were analyzed as described (20, 21). (E) Increased activity of Chk1 toward Cdc25A after treatment of cells with UV. Chk1 was immunoprecipitated from lysates of U-2-OS cells treated as in (B), and its activity was determined in an in vitro kinase assay in the presence or absence of the Chk1 inhibitor UCN-01 (1 μM) (13), with purified GST-Cdc25A used as a substrate. An equal amount of Chk1 protein was present in each kinase reaction mixture (Input). Immunoprecipitates obtained with preimmune sera contained only background activity toward GST-Cdc25A regardless of UV treatment (7). (F) UV-mediated degradation of Cdc25A is prevented by the Chk1 inhibitor UCN-01. Cells were treated and assayed as in (D), except that 300 μM UCN-01 was substituted for caffeine.

Because the G2 checkpoint pathway includes phosphorylation of Cdc25C by Chk1, a kinase capable of phosphorylating all three members of the Cdc25 family in vitro (5), we considered Chk1 as a candidate for also mediating the early effects at G1-S. Exposure of U-2-OS cells to UV induced a shift (5) of the endogenous Chk1 on SDS gels (Fig. 3C). This was accompanied by an increase in the amount of Cdc25A associated with Chk1, an effect prevented by pretreatment of the cells with caffeine (Fig. 3C), a compound that interferes with activation of Chk1 (12). Treatment with caffeine before UV irradiation also reduced the shift of the Chk1 protein (Fig. 3C), abolished the destruction and loss of activity of Cdc25A (Fig. 3D), and restored the half-life of the Cdc25A protein to the values measured in nonirradiated cells (7). Chk1 kinase assay using glutathione S-transferase (GST)–tagged Cdc25A as a substrate revealed that the activity of Chk1 increased reproducibly in cells exposed to UV (Fig. 3E). Both basal and UV-induced Chk1 activities were inhibited by the Chk1 inhibitor UCN-01 (13) (Fig. 3E), and treatment of cells with UCN-01 at the time of UV irradiation abolished the degradation of Cdc25A and allowed maintenance of the activities of both Cdc25A and cyclin E–Cdk2 (Fig. 3F). These data implicate Chk1 as an upstream regulator of Cdc25A degradation after UV-induced DNA damage.

If the Chk1-Cdc25A-Cdk2 pathway indeed represented a bona fide checkpoint response, it would be expected to arrest the cell cycle, and avoiding this response should result in enhanced DNA damage and decreased cell survival. These predictions were tested and confirmed by additional experiments. Brief treatment of proliferating U-2-OS cells with bromodeoxyuridine (BrdU) revealed that ongoing DNA synthesis was inhibited within 1 to 2 hours after UV irradiation (Fig. 4A). Kinetic analysis of G1-S progression in cells released from nocodazole arrest showed that entry into S phase was also blocked (Fig. 4B). Between 16 and 24 hours after exposure to UV, the cells resumed DNA replication and progression through the cell cycle (7), indicating that the UV-induced cell cycle arrest was reversible. An alkaline elution assay (14) was then used to measure DNA single-strand breaks that result from uninhibited activity of the DNA replication machinery in UV-damaged cells. Ectopic expression of Cdc25A enhanced formation of DNA single-strand breaks (Fig. 4C), likely resulting from collisions of replication forks with UV-induced DNA crosslinks (15). As no DNA strand breaks were induced by ectopic Cdc25A in the absence of UV irradiation (Fig. 4C), the abundance of Cdc25A appears to determine the extent of DNA synthesis upon UV-induced DNA damage. Thus, the elimination of Cdc25A evokes a cell cycle arrest permissive for repair of the DNA crosslinks caused by UV and protects the cells from formation of the DNA strand breaks. Finally, consistent with the third prediction for a checkpoint response, a 3-hour period of expression of Cdc25A to prevent down-regulation of the cellular Cdc25A activity by UV (Figs. 2, B and D, and 4C) reduced the survival of the irradiated cells examined by colony formation assays (Fig. 4D).

Figure 4

Effect of UV irradiation on DNA replication, DNA damage, and cell survival. (A) Asynchronously growing U-2-OS cells were irradiated with UV (15 J/m2). Two hours later they were incubated for 10 min with BrdU (100 μg/ml) and analyzed by multivariate flow cytometry (23) (FL1-H, BrdU signal revealed by the fluorescein isothiocyanate–conjugated antibody to BrdU; FL2-A, DNA content obtained by counterstaining the cells with propidium iodide). Numbers in parentheses indicate percentages of cells actively synthesizing DNA. (B) U-2-OS cells were arrested in mitosis by 12-hour incubation with nocodazole. Eight hours after release, the cells were exposed to UV as in (A) and incubated for another 6 hours before staining with propidium iodide and analysis of DNA content by flow cytometry (23). Numbers indicate percentages of cells that entered S phase. (C) Increased amount of UV-induced DNA single-strand breaks (SSB) in cells overexpressing Cdc25A. Expression of Cdc25A was transiently induced in U-2-OS/B3C4 cells (22) by removal of tetracycline for 3 hours before UV treatment. Thirty minutes after irradiation with the indicated doses of UV and readdition of tetracycline, formation of DNA SSB was measured by alkaline elution, exactly as described in (14). (D) Colony formation assay performed with U-2-OS/B3C4 cells exposed to the indicated doses of UV light and evaluated 12 days after irradiation. Cdc25A protein was transiently induced (−Tet) as described in (C).

Our results uncover a mechanism of cellular defense against genotoxic stress. The salient features of this G1-S checkpoint are proteasome-dependent destruction of the Cdc25A, independence of the p53-p21 pathway, involvement of Chk1 operating upstream of Cdc25A, and rapid execution of the response. The characteristics of the Chk1-Cdc25A-Cdk2 pathway, and the fact that the p53-p21 axis affects the cell cycle only several hours after UV treatment (7,16), indicate that the checkpoint response to DNA damage occurs in two waves. The Cdc25A-mediated mechanism may contribute to genomic stability by imposing a cell cycle block and preventing excessive damage of the genome before the p53-p21 pathway ensures a more sustained proliferation arrest (3, 17). Deregulation of either mechanism may cause genomic instability. The overabundance of Cdc25A found in subsets of aggressive human cancers (18) might prevent its timely degradation in response to DNA damage and thus provide a growth advantage through escape from the G1-S arrest and propagation of genetic abnormalities.

  • * To whom correspondence should be addressed. E-mail: bartek{at}


View Abstract

Navigate This Article