Main

To investigate whether degradation of Cdc25A affects the IR-induced S-phase checkpoint, we measured DNA synthesis in human U-2-OS/B3C4 cells engineered to express ectopic haemagglutinin A (HA)-tagged Cdc25A in a tetracycline-repressible manner10. Endogenous Cdc25A was degraded rapidly in response to IR, along with the dose-dependent inhibition of DNA synthesis (Fig. 1a); however, when the HA–Cdc25A protein was transiently elevated to levels high enough to prevent its degradation, it abrogated the S-phase checkpoint (Fig. 1a), resulting in radioresistant DNA synthesis (RDS) comparable to that seen in cells from ataxia-telangiectasia patients4,5,6.

Figure 1: Effects and regulation of IR-induced destruction of Cdc25A.
figure 1

a, DNA synthesis and Cdc25A abundance (inset) in γ-irradiated U-2-OS/B3C4 cells repressed (+Tet) or induced (-Tet) 1 h before IR to express ectopic HA–Cdc25A. b, Kinetics of IR-induced activities, and abundance of Cdc25A and Chk2, cyclin-E-associated H1 kinase activity and Cdk2 Tyr 15 phosphorylation in U-2-OS cells exposed to 10 Gy. c, DNA synthesis in untreated or 10-Gy γ-irradiated CD20-sorted U-2-OS/T-Rex cells expressing Cdc25A and/or Cdk2AF.

Full size image

We thought that if destruction of Cdc25A were involved in the S-phase checkpoint, it should be only transient, as the IR-induced inhibition of DNA replication lasts only several hours5. Our kinetic measurements showed that Cdc25A was almost completely downregulated by 30 min after irradiation, and recovered to the levels of non-irradiated cells 4–8 h later (Fig. 1b). The maximum downregulation of Cdc25A between 1 and 3 h after IR correlated with an increased inhibitory phosphorylation of Cdk2 on Tyr 15, a reduction of the S-phase-promoting cyclin E/Cdk2 kinase activity, and nearly 50% inhibition of DNA synthesis (Fig. 1b).

These results suggested that the main consequence of the IR-induced Cdc25A downregulation was inhibition of Cdk2, implying that interference with the Cdk2 inhibitory phosphorylation should abrogate the S-phase checkpoint. Indeed, brief conditional expression of the Cdk2AF allele14, in which the inhibitory Thr 14 and Tyr 15 are replaced by alanine and phenylalanine, respectively, mimicked the effect of Cdc25A overexpression and resulted in RDS (Fig. 1c). Simultaneous expression of both HA–Cdc25A and Cdk2AF did not further increase the degree of RDS (Fig. 1c), indicating that both proteins function in the same pathway.

Ultraviolet-light-induced downregulation of Cdc25A requires the activation of Chk1 (ref. 10), a key signal transducer that, together with Chk2, is implicated in checkpoint pathways activated by damaged or unreplicated DNA12,15,16,17,18,19. To assess whether Chk1 or Chk2 may regulate abundance of Cdc25A on IR, we measured their ability to phosphorylate Cdc25A. The activity and electrophoretic mobility of Chk1 remained unchanged until several hours after IR (Fig. 2a; and data not shown), but Chk2 became activated rapidly, as judged by its lower electrophoretic mobility and enhanced phosphorylation of GST–Cdc25A 1 h after IR (Fig. 2a). These results suggest that Chk2, but not Chk1, becomes activated with kinetics indicative of a link between IR-induced DNA damage and the rapid inhibition of S phase. Detailed time-course measurements showed that the kinetics of IR-induced Chk2 activity closely paralleled the loss of Cdc25A protein and phosphatase activity, the downregulation of cyclin E/Cdk2 by inhibitory phosphorylation, and the decrease in DNA synthesis (Fig. 1b).

Figure 2: Failure of Chk2 mutants to bind and induce degradation of Cdc25A abrogates the S-phase checkpoint.
figure 2

a, Immunoblots of Chk1 and Chk2 and their kinase activities towards GST–Cdc25A in untreated (-) or 10-Gy γ-irradiated (+) U-2-OS cells. b, Top, U-2-OS/B3C4 cells were transfected with Myc–Chk2 plasmids as indicated, stimulated to express HA–Cdc25A for 3 h, γ-irradiated (10 Gy; +) or left untreated (-), and immunoblotted for Myc–Chk2 and Cdc25A (WCE, whole-cell extract). Bottom, anti-Myc immunoprecipitates were assayed for co-precipitated Cdc25A and for in vitro kinase activity towards GST–Cdc25A (bottom). Asterisk, a threefold excess of Myc–Chk2-R145W was transfected. c, d, Various U-2-OS/Myc–Chk2 clones were analysed for Cdc25A abundance (c) and Cdk2-associated kinase activity (d) at the indicated time points after γ-irradiation (10 Gy). e, DNA synthesis in various U-2-OS/Myc–Chk2 clones measured after γ-irradiation (10 Gy). Open squares, U-2-OS/B3C4 cells with induced Cdc25A.

Full size image

To gain mechanistic insight into this link between Chk2 and Cdc25A, we assessed the IR responsiveness of Myc-tagged forms of human Chk2 transiently transfected into U-2-OS/B3C4 cells, including the engineered catalytically inactive D347A mutant12,20, and the R145W and I157T alleles13 with mutations in the putative protein-interaction FHA domain12,21,22, identified in sporadic colon cancer and as a germ-line mutation in the cancer-prone Li–Fraumeni syndrome, respectively13. On exposure to IR, the ectopic wild-type Chk2 became rapidly shifted into a more slowly migrating form, and increased its ability to interact physically with transiently elevated HA–Cdc25A in vivo and phosphorylate GST–Cdc25A in vitro (Fig. 2b). In contrast, basal activities of the three mutants of Chk2 were barely detectable in non-irradiated cells and remained low even after exposure to IR (Fig. 2b).

Notably, the D347A mutant still bound HA–Cdc25A after IR (Fig. 2b), implying that the interaction of Chk2 with its substrate(s) is mediated through its modification by an upstream regulator, independently of Chk2 autocatalytic activity. In contrast, the R145W and I157T cancer-associated mutants of Chk2 could not bind HA–Cdc25A, even though the I157T mutant preserved the IR-induced mobility shift (Fig. 2b). Thus, the tumour-associated Chk2 proteins seem defective in their ability to bind and phosphorylate substrates such as Cdc25A.

To examine whether Chk2 and Cdc25A function in a common pathway, we transfected U-2-OS cells with plasmids encoding the Chk2 alleles described in Fig. 2b, together with a gene coding for puromycin resistance. Whereas puromycin-resistant cell populations transfected with wild-type Chk2 or empty plasmid efficiently degraded Cdc25A when exposed to IR (Fig. 2c), cells expressing any of the Chk2 mutants retained substantial amounts of Cdc25A when irradiated (Fig. 2c). Consistent with these differential abilities to modulate Cdc25A protein levels, Cdk2 activity declined in cells transfected with empty vector or wild-type Chk2, but remained elevated in cells expressing the Chk2 mutants (Fig. 2d). Consequently, whereas the control cells responded by inhibiting DNA synthesis, cells expressing the Chk2 mutants failed to impose the S-phase blockade and exhibited RDS comparable to that of cells conditionally overexpressing Cdc25A (Fig. 2e). Thus, Chk2 alleles defective in catalytic activity or interaction with Cdc25A behaved as dominant-negative mutants that abrogated the S-phase checkpoint, which is consistent with Chk2 operating upstream of Cdc25A and Cdk2 in a common pathway.

In addition to Cdc25A, Chk2 also efficiently phosphorylates the p53 tumour suppressor in response to DNA damage20,23; however, the former substrate seems more relevant for S-phase arrest induced by IR, as p53 controls the G1 and sustains the G2/M checkpoints, rather than the S-phase checkpoint24,25,26, and RDS occurs independently of p53 status27. Consistent with this, rapid degradation of Cdc25A and loss of its phosphatase activity occurred in SW620 (wild-type for Chk2) but not in HCT-15 cells expressing mutant Chk2 after exposure to IR (ref. 13; Fig. 3a). Notably, both of these colon cancer cell lines express mutant p53 (ref. 13). Re-introduction of wild-type Chk2 into HCT-15 cells to a level exceeding that of the endogenous mutated protein restored the S-phase checkpoint response, including degradation of Cdc25A, increased inhibitory phosphorylation of Cdk2 and downregulation of cyclin E/Cdk2 kinase activity (Fig. 3b), and the ability to arrest S-phase progression (Fig. 3c). These results further support Chk2 as a mediator for rapidly destructing Cdc25A in response to IR, in a p53-independent S-phase checkpoint that protects cells against RDS.

Figure 3: IR-induced destruction of Cdc25A requires functional ATM and Chk2.
figure 3

a, Abundance and activities of Cdc25A and Chk2 in untreated (-) or γ-irradiated (10 Gy; +) SW620 and HCT-15 cells . b, HCT-15-derived cells were γ-irradiated (10 Gy; +) or left untreated (-) and analysed for activities and abundance of Cdc25A and Chk2, cyclin-E-associated H1 kinase activity, and Cdk2 levels (Tyr 15 phosphorylated and total). c, DNA synthesis in control vector versus Chk2-reconsitituted (Myc–Chk2) HCT-15 cells measured after IR (10 Gy). d, Normal or ataxia-telangiectasia (A-T) lymphoblasts were treated and analysed as in b.

Full size image

As RDS occurs in cells with a defective ATM gene4,5,6, as well as through deregulation of Chk2 or Cdc25A, and as ATM activates Chk2 directly (ref. 12), we next compared the response of the Chk2–Cdc25A–Cdk2 pathway in lymphoblasts derived from ataxia-telangiectasia patients to those isolated from normal individuals. Unlike normal lymphoblasts, irradiated ataxia-telangiectasia cells were unable to activate Chk2 and downregulate Cdc25A protein and activity (Fig. 3d). Consequently, exposure of ataxia-telangiectasia lymphoblasts to IR caused neither an increase in Cdk2 Tyr 15 phosphorylation nor an inhibition of cyclin E/Cdk2 kinase activity (Fig. 3d), which is consistent with the well-documented RDS phenotype in these cells5.

As signals from damaged DNA are often propagated through phosphorylation cascades, we next identified the Cdc25A residue(s) directly phosphorylated by Chk2, presumably to prime Cdc25A for rapid destruction after IR-induced DNA damage. We made glutathione S-transferase (GST)-coupled fragments derived from the Cdc25A regulatory domain and subjected them to phosphorylation by purified GST–Chk2. The fragment spanning amino acids 101–140 was strongly phosphorylated by wild-type but not by catalytically inactive Chk2 (Fig. 4a), and the sequence flanking Ser 123 of human Cdc25A, conserved in diverse mammalian species, matched the criteria for a Chk2/Chk1 consensus site identified in other proteins such as Cdc25C12,16,17,18 (Fig. 4b). Replacement of Ser 123 with alanine abolished the ability of the IR-activated Chk2 to phosphorylate the corresponding GST–Cdc25A fragment (Fig. 4c). The S123A mutation in the context of full-length Cdc25A resulted in a protein that, unlike wild-type Cdc25A, did not undergo an IR-induced shift when separated on an SDS gel (Fig. 4d), indicating that in vivo the S123A substitution abolished an important IR-dependent modification of Cdc25A. Significantly, the S123A mutation rendered Cdc25A resistant to IR-induced degradation under conditions in which a comparable amount of moderately overexpressed wild-type Cdc25A was still effectively degraded in vivo (Fig. 4e). We conclude that Chk2-dependent phosphorylation of Cdc25A on Ser 123 represents a critical step in promoting its rapid destruction in response to IR-induced DNA damage.

Figure 4: Chk2 phosphorylates Cdc25A on Ser 123 and triggers its IR-induced destruction.
figure 4

a, GST–Cdc25A fragments were incubated with purified wild type (WT) or catalytically inactive (KD) GST–Chk2. Proteins resolved by SDS–PAGE were visualized by autoradiography (top) or Coomassie staining (bottom). b, Amino-acid sequence flanking Ser 123 in human, mouse and rat Cdc25A, aligned with the Chk1/2-phosphorylated region of human Cdc25C. c, GST–Cdc25A(101–140) (WT) or GST–Cdc25A(101–140) (S123A) was incubated with Chk2 immunoprecipitates from untreated (-) or 10-Gy γ-irradiated (+) U-2-OS cells. Proteins were resolved and visualized as in b. d, U-2-OS cells transiently transfected with WT or S123A HA–Cdc25A were γ-irradiated (10 Gy; +) or left untreated (-) in the presence of proteasome inhibitor LLnL to prevent destruction, and cell lysates were analysed by immunoblotting. Arrow indicates IR-induced slower-migrating form of Cdc25A. e, U-2-OS cells transfected with WT or S123A HA–Cdc25A were γ-irradiated (10 Gy; +) or not (-) 12 h after transfection, incubated with cycloheximide (CHX; 25 µg ml-1) and immunoblotted for HA–Cdc25A at the indicated times after CHX addition.

Full size image

On the basis of this study, we propose that the ATM–Chk2–Cdc25A–Cdk2 pathway may provide the molecular explanation of the defence mechanism protecting human cells against RDS. The key components and events along this pathway in cells proficient in the S-phase checkpoint are outlined in Fig. 5 (left). The biological and pathophysiological relevance of this mechanism is further supported by the fact that tumour-associated defects of any of its main components cause RDS (Fig. 5, right), and may predispose to, or promote, tumorigenesis (refs 5, 13, 28; and this study). Cells undergoing DNA replication are particularly vulnerable to genotoxic stress including IR; the rapid degradation of Cdc25A and subsequent silencing of the Cdk2 activity may represent the initial defence barrier, which inhibits DNA synthesis to allow efficient repair. Cdc25A extends the list of important checkpoint mediators targeted by Chk2, currently encompassing the mitotic activator Cdc25C12, and the p53 (refs 20, 23) and BRCA1 (ref. 29) tumour suppressors, thereby implicating Chk2 in a complex network controlling G1, S and G2/M checkpoints, as well as DNA repair. Such a central role in protecting genome integrity is also consistent with the proposed candidacy of Chk2 for a new tumour suppressor, whose mutations may predispose to tumorigenesis in diverse tissues, as seen in patients with Li–Fraumeni syndrome13. Our finding that the tumour-associated Chk2 alleles are indeed loss-of-function mutants provides the missing functional evidence indicating that Chk2 is a genuine tumour suppressor.

Figure 5: Model of the IR-induced S-phase checkpoint pathway in normal (left) versus checkpoint-deficient (right) cells.
figure 5

Pathway components targeted in cancer are marked by an asterisk. Ub, ubiqitin chains.

Full size image

Methods

Antibodies and immunochemistry

Mouse antibody HE172 to cyclin E, as well as immunoblotting, immunoprecipitation, Cdc25A phosphatase activity and in vitro kinase assays have been described10. Rabbit antisera to Cdk2 (M2) and Chk1 (FL-476), and mouse antibody to Cdc25A (F-6, used to detect the endogenous protein) were from Santa Cruz. Rabbit antiserum to Tyr15-phosphorylated Cdk1/Cdk2 was from Calbiochem, and mouse monoclonal antibody to CD20 was from Becton Dickinson. The 9E10 antibody to the Myc epitope was a gift from G. Evan. We generated mouse monoclonal antibodies to human Chk2 (DCS-270) and Cdc25A (DCS-127) by standard hybridoma technology.

Plasmids

To construct the Myc-tagged Chk2 expression vector, the human Chk2 complementary DNA (a gift from S. Elledge) was amplified by PCR with Pfu polymerase (Stratagene) and cloned into a pcDNA3 vector (Invitrogen) containing a Myc tag. The catalytically inactive D347A12,20, and tumour-associated R145W13 and I157T13 mutants of Chk2, as well as the Cdc25A S123A mutant were generated using the QuikChange Site-Directed Mutagenesis kit (Stratagene). We expressed and purified GST–Chk2 and GST–Cdc25A10 (full length and fragments) according to standard procedures.

Cell culture and RDS assay

We grew U-2-OS cells10 and SW-620 cells in DMEM containing 10% fetal bovine serum (FBS). HCT-15 cells13 and lymphoblasts derived from normal and ataxia-telangiectasia patients (donated by Y. Shiloh) were grown in RPMI with 10% FBS. The U-2-OS-derived B3C4 clone conditionally expressing haemagglutinin (HA)-tagged Cdc25A has been described10. Expression of the transgene was induced by culturing the cells in tetracycline-free medium for the durations specified in the figure legends. The U-2-OS-derived T-Rex cell line was from Invitrogen. We induced expression of Cdc25A and Cdk2AF by adding tetracycline (1 µg ml-1) 4 h before irradiation (Fig. 1c); ectopic CD20 was expressed throughout the experiment. Isolation of CD20-positive U-2-OS/T-Rex cells with anti-CD20-coupled Dynabeads has been described30. The U-2-OS- and HCT-15-derived polyclonal cell lines stably expressing Myc-tagged wild-type or mutant Chk2 were generated by calcium phosphate transfection (U-2-OS) or electroporation (HCT-15) of cells with empty pcDNA3–Myc vector or pcDNA3–Myc vector containing wild-type or mutant Chk2. A vector encoding puromycin resistance (pBabe-puro) was co-transfected, the cells were selected for neomycin (G418; 400 µg ml-1) and puromycin (1 µg ml-1) resistance, and pooled when visible colonies emerged. Ionizing radiation was delivered by X-ray generator (RT100, Philips Medico; 100 kV, 8 mA, dose-rate 0.92 Gy min-1), and cell extracts were prepared 1 h later except where indicated otherwise. Inhibition of DNA synthesis and RDS after ionizing radiation were monitored as described6,26.