©2006 L

ANDES BIOSCI

ENCE.

DO NOT DIST

RIBUTE.

[Cell Cycle 5:18, 2153-2159, 15 September 2006]; ©2006 Landes Bioscience

Monica Savio1

Michaela Cerri2,†

Ornella Cazzalini1

Paola Perucca1

Lucia A. Stivala1

Pietro Pichierri3

AnnaPaola Franchitto3

Laurent Meijer4

Ennio Prosperi2,*1Dipartimento di Medicina Sperimentale; Patologia Generale; Università di Pavia;Pavia, Italy

2Istituto di Genetica Molecolare del CNR; Pavia, Italy;

3Istituto Superiore di Sanità; Roma, Italy;

4Cell Cycle Group; Station Biologique; Roscoff, France

†Present Address: Ematology Unit; Department Clinical and ExperimentalMedicine; University of Eastern Piedmont; Novara, Italy

*Correspondence to: Ennio Prosperi; Istituto di Genetica Molecolare del CNR, sez.;Istochimica e Citometria; Piazza Botta 10; 27100 Pavia, Italy; Tel.:+39.0382.986267; Fax: +39.0382.986430; Email: prosperi@igm.cnr.it

Original manuscript submitted: 06/22/06Revised manuscript submitted: 07/18/06Manuscript accepted: 07/26/06

Previously published online as a Cell Cycle E-publication:http://www.landesbioscience.com/journals/cc/abstract.php?id=3235

KEY WORDS

CDK2 inhibition, Roscovitine, DNA replica-tion, PCNA, S-phase checkpoint, stalledreplication forks

ACKNOWLEDGEMENTS

This work was in part supported by aCNR/CNRS joint program grant to E.P. andL.M. The Authors wish to thank J.M. Sedivyfor providing LF1 fibroblasts, and M. Stefaninifor AT4BI cells.

Report

Replication-Dependent DNA Damage Response Triggered by RoscovitineInduces an Uncoupling of DNA Replication Proteins

ABSTRACTThe cyclin-dependent kinase (CDK) inhibitor roscovitine is under evaluation in clinical

trials for its antiproliferative properties. Roscovitine arrests cell cycle progression in G1and in G2 phase by inhibiting CDK2 and CDK1, and possibly CDK7 and CDK9.However, the effects of CDK2 inhibition in S-phase cells have been not fully investigated.Here, we show that a short-term treatment with roscovitine is sufficient to inhibit DNAsynthesis, and to activate a DNA damage checkpoint response, as indicated byphosphorylation of p53-Ser15, replication protein A, and histone H2AX. Analysis ofDNA replication proteins loaded onto DNA during S phase showed that the amount ofproliferating cell nuclear antigen (PCNA), a cofactor of DNA replication enzymes, wassignificantly reduced by roscovitine. In contrast, chromatin-bound levels of DNA poly-merase δ, DNA ligase I and CDK2, were stabilized. Checkpoint inhibition with caffeinecould rescue PCNA disassembly only partially, pointing to additional effects due to CDK2inhibition and the presence of replication stress. These results suggest that in S-phasecells, roscovitine induces checkpoint-dependent and -independent effects, leading tostabilization of replication forks and an uncoupling between PCNA and PCNA-interactingproteins.

INTRODUCTIONCell cycle inhibition is an important target of pharmacological agents used for controlling

cell proliferation in various diseases.1,2 Cyclin-dependent kinases (CDKs) have been shownto be important regulators of cell cycle progression in each cell cycle compartment.3,4

Among CDKs, the type 2 enzyme has received much attention as a relevant moleculartarget regulating both G1/S phase, and G2 phase transitions.5-7 Recently, CDK2 has beenshown to be dispensable for cell cycle progression,8 probably due to the redundancy ofcellular CDK activities.3,4 Nevertheless, cell cycle inhibition by synthetic compoundsremains undoubtedly useful for antiproliferative therapy, given the physiological role ofCDKs required for cell cycle progression.9,10

Roscovitine is a purine derivative kinase inhibitor with a high degree of specificitytoward CDKs, namely CDK1, 2, 5, 7 and 9.11,12 The cell cycle effects induced by roscovitinehave been investigated on several cell lines,13,14 and this kinase inhibitor has entered clinicaltrials, thanks to its antiproliferative properties.15 In fact, roscovitine has been shown toarrest cell cycle progression at the G1 and G2+M transitions, due to inhibition of CDK1and CDK2-mediated phosphorylation of relevant targets.13,16 However, inhibition ofCDK7 and 9 resulting in impaired RNA synthesis may also contribute to arrest.17,18

Roscovitine induces also stabilization of p53,19-21 concomitatly with phosphorylation onserine 15,22 although some controversy exists concerning this modification.17

Accumulation of p53 may be induced by the inhibition of MDM2 transcription,23,24 orvia a nucleolar export mechanism.25 More recently, chemical inhibition or expression of adominant negative form of CDK2, have been found to activate an intra-S-phase check-point22,26 leading to ATM (ataxia telangiectasia mutated)/ATR (ATM and Rad3-related)-dependent p53 stabilization, and to modifications at the level of chromatin-bound MCMproteins in the prereplication complex.22,27 This is consistent with a role of CDK2 inreplication origin firing.28 It is known that an S-phase checkpoint is induced followinginhibition of DNA synthesis, due to stalling of the replication forks.29-31 However, it isnot yet clear whether p53 accumulation and S-phase checkpoint activation induced byCDK inhibitors are both triggered by the inhibition of DNA synthesis, or whether othermechanisms are involved. In fact, it has been shown that p53 accumulation may occur

www.landesbioscience.com Cell Cycle 2153

with multiple pathways,32-34 which may have little correlation withthe S-phase checkpoint.35

To gain more insight into the effects induced by roscovitine inS-phase cells, we have investigated p53 stabilization and activation ofS-phase checkpoint, and the consequences on molecular eventsoccurring during DNA replication. In particular, we have examinedthe DNA binding of replication proteins like proliferating cellnuclear antigen (PCNA), an important cofactor of DNA synthesis,36

and PCNA-interacting enzymes, like DNA polymerase δ (Pol δ) andDNA ligase I (Lig I). The latter proteins have been chosen becausetheir interaction with PCNA is regulated by CDK2.37 We havefound that roscovitine triggers a replication-dependent DNA damageresponse that appears to be only in part responsible for p53 accumu-lation, since increase in p53 protein was found to occur in all thephases of the cell cycle. In S-phase cells, roscovitine induced areduction in the levels of chromatin-bound PCNA, and an uncouplingof PCNA-interacting proteins. Checkpoint inhibition with caffeinecould not completely restore chromatin-bound PCNA to controllevels, suggesting that PCNA disassembly from chromatin was onlypartially dependent on S-phase checkpoint, and directly related toCDK2 inhibition of new origin firing.28 These results show that atcellular level, roscovitine may exert multiple cell-cycle dependenteffects that could be exploited for improving chemotherapyapproaches.

MATERIALS AND METHODSCell cultures and treatments. LF1 human embryonic lung

fibroblasts, kindly provided by Prof. J.M. Sedivy (Brown University,RI, USA), were grown in Earle’s minimal essential medium (MEM,Invitrogen) supplemented with 10% foetal bovine serum (FBS,Invitrogen), 100 units/ml penicillin and 100 µg/ml streptomycin ina 5% CO2 atmosphere. HeLa cells were cultured in Dulbecco’smodified Eagle’s medium (DMEM, Sigma) supplemented with 10%FBS (Gibco BRL), 4 mM L-glutamine (Gibco BRL), 100 units/mlpenicillin, 100 µg/ml streptomycin. ATM-defective AT4BI andAT5BI primary fibroblasts,38 kindly provided by Dr. M. Stefanini(IGM-CNR, Pavia, Italy), were grown in MEM medium supple-mented with 10% FBS. h-TERT-immortalised primary BJ fibroblastswere maintained in DMEM high-glucose (Gibco BRL) supplementedwith 2 mM L-glutamine and 15% FBS. Roscovitine (Calbiochem)was prepared as a 10 mM stock solution in DMSO and diluteddirectly in culture medium at 20 µM final concentration, previouslydetermined giving cell cycle arrest.16 PD98059 (Calbiochem) wassimilarly prepared as a 50 mM stock solution. Untreated controlsamples received a similar concentration of DMSO. Hydroxyurea(HU) and caffeine (Sigma) were prepared as 200 mM stock solutionin double-distilled (dd) H2O. In some experiments, cells were treatedfor 4 h with 20 µM 5,6-Dichloro-1-β-D-ribofuranosylbenzimidazole(DRB, Sigma). For synchronization in S phase, cells were incubatedfor 24 h in 3 µg/ml aphidicolin (Calbiochem), or in 2 mM hydrox-yurea (Sigma), followed by subsequent release in drug-free mediumfor at least 1.5 h before the addition of roscovitine. Alternatively,fibroblasts were synchronized in S phase by serum starvation for72 h, and then reincubated in complete medium for 22 h. Thedegree of S-phase synchronization, as determined by flow cytometry,16

usually ranged between 50–60% for human fibroblasts, and between80–90% for HeLa cells. In some experiments, LF1 fibroblasts wereirradiated with UV-C light (10 J/m2) using a germicidal lamp, asdescribed.39

RNA interference. RNA interference with small interfering RNA(siRNA) was used to downregulate ATR expression. Cells were trans-fected with 10 nM siRNA pool (Dharmacon) with the HiPerfectreagent (Qiagen), according to the manufacturer. As a control,h-TERT-immortalised BJ fibroblasts were transfected with siRNAsdirected against GFP, or without siRNAs. Forty-height hours aftertransfection (when ATR levels were reduced >80%), cells wereexposed to roscovitine and cell lysate prepared 4 h later by boiling inSDS sample buffer for SDS-polyacrylamide gel electrophoresis(PAGE) analysis.

Analysis of DNA synthesis by BrdU incorporation. Cells wereincubated with 30 µM bromodeoxyuridine (BrdU) during the last30 min of treatment with roscovitine, or each of the other conditionsanalysed. Cells treated with UV-C light were incubated with BrdUfor 30 min, immediately after irradiation. Cells were fixed in 70%ethanol, and immunostained with anti-BrdU monoclonal antibody(Amersham), as described.16 Samples were analysed by an Epics XLflow cytometer (Coulter Corp.). For the double immunostaining ofBrdU and histone H2AX, cells on coverslips were treated withDNase I, before antibody incubation.40

Flow cytometry and immunofluorescence analysis of p53 andPCNA. Cells were seeded in flasks, or on coverslips, and used 24 hlater. After roscovitine treatment, cells were detached by trypsin,washed in PBS and fixed in 2% formaldehyde solution in PBS for5 min at room temperature (r.t.) and then post-fixed in 70%ethanol. For immunostaining of chromatin-bound PCNA, sampleswere lysed at 4˚C with hypotonic buffer containing 10 mM TrisHCl (pH 7.4) 2.5 mM MgCl2, 0.5% Nonidet NP-40, 1 mMphenylmethylsulfonyl fluoride (PMSF), 0.2 mM Na3VO4, and acocktail of protease and phosphatase inhibitors (Sigma), asdescribed.39 Cells grown on coverslips were washed twice in PBS,dipped in cold ddH2O and lysed in situ for 10 min at 4˚C in hypo-tonic buffer as above, except that NP-40 was present at 0.1%.Thereafter, samples were washed once in hypotonic buffer, fixed in1% formaldehyde for 5 min (r.t.), and then post-fixed in 70%ethanol. After rehydration, samples were blocked in PBST buffer(PBS, containing 0.2% Tween 20) containing 1% bovine serumalbumin (BSA), and then incubated for 1 h with anti-p53 (D0-7),or anti-PCNA (PC10) monoclonal antibodies (Dako), diluted1:100 in PBST/BSA buffer. After three washes in PBST buffer, sam-ples were incubated for 30 min with Alexa 488-conjugated anti-mouse antibody (Molecular Probes) diluted 1:200 in PBST/BSAbuffer. After immunoreactions, samples for flow cytometry wereincubated for 30 min (r.t.) with 20 µg/ml Propidium Iodide (PI) inPBS containing 1 mg/ml RNase A, and then cells were analysed byflow cytometry, as above indicated. Cells on coverslips were stainedwith Hoechst 33258 dye (0.5 µg/ml) for 2 min (r.t.) and washed inPBS. Slides were mounted in Mowiol (Calbiochem) containing0.25% 1,4-diazabicyclo-[2,2,2]-octane (Aldrich) as antifading agent.Samples were viewed with an Olympus BX51 fluorescence microscope,and photographed with an Olympus C4040 digital camera.

Cell extraction procedures and Western blot analysis. For totalcontent determination of relevant proteins, cells were directly lysedin loading buffer and boiled. After very brief sonication, sampleswere loaded onto 10 or 12% gels for SDS-PAGE analysis. For analysisof chromatin-bound proteins, 5–8 x 106 cells were lysed in 1 ml ofhypotonic buffer as above, with the addition of 0.5 µM okadaic acid(Alexis), 10 mM β-glycerophosphate, and a cocktail of protease andserine/threonine phosphatase inhibitors (Sigma).39 Samples werewashed twice in lysis buffer, and then digested for 15 min at 4˚C

Uncoupling of DNA Replication Proteins by Roscovitine

2154 Cell Cycle 2006; Vol. 5 Issue 18

www.landesbioscience.com Cell Cycle 2155

Uncoupling of DNA Replication Proteins by Roscovitine

with DNase I (20 units/106 cells in 10 mM Tris-HCl, pH 7.4,containing 5 mM MgCl2, 0.2 mM PMSF, 0.5 µM okadaic acid,and protease/phosphatase inhibitors).39 Both detergent-soluble andchromatin-bound extracts were mixed with 3X SDS-loading bufferand boiled before SDS-PAGE. Equal amount of proteins weretransferred onto nitrocellulose (Amersham), blocked with 5% non-fatdry milk, followed by 1 h incubation with primary antibody. Afterextensive washings and incubation in secondary anti-mouse oranti-rabbit HRP-labelled antibodies, reaction was detected bychemiluminescence (Amersham).

Monoclonal antibodies to p53 (DO-7) and PCNA (PC10) wereobtained from Dako; antibodies to p21 (CP74), and Lig I (1A9)were from NeoMarkers; monoclonal antibodies to actin (C-40) andto Pol δ (C22), were from Sigma and BD Biosciences, respectively.The monoclonal antibody to RPA2 (9H8) was a gift of prof. M.Wold (Iowa University). The polyclonal antibodies anti-p53Ser15was obtained from Oncogene Science, anti-CDK2 (M2) from SantaCruz, and antibodies against histone γ-H2AX and histone H3 werefrom Upstate.

RESULTSDirect inhibition of DNA synthesis by Roscovitine. To define a

temporal window useful to analyse the effect of CDK2 inhibition onS-phase cells without significant changes in cell cycle distribution,LF1 fibroblasts were incubated in the presence of 20 µM roscovitinefor periods ranging from 1 to 4 h. BrdU was added during the last30 min of incubation and DNA synthesis was then determined byassessing BrdU incorporation with immunofluorescence staining.Figure 1 shows the levels of BrdU incorporation, as determined byflow cytometry, indicating that roscovitine inhibited progressivelyDNA synthesis, reaching an almost complete block after a 4-h treat-ment. At this time, distribution of cells in each phase was onlyslightly affected by roscovitine (Control, G1 = 71%, S = 19%, G2 +M = 10%; Roscovitine, G1 = 69%, S = 17%, G2+M = 14%), thusindicating that the effects observed on S-phase cells were not attrib-utable to significant modifications in cell cycle position. A similarreduction in BrdU incorporation was observed in HeLa cells treatedwith roscovitine in the same conditions (not shown).

To exclude that the observed effects on DNA synthesis were dueto inhibition of other kinases, such as Erk1 and Erk2 kinases,12 LF1fibroblasts were treated with 20 µM PD98059, a specific inhibitorof Erk kinases.11 This concentration was previously found to effec-tively arrest cell cycle in G1 phase, after a 24-h treatment period (notshown). Figure 2A shows that, as compared with roscovitine, a 4-h

treatment with 20 µM PD98059 did not inducea significant inhibition of BrdU incorporation inS-phase cells, thus suggesting that the effect ofroscovitine on DNA synthesis was not due toinhibition of Erk kinases. For comparison, UV-Cirradiation, that is known to induce DNA damageand checkpoint activation, induced a completearrest of BrdU incorporation, as observed at 4 hafter exposure. Since it is known that roscovitineinduces p53 accumulation,16 the effect ofPD98058, and of UV-C irradiation on p53protein levels was investigated in LF1 fibroblasts.The results shown in Figure 2B indicate that a 4-htreatment with roscovitine induced a significantaccumulation of p53 protein levels, and p53

phosphorylation on serine 15 (p53pS15). These changes were accom-panied by an increase in p21 protein levels. In contrast, treatment ofLF1 cells with PD98058 resulted in a modest variation in both p53and p21 cellular contents. Again, as expected for a DNA-damagingagent, UV irradiation induced an increase in p53 protein andp53pS15 levels, while p21 protein levels were significantly reducedat this time and dose, as previously described.41

Roscovitine induces p53 accumulation in all phases of the cellcycle. To clarify whether p53 accumulation induced by roscovitinewas most prominent in S-phase cells in dependence of DNA synthesisinhibition, the cell cycle distribution of p53 protein was determinedby dual-parameter flow cytometric analysis of p53 immunofluorescencevs DNA content. For comparison, parallel samples were treated withhydroxyurea (HU), an inhibitor of ribonucleotide reductase thatinduces replication fork arrest and S-phase checkpoint activation.42

Figure 3A shows that p53 immunostaining was clearly increased inroscovitine-treated cells, as compared with untreated control sample,and this event occurred to a similar extent in each phase of the cellcycle. In HU-treated cells, p53 immunofluorescence increased to alower level than that reached by roscovitine. However, this occurredpredominantly in cells located at the G1/S and in S phase. Thecontribution of checkpoint activation in the accumulation of p53protein was then investigated by abrogating the checkpoint withcaffeine, an inhibitor of ATM/ATR kinases.43 The levels of p53protein in roscovitine-treated and untreated fibroblasts were deter-mined by western blot analysis and compared with samples treatedwith HU. Figure 3B shows that caffeine was able to significantly, butnot completely, reverse the accumulation of p53 protein and p53S15phosphorylation (p53pS15) induced by roscovitine. HU inducedaccumulation of p53 protein and Ser15 phosphorylation to a lowerextent than roscovitine. However, caffeine substantially reversed p53accumulation and p53S15 phosphorylation induced by HU.

Checkpoint activation by Roscovitine is ATR-dependent. Sincep53 was accumulated in all phases of the cell cycle, we investigatedwhether DNA synthesis inhibition was due to checkpoint activationin S phase cells.30,31 The effect of roscovitine on p53 protein levelswas first analyzed in AT4BI fibroblasts in which ATM function isinactive.38 The results reported in Figure 4A show that in these cells,p53 protein was still stabilized and phosphorylated on Ser15 afterroscovitine treatment (DNA synthesis was also inhibited, notshown), thus indicating that checkpoint activation induced byroscovitine occurred independently of ATM function. To furtherextend these results, BJ fibroblasts were treated with siRNA tosuppress ATR expression. An effective reduction of ATR protein byRNAi was obtained at 48 h after transfection (not shown); at this

Figure 1. Effect of short-term treatment with roscovitine on the DNA synthesis in normal humanfibroblasts. Flow cytometric analysis of BrdU incorporation in LF1 human fibroblasts treated with20 µM roscovitine for the indicated periods of time and BrdU was added during the last 30 minof incubation. Flow cytometry dot-plots show BrdU incorporation determined by immunofluores-cence staining with anti-BrdU antibody, and DNA content determined by PI staining.

Uncoupling of DNA Replication Proteins by Roscovitine

time, cells were treated for 4 h with roscovitine. Analysisof p53pS15 levels by Western blot showed that roscov-itine greatly reduced the p53S15 signal in cells transfectedwith siRNA for ATR, while it was clearly detectable insamples transfected with control siRNA or in nontransfected cells (Fig. 4B), thus indicating that p53phosphorylation was dependent more on ATR than onATM kinase.

Roscovitine induces uncoupling of DNA replicationproteins. To investigate more in details the effectinduced by roscovitine at the levels of DNA replicationproteins, the amount of chromatin-bound PCNA inS-phase cells was revealed by immunostaining followedby both fluorescence microscopy and flow cytometricanalysis. The biparametric analysis of PCNA immuno-fluorescence vs DNA content shown in Figure 5Aindicates that roscovitine induced a drastic reduction inthe levels of chromatin-bound PCNA in S-phase cells.Fluorescence microscopy was also used to verify that lossof PCNA immunofluorescence was indeed related tothe protein bound to DNA replication foci (Fig. 5B).Treatment with the RNA synthesis inhibitor DRB(20 µM, 4 h) did not affect the chromatin-bound levelsof PCNA (not shown). To further analyze the effectsinduced by roscovitine on DNA replication proteins,LF1 fibroblasts were synchronised at the G1/Stransition by aphidicolin and then released in Sphase for 1.5 h before roscovitine treatment, toenrich the S-phase cell population. Detergent-soluble and chromatin-bound fractions wereanalysed by western blot for the distribution ofMCM2, PCNA, RPA2, CDK2, as well as actin(for loading control). Figure 6A shows that therewas no significant difference in the amount ofdetergent-soluble proteins between untreated androscovitine-treated samples. In contrast, the levelsof proteins present in the chromatin-bound fractionwere significantly modified. The amount ofchromatin-bound MCM2 was increased in theroscovitine-treated sample, and a similar behaviourwas found for CDK2. In addition, chromatin-bound RPA2 was present with multiple bandsrelated to the hyperphosphorylated forms.42 Incontrast, the amount of chromatin-bound PCNAwas significantly reduced by roscovitine treatment,in accordance with the results obtained byimmunofluorescence analysis. To further investigatewhether the different behavior of DNA replicationproteins was unique to PCNA, the amount of PCNA-interactingproteins involved in DNA replication were also analyzed. Figure 6Bshows that, similarly to CDK2, the levels of chromatin-bound Pol δ,as well as those of Lig I, were apparently increased after roscovitinetreatment, as opposed to the levels of PCNA. In order to verifywhether the effects induce by roscovitine on PCNA levels wereexclusively dependent on checkpoint activation, or they were alsodependent on CDK2 inhibition, LF1 fibroblatsts were incubated for4 h with roscovitine in the presence or in the absence of caffeine. In(Fig. 7A), flow cytometric analysis of PCNA immunofluorescenceshows that chromatin-bound levels of PCNA were not significantlymodified in the samples treated only with caffeine, as compared withthe untreated control cells, while roscovitine alone induced a drastic

reduction, as previously observed (see Fig. 5A and B). Interestingly,treatment with caffeine before roscovitine addition allowed to recoveronly modestly the chromatin-bound levels of PCNA despitecheckpoint inhibition, as verified by the disappearance of hyper-phosphorylated forms of RPA2 (not shown). The same behaviourwas seen when BrdU incorporation was determined in aliquots ofthe same samples, thus indicating that checkpoint activation wasonly partially responsible for disassembly of PCNA (Fig. 7B).

To establish whether removal of chromatin-bound PCNA wasdue to damage occurring during DNA replication, the phosphorylationof histone H2AX (γ-H2AX) on Ser139, a typical marker of DNAdamage and replication stress,44 was investigated in cells that were inS-phase at the time of roscovitine treatment. LF1 fibroblasts were

Figure 2. Specificity of DNA synthesis inhibition and p53 accumulation induced byroscovitine. (A) BrdU incorporation measured in control samples treated with solvent(Ctrl), or in samples treated for 4 h with 20 µM roscovitine (Ros), with 20 µM PD98059(PD98059), or irradiated with UV-C light at 10 J/m2 (UV). BrdU was added during thelast 30 min of incubation, or at the end of a 4 h-period post-irradiation. Dot-plots of BrdUimmunofluorescence vs DNA content are shown. (B) Western blot analysis of total proteinlevels of p53, p53 phospho-serine 15 (p53-pS15), p21 and actin, in LF1 human fibrob-last samples treated as indicated in panel A.

Figure 3. Cell cycle distribution of p53 protein accumulation induced by roscovitine. (A) Flowcytometric determination of p53 immunofluorescence vs DNA content in LF1 human fibroblaststreated for 4 h with 20 µM roscovitine (Ros), or with 2 mM HU. The levels of p53 protein weredetermined after immunofluorescence staining with anti-p53 DO-7 monoclonal antibody, andDNA content by PI staining. The level of non specific fluorescence (nsp) was determined byincubation with irrelevant primary antibody. (B) Western blot analysis of p53 protein, p53phospho-serine 15 (p53-pS15), p21 and actin, as determined in LF1 fibroblast samples treat-ed for 4 h with 20 µM roscovitine (Ros), or with 2 mM hydroxyurea (HU), both in the absenceor in the presence of 10 mM caffeine (caff).

2156 Cell Cycle 2006; Vol. 5 Issue 18

www.landesbioscience.com Cell Cycle 2157

Uncoupling of DNA Replication Proteins by Roscovitine

pulse-labeled with BrdU for 30 min, before incubation with thedrug, for the concomitant detection of BrdU incorporation andhistone γ-H2AX. Figure 8A shows that few spots of fluorescencerelated to signals of γ-H2AX, were detectable only in those cells thathad previously incorporated BrdU, thus suggesting the presence ofDNA replication stress in S-phase cells. To further confirm theseresults, the analysis was also performed by western blot on chromatin-bound extracts obtained from S-phase enriched LF1 fibroblasts. Theresults showed the appearance of detectable levels of histoneγ-H2AX in the roscovitine-treated sample, as compared with theamount of histone H3 as loading control (Fig. 8B).

DISCUSSIONIn this study we have investigated the effects induced by roscovitine

in S-phase cells after a short-term treatment, to exclude phenomenarelated to cell cycle arrest observed in long-term experiments.13,16,22,26

The effects of roscovitine on DNA synthesis and on checkpointactivation, appear to be related to inhibition of S-phase CDK2 activity.Inhibition of other protein kinases, like the MAP kinases Erk1/2,12

could be ruled out because arrest of DNA synthesis and checkpointactivation (p53 ser15 phosphorylation) were not observed in cellstreated with PD98059, a specific inhibitor of Erk1/2 kinases.12

Induction of p53 by roscovitine resulted also in an increase in p21protein levels, in agreement with previous findings,22,45 suggestingthat this effect was attributable to p53 transactivating activity.46,47

However, it is also possible that p21 levels increased because of areduced protein turn-over consequent to CDK2 inhibition,48 similarlyto that observed for CDC25A.49

It has been previously reported that CDK2 inhibition, or expressionof a dominant negative CDK2, triggers an intra-S phase checkpointdependent on both ATM and ATR pathways.22,26 Although bothATM and ATR are required to respond to double-strand breaks(DSBs),50,51 we have found that ATR is likely more involved in theresponse triggered by roscovitine, probably because no direct DSBsare produced in these conditions.28 Whatever the mechanism, theintra-S phase checkpoint could not be the only responsible for p53accumulation, given that p53 protein increased in all the phases ofthe cell cycle. Thus, the mechanisms by which roscovitine inducesp53 accumulation could be cell cycle-dependent, i.e., caused bytranscription inhibition in G1 and G2 phases, while triggered byinhibition of DNA synthesis in S-phase.46,47

The effects induced by roscovitine on the molecular machinery inthe prereplication and the replication complex of DNA replication,have confirmed that CDK2 inhibition induced an increase inchromatin-bound MCM2 protein.22,27 In addition, an increase inCDK2 protein and, consistent with checkpoint activation, theappearance of hyperphosphorylated forms of RPA2,42 wereobserved. In contrast, the levels of chromatin-bound PCNA weresignificantly reduced by roscovitine, in concomitance with DNAsynthesis inhibition. This effect occurred independently of p53accumulation, because it was also observed in HeLa cells lacking p53

Figure 6. Changes in subcellular distribution of prereplicative and replicativeproteins induced by roscovitine in LF1 human fibroblasts. (A) Samples werefractionated after treatment for 4 h with 20 µM roscovitine (Ros) as detergent-soluble (soluble) and chromatin-bound (chromatin) fractions, and analysedby western blot for DNA replication proteins. (B) Chromatin-bound levels(Chromatin) of PCNA-interacting DNA replication proteins in untreatedcontrols and in roscovitine-treated (Ros) LF1 fibroblasts.

Figure 4. Checkpoint activation induced by roscovitine depends on ATR. (A)ATM deficient AT4BI fibroblasts were incubated for 4 h with 20 µMroscovitine (Ros) and whole cell exctracts were analysed by western blot forp53 and p53 phosphorylation at Ser15 (p53-pS15). (B) BJ hTert-immortal-ized fibroblasts were transfected with siRNA pool to ATR (ATR RNAi), or toGFP (Ctrl RNAi), or not transfected (No RNAi). Forty-eight hours later cellswere treated for 4 h with 20 µM roscovitine (Ros), then lysed in SDS loadingbuffer for Western blot analysis of p53pS15.

Figure 5. Roscovitine induces disassembly of chromatin-bound PCNA at DNA replication sites. LF1 fibroblastswere incubated for 4 h with 20 µM roscovitine beforehypotonic lysis for immunofluorescence determinationof PCNA. (A) Dual-parameter flow cytometric analysisof chromatin-bound PCNA immunofluorescence vsDNA content in untreated controls (Ctrl), and in roscov-itine-treated samples (Ros). (B) In situ immunofluorescenceanalysis of chromatin-bound PCNA and DNA stainingwith Hoechst 33258 dye in LF1 fibroblasts untreated(Ctrl), or treated with roscovitine (Ros), as in (A).

Uncoupling of DNA Replication Proteins by Roscovitine

(not shown), and it was related to inhibition ofCDK2, rather than other CDKs (e.g., CDK7 orCDK9) affecting transcription, since the RNAsynthesis inhibitor DRB did not affect PCNAlevels. In striking contrast, the chromatin-boundlevels of PCNA- interacting proteins, like Pol δand Lig I, were stabilized in roscovitine-treatedsamples. The reason for this discrepancy is mostprobably due to the fact that multiple pools ofPCNA exist at DNA replication sites.36 In thesepools, PCNA is present alone, or in complex withCDK2, while in other pools, Pol δ and Lig I seemto interact with PCNA in a mutually exclusivemanner.52,53 Thus, it is possible that overall levelsof chromatin- bound PCNA were reducedbecause roscovitine inhibited new origin firingand induced disassembly of PCNA present aloneat damaged sites (e.g., histone γ-H2AX foci).However, turn-over of existing PCNA/Pol δ, orPCNA/Lig I complexes, possibly regulated byCDK2,36 was blocked by roscovitine giving riseto an apparent increase (stalled forks) in the chro-matin-bound levels of these proteins. In fact,PCNA-interacting proteins, like Pol δ and Lig I, are phosphorylatedby CDK2,37 and this modification reduces their binding affinity forPCNA.37 CDK2 inhibition by roscovitine will then result in aprolonged binding of these proteins to PCNA. This effect is consistentwith the idea that, in consequence of S-phase checkpoint activation,DNA replication forks are stabilized to avoid collapse and disassem-bly.29,50 Abrogation of the checkpoint response with caffeine couldonly partially rescue DNA synthesis inhibition, as well as thedecrease in chromatin-bound PCNA. This result may be explainedby the evidence that roscovitine halts new origin firing by CDK2inhibition, thereby blocking DNA replication at the initiationstep.28,54 Thus, checkpoint inhibition by caffeine may allow only therestart of elongating forks, but not that of new initiation events.54,55

In addition, the presence of histone γ-H2AX in S-phase cells treatedwith roscovitine suggests that DNA damage induced some PCNAdisassembly that could not be rescued by caffeine.

In conclusion, the results presented here indicate that roscovitinehas multiple effects, depending on the cell cycle position, and inhi-bition of the process utilizing the relevant CDK. This multiplicity ofeffects may be, however, exploited for potentiating the effect of existingantitumor agents, or to create new combination of drugs withincreased antiproliferative activity.2,47

References1. Senderowicz AM. Small molecule cyclin-dependent kinase modulators. Oncogene 2003;

22:6609-20.

2. Welburn JPI, Endicott JA. Inhibition of the cell cycle with chemical inhibitors: A targetedapproach. Semin Cell Dev Biol 2005; 16:369-81.

3. Hu B, Mitra J, van den Heuvel S, Enders GH. S and G2 phase roles for cdk2 revealed byinducible expression of a dominant negative mutant in human cells. Mol Cell Biol 2001;21:2755-66.

4. Sanchez I, Dynlacht BD. New insights into cyclins, CDKs, and cell cycle control. SeminCell Dev Biol 2005; 16:311-21.

5. Knockaert M, Greengard P, Meijer L. Pharmacological inhibitors of cyclin-dependentkinases. Trends Pharmacol Sci 2002; 23:417-25.

6. Meijer L, Raymond E. Roscovitine and other purines as kinase inhibitors. From starfishoocytes to clinical trials. Acc Chem Res 2003; 36:417-25.

7. Fischer PM, Gianella-Borradori A. Recent progress in the discovery and development ofcyclin-dependent kinase inhibitors. Expert Opin Investig Drugs 2005; 14:457-77.

8. Berthet C, Aleem E, Coppola V, Tessarollo L, Kaldis P. Cdk2 knockout mice are viable.Curr Biol 2003; 13:1775-85.

9. Grim JE, Clurman BE. Cycling without CDK2? Trends Cell Biol. 2003; 13:396-99.10. Fischer PM. The use of CDK inhibitors in oncology. Cell Cycle 2004; 3:742-46.11. Bain J, McLauchlan H, Elliott M, Cohen P. The specificities of protein kinase inhibitors:

An update. Biochem J 2003; 371:199-204.12. Bach S, Knockaert M, Reinhardt J, Lozach O, Schmitt S, Baratte B, Koken M, Coburn SP,

Tang L, Jiang T, Liang D, Galons H, Dierick JF, Pinna LA, Meggio F, Totzke F, SchächteleC, Lerman AS, Carnero A, Wan Y, Gray N, Meijer L. Roscovitine targets, protein kinasesand pyridoxal kinase. J Biol Chem 2005; 280:31208-19.

13. Meijer L, Borgne A, Mulner O, Chong JP, Blow JJ, Inagaki N, Inagaki M, Delcros JG,Moulinoux JP. Biochemical and cellular effects of roscovitine, a potent and selectiveinhibitor of cyclin dependent kinases cdc2, cdk2 and cdk5. Eur J Biochem 1997;243:527-36.

Figure 7. Checkpoint abrogation with caffeine does not completely rescue PCNA disassemblyand DNA synthesis inhibition induced by roscovitine. LF1 human fibroblasts were incubated for30 min in the presence or in the absence of 10 mM caffeine before addition of 20 µM roscov-itine (Ros) and further incubation for 4 h. During the last 30 min of incubation each samplereceived 30 µM BrdU. Cells were then divided in two aliquots, lysed in hypotonic buffer, fixedand immunostained for PCNA (A), or BrdU immunofluorescence (B) and DNA content determi-nation with flow cytometry.

Figure 8. Roscovitine induces DNA damage in S-phase cells. LF1 fibroblastswere treated for 4 h with 20 µM roscovitine. (A) Fluorescence micrographsof LF1 fibroblasts immunostained with anti-phosphoserine 139 of histoneH2AX (γ-H2AX), and by anti-BrdU for detection of S phase cells. DNAstaining with Hoechst 33258 is shown in blue. (B) After treatment withroscovitine (Ros), cells were fractionated to examine the detergent-insolublefraction. Samples were analysed by western blot for the presence ofphospho-serine 139 of histone H2AX (γ-H2AX), and histone H3 for loadingcontrol.

2158 Cell Cycle 2006; Vol. 5 Issue 18

www.landesbioscience.com Cell Cycle 2159

Uncoupling of DNA Replication Proteins by Roscovitine

14. McClue SJ, Blake D, Clarke R, Cowan A, Cummings L, Fischer PM, MacKenzie M,Melville J, Stewart K, Wang S, Zhelev N, Zheleva D, Lane DP. In vitro and in vivo anti-tumor properties of the cyclin dependent kinase inhibitor Cyc202 (R-roscovitine). Int JCancer 2002; 102:463-68.

15. Raynaud FI, Whittaker SR, Fischer PM, McClue S, Walton MI, Barrie SE, Garrett MD,Rogers P, Clarke SJ, Kelland LR, Valenti M, Brunton L, Eccles S, Lane DP, Workman P. Invitro and in vivo pharmacokinetic-pharmacodynamic relationships for the trisubstitutedaminopurine cyclin-dependent kinase inhibitors olomoucine, bohemine and CYC202.Clin Cancer Res 2005; 11:4875-87.

16. Alessi F, Quarta S, Savio M, Riva F, Rossi L, Stivala LA, Scovassi AI, Meijer L, Prosperi E.The cyclin-dependent kinase inhibitors olomoucine and roscovitine arrest human fibroblastsin G1 phase by specific inhibition of CDK2 kinase activity. Exp Cell Res 1998; 245:8-18.

17. Ljungman M, Paulsen MT. The cyclin-dependent kinase inhibitor roscovitine inhibitsRNA synthesis and triggers nuclear accumulation of p53 that is unmodified at Ser15 andLys382. Mol Pharmacol 2001; 60:785-89.

18. Blagosklonny MV. Flavopiridol, an inhibitor of transcription. Implications, problems andsolutions. Cell Cycle 2004; 3:1537-42.

19. David-Pfeuty T. Potent inhibitors of cyclin-dependent kinase 2 induce nuclear accumula-tion of wild-type p53 and nucleolar fragmentation in human untransformed andtumor-derived cells. Oncogene 1999; 18:7409-22.

20. David-Pfeuty T, Nouvian-Dooghe Y, Sirri V, Roussel P, Hernandez-Verdun D. Commonand reversible regulation of wild-type p53 function and of ribosomal biogenesis by proteinkinases in human cells. Oncogene 2001; 20:5951-63.

21. Kotala V, Uldrijan S, Horky M, Trbusek M, Strnad M, Vojtesek B. Potent induction ofwild-type p53-dependent transcription in tumor cells by a synthetic inhibitor ofcyclin-dependent kinases. Cell Mol Life Sci 2001; 58:1333-39.

22. Zhu Y, Alvarez C, Doll R, Kurata H, Dchebye XM, Parry D, Lees E. Intra-S phase check-point activation by direct CDK2 inhibition. Mol Cell Biol 2004; 24:6268-77.

23. Lu W, Chen L, Peng Y, Chen J. Activation of p53 by roscovitine-mediated suppression ofMDM2 expression. Oncogene 2001; 20:3206-16.

24. Demidenko ZN, Blagosklonny MV. Flavopiridol induces p53 via initial inhibition ofMdm2 and p21 and, independently of p53, sensitizes apoptosis-reluctant cells to tumornecrosis factor. Cancer Res. 2004; 64:3653-60.

25. Rubbi PC, Milner J. Disruption of the nucleolus mediates stabilization of p53 in responseto DNA damage and other stresses. EMBO J 2003; 22:6068-77.

26. Maude SL, Enders GH. Cdk inhibition in human cells compromise Chk1 function andactivates a DNA damage response. Cancer Res 2005; 65:780-86.

27. Zhu Y, Ishimi Y, Tanudji M, Lees E. Human CDK2 inhibition modifies the dynamics ofchromatin-bound minichromosome maintenance complex and replication protein A. CellCycle 2005; 4:1254-63.

28. Luciani MG, Oehlmann M, Blow JJ. Characterization of a novel ATR-dependent,Chk1-independent, intra-S-phase checkpoint that suppress initiation of replication inXenopus. J Cell Sci 2004; 117:6019-30.

29. Branzei D, Foiani M. The DNA damage response during DNA replication. Curr Opin CellBiol 2005; 17:568-75.

30. Bartek J, Lukas C, Lukas J. Checking on DNA damage in S phase. Nature Rev Mol CellBiol 2004; 5:792-804.

31. Gottifredi V, Prives C. The S phase checkpoint: When the crowd meets at the fork. Semin.Cell Dev Biol 2005; 16:355-368.

32. Ashcroft M, Taya Y, Wousden K. Stress signals utilize multiple pathways to stabilize p53.Mol Cell Biol 2000; 20:3224-33.

33. Gottifredi V, Shieh SY, Taya Y, Prives C. p53 accumulates but is functionally impairedwhen DNA synthesis is blocked. Proc Natl Acad Sci USA 2001; 98:1036-41.

34. Blagosklonny MV, Demidenko ZN, Fojo T. Inhibition of transcription results in accumu-lation of wt p53 followed by delayed outburst of p53-inducible proteins: p53 as a sensor oftranscriptional integrity. Cell Cycle 2002; 1:67-74.

35. Ho CC, Siu WY, Lau A, Chan WM, Arooz T, Poon RYC. Stalled replication induces p53accumulation through distinct mechanisms from DNA damage checkpoint pathways.Cancer Res 2006; 66:2233-41.

36. Prosperi E. The fellowship of the rings: Distinct pools of proliferating cell nuclear antigenat work. FASEB J 2006; 20:833-37.

37. Hennecke G, Koundrioukoff S, Hübscher U. Multiple roles for kinases in DNA replica-tion. EMBO rep 2003; 4:1-5.

38. Arlett CF, Green MHL, Priestley A, Harcourt SA, Mayne LV. Comparative human cellularradiosensitivity. I. The effect of SV40 immortalisation on the gamma-irradiation survivalof skin derived fibroblasts from normal individuals and from ataxia-telangiectasia patientsand heterozygotes. Int J Radiat Biol 1988; 54:911-28.

39. Scovassi AI, Prosperi E. Analysis of proliferating cell nuclear antigen (PCNA) associatedwith DNA excision repair sites in mammalian cells. Methods Mol Biol 2006; 314:457-75.

40. Cazzalini O, Perucca P, Riva F, Stivala LA, Bianchi L, Vannini V, Ducommun B, ProsperiE. p21CDKN1A does not interfere with loading of PCNA at DNA replication sites, butinhibits subsequent binding of DNA polymerase δ at the G1/S phase transition. Cell Cycle2003; 2:596-603.

41. Perucca P, Cazzalini O, Mortusewicz O, Necchi D, Savio M, Nardo T, Stivala LALeonhardt H, Cardoso MC, Prosperi E. Spatiotemporal dynamics of p21CDKN1A proteinrecruitment to DNA-damage sites and interaction with proliferating cell nuclear antigen. JCell Sci 2006; 119:1517-27.

42. Liu JS, Kuo SR, Melendy T. Comparison of checkpoint response triggered by DNA poly-merase inhibition versus DNA damaging agents. Mutat Res 2003; 532:215-26.

43. Sarkaria JN, Busby EC, Tibbetts RS, Roos P, Taya Y, Karnitz LM, Abraham RT. Inhibitionof ATM and ATR kinase activities by the radiosensitizing agent, caffeine. Cancer Res 1999;59:4375-82.

44. Ward IM, Chen J. Histone H2AX is phosphorylated in an ATR-dependent manner inresponse to replicational stress. J Biol Chem 2001; 276:47759-62.

45. Krystof V, McNae IW, Walkinshaw MD, Fischer PM, Müller P, Vojtesek B, Orsag M,Havlicek L, Strnad M. Antiproliferative activity of olomoucine II, a novel 2,6,9-trisubsti-tuted purine cyclin-dependent kinase inhibitor. Cell Mol Life Sci 2005; 62:1763-71.

46. Ljungman M, Lane DP. Transcription - The guardian of the genome by sensing DNA dam-age. Nature Rev Cancer 2004; 4:727-37.

47. Wesierska-Gadek J, Schmid G. Dual action of the inhibitors of cyclin-dependent kinases:Targeting cell-cycle progression and activation of wild-type p53. Expert Opin Invest Drugs2006; 15:23-38.

48. Zhu H, Nie L, Maki CG. Cdk2-dependent inhibition of p21 stability via a C-terminalcyclin-binding motif. J Biol Chem 2005; 280:29282-88.

49. Ducruet AP, Lazo JS. Regulation of Cdc25A half-life in interphase by cyclin-dependentkinase 2 activity. J Biol Chem 2003; 278:31838-42.

50. Trenz K, Smith E, Smith S, Costanzo V. ATM and ATR promote Mre11 dependent restartof collapsed replication forks and prevent accumulation of DNA breaks. EMBO J 2006;25:1764-74.

51. Jazayeri A, Falck J, Lukas C, Bartek J, Smith GCM, Lukas J, Jackson SP. ATM- and cellcycle-dependent regulation of ATR in response to double-strand breaks. Nature Cell Biol2006; 8:37-45.

52. Riva F, Savio M, Cazzalini O, Stivala LA, Scovassi IA, Cox LS, Ducommun B, Prosperi E.Distinct pools of proliferating cell nuclear antigen associated to DNA replication sitesinteract with the p125 subunit of DNA polymerase δ or DNA ligase I. Exp Cell Res 2004;293:357-67.

53. Sporbert A, Domaing P, Leonhardt H, Cardoso MC. PCNA acts as a stationary loadingplatform for transiently interacting Okazaki fragment proteins. Nucleic Acids Res 2005;33:3521-28.

54. Dimitrova DS, Gilbert DM. Temporally coordinated assembly and disassembly of replica-tion factories in the absence of DNA synthesis. Nature Cell Biol 2000; 2:686-94.

55. Heffernan TP, Simpson DA, Frank AR, Heinloth AN, Paules RS, Cordeiro-Stone M,Kaufmann WK. An ATR and chk1-dependent S checkpoint inhibits replicon initiation fol-lowing UVC-induced DNA damage. Mol Cell Biol 2002; 22:8552-61.