Chk1 is dispensable for G2 arrest in response to sustained ... - Nature

3 downloads 30 Views 1MB Size Report
May 2, 2011 - (DDR), which senses the damage and coordinates a multitude of different ...... (Microsoft Corp, Redmond, WA, USA) softwares. Antibodies.
Oncogene (2011) 30, 4261–4274

& 2011 Macmillan Publishers Limited All rights reserved 0950-9232/11 www.nature.com/onc

ORIGINAL ARTICLE

Chk1 is dispensable for G2 arrest in response to sustained DNA damage when the ATM/p53/p21 pathway is functional G Lossaint1,3, E Besnard2, D Fisher1, J Piette1 and V Dulic´1 1 Institut de Ge´ne´tique Mole´culaire de Montpellier, UMR 5535 CNRS-Universite´ Montpellier 1 et 2, Montpellier, France and 2Institut de Ge´ne´tique Fonctionelle, UMR 5203 CNRS, U661 INSERM, 141, rue de la Cardonille, Montpellier, France

In the presence of sustained DNA damage occurring in S-phase or G2, normal cells arrest before mitosis and eventually become senescent. The checkpoint kinases Chk1/ Chk2 and the CDK inhibitor p21 are known to have important complementary roles in this process, in G2 arrest and cell cycle exit, respectively. However, additional checkpoint roles have been reported for these regulators and it is not clear to what extent their functions are redundant. Here we compared the respective roles of Chk1, Chk2 and p21 in DNA damage-induced G2 arrest in normal human fibroblasts, normal epithelial cells and frequently used p53 proficient cancer cells. We show that in normal cells, Chk1, but not Chk2, is involved in G2 arrest whereas neither are essential. In contrast, p21 is required. However, Chk1, but not Chk2, becomes necessary for arrest in U2OS osteosarcoma cells. We find that their ATM/p53/p21 response in G2 phase is defective, like in other cancer cells with wild-type p53, and conclude that cross-talk between the Chk1 and p21 pathways allows them to switch dependency for G2 arrest onto Chk1. Using the specific ATM inhibitor KU-55933 we confirm the essential role of ATM in the induction of p21 for G2 arrest of normal cells. Efficient p21 induction is required for nuclear sequestration of inactive cyclin B1-Cdk1 complexes preceding irreversible cell cycle exit in G2. Our results demonstrate that p21 is able to fulfill the Chk1 functions in G2 arrest under continuous genotoxic stress, which has important implications for cancer chemotherapy. Oncogene (2011) 30, 4261–4274; doi:10.1038/onc.2011.135; published online 2 May 2011 Keywords: G2-M checkpoint; cyclin B1; cell cycle; Chk2; ATM

Introduction In the presence of genotoxic stress, cells activate a global signaling network, termed the DNA damage response (DDR), which senses the damage and coordinates a Correspondence: Dr J Piette or Dr V Dulic´, Institut de Ge´ne´tique Mole´culaire de Montpellier, UMR 5535 CNRS-Universite´ Montpellier 1 et 2, 1919 Route de Mende, Montpellier 34293 Herault, France. E-mail: [email protected] or [email protected] 3 Current address: Institut de Ge´ne´tique Humaine, CNRS UPR 1142, 141, rue de la Cardonille, 34396 Montpellier, Cedex 5, France. Received 3 October 2010; revised and accepted 19 March 2011; published online 2 May 2011

multitude of different pathways that either arrest the cell cycle to allow DNA repair, or prevent proliferation of potentially genetically unstable cells by inducing apoptosis or senescence (Harper and Elledge, 2007). Inactivation of this response has long been suspected to have an important role in tumor progression, a hypothesis that is strongly supported by the recent analyses of early tumor samples, which express markers of an ongoing DDR (Bartkova et al., 2005; Gorgoulis et al., 2005). Further progression of these pre-tumoral lesions to tumors would thus require the inactivation of crucial elements of the DDR checkpoint that include the p53 tumor suppressor pathway (Bartek et al., 2007). Moreover, patients bearing mutations in DDR pathways frequently show severe genetic disorders and a predisposition to cancer, for example those suffering from ataxia telangiectasia (AT, Hoeijmakers, 2001; Shiloh, 2003). ATM (AT mutated) is a member of a phosphoinositide-3-kinase-related protein kinase family that, together with ATR (ATM and Rad3-related), has a central role in coordinating the DDR, including cell cycle checkpoint control, DNA repair and apoptosis (Lavin, 2008). Among targets of ATM and ATR is p53, which has a key role in controlling DNA damage-induced G1/S and G2/M checkpoints (Taylor and Stark, 2001). The principal role of the G2/M checkpoint is to delay or prevent mitosis in the presence of DNA lesions, thus avoiding mitotic segregation of damaged chromatids, which could lead to gains, losses or other rearrangements of the genome. Therefore, not only should mitotic entry be halted promptly but also G2 arrest must be efficient and persist as long as the damage is not repaired. If this is not the case, mitotic slippage in the presence of persisting damage could lead to tetraploidy, which is considered to be an initial step towards aneuploidy, a hallmark of solid tumors (Kops et al., 2005). Cyclin B1-Cdk1, key mitotic regulator, is the ultimate target of the G2 DNA damage checkpoint. Its inactivation or, rather, prevention of its activation is a conditio sine qua non to delay or block G2/M progression in the case of genotoxic stress. According to the current paradigm, this is achieved by ATM/ATR-activated Chk1 and Chk2 kinases that, by inhibiting the Cdc25 phosphatase family, maintain cyclin B1-Cdk1 in an inactive cytoplasmic state (Bartek and Lukas, 2003; Lobrich and Jeggo, 2007; Chen and Poon, 2008; Toettcher et al., 2009). However, several studies have questioned the role of Chk2, implying that

Roles of Chk1 and p21 in G2 arrest G Lossaint et al

4262

Chk1 might be the major G2/M checkpoint regulator (Ahn et al., 2003; Jallepalli et al., 2003; Syljuasen et al., 2006; Jin et al., 2008; Zhang et al., 2008). More recent work showed that another ATM/ATR-activated kinase complex, consisting of p38MAPK and MK2 and operating mainly in p53-deficient cells, might also contribute to G2 arrest acting in parallel with Chk1 (Reinhardt et al., 2007). A parallel, and reportedly slower, pathway ensures the maintenance of G2-arrest and involves two p53-induced proteins: the CDK inhibitor (CKI) p21Waf1/Cip1 (p21) (Bunz et al., 1998; Dulic´ et al., 1998; Chan et al., 2000) and 14-3-3s (Hermeking et al., 1997). 14-3-3s was proposed to block mitosis by sequestering cyclin B1-Cdk1 in the cytoplasm (Chan et al., 1999). On the other hand, p21 was proposed to sustain G2 arrest by inhibiting cyclin B1 and Cdk1 expression (Innocente et al., 1999; Flatt et al., 2000) and by promoting cyclin B1 degradation (Gillis et al., 2009; Lee et al., 2009). More recently, Chk1 was also implicated in the maintenance of G2 arrest through repression of cyclin B1 and Cdk1 (Shimada et al., 2008). Consistent with its essential role in senescence, p21 was shown to drive irreversible cell cycle exit in G2 in normal human fibroblasts (NHF) by blocking pre-mitotic phosphorylation of pRb family pocket proteins (Baus et al., 2003; Jackson et al., 2005). Several observations in non-transformed cells suggested, however, that p21 might block G2/M progression by directly inhibiting mitotic Cdks. In NHF exposed to DNA damage after the G1/S transition, p21 not only inhibits cyclin A-Cdk1/2 (Baus et al., 2003) but also sequesters cyclin B1-Cdk1 in the nucleus, thereby preventing its activation by CAK or Cdc25 in the cytoplasm (Charrier-Savournin et al., 2004; Foijer et al., 2005). In spite of this potential second mechanism for G2 arrest, Chk1 appears sufficient for G2 arrest in U2OS cells which have wild-type p53 and which can induce p21 (Syljuasen et al., 2006). In the light of these contradictory results and the lack of evidence that, unlike Chk1/2, p21 directly inhibits mitotic Cdks in other cell models, we re-assessed the respective roles of p21, Chk1 and Chk2 in the G2 response to DNA damage. By combining time-lapse livecell imaging with small interfering (si)RNA-mediated knockdown, we show that, in non-transformed human cells, p21 has an essential role in G2 arrest in the presence of sustained DNA damage. Chk1 is, however, essential in the case of p21 depletion or inefficient induction in G2, as in several widely studied cancer cell lines.

Results

2003). This genotoxic drug enables the study of the G2/M checkpoint without previous cell synchronization that could also activate the ATR-Chk1 pathway. The DDR to ICRF, including the amount of gH2AX foci, is comparable to the DDR provoked by ionizing radiation (Charrier-Savournin et al., 2004 and Supplementary Figure S1) and its persistence confers irreversible cell cycle arrest in G2 (Baus et al., 2003). This G2 exit entails rapid onset of premature senescence, as judged by inhibition of pRb phosphorylation, downregulation of A and B1 cyclins (Figure 1a), induction of senescence-associated heterochromatin foci (SAHF; Narita et al., 2003; Supplementary Figure S2A) and b-galactosidase staining (Dimri et al., 1995, Supplementary Figure S2B). In contrast, NHF stably expressing the HPV16-E6 oncogene (NHF-E6), a potent inhibitor of the p53/p21 pathway (Baus et al., 2003), failed to exit cell cycle in G2, which resulted in aberrant mitoses and accumulation of fragmented nuclei (Supplementary Figure S2B). Firstly, we assessed the kinetics of activation of the ATM/ATR signaling pathway by comparing p21 induction vs Chk1/Chk2 phosphorylation after addition of ICRF or the radiomimetic drug bleomycin. Bleomycin provokes a greater amount of DNA damage than ICRF resulting in a stronger checkpoint response leading to both G1 and G2 arrest (Figures 1a and b; Baus et al., 2003). However, the strong DDR to bleomycin is largely due to G1 cells as the number of DNA damage-induced foci containing the PS1981-ATM signal is comparable between the two drugs (Figure 1b). As expected, both agents induced rapid and persistent phosphorylation of ATM and its targets, Chk2 and p53 (Figure 1a). Note that although Chk2 is rapidly phosphorylated on T68 (Figure 1a), the induction of the SDS–polyacrylamide gel electrophoresis (PAGE) mobility shift, corresponding to fully active enzyme (Bartek and Lukas, 2003), appears to be considerably slower (Figure 1a, arrowhead). In contrast to ATM, ATR phosphorylation was unaffected by DNA damage. However, both drugs stimulated phosphorylation of its target Chk1, which was transient and occurred in parallel with p21 induction only few hours after drug addition (Figure 1a). Interestingly, despite unequal intensity of ATM activation by ICRF and bleomycin, p21 induction by either drug was comparable. Finally, p21 strongly associated with cyclin B1-Cdk1 (Supplementary Figure S3) leading to its pre-mitotic nuclear retention (Figure 1c). These results suggest that p21, Chk1 and Chk2 act in parallel and that direct inhibition of mitotic cyclin-Cdk complexes by p21 might have an important role in G2 arrest.

DDR in NHF involves simultaneous Chk1/Chk2 activation and p21 induction leading to pre-mitotic nuclear sequestration of cyclin B1 and cell cycle exit in G2 To determine the relative importance of Chk1, Chk2 and p21 in early stages of DNA damage-induced G2 arrest, we investigated NHFs exposed to ICRF-193 (ICRF), a DNA topoisomerase II inhibitor that induces a G2 cell cycle arrest (Downes et al., 1994; Baus et al.,

In non-transformed human fibroblasts and epithelial cells DNA damage-induced G2 arrest requires Chk1 and p21 but not Chk2 Next, we directly tested the contributions of p21, Chk1 or Chk2 in DNA damage-induced G2 arrest by using siRNA-mediated knockdown (KD; Figure 2a) and timelapse live-cell imaging to monitor both the frequency

Oncogene

Roles of Chk1 and p21 in G2 arrest G Lossaint et al

4263

Figure 1 In NHF, p21 is induced in parallel with Chk1/Chk2 phosphorylation in response to DNA damage (a). Immunoblots showing phosphorylation status or proteins levels of different cell cycle and checkpoint regulators in the presence of ICRF or bleomycin. Asynchronously growing NHFs were exposed to genotoxic drugs for the indicated times. Arrowhead points at phosphorylation-induced SDS–PAGE mobility shift of Chk2. LC, loading control. Upper panel: NP40 soluble extracts. Lower panel: NP40 non-soluble extracts containing chromatin-associated proteins. LE, longer exposure. (b) Immunofluorescence analysis showing PS1981-ATM and gH2AX in non-treated control (Cont) and NHF exposed for 12 h to ICRF and bleomycin. Bar, 10 mM. (c) Immunofluorescence analysis showing co-localization of cyclin B1 (cycB1) and p21 in non-treated control (Cont) and NHF arrested in G2 by ICRF (16 h). Arrows indicate cells in G2 phase and prophase (pro) in non-treated cultures. Bar, 10 mM.

and the kinetics of mitotic entry in the presence of ICRF. Although Chk1 KD, Chk2 KD or Chk1/Chk2 double KD failed to abrogate G2 arrest, triple p21/

Chk1/Chk2 knockdown (TKD) efficiently promoted mitosis (Figure 2b, Supplementary Figure S4A and unpublished data). Knockdown of p21 alone also promoted mitosis, albeit with a delay of several hours in respect to TKD. In this case, however, DNA damageinduced Chk1 phosphorylation was invariably increased (Figure 2a, arrow). These results show not only that p21 is required for G2 arrest maintenance, as expected (Bunz et al., 1998), but also that its induction is sufficient to block mitosis in the absence of Chk1 and Chk2, which is consistent with its strong association with cyclin B1Cdk1. Moreover, they suggest that p21 may cooperate with checkpoint kinases under normal circumstances in the establishment of a stable G2 arrest. To determine which checkpoint kinase mediates G2 arrest, we first generated Chk1 and Chk2 KD in NHF expressing HPV16-E6 (NHF-E6; Figure 2c). In these cells, Chk1 KD, but not Chk2 KD, accelerated G2/M progression in the presence of ICRF or partially abolished G2 arrest by the radiomimetic drug bleomycin (Figure 2d and Supplementary Figure S5A). Moreover, UCN-01, a potent Chk1 inhibitor (Graves et al., 2000), abrogated G2 arrest only if p21 was depleted (Figure 2e and Supplementary Figure S4B) or absent, as in NHF-E6 (Supplementary Figure S5B). These results show that, in NHF, in the absence of p21, DNA damage-induced G2 arrest mainly relies on Chk1; second, both Chk1 and p21 can arrest cells in G2, but sustained arrest requires p21. To test whether these findings can be generalized to epithelial cells, we studied non-transformed human mammary epithelial cells (HMEC) immortalized by hTERT. Similarly to NHF, HMEC arrested in G1 and G2 with bleomycin, and in G2 when treated with ICRF (Supplementary Figure S6A). Although in these cells ICRF induced a weaker DDR than bleomycin, both drugs similarly induced p21 levels and blocked pRb phosphorylation leading to irreversible cell cycle arrest (Figures 3a and b and data not shown). Importantly, p21 was strongly induced in S and G2 phases, as all drug-treated cells expressing cyclin A also displayed nuclear p21 (Figure 3c, upper panel) resulting in cyclin A-Cdk1/2 inactivation (Figure 3d). As in NHF, p21 also bound cyclin B1-Cdk1 (Figure 3e) and ICRF-arrested cells accumulated nuclear cyclin B1 coincidently with strong p21 induction (Figure 3c, lower panel). Accordingly, in ICRF-treated HMEC-E6 (which have no detectable p21 even after DNA damage) cyclin B1 was exclusively cytoplasmic (Figure 3f). Finally, as in fibroblasts, Chk1 knockdown promoted mitosis in ICRF-treated HMEC only if p21 was also depleted (Figures 3g and h and Supplementary Figure S6B). Surprisingly, however, unlike in NHF (Figure 1b) p21 knockdown alone failed to abrogate G2 arrest in HMEC (Figure 3h and Supplementary Figure S6). This is possibly due to stronger Chk1 activation (Figure 3g, arrow). Collectively, our results suggest that, in nontransformed cells, p21 and Chk1 cooperate to induce DNA damage-induced G2 arrest. Moreover, the absence of p21 augments Chk1 activation, suggesting negative feedback between Chk1 and p21. Oncogene

Roles of Chk1 and p21 in G2 arrest G Lossaint et al

4264

Figure 2 In NHF, DNA damage-induced arrest requires Chk1 and p21 but not Ck2. (a) Immunoblots showing phosphorylation status (Chk1, Chk2) or protein levels (Chk1, Chk2, p21) in non-treated (Ctl-luc) and ICRF-treated (16 h) NHF in which p21, Chk1 and Chk2 (Ch1 þ 2) or p21, Chk1 and Chk2 (triple knock-down, TKD) were previously depleted by siRNA. Arrowhead points at phosphorylation-induced SDS–PAGE mobility shift of Chk2. Arrow, PS317-Chk1. Luc, luciferase. (b) Time-lapse data showing entry into mitosis (percentage of total cells) of non-treated (Ctl-luc) and ICRF-treated control (Luc), p21 knockdown (sip21), double Chk1 þ Chk2 KD and triple p21 þ Chk1 þ Chk2 KD (TKD) in NHF cultures (described in (a): see Materials and methods). New mitotic cells were counted within 10-h intervals (mean and mean deviation of five fields in two separate experiments are given). In nontreated cultures, within the presented time interval (20 h), Chk1 and Chk2 knockdown did not significantly influence cell cycle progression (data not shown). For kinetics see Supplementary Figure S4A. (c) Chk2 and PS317-Chk1 levels (western blots) in extracts of ICRF-treated NHF-E6 in which Chk1, Chk2 or both kinases (Chk1 þ 2) were depleted by siRNA. Arrowhead points at phosphorylation-induced SDS–PAGE mobility shift of Chk2. (d) Time-lapse data showing entry into mitosis of ICRF- and bleomycin (Bleo)-treated NHF-E6 in which Chk1, Chk2 or both kinases (Chk1 þ 2) were depleted by siRNA. Ctl-Luc, untreated control cells. Other experimental conditions were as described in (b). For kinetics see Supplementary Figure S5A. (e) Time-lapse data showing entry into mitosis of ICRF-treated control (Luc) and p21 KD NHF in the presence of Chk1 inhibitor UCN-01 (mean and mean deviation of six fields in two separate experiments are given). For kinetics see Supplementary Figure S4B and 5B (NHF-E6 cells).

Chk1 is essential in G2 arrest of U2OS cells as p21 accumulation is impaired These data contrast with results from human osteosarcoma U2OS cells, in which Chk1 inhibition/ablation alone promoted mitosis in the cells arrested in G2 by ionizing radiation (IR; Syljuasen et al., 2006). One possibility was that the G2 DDR uses distinct mechanisms in U2OS cells and non-transformed cells, or that sustained treatment with DNA damaging agents does not induce the same response as IR. To distinguish these possibilities, we tested whether Chk1 depletion could abrogate G2 arrest also in the presence of ICRF or bleomycin in U2OS cells (Figures 4a and b). Time-lapse data showed that Chk1 KD, but not Chk2 KD, Oncogene

efficiently abrogated G2 arrest induced by both drugs (Figure 4c and Supplementary Figure S7). This suggests that IR and DNA-damaging drugs provoke similar response mechanisms (cf. Supplementary Figure S8), but their effects differ between U2OS and non-transformed cells. Notably, these data imply that p21 does not have an important role in G2 arrest in U2OS cells. To explore these differences, we compared the response of NHF and U2OS cells to ICRF and bleomycin. Firstly, we found that, in contrast to NHF and HMEC, in G2 arrested U2OS cells cyclin B1 was largely cytoplasmic, which coincided with the absence of p21 (Figure 5a). Indeed, the immunoblot data revealed that p21 induction was strongly impaired in U2OS cells,

Roles of Chk1 and p21 in G2 arrest G Lossaint et al

4265

Figure 3 In the presence of p21, Chk1 is dispensable for G2 arrest in HMEC but it is hyper-activated in its absence. (a) Immunoblots showing phosphorylation status and/or proteins levels of different cell cycle and checkpoint regulators in response to sustained DNA damage. Asynchronously growing HMEC were exposed to bleomycin (Bleo) or ICRF for the indicated times. Ctl, non-treated cultures. Asterisk, weak PS317-Chk1 band in ICRF-treated cells. Upper panel: NP40 soluble extracts. Lower panel: NP40 non-soluble extracts containing chromatin-associated proteins. Arrowhead indicates phosphorylation-induced SDS–PAGE shift of Chk2. Arrowheads, phosphorylation-induced Chk2 SDS–PAGE mobility shift. LC, loading control. (b) Immunofluorescence analysis showing PS1981-ATM and gH2AX in non-treated control (Cont) and HMEC exposed for 12 h to ICRF and bleomycin. Bar, 10 mM. (c) Immunofluorescence analysis showing cyclin A/p21 (upper panel) and cyclin B1/p21 co-staining (lower panel) in non-treated control (Cont), bleomycin (Bleo; cyclin A)- or ICRF (cyclin B1)-treated HMEC. Asynchronous HMEC cultures were exposed to genotoxic agents for 16 h before fixation. Arrows indicate G2-arrested cells that accumulate both p21 and cyclin A. Bar, 10 mM. (d) Histone H1 kinase (H1K) and immunoblot analysis of cyclin A immunoprecipitates from non-treated (–) and bleomycin-treated ( þ ) HMEC (16 h). (e) Immunoblot analysis showing cyclin B1 and Cdk1 in p21 immunoprecipitates isolated from the extracts of non-treated (Ctl), bleomycin- and ICRFtreated (12 h) HMEC. (f) Immunofluorescence showing cyclin B1/p21 co-staining in ICRF-treated HMEC-E6 ( þ E6). Asynchronous HMEC-E6 cultures were exposed to ICRF for 16 h. Bar, 10 mM. (g) Immunoblots showing indicated proteins in cell extracts from non-treated (Ctl-Luc) and ICRF-treated (16 h) HMEC in which Chk1, p21 or p21 and Chk1 (double knock-down; p21/Ch1) were previously depleted by siRNA oligonucleotides (24 h prior to drug addition). Arrow points to increased PS317-Chk1 signal in ICRF-treated p21 knockdown HMEC. LC, loading control. (h) Entry into mitosis (percentage of total cells) of non-treated (Ctl-luc) and ICRF-treated control (Luc), p21 KD, Chk1 KD and p21 þ Chk1 KD HMEC was scored by inspection of video-microscopy sequences. New mitotic cells were counted within 10-h intervals (mean and mean deviation of nine separate fields in three separate experiments). For kinetics see Supplementary Figure S6B.

suggesting a deficient ATM activation (Figure 5a). This was confirmed by immunoblots showing diminished phosphorylation of both ATM (PS1981-ATM) and

chromatin-bound p53 (PS15-p53) in the presence of bleomycin (Figure 5a), which could also explain the absence of G1 arrest in U2OS cells (Figure 4a). Oncogene

Roles of Chk1 and p21 in G2 arrest G Lossaint et al

4266

Figure 4 In U2OS cells, Chk1 alone is necessary to induce G2 arrest. (a) FACS analysis showing DNA content of non-treated (Cont) U2OS cells, the cells exposed for 24 h to ICRF or bleomycin (Bleo), and the cells exposed to ionizing radiation (IR, 5 Gy) 24 h before analysis. Numbers indicate the percentage of 2N and 4N cell populations. (b) Immunoblots showing Chk2 and Chk1 levels in nontreated control (Ctl-luc) and drug-treated U2OS cells depleted for Chk1 (Ch1), Chk2 (Ch2) or both (Ch1 þ 2) kinases. Twenty-four hours after transfection, asynchronous U2OS cultures were exposed to ICRF or bleomycin (Bleo) for 16 h. LC, loading control; Luc, luciferase. (c) Time-lapse data showing entry into mitosis (percentage of total cells) of non-treated U2OS cells (Ctl-Luc) and ICRF- or bleomycin-treated control (Luc), Chk1 KD, Chk2 KD and Chk1 þ Chk2 KD (siChk1 þ 2) cells (described in a). New mitotic cells were counted within 10-h intervals during 30 h (mean of two fields in two experiments). Within the first 24 h, Chk1 and Chk2 knockdown did not significantly influence cell cycle progression in non-treated cultures (data not shown). This video-microscopy experiment was carried out in parallel with the one described in (b). For kinetics see Supplementary Figure S7.

Moreover, U2OS cells exhibited impaired Chk2 phosphorylation, notably with respect to the SDS–PAGE mobility shift (see also Supplementary Figure S8). In contrast, both drugs induced strong and persistent Chk1 phosphorylation. This is reminiscent of the results obtained in NHF and HMEC, in which p21 depletion upregulates Chk1 activation, suggesting that increased Chk1 activity might have compensated for defective p21 induction. However, G2-arrested U2OS cells maintained hyper-phosphorylated pRb and high cyclin A levels even after prolonged drug treatment (Figure 5a and Supplementary Figure S8). Thus, although hyper-active Chk1 can maintain G2 arrest for more than 30 h, inefficient p21 induction strongly compromised cell-cycle exit into senescence (see also Supplementary Figure S8). Deficient p21 induction in G2 coincides with an impaired ATM response in carcinoma cells Next, we tested whether pre-mitotic p21 induction might be also defective in other p53-proficient transformed cell Oncogene

lines (MCF7, HCT116) that were frequently used to study the DDR (Bunz et al., 1998). Indeed, despite strong overall p21 induction in the presence of ICRF or bleomycin (see below), immunofluorescence data showed that p21 was either completely absent (MCF7: Figures 6a and b) or inefficiently induced (HCT116: Figures 6c and d) in G2. Moreover, this invariably coincided with the absence of pre-mitotic nuclear retention of cyclin B1 (Figures 6b and d) and, most importantly, with an impaired G2/M checkpoint. This was especially flagrant in ICRF-treated MCF7 and HCT116 cultures that accumulated numerous mitotic cells (Figures 6e and g and Supplementary Figure S9). Why should pre-mitotic induction of p21 be impaired in these carcinoma cells? One possibility is a deficient ATM signaling as suggested by the results in U2OS cells. Indeed, in MCF7 and HCT116 cells both drugs stimulated phosphorylation of ATM (PS1981) or its substrate Chk2 much less efficiently (Figures 6e–h and Supplementary Figure S10A). Interestingly, although in HCT116 cells bleomycin strongly stimulated Chk2

Roles of Chk1 and p21 in G2 arrest G Lossaint et al

4267

Figure 5 In U2OS cells, Chk1 is hyperactive compensating for a deficient ATM pathway and delayed p21 induction. (a) Immunofluorescence analysis showing co-staining of cyclin B1 (cycB1) and p21 in bleomycine (Bleo) or ICRF-treated U2OS cells. Arrows indicate prophase (pro) and G2 cells. (b) Immunoblots comparing phosphorylation status or protein levels of different cell cycle or checkpoint regulators between NHF and U2OS cells exposed to bleomycin (Bleo) and ICRF (IC) for the indicated times. Upper panel: Triton soluble extracts. Lower panel: Triton insoluble extracts containing chromatin-associated proteins. Arrowheads indicate phosphorylation-induced SDS–PAGE mobility shift of Chk2. LC, loading control.

phosphorylation on T68, it failed to induce the SDS– PAGE shift, suggesting that the latter event is dependent more on ATM than the former (arrowhead, Figure 6h). Quite surprisingly, although phosphorylation of p53 was dramatically impaired in MCF7 cells (Figure 6F), both S15 and S20 were efficiently phosphorylated in HCT116 cells (Figure 6h). However, unlike MCF7 cells, whose ATR signaling, as judged by Chk1 phosphorylation, was also defective (Supplementary Figure S10B), in HCT116 cells bleomycin—but not ICRF—strongly stimulated Chk1 (Figure 6h and Supplementary Figure S10A). Altogether, these results suggest that impaired p21 induction in G2 in p53-proficient cancer cell lines might be due to a deficient ATM signaling pathway leading to a compromised G2/M checkpoint.

ATM inhibition blocks pre-mitotic p21 induction and cyclin B1 nuclear retention abrogating G2 arrest To examine this possibility, we tested the requirement of ATM for G2-specific functions of p21 in non-transformed cells by using caffeine or the specific ATM inhibitor KU-55933 (Hickson et al., 2004). As KU55933 induced a strong stress response in HMEC its effects were preferentially studied in NHF whose proliferation was unperturbed by this drug (unpublished data). As shown in Figure 7a, in ICRF-treated NHF both drugs readily inhibited ATM auto-phosphorylation and its appearance at the DNA damage-induced foci (Figure 7b; Berkovich et al., 2007). They also reduced p53 (S15) but not Chk2 phosphorylation on T68 suggesting that PS15-p53 but not PT68-Chk2 is due to Oncogene

Roles of Chk1 and p21 in G2 arrest G Lossaint et al

4268

Oncogene

Roles of Chk1 and p21 in G2 arrest G Lossaint et al

4269

ATM activation. Surprisingly, although KU-55933 also inhibited ATM and p53 phosphorylation in the presence of bleomycin, caffeine was much less efficient albeit it did block p21 induction. On the other hand, KU-55933 and caffeine had no effect on phosphorylation of ATR and DNA-dependent protein kinase (DNA-PK; Figure 7a and Supplementary Figure S11). Next, we asked how KU-55933 and caffeine affect different regulators of the G2/M checkpoint in synchronized NHF. Video-microscopy data showed that both agents readily abrogated G2 arrest by ICRF (Figure 7c) while they were somewhat less efficient in the presence of bleomycin (Supplementary Figure S12). In addition, caffeine abrogated ICRF-induced G2 arrest in HMEC (Supplementary Figure S13). However, after mitosis the cells stopped dividing indicating that ATM inhibition did not prevent post-mitotic cell cycle arrest. Immunoblot analysis in synchronized cells confirmed similarities observed in the action of caffeine and KU-55933 (Figure 7d). As predicted, by inhibiting ATM-p53 pathway both agents prevented p21 accumulation in G2 thus completely abolishing pre-mitotic cyclin B1 nuclear retention in NHF and in HMEC (Figures 7e and f). Moreover, although blocking the Chk2 mobility shift they did not abolish T68 phosphorylation, indicating that ATM is dispensable for this modification. This conclusion is supported by the observations in ATMdeficient HMEC (Arlander et al., 2008) or NHF (Tomimatsu et al., 2009). On the other hand, KU55933 and caffeine had a limited effect on the ATR pathway, as they did not block Chk1 activation (Figure 7d). This could explain the observed delay in mitotic entry in ICRF-treated cells (Figure 7c) as well as partial abrogation of bleomycin-induced G2 arrest (Supplementary Figure S12). Finally, re-enforcing the key role of p21 in the onset of senescence in G2 arrested cells, ATM inhibition blocked pre-mitotic accumulation of hypo-phosphorylated pRb (Figure 7d, arrows). In contrast, ATM inhibition did not prevent post-mitotic p21 induction,

and cells could still arrest in G1 (Figures 7c and d, ICRF-24 h) as documented by the absence of cyclin A. In conjunction with our observations in cancer cell lines, these results suggest that fully active ATM is absolutely required for p21 induction in G2 and efficient inhibition of mitotic Cdks in response to DNA damage. Discussion According to the currently accepted model (Lobrich and Jeggo, 2007; Chen and Poon, 2008; Reinhardt and Yaffe, 2009; Toettcher et al., 2009), DNA damageinduced G2 arrest and inactivation of the key mitotic regulator, cyclin B1-Cdk1, requires action of two checkpoint kinases, Chk1 and Chk2. Although the role of Chk2 in G2 arrest has been contested (Ahn et al., 2003; Jallepalli et al., 2003; Jin et al., 2008), an everincreasing family of Chk1 targets (Dai and Grant, 2010) reinforces the notion that Chk1 is the major player of the DNA damage checkpoint and hence a promising target in combined anticancer therapy (Merry et al., 2010). Here we show that, in the presence of a sustained genotoxic stress, Chk1 is not essential for G2 arrest if p21 is promptly and properly induced. This is the case in non-transformed human fibroblasts (NHF) and epithelial cells (HMEC) but not in the p53-proficient U2OS osteosarcoma cell line in which p21 induction is defective due to impaired function of the ATM pathway and in which G2 arrest relies entirely on Chk1. In contrast, Chk2 seems to be dispensable for G2 arrest under both circumstances. We also found that G2 functions of p21 are impaired in several commonly studied carcinoma cell lines (MCF7, HCT116) and this invariably coincided with a defective ATM signaling pathway and compromised G2/M checkpoint. However, the extent of those defects depends greatly on the genotoxic agent (bleomycin vs ICRF-193) and the cell line (Figure 8). In MCF7 cells,

Figure 6 Deficient p21 induction in G2 coincides with an impaired ATM response in carcinoma cells. (a) Immunofluorescence analysis showing co-staining of cyclin A (cycA) and p21 in non-treated (Cont) and bleomycin (Bleo)-treated MCF7 cultures (16 h). Note the absence of p21 in cyclin A-expressing cells (arrowheads). DNA was visualized by Hoechst-33258 staining. Bar, 10 mM. (b) Immunofluorescence analysis showing co-staining of cyclin B1 (cycB1) and p21 in non-treated (Cont) and ICRF-treated MCF7 cultures (16 h). Asterisk (*) indicates mitotic cells. Bar, 10 mM. (c) Immunofluorescence analysis showing co-staining of cyclin A (cycA) and p21 in bleomycin-treated HCT116 cultures (16 h). Note the low p21 signal in a large population of cyclin A-expressing cells (arrowheads). Bar, 10 mM. (d) Immunofluorescence analysis showing co-staining of cyclin B1 and p21 in ICRF-treated HCT116 cultures (16 h). Bar, 10 mM. (e) Immunofluorescence analysis showing PS1981-ATM (P-ATM) signal in control, bleomycin (Bleo)- and ICRF-193-treated MDA MB435 (upper panel, positive control) and MCF7 (lower panel) breast carcinoma cells. Asynchronous cultures were exposed to genotoxic agents for 16 h. Arrows indicate mitotic cells that accumulate due to a defect in G2/M checkpoint (see also Supplementary Figure S9). Bar, 10 mM. (f) Immunoblots comparing phosphorylation status or protein levels of different checkpoint regulators between drug-treated MDA MB435 and MCF7 breast carcinoma cells. The extracts were prepared from asynchronous non-treated (Cont) cells or cultures exposed to bleomycin (Bleo) or ICRF for 16 h. Arrowheads indicate phosphorylation-induced SDS–PAGE mobility shifts of Chk2. (g) Immunofluorescence analysis showing PS1981-ATM (P-ATM) in non-treated control (Cont), bleomycin (Bleo) and ICRF-treated colon carcinoma cells HCT116 and NHFs. Note a strong cytoplasmic PS1981-ATM signal in prophase cell (asterisk). Arrows indicate mitotic cells that accumulated due to a defect in G2/M checkpoint (see also Supplementary Figure S9). Bar, 10 mM. (h) Immunoblots comparing phosphorylation status or protein levels of different checkpoint regulators between drug-treated HCT116 cells and NHF. The extracts were prepared from asynchronous non-treated (Cont) cells or cultures exposed to bleomycin (Bleo) or ICRF for 12 h. See Supplementary Figure S10A for the immunoblots showing Chk1 and Chk2 phosphorylation after 24 h exposure to the same drugs. Upper panel: Triton soluble extracts. Lower panel: Triton insoluble extracts containing chromatin-associated proteins. Arrowhead indicates phosphorylation-induced SDS–PAGE mobility shift of Chk2. LC, loading control. Oncogene

Roles of Chk1 and p21 in G2 arrest G Lossaint et al

4270

Figure 7 ATM inhibition blocks p21 induction in G2 and cyclin B1 nuclear retention abrogating G2 but not post-mitotic G1 arrest.(a) Immunoblots showing effects of KU-55933 ( þ Kud) or caffeine (Caf) on ATM, ATR, Chk2 or p53 phosphorylation and p21 protein levels in ICRF- and bleomycin-treated NHF (8 h). Except in *(NP40 non-soluble extracts containing chromatin-bound proteins), cells were lysed by a direct application of the Laemmli SDS–PAGE buffer. Arrowheads, Chk2 mobility shift; Cont, non-treated cells; LC, loading control. The same extracts were analyzed for P-DNA-PKcs (Supplementary Figure S11). (b) Immunofluorescence analysis showing PS1981-ATM in non-treated (Cont) and ICRF-treated NHF in the absence (–) and the presence of KU-55933 ( þ Kud) or caffeine ( þ Caf). Asynchronous NHF cultures were exposed to ICRF for 16 h before fixation. Note the absence or decrease of PS1981-ATM signal in DNA-damage induced foci with KU-55933 or caffeine, respectively, in ICRF-treated cells. PS1981-ATM-specific nucleoplasmic signal was invariably detected in ICRF-treated cells in the presence of caffeine. Bar, 10 mM. (c) Entry into mitosis (percentage of total cell number) of non-treated (Cont) and ICRF-treated synchronized NHF in the absence () or presence of Caffeine ( þ Caf) or KU-55933 ( þ Kud). ICRF was added 22 h after replating of contact-inhibited cells when majority of cells progressed into S-phase. Caffeine or KU-55933 was added 1 h before ICRF. Cont, non-treated cells. Mitoses were counted at 2-h interval during 30 h, as described in Materials and methods. Effects of caffeine or KU-55 933 on bleomycin-treated NHF are shown in Supplementary Figure S12. (d) Immunoblots showing effects of KU-55 933 (Kud) and caffeine (Caf) on phosphorylation status and protein levels of indicated cell cycle or checkpoint regulators in non-treated control (Cont) and ICRF- or bleomycin-treated synchronized NHF (as in c). The cell extracts were analyzed 4 h ( ¼ G2 phase) and 24 h after drug addition. Arrows, hypophosphorylated pRb; arrowheads, phosphorylation-induced Chk2 SDS–PAGE mobility shift; LC, loading control. (e) Immunofluorescence analysis showing co-staining of cyclin B1 (cycB1) and p21 in ICRF-treated NHF in the absence (Ct) or presence of KU-55993 ( þ Kud). Note pre-mitotic and p21-dependent nuclear cyclin B1 accumulation in ICRF-treated cells that was abolished by KU-55993. Bar, 10 mM. (f) Immunofluorescence analysis showing co-staining of cyclin B1 (cycB1) and p21 in ICRF-treated HMEC in the absence (Ct) or presence of caffeine ( þ Caf). Note pre-mitotic and p21-dependent nuclear cyclin B1 accumulation in ICRF-treated cells that was abolished by caffeine. Bar, 10 mM. Data showing abrogation of ICRF-induced G2 arrest by caffeine in HMEC are shown in Supplementary Figure S13B and C.

Oncogene

Roles of Chk1 and p21 in G2 arrest G Lossaint et al

4271

Figure 8 Roles of p21 in early stages of G2 arrest, its maintenance and entry into senescence from G2. Model summarizing multiple roles of p21 in G2 arrest, which are defective in frequently studied p53 wild-type cancer cell lines. p21, like Chk1, directly (and irreversibly) inhibits activation of cyclin B1-Cdk1 by Cdc25 phosphatases and CAK, Cdk-activating kinase (early response). Its G2 maintenance role involves blocking expression of mitotic regulators by inhibiting Cdk-dependent phosphorylation of pRb family pocket proteins and inducing degradation of cyclin B1. This eventually leads to irreversible cell cycle exit in G2 (senescence). In carcinoma (MCF7, HCT116) or U2OS sarcoma cell lines these G2 roles of p21 are impaired due to a defective ATM-p53-p21 pathway. Our data do not support the current model in which Chk2 is the major regulator of the G2/M checkpoint. For the sake of simplicity, this model does not take into account roles of Chk1 in maintenance of the G2 arrest (Shimada et al., 2008; Shibata et al., 2010) neither the role of the p38MAPK-MK2 pathway in G2/M checkpoint (Reinhardt et al., 2007). Dotted lines (crosses) indicate defects in ATM and/or ATR pathways in different cancer cell lines. In HCT116 cells, in contrast to bleomycin, ICRF does not activate Chk1.

both ATM (p53, p21) and ATR (Chk1) pathways are defective leading to a complete lack of G2 arrest in response to either drug. By contrast, in HCT116, bleomycin strongly activates both ATM (albeit less than in NHF) and ATR, resulting in G2 arrest, whereas ICRF is completely inefficient leading to accumulation of mitotic cells. The latter defect could be related to a deficiency in MRE11 (MRN complex) affecting the ATM pathway (Takemura et al., 2006). Finally, in U2OS cells, despite relative efficient phosphorylation of ATM and p53 (S15) by both genotoxic agents, p21 induction is strongly delayed. This suggests that inefficient p21 induction might be due to another defect in the complex process of p53 activation, such as Cterminal acetylation (Meek and Anderson, 2009). In addition, it would be interesting to test the functionality of Che-1, an ATM-activated transcription factor that is required for p21 induction and that has a major role in the G2/M checkpoint (Bruno et al., 2006). The crucial role of ATM activation in G2 arrest is demonstrated by our observation that pharmacological inhibition of ATM blocks p53 phosphorylation (S15), pre-mitotic p21 induction and abrogates G2 arrest. We also show that ATM inhibitors do not interfere with activation of other members of PI3K family, particularly ATR or DNA-PKc. Thus, a prerequisite for efficient therapeutic activity of Chk1 inhibitors most

likely implicates a deficient ATM/p53/p21 pathway. This is the case in the majority of tumors due to mutations in p53 or other participant(s) in the pathway (Palmero et al., 2010). In conjunction with earlier observations (Baus et al., 2003), our results suggest that, in non-transformed cells, in addition to inhibiting cyclin A-Cdk1/2, p21 exercises an equivalent role to Chk1 and reinforces its action in G2 arrest by blocking cyclin B1-Cdk1 in an inactive state (see Figure 8). By associating with cyclin B1-Cdk1 and sequestering it in the nucleus, p21 prevents Cdk1 activation either by CAK (Smits et al., 2000) or Cdc25 (Charrier-Savournin et al., 2004) in the cytoplasm (Lindqvist et al., 2005; Gavet and Pines, 2010), thus participating directly in G2 arrest (Figure 8). Moreover, as in the case of sustained or unrepaired DNA damage, permanent Cdk1 inactivation by p21 would preclude either the resumption of cell cycle progression into mitosis, if the signaling is turned off (checkpoint silencing or adaptation; (Deckbar et al., 2007; Jurvansuu et al., 2007; van Vugt and Yaffe, 2010)), or endoreplication, if the checkpoint signaling persists (Toettcher et al., 2009; Davoli et al., 2010). Finally, in addition to contributing to the maintenance of G2 arrest (Bunz et al., 1998; Chan et al., 2000), we propose that p21mediated Cdk1 inhibition together with pRb phosphorylation block, may be a prerequisite for senescent-like cell cycle exit in G2 (Baus et al., 2003). Intriguingly, deficient p21 induction (U2OS) or p21 knockdown (NHF, HMEC) is accompanied by increased/prolonged Chk1 activation, which could compensate for p21 deficiency and thus contribute to G2 arrest maintenance. The mechanism of this feedback control is not known, but it could involve stabilization of Chk1 that is normally rapidly degraded by the proteasome (Zhang et al., 2005; Merry et al., 2010) or its increased expression (Gottifredi et al., 2001). Our finding that p21 depletion or impaired induction strongly stimulated Chk1 activation by DNA damage might explain why p53- or p21-deficient cells arrest temporarily in G2 (Levedakou et al., 1995; Bunz et al., 1998; Andreassen et al., 2001) and why Chk1 depletion promoted mitosis in g-irradiated p53-proficient U2OS cells (Syljuasen et al., 2006). In addition, they are fully consistent with results showing that p53-deficient cells could overcome a dependence on p21 for G2 arrest by switching to dependence on Chk1 and p38MAPK/MK2 (Reinhardt et al., 2007). An important corollary is that cancer cells with an impaired ATM pathway could also compensate for the absence of strong p21 induction in G2 by Chk1 hyperactivation. Moreover, our results provide a plausible explanation for why the role of p21 in inactivation of mitotic Cdks was previously considered marginal in comparison to that of Chk1/Chk2. Our work also showed that, despite its rapid phosphorylation on T68, Chk2 is not required for G2 arrest in non-transformed cells corroborating earlier results obtained in cancer cells (Ahn et al., 2003; Jallepalli et al., 2003; Jin et al., 2008). This is consistent with the data showing that Chk2 activation is defective in currently used p53-proficient cancer cells Oncogene

Roles of Chk1 and p21 in G2 arrest G Lossaint et al

4272

cf. (Takemura et al., 2006) and that full induction of Chk2 activity is slow even in cells with an intact ATM pathway. These observations are intriguing as Chk2 is considered as a bona fide negative regulator of G2/M transition in the DDR (Lobrich and Jeggo, 2007; Toettcher et al., 2009; van Vugt and Yaffe, 2010; van Vugt et al., 2010). On the other hand, they should be reconciled with the widely recognized role of Chk2 in DDR, senescence (Gire et al., 2004) and in carcinogenesis (Bartek and Lukas, 2003). One possibility is that either Chk2’s role is redundant if other checkpoint regulators are active (Meng et al., 2009) or that it might intervene in other pathways controlling cell proliferation, such as pRb phosphorylation (Inoue et al., 2007). In conclusion, our results show that both Chk1 and p21 are required for G2 arrest in response to a sustained DNA damage and that Chk1 is dispensable in the presence of a functional ATM/p53/p21 pathway. The p21’s pivotal role in inactivating mitotic Cdks is impaired in several p53-proficient cancer cells due to compromised ATM activity. We propose that p21 acts as an ultimate G2 phase gatekeeper whose deficient induction compromises the DDR and permanent cell cycle exit with the ensuing risks of tumor development.

Materials and methods Cell lines, drug treatments and cell cycle analysis HMECs were obtained from reduction of mammoplasty tissues of anonymous healthy donors as previously described (Stampfer et al., 1980) and grown in MEGM medium (Lonza, Basel, Switzerland). HMEC were infected by retroviral transfer of the HPV16 E6 oncogene in the PLXSN vector, provided generously by Dr D Galloway (Seattle, USA) and selected for resistance to G418 or by retroviral transfer of the hTERT gene in the pLPC vector (Clontech, Palo Alto, CA, USA) and selected for resistance to puromycin. Normal human (foreskin) fibroblasts (NHF) were described previously (Baus et al., 2003). The human osteosarcoma U2OS, HeLa and human breast carcinoma MCF7 and MDA-MB435 cell lines were from the ATCC (Manassas, VA, USA). Human colon carcinoma cells HCT116 was a generous gift of Dr B Vogelstein (Baltimore, USA). Except HCT116 (McCoy medium) all cell lines were cultured in DMEM supplemented with 10% fetal calf serum. Bleomycin (10 mg/ml) and ICRF-193 (bis(2,6-dioxopiperazin, 2 mg/ml) were added to synchronized (NHF) or asynchronously growing (HMEC, U2OS) cells as described previously (Baus et al., 2003). Shortly, G0-arrested fibroblasts (by contact inhibition) were released into the cell cycle by dilution and the drugs were added at the time when most of cells entered S-phase (22 h). The ATM inhibitors, caffeine (Sigma-Aldrich, St Louis, MO, USA) and KU-55933 (Kudos Pharmaceuticals, Cambridge, UK), were added 1 h before genotoxic agents, at concentrations of 5 mM and 10 mM, respectively. Where indicated, cells were irradiated (5 Gy) using a 137Cesium source. UCN-01 (Merck Chemicals, Darmstadt, Germany) was used at concentration of 250 nM. Cell cycle distribution was determined by flow cytometric analysis (FACS) of propidium iodide-stained cells. Video-microscopy experiments For time-lapse experiments, transfection with siRNA (see below) was carried out in six-well or 12-well Petri dishes that Oncogene

were placed, after 24 h and upon drug addition, under inverted wide-field microscope (Leica DMIRE2, objective Leica 10  HC PL FLUOTAR 0.3 PH1, Leica, Wetzlar, Germany and Camera Micromax YHS 1300, Photometrics, Tucson, AZ, USA) equipped with CO2-, temperature- and humiditycontrolled chamber (Montpellier RIO Imaging facility-MRI). For the study of ATM or Chk1 inhibitors, we used asynchronously proliferating or synchronized cells. Mitoses (rounded cells) were scored by inspection of video-microscopy sequences (Metamorph software, Molecular Devices, Sunnyvale, CA, USA). Each mitosis was counted only once. The images were taken at 10–20 min interval for the duration of at least 48 h. Three fields for each situation were analyzed and normalized for the cell number at the beginning of the time-lapse sequence (percentage of cells at the start). For mitosis-entry kinetics, total number of mitotic cells during the given interval was plotted. For each experiment, all the conditions, including controls with untreated cells transfected with different siRNAs, were tested in parallel. Immunoblotting and immunoprecipitation Preparation of cell lysates, conditions for immunoprecipitation, p21 depletion experiments, histone H1 kinase assays and immunoblotting experiments have been described previously (Dulic´ et al., 1998; Baus et al., 2003). To assess cyclin-associated kinase activity (32P-ATP incorporation) histone H1 bands were excised from 11% SDS–PAGE gels and counted by Cerenkov; to detect PS317-Chk1 and phosphorylation-induced mobility shifts of Chk2, pRb and p130, extracts were analyzed on 7.5% SDS–PAGE gels. Other cell cycle regulators and PT68-Chk2 were analyzed on 12.5% SDS– PAGE gels. For certain experiments presented in Figure 7 and Supplementary Figure S11 (P-DNA-PKcs) the cells were directly lysed in the Laemmli sample buffer. To detect PS1981-ATM, PS428ATR and PS2056-DNA-PKcs we analyzed chromatin enriched cell extracts (NP40 non-soluble) on 7% SDS–PAGE gels. Immunofluorescence. Experimental conditions for immunofluorescence were described previously (Baus et al., 2003; Charrier-Savournin et al., 2004). Immunofluorescence and phase contrast photomicrographs were image captured (microscope Leica CTR6000, objective Leica 40  HCX PL APO 1.25–0.75 oil, camera Coolsnap HQ2, Photometrics) and a composite was generated using Adobe Photoshop (Adobe Systems Inc., San Jose, CA, USA) and Microsoft PowerPoint (Microsoft Corp, Redmond, WA, USA) softwares. Antibodies. Most of the primary antibodies were described in our earlier publications (Baus et al., 2003; Charrier-Savournin et al., 2004; Gire et al., 2004). In addition, we have used phospho-specific mouse (S139-H2A.X and S1981-ATM) and rabbit polyclonal (S428-ATR, T68-Chk2, S317-Chk1, S20-p53 and S15-p53) from Cell Signaling (Franklin Lakes, NJ, USA). For certain experiments we used anti-PS1981-ATM antibodies from Rockland (Gilbertsville, PA, USA) and Millipore (Billerica, MA, USA). To detect DNA-PKcs we have used phospho-specific rabbit polyclonal (PS2056, ab18192) and mouse anti-DNA-PKcs (ab1832) by Abcam (Cambridge, UK). Secondary antibodies for immunoblot analysis were anti-mouse (DAKO, Glostrup, Denmark) and anti-rabbit immunoglobulin G horseradish peroxidase-conjugates (Promega, Madison, WI, USA) and protein AG coupled to horseradish peroxidase (Pierce, Rockford, IL, USA). Proteins were visualized by enhanced chemiluminiscence detection according to the manufacturer (Amersham, Piscataway, NJ, USA). Secondary antibodies for immunofluorescence, anti-mouse Texas Red and anti-rabbit Alexa 488 were from Molecular Probes (Carlsbad, CA, USA).

Roles of Chk1 and p21 in G2 arrest G Lossaint et al

4273 siRNA experiments Chk1-, Chk2- and p21-specific siRNA corresponding to a mix of four specific sequences (SmartPool) were purchased from Dharmacon Research (Lafayette, CO, USA). Asynchronously growing NHF, HMEC or U2OS cells were transfected using a standard Ca-phosphate protocol (40 nM of siRNA). Similar conditions were used for triple Chk1/Chk2/p21 knockdown in NHF. As control, we used siRNA for luciferase, 50 -ACUGA CGACUCUGCUACUC-30 (Eurogentec, Seraing, Belgium). Twenty-four hours after transfection, the cells were exposed (or not) to different genotoxic agents and harvested for biochemical or immunofluorescence analysis at indicated times (usually after 16 or 48 h) or monitored by time-lapse microscopy.

Conflict of interest The authors declare no conflict of interest.

Acknowledgements This work was supported by grants of ARC (grant N1 3631 to JP and N13793 to VD) the Cance´ropole du Grand Sud Ouest and the Ministe`re de l’Education Nationale et de la Recherche (MENR, EB). The team (GL, VD and DF) is ‘Equipe labellise´’ by the Ligue Nationale Contre le Cancer (LNCC, N1 EL2010.LNCC/DF). GL was recipient of a PhD fellowship from LNCC. We are grateful to KUDOS Pharmaceuticals for gift KU-55933, Dr AM Creighton for gift ICRF-193, Dr D Galloway (Seattle, USA) for the E6 retroviral vector and Dr A Constantinou for DNA-PKcs-specific antibodies. We thank Dr D Fraillery for helping us in analysis of HeLa and U2OS cells, Drs V Gire and J Loncarek for the retroviral transduction of the HMEC-derived cell lines, Drs E Julien, C Sardet, A Camasses, P Coopman and N Taylor for critical reading of the manuscript and Dr A Constantinou for encouragement. We acknowledge the Montpellier RIO imaging facility (Head Dr P Travo) for their help. Finally, we thank the reviewers for insightful suggestions.

References Ahn J, Urist M, Prives C. (2003). Questioning the role of checkpoint kinase 2 in the p53 DNA damage response. J Biol Chem 278: 20480–20489. Andreassen PR, Lacroix FB, Lohez OD, Margolis RL. (2001). Neither p21(WAF1) Nor 14-3-3sigma prevents G(2) progression to mitotic catastrophe in human colon carcinoma cells after DNA damage, but p21(WAF1) induces stable G(1) arrest in resulting tetraploid cells. Cancer Res 61: 7660–7668. Arlander SJ, Greene BT, Innes CL, Paules RS. (2008). DNA protein kinase-dependent G2 checkpoint revealed following knockdown of ataxia-telangiectasia mutated in human mammary epithelial cells. Cancer Res 68: 89–97. Bartek J, Bartkova J, Lukas J. (2007). DNA damage signalling guards against activated oncogenes and tumour progression. Oncogene 26: 7773–7779. Bartek J, Lukas J. (2003). Chk1 and Chk2 kinases in checkpoint control and cancer. Cancer Cell 3: 421–429. Bartkova J, Horejsi Z, Koed K, Kramer A, Tort F, Zieger K et al. (2005). DNA damage response as a candidate anti-cancer barrier in early human tumorigenesis. Nature 434: 864–870. Baus F, Gire V, Fisher D, Piette J, Dulic´ V. (2003). Permanent cell cycle exit in G2 phase after DNA damage in normal human fibroblasts. EMBO J 22: 3992–4002. Berkovich E, Monnat Jr RJ, Kastan MB. (2007). Roles of ATM and NBS1 in chromatin structure modulation and DNA double-strand break repair. Nat Cell Biol 9: 683–690. Bruno T, De Nicola F, Iezzi S, Lecis D, D’Angelo C, Di Padova M et al. (2006). Che-1 affects cell growth by interfering with the recruitment of HDAC1 by Rb. Cancer Cell 10: 473–486. Bunz F, Dutriaux A, Lengauer C, Waldman T, Zhou S, Brown JP et al. (1998). Requirement for p53 and p21 to sustain G2 arrest after DNA damage. Science 282: 1497–1501. Chan TA, Hermeking H, Lengauer C, Kinzler KW, Vogelstein B. (1999). 14-3-3Sigma is required to prevent mitotic catastrophe after DNA damage. Nature 401: 616–620. Chan TA, Hwang PM, Hermeking H, Kinzler KW, Vogelstein B. (2000). Cooperative effects of genes controlling the G(2)/M checkpoint. Genes Dev 14: 1584–1588. Charrier-Savournin FB, Chateau MT, Gire V, Sedivy J, Piette J, Dulic´ V. (2004). p21-Mediated nuclear retention of cyclin B1-Cdk1 in response to genotoxic stress. Mol Biol Cell 15: 3965–3976. Chen Y, Poon RY. (2008). The multiple checkpoint functions of CHK1 and CHK2 in maintenance of genome stability. Front Biosci 13: 5016–5029.

Dai Y, Grant S. (2010). New insights into checkpoint kinase 1 in the DNA damage response signaling network. Clin Cancer Res 16: 376–383. Davoli T, Denchi EL, de Lange T. (2010). Persistent telomere damage induces bypass of mitosis and tetraploidy. Cell 141: 81–93. Deckbar D, Birraux J, Krempler A, Tchouandong L, Beucher A, Walker S et al. (2007). Chromosome breakage after G2 checkpoint release. J Cell Biol 176: 749–755. Dimri GP, Lee X, Basile G, Acosta M, Scott G, Roskelley C et al. (1995). A biomarker that identifies senescent human cells in culture and in aging skin in vivo. Proc Natl Acad Sci USA 92: 9363–9367. Downes CS, Clarke DJ, Mullinger AM, Gimenez-Abian JF, Creighton AM, Johnson RT. (1994). A topoisomerase II-dependent G2 cycle checkpoint in mammalian cells. Nature 372: 467–470. Dulic´ V, Stein GH, Far DF, Reed SI. (1998). Nuclear accumulation of p21Cip1 at the onset of mitosis: a role at the G2/M-phase transition. Mol Cell Biol 18: 546–557. Flatt PM, Tang LJ, Scatena CD, Szak ST, Pietenpol JA. (2000). p53 regulation of G(2) checkpoint is retinoblastoma protein dependent. Mol Cell Biol 20: 4210–4223. Foijer F, Wolthuis RM, Doodeman V, Medema RH, te Riele H. (2005). Mitogen requirement for cell cycle progression in the absence of pocket protein activity. Cancer Cell 8: 455–466. Gavet O, Pines J. (2010). Activation of cyclin B1-Cdk1 synchronizes events in the nucleus and the cytoplasm at mitosis. J Cell Biol 189: 247–259. Gillis LD, Leidal AM, Hill R, Lee PW. (2009). p21Cip1/WAF1 mediates cyclin B1 degradation in response to DNA damage. Cell Cycle 8: 253–256. Gire V, Roux P, Wynford-Thomas D, Brondello JM, Dulic´ V. (2004). DNA damage checkpoint kinase Chk2 triggers replicative senescence. EMBO J 23: 2554–2563. Gorgoulis VG, Vassiliou LV, Karakaidos P, Zacharatos P, Kotsinas A, Liloglou T et al. (2005). Activation of the DNA damage checkpoint and genomic instability in human precancerous lesions. Nature 434: 907–913. Gottifredi V, Karni-Schmidt O, Shieh SS, Prives C. (2001). p53 downregulates CHK1 through p21 and the retinoblastoma protein. Mol Cell Biol 21: 1066–1076. Graves PR, Yu L, Schwarz JK, Gales J, Sausville EA, O’Connor PM et al. (2000). The Chk1 protein kinase and the Cdc25C regulatory pathways are targets of the anticancer agent UCN-01. J Biol Chem 275: 5600–5605. Harper JW, Elledge SJ. (2007). The DNA damage response: ten years after. Mol Cell 28: 739–745. Oncogene

Roles of Chk1 and p21 in G2 arrest G Lossaint et al

4274 Hermeking H, Lengauer C, Polyak K, He TC, Zhang L, Thiagalingam S et al. (1997). 14-3-3 sigma is a p53-regulated inhibitor of G2/M progression. Mol Cell 1: 3–11. Hickson I, Zhao Y, Richardson CJ, Green SJ, Martin NM, Orr AI et al. (2004). Identification and characterization of a novel and specific inhibitor of the ataxia-telangiectasia mutated kinase ATM. Cancer Res 64: 9152–9159. Hoeijmakers JH. (2001). Genome maintenance mechanisms for preventing cancer. Nature 411: 366–374. Innocente SA, Abrahamson JL, Cogswell JP, Lee JM. (1999). p53 regulates a G2 checkpoint through cyclin B1. Proc Natl Acad Sci USA 96: 2147–2152. Inoue Y, Kitagawa M, Taya Y. (2007). Phosphorylation of pRB at Ser612 by Chk1/2 leads to a complex between pRB and E2F-1 after DNA damage. EMBO J 26: 2083–2093. Jackson MW, Agarwal MK, Yang J, Bruss P, Uchiumi T, Agarwal ML et al. (2005). p130/p107/p105Rb-dependent transcriptional repression during DNA-damage-induced cell-cycle exit at G2. J Cell Sci 118: 1821–1832. Jallepalli PV, Lengauer C, Vogelstein B, Bunz F. (2003). The Chk2 tumor suppressor is not required for p53 responses in human cancer cells. J Biol Chem 278: 20475–20479. Jin J, Ang XL, Ye X, Livingstone M, Harper JW. (2008). Differential roles for checkpoint kinases in DNA damage-dependent degradation of the Cdc25A protein phosphatase. J Biol Chem 283: 19322–19328. Jurvansuu J, Fragkos M, Ingemarsdotter C, Beard P. (2007). Chk1 instability is coupled to mitotic cell death of p53-deficient cells in response to virus-induced DNA damage signaling. J Mol Biol 372: 397–406. Kops GJ, Weaver BA, Cleveland DW. (2005). On the road to cancer: aneuploidy and the mitotic checkpoint. Nat Rev Cancer 5: 773–785. Lavin MF. (2008). Ataxia-telangiectasia: from a rare disorder to a paradigm for cell signalling and cancer. Nat Rev Mol Cell Biol 9: 759–769. Lee J, Kim JA, Barbier V, Fotedar A, Fotedar R. (2009). DNA damage triggers p21WAF1-dependent Emi1 down-regulation that maintains G2 arrest. Mol Biol Cell 20: 1891–1902. Levedakou EN, Kaufmann WK, Alcorta DA, Galloway DA, Paules RS. (1995). p21CIP1 is not required for the early G2 checkpoint response to ionizing radiation. Cancer Res 55: 2500–2502. Lindqvist A, Kallstrom H, Lundgren A, Barsoum E, Rosenthal CK. (2005). Cdc25B cooperates with Cdc25A to induce mitosis but has a unique role in activating cyclin B1-Cdk1 at the centrosome. J Cell Biol 171: 35–45. Lobrich M, Jeggo PA. (2007). The impact of a negligent G2/M checkpoint on genomic instability and cancer induction. Nat Rev Cancer 7: 861–869. Meek DW, Anderson CW. (2009). Posttranslational modification of p53: cooperative integrators of function. Cold Spring Harb Perspect Biol 1: a000950. Meng S, Arbit T, Veeriah S, Mellinghoff IK, Fang F, Vivanco I et al. (2009). 14-3-3sigma and p21 synergize to determine DNA damage response following Chk2 inhibition. Cell Cycle 8: 2238–2246. Merry C, Fu K, Wang J, Yeh IJ, Zhang Y. (2010). Targeting the checkpoint kinase Chk1 in cancer therapy. Cell Cycle 9: 279–283.

Narita M, Nunez S, Heard E, Lin AW, Hearn SA, Spector DL et al. (2003). Rb-mediated heterochromatin formation and silencing of E2F target genes during cellular senescence. Cell 113: 703–716. Palmero EI, Achatz MI, Ashton-Prolla P, Olivier M, Hainaut P. (2010). Tumor protein 53 mutations and inherited cancer: beyond Li-Fraumeni syndrome. Curr Opin Oncol 22: 64–69. Reinhardt HC, Aslanian AS, Lees JA, Yaffe MB. (2007). p53-deficient cells rely on ATM- and ATR-mediated checkpoint signaling through the p38MAPK/MK2 pathway for survival after DNA damage. Cancer Cell 11: 175–189. Reinhardt HC, Yaffe MB. (2009). Kinases that control the cell cycle in response to DNA damage: Chk1, Chk2, and MK2. Curr Opin Cell Biol 21: 245–255. Shibata A, Barton O, Noon AT, Dahm K, Deckbar D, Goodarzi AA et al. (2010). Role of ATM and the damage response mediator proteins 53BP1 and MDC1 in the maintenance of G(2)/M checkpoint arrest. Mol Cell Biol 30: 3371–3383. Shiloh Y. (2003). ATM and related protein kinases: safeguarding genome integrity. Nat Rev Cancer 3: 155–168. Shimada M, Niida H, Zineldeen DH, Tagami H, Tanaka M, Saito H et al. (2008). Chk1 is a histone H3 threonine 11 kinase that regulates DNA damage-induced transcriptional repression. Cell 132: 221–232. Smits VA, Klompmaker R, Vallenius T, Rijksen G, Makela TP, Medema RH. (2000). p21 inhibits thr161 phosphorylation of cdc2 to enforce the G2 DNA damage checkpoint. J Biol Chem 275: 30638–30643. Stampfer M, Hallowes RC, Hackett AJ. (1980). Growth of normal human mammary cells in culture. in vitro 16: 415–425. Syljuasen RG, Jensen S, Bartek J, Lukas J. (2006). Adaptation to the ionizing radiation-induced G2 checkpoint occurs in human cells and depends on checkpoint kinase 1 and Polo-like kinase 1 kinases. Cancer Res 66: 10253–10257. Takemura H, Rao VA, Sordet O, Furuta T, Miao ZH, Meng L et al. (2006). Defective Mre11-dependent activation of Chk2 by ataxia telangiectasia mutated in colorectal carcinoma cells in response to replication-dependent DNA double strand breaks. J Biol Chem 281: 30814–30823. Taylor WR, Stark GR. (2001). Regulation of the G2/M transition by p53. Oncogene 20: 1803–1815. Toettcher JE, Loewer A, Ostheimer GJ, Yaffe MB, Tidor B, Lahav G. (2009). Distinct mechanisms act in concert to mediate cell cycle arrest. Proc Natl Acad Sci USA 106: 785–790. Tomimatsu N, Mukherjee B, Burma S. (2009). Distinct roles of ATR and DNA-PKcs in triggering DNA damage responses in ATMdeficient cells. EMBO Rep 10: 629–635. van Vugt MA, Gardino AK, Linding R, Ostheimer GJ, Reinhardt HC, Ong SE et al. (2010). A mitotic phosphorylation feedback network connects Cdk1, Plk1, 53BP1, and Chk2 to inactivate the G(2)/M DNA damage checkpoint. PLoS Biol 8: e1000287. van Vugt MA, Yaffe MB. (2010). Cell cycle re-entry mechanisms after DNA damage checkpoints: giving it some gas to shut off the breaks!. Cell Cycle 9: 2097–2101. Zhang WH, Poh A, Fanous AA, Eastman A. (2008). DNA damageinduced S phase arrest in human breast cancer depends on Chk1, but G(2) arrest can occur independently of Chk1, Chk2 or MAPKAPK2. Cell Cycle 7: 1668–1677. Zhang YW, Otterness DM, Chiang GG, Xie W, Liu YC, Mercurio F et al. (2005). Genotoxic stress targets human Chk1 for degradation by the ubiquitin-proteasome pathway. Mol Cell 19: 607–618.

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

Oncogene