Oncogene (2006) 25, 7361–7372 & 2006 Nature Publishing Group All rights reserved 0950-9232/06 $30.00 www.nature.com/onc ORIGINAL ARTICLE Analysis of B1 and CDK activity during induced by camptothecin treatment

A Borgne1,4,6, I Versteege1,5,6, M Mahe´ 1, A Studeny1,SLe´ once1, I Naime2, M Rodriguez2, JA Hickman1, L Meijer3 and RM Golsteyn1

1Institut de Recherches Servier, Cancer Drug Discovery, Croissy-sur-Seine, France; 2Institut de Recherches Servier, Molecular Pharmacology and Physiopathology, Croissy-sur-Seine, France and 3CNRS, Station Biologique, Roscoff, France

We have studied the role of and cyclin-dependent metazoans. In an individual, however, if either of these kinase (CDK) activity in apoptosis induced by campto- processes is not regulated properly, the cell number can thecin (CPT). In this model, 22% of the cells stain for increase inappropriately to cause tumour formation annexin-V at 24 h and then proceed to be 93% positive by (Hartwell and Kastan, 1994). During apoptosis, cells die 72 h. This time window permits the analysis of cyclins in by following an ordered intracellular mechanism in cells that are committed to apoptosis but not yet dead. which and DNA are degraded. During the cell We provide evidence that cyclin levels and then cycle, DNA is precisely duplicated rather than degraded associated kinase levels increase after CPT treatment. and proteins are synthesized and degraded to assure Strikingly, cyclin B1 and proteins are present at progression through each phase of the cycle. Despite this the same time in CPT treated HT29 cells. Although cyclin fundamental difference between apoptosis and the cell B1 and E1 CDK complexes are activated in CPT treated cycle, there are increasing amounts of evidence that cells, only the cyclin B1 complex is required for apoptosis at some point these two processes use a common bio- since reduction of cyclin B1 by RNAi or roscovitine chemical pathway and yet achieve a different result. treatment reduces the number of annexin-V-stained cells. Identifying and understanding this apoptotic pathway We have detected poorly organized and will be important in characterizing compounds for phosphorylated histone H3 epitopes at the time of treating cancer. maximum cyclin B1/CDK kinase activity in CPT-treated The role of cyclin-dependent kinase (CDKs) in the cells, which suggests that these cells enter a mitotic is well understood and different phases of the catastrophe. Understanding which CDKs are required for cell cycle can be characterized by the presence and apoptosis may allow us to better adapt CDK inhibitors for the activity of different CDKs and cyclins. Somewhat use as anti-cancer compounds. surprisingly, CDK activity has also been described in Oncogene (2006) 25, 7361–7372. doi:10.1038/sj.onc.1209718; cells undergoing apoptosis (Shi et al., 1994); for reviews published online 19 June 2006 see (Guo and Hay, 1999; Borgne and Golsteyn, 2003). In contrast to the cell cycle, it is not known under which Keywords: mitotic catastrophe; cyclins; cyclin- conditions CDKs function during apoptosis. This is the dependent kinases; roscovitine; CYC202; Seliciclib; subject of the study described here. camptothecin A role for CDKs in apoptosis has been identified in different cell types by using many different experimental approaches. Apoptosis can be blocked in certain models of cultured human and murine cell lines by chemical Introduction inhibitors of CDKs, such as roscovitine (De Luca et al., 1997; Choi et al., 1999; Hsu et al., 1999) and structurally Apoptosis and the cycle are two cellular related purine type compounds such as olomoucine processes that are essential for normal development of (Hsu et al., 1999) and purvalanol (Adachi et al., 2001) or a flavonoid compound, flavopiridol (Park et al., 1996). Despite their chemical diversity, these small molecules Correspondence: Dr RM Golsteyn, Institut de Recherches Servier, bind to the ATP-binding pocket of the catalytic domain Cancer Drug Discovery, 125 chemin de Ronde, 78290 Croissy-sur- and reduce the activity of the enzyme (Vesely et al., Seine, France. E-mail [email protected] 1994; Knockaert et al., 2002). This suggests that CDK 4Current address: Theraptosis, Pasteur Biotop, Institut Pasteur, 25-28, activity is required for apoptosis, as it is for the cell rue du Docteur Roux, 75015 Paris, France. cycle. The effects of small molecule inhibitors can be 5Current address: Crucell Holland B.V., Archimedesweg 4, PO Box mimicked by naturally occurring small inhibitor pro- 2048, Leiden, The Netherlands. cip1 kip1 ink4 6These authors contributed equally to this study. teins, such as , p27 , and , which Received 7 February 2006; revised 21 April 2006; accepted 28 April 2006; can inhibit the activity of CDK2 (Coqueret, 2003) published online 19 June 2006 and block apoptosis in HeLa cells (Harvey et al., 1998), A role for cyclin B1 in apoptosis A Borgne et al 7362 in fibroblasts (Hiromura et al., 1999) and in differentiat- Materials and methods ing myocytes (Wang and Walsh, 1996). Additional supportive experimental evidence for the role of CDK Cell lines activity in apoptosis comes from studies that use ‘kinase The human cell lines HT29 (colon adenocarcinoma) and HeLa dead’ mutants of CDK1 or CDK2 (Meikrantzand were obtained from the American Type Culture Collection. HT29 cells were maintained in RPMI 1640 medium supple- Schlegel, 1996; Harvey et al., 2000; Castedo et al., mented (RPMIc) with 10% decomplemented fetal calf serum 2002b). Finally, the FT210 murine cell line, which (FCS), 2 mML-glutamine (Invitrogen, Cergy-Pontoise, harbours a temperature sensitive mutation in the CDK1 France), and 10 mM HEPES, pH 7.4. HeLa cells were , will not enter apoptosis when cultivated at the maintained in Dulbecco’s modified Eagle’s medium (DMEM) restrictive temperature and treated with apoptosis- with 10% decomplemented fetal calf serum, 2 mML-glutamine, inducing agents (Shi et al., 1994). Thus, inhibition of and 10 mM (N-2-hydroxyethlypiperazine-N02 ethane sulphonic CDK activity by methods using chemical inhibitors, acid) (HEPES), pH 7.4. protein inhibitors and dominant negative genetics show that CDKs participate in apoptosis. Flow cytometric detection of apoptosis by annexin-V labelling, Although CDK inhibitors can protect cells from cell cycle phase and cyclins apoptosis, under certain conditions they can also induce HT29 cells in exponential phase of growth were treated with apoptosis (Knockaert et al., 2002). This property has CPT or not, the medium containing cells in suspension was lead to the development of inhibitors that are candidates collected and adherent cells were trypsinized. Cells were pooled 6 for evaluation in preclinical and clinical studies. The and resuspended at a density of 10 cells/ml in culture medium 1 biological mechanism of action of these compounds is containing 20% FCS. After 1 h incubation at 37 C, cells were washed twice with cold PBS and resuspended in 500 mlof not precisely known, however, it has been proposed that binding buffer (100 mM HEPES, pH 7.4, 14 mM NaCl, and inhibition of distant Cdk1 family members such as Cdk7 25 mM CaCl2) containing 5 ml of annexin-V-fluorescein iso- and Cdk9 may lead to apoptosis (MacCallum et al., thiocyanate (FITC) (Immunotech, Beckman, Marseille, 2005). Further research is required to uncover the France) and 10 mg/ml propidium iodide (PI) for 15 min at specific role of each Cdk complex in apoptosis. 41C in the dark. For each sample, approximately 105 cells In cells undergoing apoptosis, the assignment of a were analysed by flow cytometry. FITC and PI emissions cyclin to any specific event has not yet been made, were collected through 520 and 630-nm bandpass filters, although, many different examples of cyclin expression respectively. have been described. protein induction has For PI labelling, cells were fixed with ethanol and incubated been identified in IM-9, MOLT-4 and RPMI 8226 cell for 30 min at room temperature in 1 ml PBS containing 100 mg/ ml RNase and 50 mg/ml PI (Sigma, Saint Quentin Fallavier, lines undergoing apoptosis (Mazumder et al., 2000), France). Cells were analysed on an Epics XL/MCL flow neuronally derived cells (Padmanabhan et al., 1999), as cytometer (Beckman Coulter, Cergy-Pontoise, France). To well as in HT29 cells treated with a novel acronycine analyse cyclin proteins, samples were fixed with ethanol and derivative (Le´ once et al., 2001). B-type cyclins associated washed twice with PBS, and incubated for 5 min in PBS with CDK1 have been identified in HL60 cells as they containing 0.5% Triton X-100 at 41C. After two washes with enter apoptosis after treatment with the topoisomerase I PBS, cells were incubated for 2 h at room temperature with 5 ml inhibitor camptothecin (CPT) (Shimizu et al., 1995) and of monoclonal anti-cyclin E1 (HE12, BD Pharmingen, San in epithelial-derived cell lines (Scatena et al., 1998), a Diego, CA, USA), or 20 ml of FITC-conjugated monoclonal HeLa syncytia model (Castedo et al., 2002b) as well as anti-cyclin B1 (GNS1, BD Pharmingen). Cyclin E1 was neuronal cells (Shirvan et al., 1998) suggesting that this detected by incubation for 1 h with 20 ml of FITC-conjugated goat anti-mouse IgG (Santa CruzBiotechnology, Le Perrayen activity might be required for apoptosis in many Yvelines, France). different cell types (for review see Castedo et al. (2002a)). In a similar manner, reports have also described that protein levels and kinase activity Immunofluorescence analysis 5 increase during apoptosis (Meikrantz et al., 1994; Shi 2.5 Â 10 cells were cultured for 24 h before treatment with et al., 1996; Harvey et al., 1998; Hakem et al., 1999). 1 mM CPT for various times. After one wash with PBS, cells were fixed in 3% formaldehyde for 20 min at room tempera- A role of individual cyclins in apoptosis needs to be ture and permeabilized for 5 min in 0.2% Triton X-100. Cells defined as has been done for cyclins during the cell cycle were incubated with polyclonal anti-cyclin B1 (H-433, Santa in order to understand how CDKs regulate apoptosis. CruzBiotechnology, 1:50) and/or monoclonal anti-cyclin E1 This task is not simple because studies of cyclins in (13A3, Novocastra, Le Perrayen Yvelines, France, 1:80) or cells engaged in apoptosis are confounded by a back- antiphospho-histone H3 (6G3, Cell Signalling, Saint Quentin, ground of cyclin function in the phase of the cell cycle France, 1:50) for 2 h at room temperature. After three washes during which apoptosis is occurring. We present an with PBS, Texas Red anti-rabbit (for cyclin B1), Texas Red experimental model of HT29 cells engaged to enter anti-mouse (for phospho-histone H3) or FITC anti-mouse (for apoptosis in which we analyse cyclin B1 and cyclin E1 cyclin E1) secondary antibodies were added for 2 h. Nuclei kinase complexes. These kinase activities have been were marked with DAPI. Cells were observed on a Zeiss Axioplan 2 microscope driven by Axiovision 3.1 software in separated from their cell cycle phase-specific activity, which the linear range of the Zeiss HR camera was set first and despite that the activity of both kinase complexes with non-treated cells. The number of cells that stained correlate well with apoptosis, we propose that only positive for cyclin B1, cyclin E1 or phospho-histone H3 cyclin B1 is required for apoptosis in this experimental was determined from photographs. A total of 200 cells system. were counted and the percent positive was calculated. The

Oncogene A role for cyclin B1 in apoptosis A Borgne et al 7363 experiments were performed twice. Images were prepared with Preparation and transfection of small interference RNA Photoshop 7.0 software without changing the linear dynamic Double-stranded small interference RNA (siRNA) corre- range. Confocal images were collected on a Zeiss LSM 510 sponding to the cDNA sequences of human cyclin E1 microscope within the linear dynamic range, free from signal (AATGGCCAAAATCGACAGGAC, nucleotides 90–110), crossover by separate excitation with either an Argon 488 laser human cyclin B1 (AACAGCTCTTGGGGACATTGG, nu- or an HeNe 543 laser. DAPI signals were collected by cleotides 126–146) and human cyclin B1 with four mutations excitation with a 405 nM diode source. (AACAACTATTGGGAACGTTGG, nucleotides 126–146) were designed as recommended by (Elbashir et al., 2002). Measure of cyclin mRNA levels by real-time reverse The duplexes were purchased from Dharmacon Research Inc., transcriptase–polymerase chain reaction Lafayette, CO, USA (for cyclin E1) and Proligo, Paris, France RNA (1 mg) from cells treated or not with CPT were subjected (for cyclin B1). A scrambled sequence (CAGTCGCGTTTGC to a reverse transcription step using the cDNA archive Kit GACTGG) was also included in control experiments (Dhar- (Applied Biosystems, USA). Single-stranded cDNA products macon). Cells were cultured in six-well plastic plates in RPMIc were then analysed by an ABI PRISM 7900HT Sequence medium and transfected by adding 12 ml of lipofectamine 2000 Detection System using Cyclin B1 (Hs00259126_m1), (Invitrogen) and siRNA in Optimem (500 ml final volume). The (Hs00277039_m1), Cyclin E1 (Hs00233356_m1) and 18S rRNA minimum active concentration for each siRNA duplex was (Hs99999901_s) mRNAs. In each case, triplicate threshold cycle determined emperically before use in experiments. Cells were rinsed 6 h post-transfection and maintained in culture for 24 h (Ct) values were obtained and averaged; then expression levels were evaluated by the 2ÀDDCt method (Livak and Schmittgen, before addition of 1 mM CPT for additional 24 h before FACS 2001). The fold change in cyclin B1, D1 or E1 was normalized and Western blot analysis. to the 18S rRNA and compared to the untreated control (calibrator sample), using the following formula: ¼ 2ÀDDCt ; where DDCt ¼ (CtÀtargetÀCtÀreference)treatedÀsampleÀ(CtÀtargetÀ CtÀreference)calibratorÀsample. Calibrator-sample represents the Results expression level (1 Â ) of the target gene normalized to 18S rRNA. Characterisation of cell cycle profile and apoptosis in cells treated with CPT Cyclin immunoprecipitation We treated human colon carcinoma cells (HT29) with Cells (107) were resuspended in 1 ml extraction of buffer CPT in order to define experimental conditions that (25 mM Tris-HCl, pH 7.5, 25 mM NaCl, 5 mM EDTA, 1% would permit the analysis of CDK activity in cells that Nonidet P-40, protease inhibitors cocktail). After a 40 s are committed to apoptosis but not yet dead. Cells were 1 sonication at 4 C, the suspension was centrifuged at 12 000 g incubated with various concentrations of CPT for 24 h for 10 min at 41C. Extracts were either analysed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS– and then analysed for annexin-V and PI labelling PAGE) or used for isolation of CDK/cyclin complexes by (Figure 1a). In comparison to other concentrations immunoprecipitation. Total cell lysate (0.5–1.0 mg) was tested, 1 mM CPT gave the greatest number of cells incubated with antibodies directed against cyclin E1 (13A3, positive for annexin-V labelling without PI staining Novocastra, 8 ml), (H-432, Santa Cruz, 5 ml) or cyclin (19%). These cells were considered to be in primary B1 (GNS1, Santa Cruz, 5 ml) at 41C and complexes were apoptosis. Treatments with higher or lower concentra- recovered with protein A/G-sepharose beads. The beads were tions of CPT gave either fewer cells in primary apoptosis washed three times with bead buffer and twice with Buffer C or a mixture of cells that stained for annexin-V and PI (60 mM b-glycerophosphate, 15 mM p-nitrophenylphosphate, (secondary apoptosis). After 72 h of treatment, nearly 25 mM MOPS (pH 7.2), 5 mM EGTA, 15 mM MgCl2,1mM 94% of the HT29 cells were positive for annexin-V DTT, 1 mM sodium vanadate, 1 mM phenylphosphate) (Meijer et al., 1989). labelling indicating that more than 70% of the non- apoptotic population at 24 h was indeed committed to apoptosis (Figure 1b). Histone H1 kinase assay The CDK complex beads were incubated in a reaction buffer Cells treated with 1 mM CPT were analysed by flow containing 0.2 mg/ml histone H1 (Calbiochem, Fontenay-sous- cytometry to determine their position in the cell cycle. Bois, France), 15 mM ATP, 0.2 mCi [g-33P]ATP (Amersham GE, The profile of treated cells at 24 h was similar to that Saday, France) in Buffer C for 20 min at 301C. The reaction of non-treated cells, revealing that the treated cells did was stopped by addition of 20 ml2Â Laemmli sample buffer. not accumulate at a specific phase of the cell cycle Samples (20 ml) were loaded on gels. Signals were captured by (Figure 1c). Whereas lower concentrations of CPT lead an Amersham PhosphorImager Typhoon 9200 and analysed to a G2–M-phase arrest ((Shao et al., 1997), not shown), by ImageQuant 5.2 software (Molecular Dynamics, Amer- this ‘freezing’ of the cell cycle was considered to be sham GE, Saday, France). Kinase activities are given in useful in that any changes in the expression of cyclins arbitrary units. associated with cells committed to apoptosis would be independent of a classical cell cycle arrest. Electrophoresis and Western blotting We then examined cyclin A2, B1 and E1 protein levels After standard PAGE and transfers, membranes were by Western blotting of total cell extracts prepared from incubated with the primary antibody in milk as follows: polyclonal anti-cyclin A2 (H-432, Santa Cruz, 1:1000), mono- HT29 cells treated for various times with CPT (Figure 2). clonal anti-cyclin B1 (GNS1, Santa Cruz, 1:500), monoclonal Under these conditions, all cyclin protein levels in- anti-cyclin E1 (HE12, BD Pharmingen, 1:500). The mem- creased, although the timing and the duration of the branes were treated with horseradish peroxidase-coupled increase was different for each cyclin. Cyclin E1 protein anti-mouse IgG (Amersham, 1:2000) for ECL detection. levels increased as early as 8 h after treatment, whereas

Oncogene A role for cyclin B1 in apoptosis A Borgne et al 7364 a Non-treated 24 h CPT 1 µM 24 h CPT 24 h 40 30 2 % 3 % 20 10 % of cells PI fluorescence 0 1 % PI fluorescence 19 % 0.1 0.25 0.5 1 2 CPT (µM) Annexin-V FITC Annexin-V FITC

b Non-treated 72 h CPT 1 µM 72 h c Non-treated CPT 1 µM

6 % 61% - -

- - 4 % 32% Number of cells Number of cells PI fluorescence PI fluorescence Annexin-V FITC Annexin-V FITC PI fluorescence PI fluorescence Figure 1 Characterization of annexin-V, PI staining and cell cycle in CPT treated HT29 cells. (a) Human colon carcinoma cells (HT29) were cultivated in the presence of CPT at various concentrations. At 24 h, cells were collected and analysed for annexin-V and PI labelling. The percentage of cells (black bars) that were labelled with annexin-V (primary apoptosis) and with annexin-V plus PI (secondary apoptosis) was measured. The fraction of the total percentage that showed only annexin-V labelling is shown with grey bars, and the fraction of the total percentage that showed annexin-V plus PI labelling is shown with white bars. The distribution of cells by biparametric annexin-V and PI staining with or without 1 mM CPT is shown. (b) Cells were treated with CPT or not (Non-treated) for 72 h and then analysed by flow cytometry for annexin-V and PI labelling. (c) Cells were treated with CPT or not (Non-treated) for 24 h and then analysed by flow cytometry for DNA content.

CPT 1 µM time (hr) mRNA levels may, in part, contribute to the increase

in protein levels. The increase in cyclin protein levels NT 8 16 32 40 44 48 Noco was independent of changes in the cell cycle phase, as shown in Figure 1c, although it was correlated with Cyclin A2 cells that were either dying or committed to apoptosis (Figure 1b).

Cyclin B1 Cyclin B1 and cyclin E1 are expressed in the same cell at the same time during CPT-induced apoptosis We examined cyclin B1 and cyclin E1 proteins in HT29 cells by confocal microscopy and by FACS to determine Cyclin E1 which cells express increased levels of these proteins after CPT treatment. Strong cyclin E1 staining was detected in the nuclei of many cells (Figure 3). The Figure 2 Western blot analysis of cyclin proteins during CPT merged image showed the striking simultaneous expres- treatment. HT29 cells were either not treated (NT) or treated with CPT for various times up to 48 h and analysed by Western blotting sion of cyclin B1 and cyclin E1 in the same cell. In with anti-cyclin A2, cyclin B1 and cyclin E1 antibodies. Cells addition, several different staining patterns could be treated with nocodazole (Noco) were included as a control. detected, including cyclin B1 nuclear staining and cells Equivalent protein loading for each sample was confirmed by that showed no cyclin B1 or cyclin E1 staining at all. At Coomassie Blue staining (not shown). this timepoint, 15% of the cells counted co-expressed cyclin B1 and cyclin E1 (Figure 3b), whereas co- expression was rarely detected in non-treated cells cyclin A2 and B1 levels showed a maximum level at 32 h. (image not shown). We then examined HeLa cells by After 32 h cyclins A, B1 and E1 generally persisted confocal microscopy after treatment by CPT. We were throughout the duration of the experiment. The mRNA able to detect the simultaneous expression of cyclin B1 levels of cyclin D1, cyclin B1 and cyclin E1 were then and cyclin E1 proteins indicating that this unusual analysed by quantitative PCR from total RNA isolated staining pattern was not limited to HT29 cells. The from cells treated with CPT for 24 h and these were expression of cyclin B1 and E1 was also present in both compared with the levels of cyclin mRNAs in exponen- cell types at 16 and at 32 h and rarely seen in non-treated tially growing cells. Cyclin B1 mRNA levels increased cells (image not shown). 1.8-fold and cyclin E1 mRNA levels increased 2.7-fold The expression of cyclin B1 and cyclin E1 within the after CPT treatment. By contrast, the level of cyclin same cell suggested that at least one of these two cyclins D1 mRNA at 24 h CPT treatment was the same as was likely expressed outside of its normal cell cycle that of untreated cells. The increase in cyclin B1 and E1 phase. To examine this possibility, we analysed cells by

Oncogene A role for cyclin B1 in apoptosis A Borgne et al 7365 flow cytometry at 4, 8 and 24 h after treatment with CPT (top row), cyclin E1 was strongly expressed in cells in to determine their position in the cell cycle relative to G1/S-phase as expected, but also in G2/M-phase cells at cyclin B1 and cyclin E1 protein expression (Figure 4). 8 h and more strongly at 24 h at which time 35% of the In this experiment, in which 30% of the cells were cells expressed cyclin E1 outside of the (middle positive for annexin-V labelling by 24 h CPT treatment row). Similarly, cyclin B1 was expressed in G2/M-phase cells as expected, but also in 36% of cells that were in the G1-phase of the cell cycle at 24 h after treatment (bottom row). These results complemented the observa- a tions made by confocal microscopy and show that under conditions that lead to apoptosis, cyclin B1 and cyclin E1 can be expressed outside of their classical cell cycle phase.

Cyclin A, B1 and E1 form active protein kinase complexes in CPT-treated cells The strong expression of both cyclins outside of their normal cell cycle phase led us to examine if cyclins were part of active CDK complexes or not. Cells were treated with CPT and then treated with the CDK inhibitor (R)- roscovitine (CYC202, Seliciclib) (Meijer et al., 1997) and analysed by flow cytometry for annexin-V labelling and PI labelling. Consistent with what had been reported in other models, roscovitine reduced the number of annexin-V staining cells from 47% in CPT treated cells to 28 and 12% (Figure 5a). When tested separately, roscovitine did not induce apoptosis. The ability of a CDK inhibitor to reduce the number of annexin-V stained cells suggested that active CDK complexes were probably contributing the apoptotic programme in this model. We then tested the protein kinase activity associated with the cyclin complexes in cells treated with CPT at various times between 8 and 48 h after treatment. Cyclin A and cyclin B1 complexes showed a strong increase in associated kinase activity (Figure 5b and c) that appeared to be similar to increases in protein levels as detected by Western blotting (Figure 2). By 32 h after treatment, relatively high levels of cyclin A and cyclin B1 kinase activity could be detected. Cyclin E1 complex activity levels were also measured and com- pared during treatment with CPT. The activity in- creased, as compared to non-treated cells, and remained HeLa cells CPT 24h HT 29 CPT 24h active up to the last time point (Figure 5d). These results suggested that three cyclin complexes tested are activated in CPT treated cells and support the inter- pretation that CDK activity is required in this apoptosis model.

b 15 HT29 Figure 3 Confocal microscopy analysis of CPT treated HT29 and HeLa HeLa cells reveal simultaneously expression cyclin B1 and cyclin 10 E1. (a) HT29 cells (top panel) or HeLa cells (bottom panel) were cultivated on glass coverslips and then either not treated or treated with 1 mM CPT for 24 h. Cells were fixed and permeabilized, incubated with cyclin B1 and cyclin E1 antibodies and DAPI and 5 then examined by confocal fluorescence microscopy. In each group, single staining and the merged image of all three stains are shown. The scale bar represents 20 mm. (b) HT29 and HeLa cells were

% cells with cyclin B1 and E 0 treated with 1 mM CPT for various times. The coverslips were 0162432 observed by microscopy and the number of cells that were positive Time CPT (h) for both cyclin B1 and cyclin E1 are expressed as percentage.

Oncogene A role for cyclin B1 in apoptosis A Borgne et al 7366 Not treated CPT 4hCPT 8h CPT 24h

2 % 2 % 2 % 6 %

1 % 1 % 2 % 30 % PI fluorescence Annexin-V FITC

0 % 0 %1 % 3 % 19 % 19 %32 % 35 %

Cyclin E1 Cyclin E1-FITC PI fluorescence

Cyclin B1

0 %32 % 1 %29 % 2 % 31 % 36 % 50 % Cyclin B1-FITC PI fluorescence Figure 4 CPT treated HT29 cells express cyclin B1 and cyclin E1 outside a standard cell cycle phase. HT29 cells were either not treated or treated with CPT for 4, 8 or 24 h and then analysed by flow cytometry for PI and annexin-V labelling (top row). In this experiment, 30% of the cells were in primary apoptosis and 6% in secondary apoptosis by 24 h treatment. A subset of these cells were then analysed to determine cyclin E1 or cyclin B1 protein expression relative to the cell cycle position. The intensity scale for cyclin E1 is logarithmic and the intensity scale for cyclin B1 is linear.

Reduction of cyclin B1 but not cyclin E1 reduces the cyclin B1 siRNA cells showed only 10% annexin-V number of cells that stain for annexin-V after treatment positive cells whereas scrambled and transfected only with CPT cells showed 26 and 31% positive cells, respectively. By To address directly the role of cyclin expression in the 48 h (Figure 6d), only 39% of the cells were in primary commitment to apoptosis, we wanted to reduce cyclin apoptosis and 14% were in secondary apoptosis (53% protein levels and then determine whether the kinetics of total), whereas transfected only and scrambled cells apoptosis entry were altered. We examined the levels of showed 64 and 53% cells in primary apoptosis and 26 cyclin B1 in cells that were treated with roscovitine, or and 34% in secondary apoptosis (90 and 87% total) with roscovitine plus CPT. As shown in Figure 6a, respectively. Cells in which the levels of cyclin B1 were cyclin B1 protein levels increased after CPT treatment. lowered showed a consistent reduction in CPT-induced In the presence of roscovitine, however, cyclin B1 levels apoptosis. remained relatively low even after CPT treatment. The Cyclin E1 protein levels and activity were also lack of cyclin B1 complex, as well as the inhibition of elevated in CPT-treated cells, therefore, we tested if it cyclin B1 activity may participate in the block of the was required for apoptosis. As had been carried out for apoptosis program as described Figure 4. We explored cyclin B1, cells were treated with siRNAs against human this interpretation further by using RNA interference cyclin E1 and protein levels were analysed (Figure 7a). against cyclin B1 and measuring entry into apoptosis. The levels of cyclin E1 were strongly reduced in cells Cells were treated with siRNAs specific against that were treated with siRNAs but not in cells that were human cyclin B1 and with control siRNAs containing treated with siRNAs against cyclin B1 or scrambled a scrambled sequence. At 24 and 48 h after transfection, sequences. The absence of cyclin E1 had little effect cells were analysed by FACS and by Western blotting to upon the cell cycle profile (not shown). The number of determine cyclin B1 protein levels (Figure 6b). The siRNA cyclin E1 cells that stained for annexin-V either cyclin B1 protein could not be detected in siRNA cyclin in primary or secondary apoptosis after CPT were B1 treated cells at 48 h, even in cells that were co-treated similar to the control cells. This suggested that despite with CPT, whereas a strong induction of cyclin B1 the strong induction of cyclin E1 protein and its protein levels were detected in the control. Cells that did associated protein kinase activity, this cyclin complex not express cyclin B1 protein were more likely to be in is not required for apoptosis in this model. the G2/M phase of the cell cycle (38%) as compared to the control cells. By 48 h, there was a reduction in the number of cells in G2/M-phase (14%) and a corre- CPT-treated cells enter mitotic catastrophe at the time sponding increase in the number of cells with sub-G1 when cells have high cyclin B1 kinase activity levels of DNA. The requirement for cyclin B1 protein and activity in We then compared the effects of CPT upon primary apoptosis led us to examine if these cells enter apoptosis (annexin-V positive and PI negative) in cells before dying. Cells were treated with CPT for varying that were negative for cyclin B1. At 24 h (Figure 6c), times and then examined for mitotic figures (Figure 8a).

Oncogene A role for cyclin B1 in apoptosis A Borgne et al 7367 µ a Not treated CPT 1 M b Cyclin A 1 % 8 % CPT time h

8 16 32 40 44 48

IPCont Nt Noco Histone H1 (33P) 3 % 39 % 5 PI fluorescence 4 Annexin-V FITC 3 2 CPT+R 1 µM CPT + R 5 µM 1 Arbitray Units 0 6 % 4 % 08 1632404448 Time CPT (h)

c Cyclin B1 22 % 8 % CPT time h 44

I 8 16 32 40 48

Nt IPCont Noco R5 µM Histone H1 (33P) 1 % 6 4 2

Arbitray Units 0 1 % 08 1632404448 Time CPT (h)

d Cyclin E1 t CPT time h

8 16 32 40 44 48

Noco Nt IPCont Histone H1 (33P) 6

4

2

Arbitray Units 0 08 1632404448 Time CPT (h)

Figure 5 CPT treatment induces CDK activity in HT29 cells. (a) Cells were either not treated or treated for 24 h with 1 mM CPT or CPT and 1 or 5 mM roscovitine (R) before analysis by flow cytometry for PI and annexin-V labelling. Cells treated only with 5 mM roscovitine were also analysed. The percentage of cells in primary and in secondary apoptosis is given for each test condition. (b) Extracts were prepared from cells that were not treated (Nt) or treated for different times (h) with CPT. Cyclin A complexes were isolated by immunoprecipitation and the associated protein kinase activity was measured in histone H1 kinase assays by phosphorImager analysis. IP cont indicates activity associated with a non-specific antibody. Noco indicates extracts prepared from mitotic cells arrested by nocodazole treatment. Values are given in arbitrary units as the average of two experiments. (c) Extracts were prepared from cells and protein kinase activity associated with cyclin B1 complexes was measured as in (b). (d) Extracts were prepared from cells and protein kinase activity associated with cyclin E1 complexes was measured as described in (b).

Cells that displayed chromosomes in either mitotic catastrophe, and that this event was related to or configurations were easily detected with the requirement for activity. anti-phosphorylated histone H3 antibodies (top row) in control cells. Approximately 6% of cells in an exponen- tially growing culture can be detected by this method. The number of cells that were positive for phospho- Discussion histone H3 staining were then counted at various times after CPT treatment (Figure 8b). At 16 h o1% of the We have developed an experimental model to study the total cells were positive, whereas the number increased potential role of cyclins in apoptosis. HT29 cells treated to 10% by 32 h, in a pattern that was similar to cyclin B1 with CPT enter an apoptosis program in which progress kinase activity (Figure 5c). At 32 and 40 h, chromo- through the cell cycle is blocked in all phases. By using somes and DNA that co-stained with phospho-histone this model we have obtained evidence that cyclin levels H3 appeared to be poorly organized as compared to the and activity increase during the period when cell engage well-formed mitotic figures seen in control cells. The the apoptosis program. Of these, cyclin B1 is partially chromosomes shown in the enlarged images resembled required for the apoptosis program because the reduc- that of chromosomes in cells in mitotic catastrophe. tion of cyclin B1 protein by RNAi or by chemical These results suggested that CPT-treated cells enter a inhibition decreases the number of apoptotic cells.

Oncogene A role for cyclin B1 in apoptosis A Borgne et al 7368 To study the role of cyclins in apoptosis, it was The expression of cyclin B1 outside of the G2/M-phase necessary to identify chemical treatments that commit of the cell cycle has been previously reported in cells to apoptosis within a time scale amenable to further non-dividing neuronal cells, haematopoietic cell lines experimental manipulation. We chose to use the (Porter et al., 2000) and cells taken from human topoisomerase I inhibitor, CPT, at 1 mM by which 93% tumours (Shen et al., 2002). The levels of cyclin B1 of HT 29 cells are in apoptosis in 72 h. At concentrations protein and associated activity decreased immediately of around 100 nM, CPT induces a G2/M-phase block in after CPT treatment, as had been previously the cell cycle by activating a DNA damage checkpoint reported (Maity et al., 1996). However, an increase in ((Tanizawa et al., 1994), data not shown). Whereas cyclin B1 kinase activity at approximately 32 h after these cells also enter apoptosis, it is more difficult treatment with CPT occurred when the majority of cells to characterize a putative apoptosis-specific activity of were not in the G2/M-phase as measured by DNA cyclin B1/CDK1 complex in G2/M-phase cells because content. cyclin B1 protein levels are already elevated as compared Cells that express cyclin E1 outside of S-phase were to G1 cells where cyclin B1 levels are very low (Pines and detected by flow cytometry and the simultaneous Hunter, 1989). Therefore, we used higher concentrations expression of cyclin B1 and cyclin E1 in cells was also of CPT, as have been previously reported in the detected by confocal immunofluorescence microscopy. literature (Smits et al., 2000; Furuta et al., 2003), at As in the case of cyclin B1, examples of cyclin E1 which cells were blocked in all major phases of the cell expression outside of the normal cell cycle phase have cycle. The reason for this cell cycle ‘freezing’ is not been reported in pluripotent murine embryonic cells known but it provided a convenient experimental (Stead et al., 2002) and in cells undergoing apoptosis method that permitted us to observe the expression of (Mazumder et al., 2000). The cyclin E1 complex was cyclin E1 and cyclin B1 outside of their normal phase of active at 16 h after treatment, which was earlier than the cell cycle. cyclin A or cyclin B1 complexes. In contrast to previous We detected cyclin B1 protein expression both by flow reports, we did not detect small p18 forms of cyclin E cytometry and by immunofluorescence microscopy to that might induce activated CDK complexes (Porter confirm that the expression profile was not directly et al., 2001; Mazumder et al., 2002) although the same linked to the position of the cell in the cell cycle. antibody was used.

a

CPT time (h) Noco Rosc./CPT C 16 24 38 Rosc. Cyclin B1

b 48h transfection 256 60 % 256 59 %

1 Events Events 25 % 22 % 15 % 18 % 0 0 0PI 1023 0PI 1023 Transf siRNA cont

siRNA cont.

siRNA B1 + CPT

Transf. + CPT siRNA cont. + CPT

Transf. siRNA B1 256 256

Cyclin B1 42 % 36 % Events Events 16 % 30 % 4 % 38 % 20 % 14 % 0 0 0PI 1023 0 PI 1023 siRNA B1 24h siRNA B1 48h Figure 6 Treatment with cyclin B1 siRNAs strongly reduces CPT induced apoptosis. (a) Cells were either not treated (c; control) or treated for various times h with CPT or roscovitine (Rosc) or treated with both roscovitine and CPT (Rosc/CPT). After treatment, extracts were analysed by Western blotting with cyclin B1 antibodies. Equivalent protein loading for each sample was confirmed by Coomassie Blue staining (not shown). Cells treated with nocodazole (Noco) were included as a control for cyclin B1 expression. (b) Cells were treated by mock transfection (transf) or by scrambled siRNA sequences (siRNA cont) or by siRNAs directed against cyclin B1 (siRNA B1). Cells were then either not treated or treated with CPT ( þ CPT). 48 h after transfection, extracts were analysed by Western blotting with cyclin B1 antibodies (left panel). Equivalent protein loading for each sample was confirmed by Coomassie Blue staining (not shown). Cells treated by the same transfection conditions (but without CPT) were analysed by flow cytometry for DNA content (right panel). The sub-G1, G1, S, (G2–M) phase distribution is shown in percentage. (c) Cells were treated with siRNA against cyclin B1 as described in (b) and then either not treated (NT) or treated with CPT. After 24 h, cells were collected and analysed for annexin-V and PI labelling. The percentage of positive cells is given. (d) Cells were treated with siRNA against cyclin B1 as described in (b) and then either not treated (NT) or treated with CPT. After 48 h, cells were collected and analysed as described in (c).

Oncogene A role for cyclin B1 in apoptosis A Borgne et al 7369 c Transfected only Cont.siRNA siRNA Cyclin B1 NT 24h 7 % 8 % 7 % PI

4 % 4 % 4 %

Annexin V + CPT 24h 6 % 9 % 9 % PI

26 % 31 % 10 %

Annexin V

d Transfected only Cont.siRNA siRNA Cyclin B1 NT 48 h 6.9 10.0 9.2 PI

4.6 6.0 7.6

Annexin V + CPT 48 h 26.2 34.3 14.1 PI

63.9 52.8 39.0

Annexin V Figure 6 Continued

Several lines of evidence support the hypothesis that A reduction in cyclin B1 protein levels after roscovitine active CDK complexes participate in CPT-induced and CPT treatment had been previously reported for apoptosis. We found that cyclin A, B1 and E1 cyclin/ synchronised HT29 cells (Abal et al., 2004). The CDK complexes were active after treatment with CPT. reduction is likely indirectly due to inhibition of Consistent with a role in apoptosis, (R)-roscovitine, CDK7/CDK9 complexes, whose activities are required which inhibits CDK1 and CDK2 catalytic subunits for transcription of a variety of mRNAs including cyclin (Knockaert et al., 2002) also blocked the apoptotic B1 and anti-apoptotic (MacCallum et al., 2005). program, although it does not permit one to identify We then tested the role of cyclin B1 and cyclin E1 which cyclin CDK complex is specifically required directly by blocking the expression by RNAi. siRNA for amongst the active complexes present. In addition to cyclin B1 reduced the amount of cyclin B1 protein and the known ATP competition activity of (R)-roscovitine, reduced the number of cells in primary apoptosis at 24 we found that cyclin B1 levels were reduced in the and 48 h. It has been previously reported that reduction presence of this compound, even after CPT treatment. of cyclin B1 protein by RNAi (without further treatment

Oncogene A role for cyclin B1 in apoptosis A Borgne et al 7370 a Figure 2). The role of cyclin A is more difficult to evaluate because its level increases during both S-phase Cyclin E1 and during G2/M-phase of the cell cycle; we decided to focus our studies upon cyclin E1 and cyclin B1 because their expression is temporally distinct under prolifera-

Control tion conditions. siRNA E1 siRNA B1 The question of whether or not cells pass via a ‘failed mitosis’ or mitotic catastrophe during apoptosis has b Control siRNA Cyclin E1 been prompted by the discovery of active CDK complexes during apoptosis (Fotedar et al., 1995; 8 % 10 % Castedo et al., 2002a). A link between entry into mitosis - CPT and apoptosis has been described in a different model in which HeLa Env/CD4 syncytia express active cyclin 5 % 6 % PI fluorescence B/CDK1 complexes. In these cells, the nuclear lamina Annexin-V FITC disappears, microtubules are reorganised into multi- polar spindles and DNA condenses into like structures, which is reminiscent of a mitotic 13 % 14 % catastrophe (Castedo et al., 2002b). Mitotic catastrophe + CPT has also been identified in HT29 cells treated with C-1311, a putative topoisomerase II inhibitor (Hyzy 22 % 23 % et al., 2005). We directly observed CPT treated cells by confocal microscopy and detected cells with chromo- Figure 7 Treatment with cyclin E1 siRNAs does not block CPT somes that were poorly organized relative to non-treated induced apoptosis. (a) Cells were treated with siRNAs directed cells (Figure 8). These cells also expressed phospho- against cyclin E1, cyclin B1 or scrambled (Control) sequences. 24 h after transfection, extracts were prepared and samples were histone H3, a phosphorylated epitope that is expressed analysed by Western blotting for cyclin E1 protein. (b) Cells were during mitosis in proliferating cells (Sauve et al., 1999). treated with scrambled siRNA (control) or cyclin E1 siRNA and The coexpression of phospho-histone H3 in cells with then either not treated with CPT (ÀCPT) or treated with CPT condensed chromosomes, at the time when cyclin B1 is ( þ CPT). After 24 h, cells were collected and analysed for annexin- active, strongly supports the idea that these cells have V and PI labelling. The percentage of positive cells is given. entered a mitotic catastrophe. The activation of a cyclin B1 complex and its eventual outcome on the cell might be determined by the cellular context. For example, the expression of cyclin B1 gene with cytotoxic agents such as CPT) increases the and production of protein outside of the G2/M-phase of percentage of cells in G2/M-phase and initiates apop- the cell cycle may create a cellular context in which tosis by 48 h (Yuan et al., 2004). In the results reported cyclin B1 can be used for apoptosis rather than cell here, we also detected a cell cycle arrest and an increase division (Shen et al., 2004). The localization of cyclin B1 in sub-G1 cells after siRNA treatment at 48 h has an important role in apoptosis in haemopoietic- (Figure 6b). The induction of a cell death pathway, derived cells (Porter et al., 2003). There also may be a independent of CPT, might explain why cyclin B1 different accessibility to substrates of cyclin B/CDK1 siRNAs could not completely protect cells from CPT that would change the outcome of its activity. For dependent apoptosis. Our results suggest that cyclin B1 example, the pro-apoptotic protein BAD is phosphory- has an important role in CPT dependent apoptosis lated by cyclin B1/CDK1 both in vitro and in vivo during and that there is a requirement for specific cyclin apoptosis in postmitotic neurons (Konishi et al., 2002). B1-dependent kinase activity rather than a generic We were unable to detect Bad phosphorylation in our CDK activity as might be supplied by active cyclin model. In view of the increasing amount of evidence that E complexes. CDK1 (and other CDKs) participate in apoptosis, it is Although the expression and corresponding activity possible that CDKs are activated and function in a of cyclin E1 correlated well with the induction of the controlled manner during apoptosis rather than their apoptosis program, we were unable to reduce the misregulation or ‘unscheduled’ activity being the cause number of cells in apoptosis with cyclin E1 siRNA. of apoptosis. With this view in mind, we have The reduction of cyclin E1 protein by RNAi is toxic to characterized the CPT-induced apoptosis model because some cell lines after 96 h of culture (Li et al., 2003). In relatively long period before cell death will permit the model described here, we focused on activities within detailed analysis of CDK complexes in apoptosis. 48 h, at times when cyclin E levels were reduced. The An important difference between cyclin B1 in expression of cyclin E1 might be related to other cellular apoptosis and cyclin B1 in mitosis is that cyclin processes such as stalled replication forks or an S-phase B1/CDK1 is required for progression into the (Bartek and Lukas, 2001). However, these in all species, whereas there are examples in which activities would not be detected in tests designed apoptosis functions without active cyclin B1, or more to identify progression through apoptosis. Cyclin A for review rarely, without a CDK complex (Castedo protein levels also increased after CPT treatment (see et al., 2002a; for review Borgne and Golsteyn, 2003;

Oncogene A role for cyclin B1 in apoptosis A Borgne et al 7371 a DAPI Histone H3-P Merge Enlargement NT 40 h 32 h 24 h

b 10 8 6 4

H3 positive 2

% Cells histone 0 016243240 Time CPT (h) Figure 8 Cells treated with CPT enter a mitotic catastrophe at the time of cyclin B1/Cdk1 kinase activity. (a) HT29 cells were cultivated on glass coverslips and either not treated (NT) or treated with CPT for 24, 32 or 40 h. Cells were fixed and permeabilized, incubated with phospho-histone H3 antibodies and DAPI and then examined by confocal fluorescence microscopy. In each group of images, single staining and the merged image of both stains are shown. An enlarged image from each time point is shown in the right column. The scale bar represents 20 mm. (b) HT29 cells were treated with CPT as described in (a) for various times. The coverslips were observed by microscopy and the number of phospho-histone H3 positive cells were counted and presented as percent of total cells.

Golsteyn, 2005). As only a subset of cell survival signals The multiple effects of CDK inhibitors upon various act on the CDK1 pathway, an important question for members of the CDK protein family may, in some further study is to identify what types of cellular stresses instances, determine the anti- or proapoptotic outcome and survival signals are involved. This study might have of certain inhibitors. For example, roscovitine may important implications in the clinical treatment of enhance the activity of CPT derivatives in vivo (Abal human tumours. Indeed, CDK inhibitors may have et al., 2004), via its effects on transcription in addition to protective roles in non-dividing cells, as suggested by its effects upon CDK1 and CDK2 (MacCallum et al., data obtained with neuronal cells submitted to various 2005). chemical insults (Knockaert et al., 2002). In addition, In one case, there is a need to reduce the proliferative CDK inhibitors could be used to protect cells from the index of tumours, yet in another case, it is necessary to undesired effects of classical chemotherapeutic agents, promote apoptosis of tumours cells. If CDKs have a as illustrated by the protective effects of roscovitine role both in apoptosis as well as proliferation, it will be upon renal cells treated with cisplatin (Price et al., 2004). important to identify the apoptotic pathways in which

Oncogene A role for cyclin B1 in apoptosis A Borgne et al 7372 CDKs function. This would permit a more rational use Acknowledgements of CDK inhibitors as potential anti-tumour agents. This work was supported by the Institut de Recherches Servier Abbreviations as part of a program ‘Alliance Strate´ gique’ with the CNRS. We thank our colleagues in the Cancer Drug Discovery CDK, cyclin-dependent kinase; CPT, camptothecin; FCS, fetal division at the IdRS for valuable discussions and Marie calf serum; PI, propidium iodide. Knockaert for pg beads and advice.

References

Abal M, Bras-Goncalves R, Judde JG, Fsihi H, De Cremoux Mazumder S, Gong B, Almasan A. (2000). Oncogene 19: P, Louvard D et al. (2004). Oncogene 23: 1737–1744. 2828–2835. Adachi S, Obaya AJ, Han Z, Ramos-Desimone N, Wyche JH, Mazumder S, Gong B, Chen Q, Drazba JA, Buchsman JC, Sedivy JM. (2001). Mol Cell Biol 21: 4929–4937. Almasan A. (2002). Mol Cell Biol 22: 2398–2409. Bartek J, Lukas J. (2001). FEBS Lett 490: 117–122. Meijer L, Arion D, Golsteyn R, Pines J, Brizuela L, Hunt T Borgne A, Golsteyn RM. (2003). Progress in Cell Cycle et al. (1989). EMBO J 8: 2275–2282. Research, Vol 5 In: Meijer L, Je´ ze´ quel A and Roberge M Meijer L, Borgne A, Mulner O, Chong JP, Blow JJ, Inagaki N (eds). Life in Progress Editions: Roscoff, France, et al. (1997). Eur J Biochem 243: 527–536. pp 453–459. MeikrantzW, Gisselbrecht S, Tam SW, Schlegel R. (1994). Castedo M, Perfettini J-L, Roumier T, Kroemer G. (2002a). Proc Natl Acad Sci 91: 3754–3758. Cell Death Diff 9: 1287–1293. MeikrantzW, Schlegel R. (1996). J Biol Chem 271: Castedo M, Roumier T, Blanco J, Ferri KF, Barretina J, 10205–10209. Tintignac LA et al. (2002b). EMBO J 15: 4070–4080. Padmanabhan J, Park DS, Greene LA, Shelanski ML. (1999). Choi KS, Eom YW, Kang Y, Ha MJ, Rhee H, Yoon J-W et al. J Neurosci 19: 8747–8756. (1999). J Biol Chem 274: 31775–31783. Park DS, Farinelli SE, Greene LA. (1996). J Biol Chem 271: Coqueret O. (2003). Trends Cell Biol 13: 65–70. 8161–8169. De Luca A, De Maria R, Baldi A, Trotta R, Facchiano F, Pines J, Hunter T. (1989). Cell 58: 833–846. Giordano A et al. (1997). J Cell Biochem 64: 579–585. Porter DC, Zhang N, Danes C, McGahren MJ, Harwell RM, Elbashir SM, Harborth J, Weber K, Tuschl T. (2002). Methods Faruki S et al. (2001). Mol Cell Biol 21: 6254–6269. 26: 199–213. Porter LA, Cukier IH, Lee JM. (2003). Blood 101: Fotedar R, Flatt J, Gupta S, Margolis RL, Fitzgerald P, 1928–1933. Messier H et al. (1995). Mole Cell Biol 15: 932–942. Porter LA, Singh G, Lee JM. (2000). Blood 95: 2645–2650. Furuta T, Takemura H, Liao ZY, Aune GJ, Redon C, Price PM, Safirstein RL, Megyesi J. (2004). Am J Physiol Sedelnikova OA et al. (2003). JBiolChem278: 20303–20312. Renal Physiol 286: 378–384. Golsteyn RM. (2005). Cancer Lett 217: 129–138. Sauve DM, Anderson HJ, Ray JM, James WM, Roberge M. Guo M, Hay BA. (1999). Curr Opin Cell Biol 11: 745–752. (1999). J Cell Biol 145: 225–235. Hakem A, Sasaki T, Kozieradzki I, Penninger JM. (1999). Scatena CD, Stewart ZA, Mays D, Tang LJ, Keefer CJ, Leach J Exp Med 189: 957–967. SD et al. (1998). J Biol Chem 273: 30777–30794. Hartwell LH, Kastan MB. (1994). Science 266: 1821–1828. Shao RG, Cao CX, Shimizu T, O’Connor PM, Kohn KW, Harvey KJ, Blomquist JF, Ucker DS. (1998). Mole Cell Biol Pommier Y. (1997). Cancer Res 57: 4029–4035. 18: 2912–2922. Shen M, Feng Y, Gao C, Tao D, Gong J. (2002). Chin J Oncol Harvey KJ, Lukovic D, Ucker DS. (2000). J Cell Biol 148: 24: 215–218. 59–72. Shen M, Feng Y, Gao C, Tao D, Hu J, Reed E et al. (2004). Hiromura K, Pippin JW, Fero ML, Roberts JM, Shankland Cancer Res 64: 1607–1610. SJ. (1999). J Clin Invest 103: 597–604. Shi L, Chen G, He D, Bosc DG, Litchfield DW, Greenberg Hsu S, Yin S, Liu M, Reichert U, Ho WL. (1999). Exp Cell AH. (1996). J Immunol 157: 2381–2385. Res 252: 332–341. Shi L, Niskioka WK, Th’ng J, Bradbury EM, Litchfield DW, Hyzy M, Bozko P, Konopa J, Skladanowski A. (2005). Greenberg AH. (1994). Nature 263: 1143–1145. Biochem Pharmacol 69: 801–809. Shimizu T, O’Connor PM, Kohn K, Pommier Y. (1995). Knockaert M, Greengard P, Meijer L. (2002). Trends Cancer Res 55: 228–231. Pharmacol Sci 9: 417–425. Shirvan A, Ziv I, Zilkha-Falb R, Machlyn T, Barzilai A, Konishi Y, Lehtinen M, Donovan N, Bonni A. (2002). Mol Melamed E. (1998). Neurochem Res 23: 767–777. Cell 9: 1005–1016. Smits VA, Klompmaker R, Arnaud L, Rijksen G, Nigg EA, Le´ once S, PerezV, Lambel S, Peyroulan D, Tillequin F, Medema RH. (2000). Nat Cell Biol 2: 672–676. Michel S et al. (2001). Mol Pharmacol 60: 1383–1391. Stead E, White J, Faast R, Conn S, Goldstone S, Rathjen J Li K, Lin SY, Brunicardi FC, Seu P. (2003). Cancer Res 63: et al. (2002). Oncogene 21: 8320–8333. 3593–3597. Tanizawa A, Fujimori A, Fujimori Y, Pommier Y. (1994). Livak KJ, Schmittgen TD. (2001). Methods 25: 402–408. J Natl Cancer Inst 86: 836–842. MacCallum DE, Melville J, Frame S, Watt K, Anderson S, Vesely J, Havlicek L, Strnad M, Blow JJ, Donella-Deana A, Gianella-Borradori A et al. (2005). Cancer Res 65: Pinna L et al. (1994). Eur J Biochem 224: 771–786. 5399–5407. Wang J, Walsh K. (1996). Science 273: 359–361. Maity A, Hwang A, Janss A, Phillips P, McKenna WG, Yuan J, Yan R, Kramer A, Eckerdt F, Roller M, Kaufmann Muschel RJ. (1996). Oncogene 13: 1647–1657. M et al. (2004). Oncogene 23: 5843–5852.

Oncogene