Mr

JD

a

A

R

R

2

A

P

K

S

C

D

A

p

D

1

Ctagcpctp

1d

d n a r e p a i r 8 ( 2 0 0 9 ) 29–39

avai lab le at www.sc iencedi rec t .com

journa l homepage: www.e lsev ier .com/ locate /dnarepai r

ammalian SWI/SNF chromatin remodeling complexes areequired to prevent apoptosis after DNA damage

i-Hye Park, Eun-Jung Park, Shin-Kyoung Hur, Sungsu Kim, Jongbum Kwon ∗

epartment of Life Science and Division of Life and Pharmaceutical Sciences, Ewha Womans University, Seoul, 120-750, Republic of Korea

r t i c l e i n f o

rticle history:

eceived 26 February 2008

eceived in revised form

8 August 2008

ccepted 28 August 2008

ublished on line 7 October 2008

eywords:

WI/SNF

hromatin remodeling

NA damage response

poptosis

53

NA damage checkpoint

a b s t r a c t

Although SWI/SNF chromatin remodeling complexes play important roles in transcription,

recent studies suggest that they also participate directly in DNA repair. In yeast, SWI/SNF

and related RSC complexes have been shown to be recruited to the sites of DNA double

strand breaks (DSBs) to facilitate DNA repair. We recently have shown that mammalian

SWI/SNF complexes contribute to DBS repair by direct mechanisms of stimulating the phos-

phorylation of histone H2AX at DSB-surrounding chromatin. Here we investigated the role

of mammalian SWI/SNF complexes in cell survival after DNA damage. When SWI/SNF was

inactivated by means of dominant negativity or its catalytic subunit BRG1 was knockdowned

by small interfering RNA, cells became highly susceptible to DNA damage-induced apopto-

sis. SWI/SNF inactivation had no effect on the activation and establishment of G2/M DNA

damage checkpoint. However, SWI/SNF-defective cells could not sustain the G2/M check-

point long enough to survive DNA damage, and rather underwent apoptosis before entering

mitosis. We also found that, although the basal state and DNA damage-triggered activa-

tion of p53 were normal, the kinetics of p53 downregulation was significantly delayed in

SWI/SNF-defective cells. Finally, the sustained p53 activation in SWI/SNF-defective cells

was accompanied by accumulation of unrepaired DSBs owing to inefficient DNA repair.

These results suggest that mammalian SWI/SNF complexes prevent DNA damage-induced

apoptosis in part by facilitating efficient repair and thereby ensuring timely elimination of

unrepaired DSBs that could otherwise lead to excessive prolongation of p53 activation.

Once DSBs are eliminated by repair, the DNA damage check-

. Introduction

ells must overcome the constant threatening of damage toheir DNA for healthy survival. Double strand breaks (DSBs)re the most deleterious form of DNA damage, which areenerated by exposure to ionizing radiation (IR) or genotoxichemicals, and can also occur during the normal cellularrocesses such as DNA replication. If left unrepaired, DSBs

an result in cell death as well as loss of genetic informa-ion, chromosomal translocations and genome instability, allotentially leading to cancer development. Cells have evolved

∗ Corresponding author. Tel.: +82 2 3277 4334; fax: +82 2 3277 3760.E-mail address: jongkwon@ewha.ac.kr (J. Kwon).

568-7864/$ – see front matter © 2008 Elsevier B.V. All rights reserved.oi:10.1016/j.dnarep.2008.08.011

© 2008 Elsevier B.V. All rights reserved.

an elaborate mechanism to cope with this life-threateningDNA damage [1].

Upon DSB generation, DNA damage sensors detect the DNAlesions and provoke the activation of the coordinated cel-lular signaling pathways of DNA damage responses, whicheventually culminates in the execution of DNA repair aswell as the arrest of cell cycle via DNA damage checkpoint.

point becomes inactivated to allow for resuming the cellcycle progression. Unless DSBs are efficiently repaired, thearrested cells cannot resume the cell cycle, but instead tend

8 ( 2

30 d n a r e p a i r

to undergo programmed cell death termed as DNA damage-induced apoptosis [2,3].

The tumor suppressor p53 plays a central role in DNAdamage-induced apoptosis as well as in checkpoint-mediatedcell cycle arrest. In response to DSB generation, p53 is phos-phorylated by PI3 kinase-like kinases, ATM /ATR, which leadsto stabilization and cellular accumulation of itself as well asstimulation of its transcriptional activity. The activated p53elevates the expression of a number of its target genes respon-sible for the activation of the cellular pathways leading toapoptosis. Once DNA repair is completed, the p53 activationshould be turned off to allow for resuming the cell cycle as wellas to extinguish the apoptotic signals. Therefore, not only theactivation of p53 following DNA damage but also its downreg-ulation after completion of DNA repair is important for cellsto survive DNA damage [4–6].

Detection and repair as well as generation of DSBs ineukaryotes occur in the context of chromatin in which DNAis tightly bound to histones, which creates a natural barrierto protein access. It has been increasingly clear that histonemodifications and ATP-dependent chromatin remodeling, thetwo major mechanisms for modulating chromatin structure,both play important roles in DSB repair [7–9]. For example, thephosphorylation of H2AX (�-H2AX), one of the best character-ized histone modifications that occur on the chromatin at thesites of DSBs forming visible nuclear foci, is important for effi-cient DSB repair and DNA damage checkpoint as well as forsuppressing cancer development [10].

Recent studies suggest that the Snf2-based ATP-dependentchromatin remodeling complexes play a direct role in DSBrepair [8,11]. Most of these studies were focused on usingyeast system. The yeast INO80 complex was first found tobe recruited to a DSB via interaction with �-H2AX, where itis thought to induce histone eviction so as to facilitate theaccess of repair proteins to the site of DSB [12–15]. In a sim-ilar manner to the INO80 complex, the yeast SWR1 complex,another member of the INO80 subfamily, has been shown tomake a direct contribution to DSB repair via association with�-H2AX [12,14,16]. Subsequently, the yeast SWI/SNF complex,the founding member of the Snf2 chromatin remodeling com-plex family, has also been shown to be recruited to a DSBalthough the mechanism underlying its recruitment to theDSB site remains unknown [17]. Further, the SWI/SNF-relatedyeast RSC complex has been shown to be recruited to a DSB bythe mechanisms dependent on the DNA repair proteins suchas Mre11 and yKu, and suggested to remodel nucleosomessuch a manner to facilitate the access of the repair machineryto the site of DSB [18,19].

Recently, we have reported that mammalian SWI/SNF com-plexes also participate in DSB repair directly, emphasizingthe evolutionary importance of this subfamily of chromatinremodelers in DNA repair [20]. This study showed that theSWI/SNF complexes bind to the chromatin at the sites ofDSBs via interaction with �-H2AX, which further increasesthe phosphorylation of H2AX and its nuclear focus formation.In addition, the study also showed that SWI/SNF inactiva-

tion results in defect of DSB repair as well as a decrease ofclonogenic ability to divide and give rise to formation of vis-ible colonies. However, the role of SWI/SNF in cell survivalin response to DNA damage has not been investigated. For

0 0 9 ) 29–39

example, it was not known whether the compromised clono-genic ability by SWI/SNF inactivation is due to apoptosis ormerely a result of inefficient damage repair that prevents fur-ther cell division. In the present study, we show that SWI/SNFcontributes to cell survival after DNA damage by protectingagainst apoptosis.

2. Materials and methods

2.1. Cells, DNA damage and antibodies

tet-VP16 and B05-1 cells have been previously described [21].Cells were maintained in the medium containing 2 �g/ml oftetracycline, 75 �g/ml of G418 (for tet-VP16 and B05-1 cells)and 350 �g/ml of hygromycin B (for B05-1 cells). To obtainthe maximum induction of the Flag-tagged protein expres-sion, cells were cultured for 4 days under the conditions withno tetracycline. For DSB generation, doxorubicin or etoposide(Sigma), dissolved in DMSO, were added to cell culture at indi-cated concentrations for various time, and DMSO as a vehiclewas added to culture for controls. IR exposure was performedby using �-irradiator (137Cs; Cell Irradiation System, GC 3000Elan-Model �; MDS Nordion, Ontario, Canada). The sources ofantibodies are as follows: the antibodies against �-H2AX (cloneJBW301), H2A (acidic patch), and phospho-H3(Ser-10) fromUpstate; the antibodies against actin, p53, phospho-p53(Ser-18), BRG1, Cdc2, phospho-Cdc2(Tyr-15), and p21 from SantaCruz; the antibody against Flag from Sigma.

2.2. Apoptosis assays

2.2.1. Tryphan blue exclusionCells were harvested, washed twice with PBS, and resus-pended in 0.4% (w/v) of tryphan blue (Sigma). Live cells weredetermined based on their tryphan blue exclusion and threeindependent counts were taken for evaluating the viability.

2.2.2. Annexin-V stainingAnnexin-V/PI double staining was performed according to themanufacture’s protocols (BD Bioscience Parmingen). 1 × 106

cells were washed twice with cold PBS and resuspended in1 ml of binding buffer. 100 �l of the suspension (1 × 105 cells)were mixed with 5 �l of Annexin-V-FITC and 5 �l of propodiumiodide (PI, Calbiochem). After gentle vortexing, cells were incu-bated for 15 min at room temperature (RT) in the dark beforebeing subjected to flow cytometric analysis (FACScan, BectonDickinson BD, Mountain View, CA, USA).

2.2.3. TUNEL assaysTUNEL assays were performed according to the manufacture’sprocedures (APO-BrdU TUNEL assay kit, Molecular Probe).Cells were harvested and suspended in 5 ml of 1% (w/v)paraformaldehyde in PBS. After incubation on ice for 15 min,cells were collected by centrifugation at 300 × g for 5 min at4 ◦C, washed with PBS two times. Cells were then resuspended

in 5 ml of ice-cold 70% (v/v) ethanol followed by incubationat −20 ◦C for 12–18 h. Fixed cells were pelleted, washed withthe wash buffer and added by 50 �l of DNA-labeling solution(10 �l reaction buffer, 0.75 �l TdT enzyme, 8.0 �l BrdUTP and

( 2 0

3rsasofa

2

potfat3ti

2

CsaroD(tp

2

AppNff1awbltmcou(0a1

2

Cfi

d n a r e p a i r 8

1.25 �l dH2O). Cells were then incubated for 4 h at 37 ◦C, andinsed with 1 ml of the rinse buffer two times. 100 �l of thetaining solution (5 �l of Alexa Fluor 488 dye-labeled anti-BrdUntibody plus 95 �l of the rinse buffer) were added to eachample and incubated for 30 min at RT. To this reaction, 0.5 mlf the PI/RNase A staining buffer were added and incubatedor additional 30 min at RT before being subjected to FACSnalysis.

.3. pSuper-BRG1 and transfection

Super-BRG1 construct was generated by inserting annealedligonucleotides into BglII-HindIII cloning sites of pSuper vec-or (Oligoengine). The sequences of the oligonucleotides areollows: sense strand, 5′-gatccccgttggagctgttggcgtagttcaagag-ctacgccaacagctccaactttttggaaa-3′; antisense strand, 5′-agcttt-ccaaaaagttggagctgttggcgtagtctcttgaactacgccaacagctccaacggg-′. Human embryonic kidney 293T cells were transfected withotal 5 �g of pSuper or pSuper-BRG1 plasmids per 1 × 106 cellsn a 60-mm dish by calcium phosphate method.

.4. Analysis of checkpoint responses

ell cycle profile was analyzed by PI staining and FACS analy-is. Cells were washed with PBS, fixed in 70% ice-cold ethanol,nd stored at −20 ◦C at least for 30 min. Cells were thenesuspended in PBS containing 50 �g/ml of PI and 50 �g/mlf Ribonuclease A (Sigma), and subjected to FACS analysis.ata were transformed to a plot by using CellQuest software

Becton Dickinson). G2-M checkpoint was analyzed by iden-ifying cells in mitosis by costaining with PI/phospho-H3 asreviously described [20].

.5. Histone extraction and immunoblot analysis

cid extraction of histones for immunoblot analysis waserformed as follows. Cells were washed with PBS and resus-ended in the buffer containing 20 mM Tris–Cl (pH 8.0), 150 mMaCl, 1 mM EDTA (pH 8.0) and 0.5% NP-40, and incubated on ice

or 10 min. Nuclei were collected by centrifugation at 6000 × gor 5 min at 4 ◦C, resuspended in 0.1 M HCl and incubated for0 min at RT. Histone extracts were obtained by centrifugationt 6000 × g for 5 min at 4 ◦C, and concentrations of histonesere measured by Bradford method. Following separation

y 15% SDS-PAGE, proteins were transferred onto nitrocellu-ose membrane (Hybond: Amersham Biosciences, Inc.) usinghe transfer buffer containing 50 mM CAPS (pH 10) and 20%

ethanol. Signals were detected by enhanced chemilumines-ence (ECL; Amersham Life Science). Immunoblot analysis forther cellular proteins were performed by standard methodsing cell extracts prepared in lysis buffer (50 mM Tris–HCl

pH 7.5), 5 mM EDTA, 150 mM NaCl, 1% NP-40, 1 mM DTT,.5 mM PMSF, 50 �g/ml pepstatin, 5 �g/ml leupeptin, 5 �g/mlprotinin, 100-fold dilution of phosphatase inhibitor cocktail(Sigma)).

.6. Immunohistochemistry

ells were seeded onto glass coverslip or 30 mm culture dish,xed with 4% paraformaldehyde for 15 min, and washed for

0 9 ) 29–39 31

10 min in PBS three times. After quenching in 50 mM NH4Clfor 10 min, cells were made permeable by incubating in 0.2%Triton X-100 for 5 min, and blocked with 1% BSA for 1 h atRT. Cells were washed for 10 min in PBS four times, and incu-bated with the �-H2AX antibodies overnight at 4 ◦C followedby incubation with Alexa Fluoro 568-conjugated anti-mouse(Molecular Probes) or FITC-conjugated anti-rabbit (Zymed)antibodies for 30 min at 30 ◦C. Cells were washed for 10 minwith PBS four times and mounted by using Vectashield mount-ing medium with 4,6-diamidino-2-phenylindole (DAPI, VectorLaboratories). Fluorescence images were photographed byusing Axiovert 100 M.

2.7. Single cell gel electrophoresis (comet assay)

Comet assays were performed under neutral conditions usingthe Trevigen’s CometAssay kit (4250-050-K) as previouslydescribed [20].

3. Results

3.1. Inactivation of the SWI/SNF complexes results ina large increase of cell death after DNA damage

Since, in our previous study, the effect of SWI/SNF inactiva-tion on cell survival was examined by evaluating the abilityto form visible colonies about two weeks after DNA dam-age [20], it was not known whether the reduced viability ofSWI/SNF-defective cells is owing to activation of apoptosis orsimply a result of inefficient damage repair that prevents fur-ther cell division. To address this issue, we employed the sameNIH-3T3 cells that we used in the previous study, in whichthe flag-tagged ATPase-defective dominant negative form ofBRG1, the catalytic subunit of the SWI/SNF complexes, canbe inducibly expressed by deprivation of tetracycline [20]. Theexpression of flag-tagged proteins was confirmed to be nicelyinduced when B05-1 cells were cultured in the medium with-out tetracycline (−tet), whereas little flag-tagged proteins weredetected from the cells cultured in the medium with tetra-cycline (+tet) (Fig. 1A). The control cells, tet-VP16, containingempty vector expressed no flag-tagged proteins regardless oftetracycline depletion as expected (Fig. 1A).

First, we examined whether SWI/SNF inactivation has aneffect on cell survival in relatively short period of time afterDNA damage. We treated cells with doxorubicin at the concen-trations of 0.5 �M, one of the most widely used DSB-generatingchemicals, and counted viable cells after various time up to36 h using tryphan blue exclusion method. More than 10%of B05-1(−tet) cells were detected dead as early as 12 h afterdrug treatment, with approximately 50% and 80% dead at24- and 36-h time points, respectively (Fig. 1B). In contrast,much less B05-1(+tet) and little tet-VP16 cells were dead atthe corresponding time points (Fig. 1B), indicating that theincreased cell death of B05-1(−tet) after DNA damage is theresult of SWI/SNF inactivation. The effects of doxorubicin on

cell death were dose-dependent (Fig. 1C), further showingthat the increased cell death of B05-1(−tet) was specificallyattributed to the drug treatment. When etoposide, anothertype of DSB-generating chemical, was used as a DNA dam-

32 d n a r e p a i r 8 ( 2 0 0 9 ) 29–39

Fig. 1 – SWI/SNF inactivation results in a large increase of cell death after DNA damage. (A) A representative gel showing theinduction of flag-tagged dominant negative BRG1 by tetracycline depletion. tet-VP16 and B05-1 cells were cultured in themedium with (+) or without (−) tetracycline for 4 days, and cell lysates were analyzed by immunoblot for the expression offlag-tagged proteins. The expression of actin was also analyzed as an internal control. (B) After grown in the medium with(+tet) or without tetracycline (−tet) for 4 days, tet-VP16 and B05-1 cells were treated with vehicle (DMSO, indicated by 0 h), or0.5 �M of doxorubicin for 6, 12, 24, and 36 h. Viable cells were then determined by tryphan blue exclusion method.Percentages of viable cells were depicted as graph with each point presented as mean ± standard deviation (S.D.) fromtriplicates. (C) Cells grown as per in (A) were treated with vehicle (indicated by 0 �M) or doxorubicin at the concentrations of0.1, 0.5 and 1 �M for 24 h. Viable cells were then determined by tryphan blue exclusion method. Average numbers of thepercentages of dead cells were depicted as graph; error bar indicates mean ± S.D. from triplicates. (D) Cells grown as per in(A) were treated with vehicle (DMSO, indicated by 0 �M), or etoposide at the concentrations of 1, 5, 10 and 25 �M for 24 h.Viable cells were then determined by tryphan blue exclusion method. Average numbers of the percentages of dead cellswere depicted as graph; error bar indicates mean ± S.D. from triplicates. (E) Cells grown as per in (A) were left untreated orexposed to 2, 4, 8 and 12 Gy of �-ray, and after 4 days, viable cells were determined by tryphan blue exclusion method.Average numbers of the percentages of dead cells were depicted as graph; error bar indicates mean ± S.D. from

triplicates.

aging agent, the similar results were obtained (Fig. 1D). Tofurther confirm these results, we performed similar exper-iments by exposing cells to 2–12 Gy of ionizing radiation(IR). B05-1(−tet) cells were highly susceptible to IR-inducedcell death compared to control cells, although the kineticswas greatly delayed in such a way that the levels of celldeath comparable to those obtained by drug treatment could

be achieved after 4 days post-irradiation (Fig. 1E). All theseresults demonstrate that SWI/SNF is important for main-taining cells alive in response to the generation of DSBs ingeneral.

3.2. The increase of DNA damage-induced cell deathby SWI/SNF inactivation is due to apoptosis

To examine whether the increased cell death of SWI/SNF-defective cells following DNA damage is due to apoptosis,we performed Annexin-V staining experiments to determinethe loss of membrane polarity, the hallmark of apoptosis.

When we treated cells with doxorubicin, we found that thepercentage of apoptotic B05-1(−tet) cells was approximatelythree times more than that of control cells (Fig. 2A). At 24 hafter drug treatment, total 27.8% of B05-1(−tet) cells were

d n a r e p a i r 8 ( 2 0 0 9 ) 29–39 33

Fig. 2 – DNA damage-triggered cell death in SWI/SNF-defective cells is due to apoptosis. (A) tet-VP16 and B05-1 cells grownin the medium without tetracycline for 4 days were treated with vehicle (indicated by 0 h), or 0.5 �M of doxorubicin for 12and 24 h. Cells were collected and subjected to double staining by PI and Annexin-V followed by FACS analysis. Averagenumbers of the percentages of apoptotic cells (early and late apoptotic cells; population of cells in lower right plus upperright of FACS data) were depicted as graph; error bar indicates mean ± S.D. from triplicates. (B) A representative FACS resultcorresponding to 24-h time point in (A). (C) B05-1 cells grown in the medium with or without tetracycline for 4 days weretreated with vehicle (indicated by 0 h), or 0.5 �M of doxorubicin for 12 or 24 h. Cells were collected and subjected to TUNELassay using APO-BrdU TUNEL Assay Kit. Average numbers of the percentages of BrdU-positive cells were depicted as graph;error bar indicates mean ± S.D. from triplicates. (D) A representative FACS result corresponding to 24-h time point in (C). (E)tet-VP16 and B05-1 cells grown in the medium without tetracycline for 4 days were treated with vehicle (data not shown) or10 �M of etoposide for 24 h. Cells were collected and subjected to Annexin-V staining assays as per in (A). (F) tet-VP16 andB or es

untaBamiBmaAer

05-1 cells grown as (E) were left untreated (data not shown)ubjected to Annexin-V staining assays as per in (A).

ndergoing or dead of apoptosis whereas 7.2% were dead ofecrosis (Fig. 2B), indicating that the majority of DNA damage-

riggered cell death of SWI/SNF-defective cells is a result ofpoptosis. To further confirm these results, we carried out Tdt-rdU incorporation assays to measure DNA fragmentation,nother hallmark of apoptosis. At 24 h after doxorubicin treat-ent, 17.5% of B05-1(−tet) cells were detected BrdU-positive

n contrast to only 3.9% of B05-1(+tet) cells being detectedrdU-positive, showing that SWI/SNF-defective cells are muchore susceptible to DNA fragmentation following DNA dam-

ge than control cells (Fig. 2C and D). When we performednnexin-V staining experiments with the cells treated withtoposide or the cells exposed to IR, we observed the similaresults that the majority of B05-1(−tet) cells were dead of apop-

xposed to 80 Gy of �-ray. After 24 h, cells were collected and

tosis (Fig. 2E and F). It is noted that a very high dose of IR (80 Gy)was required to achieve the levels of apoptosis comparable tothose achieved by treatment of doxorubicin or etoposide.

To further investigate whether the SWI/SNF complexes arespecifically responsible for preventing apoptosis after DNAdamage, BRG1 was knockdowned by siRNA. The expressionof BRG1 was efficiently downregulated when human embry-onic kidney 293T cells were transfected with pSuper-BRG1(Fig. 3A). When we subjected these cells to DNA damageunder the same conditions as described above, we found that

BRG1-knockdowned cells exhibited 2–3 times more apoptosisthan control cells (Fig. 3B). It is noted that neither SWI/SNFinactivation nor BRG1-knockdown has a significant effecton spontaneous apoptosis (Figs. 1–3). Taken together, these

34 d n a r e p a i r 8 ( 2 0 0 9 ) 29–39

Fig. 3 – siRNA knockdown of BRG1 results in increase of DNA damage-induced apoptosis. (A) Human embryonic kidney293T cells were transfected with pSuper or pSuper-BRG1 vectors for 48 h, and cell lysates were analyzed by immunoblot forthe expression of BRG1 and actin as an internal control. (B) After 48-h transfection with pSuper or pSuper-BRG1, cells wereleft untreated, exposed to 50 Gy of �-ray, treated with 10 �M of etoposide, or 0.5 �M of doxorubicin. After 24 h, cells were

taini

collected and subjected to apoptosis assays by Annexin-V s

data demonstrate that SWI/SNF is critical for protecting cellsagainst apoptosis after DNA damage.

3.3. SWI/SNF inactivation causes apoptosis at G2without affecting G2/M DNA damage checkpointactivation

We wished to understand how SWI/SNF prevents apoptosisafter DNA damage. First, we analyzed the cell cycle profile atvarious time points after doxorubicin treatment. B05-1(−tet),like control cells, continued to increase in their G2/M popula-tion up to 12 h after drug treatment, indicating that the G2/MDNA damage checkpoint is normally activated in these cells(Fig. 4A and B). However, at 24 h after drug treatment, the G2/Mpercentage of B05-1(−tet) cells was markedly decreased withconcomitant increase of apoptotic sub-G1 cells, whereas theG2/M population of control cells was still maintained at signif-icantly high levels at 24 h and even at 36 h time points (Fig. 4Aand B). These results suggested that SWI/SNF-defective cellscannot sustain G2/M DNA damage checkpoint sufficiently longenough to protect against apoptosis.

Since B05-1(−tet) cells after DNA damage exhibit massiveapoptosis at G2/M, we asked if apoptosis occurs after enter-ing mitosis as known for the case of mitotic catastrophe [22].For this, we treated cells with doxorubicin for various timeup to 36 h and harvested them for analyzing mitotic cells bythe phospho-H3 staining method. B05-1(−tet) cells, like con-trol cells, showed very low percentages of mitotic cells at 6 hafter drug treatment (Fig. 4C), again verifying that SWI/SNFis dispensable for the activation of G2/M checkpoint. TheG2 arrest of B05-1(−tet) cells was maintained well afterwarduntil 36 h post-drug treatment although moderately leaky atlater time points (Fig. 4C). B05-1(+tet) cells showed the similarG2/M checkpoint patterns as B05-1(−tet) cells (Fig. 4C). Theseresults strongly suggest that B05-1(−tet) cells after DNA dam-age undergo apoptosis at G2 phase without entering mitosis.

In support of this, we hardly observed any indications of thedeath in mitosis for B05-1(−tet) cells such as micronucleationor chromosomal fragmentation, whether or not subjected toDNA damage (Fig. 4D and Supplementary Fig. S1).

ng.

To investigate the mechanisms accounting for the G2-phase apoptosis of SWI/SNF-defective cells, we analyzed Cdc2,the master kinase that controls the cell cycle progression fromG2 to mitosis. In response to DNA damage, Cdc2 is phospho-rylated at Tyr-15 (Tyr-18 in case of mouse) to be inactivatedand the cell cycle is arrested at G2 [23]. As expected from thefact that the G2/M checkpoint activation is intact in SWI/SNF-defective cells, the phosphorylation of Cdc2 in B05-1(−tet)increased at 6 h after doxorubicin treatment just like in thecontrol cells (Fig. 4E, lanes 1, 2, 6 and 7). Interestingly, thephosphorylation of Cdc2 was largely decreased at 12 h andlater time points similarly in B05-1(+tet) and B05-1(−tet) cells(Fig. 4E, lanes 3–5 and lanes 8–10). These results, togetherwith the p-H3 analysis data, suggest that both control andSWI/SNF-defective cells mostly remain at G2 although Cdc2becomes reactivated after 12 h post-drug treatment. Nonethe-less, the SWI/SNF-defective cells undergo apoptosis muchmore severely than the control cells. Taken together, theresults suggest that, although able to activate the G2/M DNAdamage checkpoint, SWI/SNF-defective cells cannot sustain itas efficiently as wild type cells, and hence undergo apoptosisat G2 before entering mitosis.

3.4. SWI/SNF inactivation causes a prolongedactivation of p53 following DNA damage

Since p53 plays a central role in DNA damage-induced apop-tosis, we asked whether the increased susceptibility of theSWI/SNF-defective cells to DNA damage-induced apoptosisis due to dysregulation of p53 activation. We examined theeffects of SWI/SNF inactivation on p53 activation by determin-ing the protein stability using immunoblottings. The cellularlevels of p53 under normal conditions were maintained verylow equally in B05-1(+tet) and B05-1(−tet) cells (Fig. 5A, lanes 1and 6; Fig. 5B and C, lanes 1 and 8). When the cells were treatedwith doxorubicin, the p53 levels were greatly increased and

the extent of their increase was not different between B05-1(+tet) and B05-1(−tet) cells at least until several hours afterdrug treatment (Fig. 5A, lanes 2–5 and lanes 7–10). Confirmingthese results, the basal expression as well as DNA damage-

d n a r e p a i r 8 ( 2 0 0 9 ) 29–39 35

Fig. 4 – SWI/SNF inactivation causes apoptosis at G2 phase without affecting G2/M DNA damage checkpoint activation. (A)After grown in the medium with (+) or without (−) tetracycline for 4 days, tet-VP16 and B05-1 cells were left untreated(indicated by 0 h) or treated with 0.5 �M of doxorubicin for 6, 12, 24 or 36 h. Cells were then collected for the analysis of cellcycle profile; M1 = sub-G1, M2 = G1, M3 = S, M4 = G2/M. (B) The percentages of the population in sub-G1, G1, S and G2/M of theFACS results in (A) were quantitated and depicted as graph. (C) B05-1 cells grown in the medium with (+) or without (−)tetracycline for 4 days were left untreated (0 h) or treated with 0.5 �M of doxorubicin for 6, 12, 24 and 36 h. Cells were thensubjected to FACS analysis using anti-phospho-H3 antibodies. Percentages of mitotic cells (small square) and apoptotic cellswere shown at the top and bottom of the FACS data, respectively. (D) B05-1 cells were prepared similarly as described in (C)and the nuclei were stained with DAPI (Supplementary Fig. S1). Representative images at time point 24 h were shown. (E)B05-1 cells grown as per in (C) were untreated (0 h) or treated with 0.5 �M of doxorubicin for 6, 12, 18 or 24 h. Cells were thenc flag-a

iw(

eopBHp1(Fpahdtb

ollected and analyzed by immunoblot for the expression ofn internal control.

nduced accumulation of p21, one of the major targets of p53,as detected at the similar levels between these two cells

Fig. 5A).Next, we examined the kinetics of p53 activation for an

xtended period of time after DNA damage. We found that,nce activated by DNA damage, the elevated levels of p53roteins were maintained similarly between B05-1(+tet) and05-1(−tet) cells until around 8 h post-drug treatment (Fig. 5B).owever, the kinetics of p53 downregulation after this timeoint was different, with the rate of p53 decrease in B05-(−tet) cells significantly slower than that in B05-1(+tet) cellsFig. 5B; compare lanes 6 and 7 with lanes 13 and 14, andig. 5C; compare lanes 3–7 with lanes 10–14). The delay of53 downregulation in B05-1(−tet) cells was confirmed bynalyzing the phosphorylation of p53 at Ser18, which is the

allmark modification of p53 activation in response to DNAamage (Fig. 5B and C). Therefore, although not required forhe maintenance of the basal state of p53 or its activationy DNA damage, SWI/SNF appears to be important for the

tagged proteins, phospho-Cdc2 at Tyr15, Cdc2, and actin as

downregulation of p53, suggesting that the increased DNAdamage-induced apoptosis of SWI/SNF-defective cells couldbe due to the sustained p53 activity.

3.5. �-H2AX defect and inefficient DSB repair inSWI/SNF-defective cells after doxorubicin treatment

It was previously shown that SWI/SNF-defective cells exhibit�-H2AX defect and inefficient DSB repair after exposure to IR[20]. We wanted to see if we could obtain similar results whenDSBs are created by chemical drugs such as doxorubicin. Cellswere treated with doxorubicin for 1 h, and, after wash, furthercultured in the medium without drug. After various time up to24 h, the cells were subjected to single cell gel electrophoresisunder neutral conditions to measure amount of unrepaired

DSBs. Immediately after removing drug from the medium, asimilar amount of DSBs was detected from B05-1(+tet) andB05-1(−tet) cells (Fig. 6A and B, 0-h time point). After 12 h post-drug treatment, the amount of unrepaired DSBs in B05-1(−tet)

36 d n a r e p a i r 8 ( 2 0 0 9 ) 29–39

Fig. 5 – SWI/SNF inactivation causes a prolonged p53 activation following DNA damage. After grown in the medium with (+)or without (−) tetracycline for 4 days, tet-VP16 and B05-1 cells were left untreated (indicated by 0 h) or treated with 0.5 �M ofdoxorubicin for indicated time. Cells were collected at indicated time points and analyzed by immunoblot for the expressionof flag-tagged proteins, p53, phosphor-p53 at Ser18 (corresponding to Ser15 in human) and p21. The expression of actin

lt of

was also analyzed for internal control. A representative resu

cells was much higher than in B05-1(+tet) cells (Fig. 6A–C),indicating that DSB repair is inefficient in SWI/SNF-defectivecells. Thus, the massive apoptosis of SWI/SNF-defective cellsafter doxorubicin treatment is possibly due to the accumu-lated unrepaired DSBs that evoke prolonged p53 activation.We also observed that the �-H2AX induction in response todoxorubicin treatment was severely compromised in B05-1(−tet) cells as previously shown for the cells exposed to IR(Fig. 6D–F), suggesting that the SWI/SNF complexes adopta similar mechanism for facilitating the repair of DBSswhether the DSBs are created by IR exposure or chemicaltreatment.

4. Discussion

Studies have shown that mammalian SWI/SNF complexes areinvolved in a variety of biological processes including cellcycle control, stress responses, differentiation, developmentand tumorigenesis [24,25]. Adding to this list, we previouslyhave shown that SWI/SNF directly participates in DSB repairvia interaction with �-H2AX [20]. Relevant to this functionwas the finding that inactivation of the SWI/SNF complexesseverely compromises the cells’ clonogenic ability after DNAdamage, but how SWI/SNF contributes to this post-damagecell survival has not been investigated. In the present study,we clearly show that SWI/SNF inactivation renders cells to

become highly susceptible to DNA damage-induced apopto-sis. Thus, it appears that the compromised clonogenic abilityof SWI/SNF-defective cells following DNA damage is primarilydue to their increased apoptosis rather than simply a result of

two (A, B) or three (C) independent experiments is shown.

inefficient damage repair that prevents further cell division.Our results suggest that, although dispensable for the

activation and initial establishment of G2/M DNA damagecheckpoint, SWI/SNF is required for sustaining the activatedG2/M checkpoint and thereby preventing cells from undergo-ing apoptosis. These results, together with the data showingthat little of B05-1(−tet) mitotic cells were detected when theyundergo massive apoptosis following DNA damage, suggestthat SWI/SNF-defective cells after DNA damage undergo apop-tosis at G2 without entering mitosis. In keeping with this,B05-1(−tet) cells, whether or not subjected to DNA damage,did not exhibit any indications of mitotic cell death or mitoticcatastrophe such as micronuclei, polyploidy nuclei and cytoki-nesis failure (S.K.H. and J.K., unpublished observations andSupplementary Fig. S1). In this regard, it is notable that arecent work revealed that tumors from BRG1 heterozygousmice exhibit genomic instability such as DNA copy numberchanges of various chromosomal segments, but not polyploidy[26].

Although it remains to be further investigated, SWI/SNFdoes not appear to prevent DNA damage-induced apoptosisby directly regulating the apoptosis pathways. Our data showthat, under normal conditions, SWI/SNF inactivation neitherhas an effect on the basal state of p53 (Fig. 5) nor causesabnormal expression of the p53 target genes involved in theregulation of apoptosis such as Bax, Puma and Bcl2 (data notshown). In addition, there is no aberration in the activation

of p53 by DNA damage in SWI/SNF-defective cells. How-ever, SWI/SNF-defective cells exhibit a significantly delayedkinetics of p53 downregulation compared to control cells, sug-gesting that the increased susceptibility of SWI/SNF-defective

d n a r e p a i r 8 ( 2 0 0 9 ) 29–39 37

Fig. 6 – �-H2AX defect and inefficient DSB repair in SWI/SNF-defective cells after doxorubicin treatment. (A) Comet assayswere performed to examine the effects of SWI/SNF inactivation on DSB repair. (top) B05-1 cells grown in the medium with(+) or without (−) tetracycline for 4 days were treated with 0.5 �M of doxorubicin for 1 h. Cells were washed with PBS andreplenished with the medium without drug, and harvested immediately (0 h) or further cultured for 6, 12, 18 and 24 h. Cellswere collected and then subjected to neutral comet assays. Average tail moments were obtained by counting at least 80 cellsand depicted as graph. (B) Representative comet images for each time point are shown. (C) The quantitative distribution oftail moment for 12 h time point is shown. (D) B05-1 cells grown as in (A) were treated with 0.5 �M of doxorubicin. After 0.5, 1and 5 h, the expression of �-H2AX was analyzed by immunoblottings. The expression of H2A was also analyzed for loadingcontrol. (E) B05-1 cells grown as per in (A) were treated with 0.5 �M of doxorubicin, and the cells were fixed after 1 h forstaining with �-H2AX antibodies. The graph shows the average number of �-H2AX foci determined by counting about 100nuclei per sample. The error bar indicates mean ± S.D. from three independent experiments. (F) A representative of thefl ting

ctpicatefui

tittncirim

uorescence microscopic images used for �-H2AX foci coun

ells to DNA damage-induced apoptosis could be attributedo the sustained p53 activity rather than of misregulation of53 activation. These results, together with the data show-

ng the increased accumulation of DSBs in SWI/SNF-defectiveells owing to inefficient repair, suggest that SWI/SNF preventspoptosis after DNA damage by facilitating efficient repair andhereby ensuring timely elimination of DSBs that could oth-rwise evoke sustained p53 activation. However, we cannotormally exclude the possibility that SWI/SNF directly reg-lates p53 downregulation since p53 has been reported to

nteract with SWI/SNF [27].Since the defect of DNA damage-induced apoptosis con-

ributes to tumorigenesis, our results, showing that BRG1nterferes with DNA damage-induced apoptosis, may seemo be apparently in conflict with the fact that BRG1 func-ions as a tumor suppressor [24]. However, BRG1 defect haso significant effect on spontaneous apoptosis in culturedells (Figs. 1–3), further supporting that BRG1 is not directly

nvolved in the regulation of apoptosis. Extrapolating theseesults to in vivo situations at the organism levels, BRG1s not likely to function to suppress apoptosis under nor-

al conditions, but rather protect against apoptosis upon

in (E) is shown.

DNA damage simply by facilitating DNA repair. We hypoth-esize that, in this way, BRG1 can ensure the maintenanceof genome integrity against DNA damage as well as healthycell survival, which eventually serves to suppress cancerdevelopment.

Recent studies have implicated Snf5/ini1, another coresubunit of mammalian SWI/SNF complexes, in DNA dam-age responses as well as mitotic checkpoint [28–30]. Thestudies showed that, like BRG1-defective cells, cells lack-ing Snf5 are hypersensitive to DNA damage [28], which maynot be surprising in that BRG1 and Snf5 function togetherexisting in the same complexes. However, in contrast to BRG1-defective cells, the cells lacking Snf5 exhibit an increasedconstitutive p53 activity, leading to spontaneous apopto-sis and thereby proliferative defect [28,29]. The constitutiveapoptosis of the cells lacking Snf5 has been suggested tobe attributed to mitotic catastrophe that is induced byspontaneous DNA damage. In addition, in contrast to BRG1-

defective cells, Snf5-defective cells exhibit mitotic checkpointdefect and poly- and aneuploidy [28–30]. These results sug-gest that BRG1 and Snf5 have distinct biological functionsalthough existing in the same complexes. In keeping with

8 ( 2

r

38 d n a r e p a i r

this, Snf5 has been shown to be not required for assem-bly of SWI/SNF complex or BRG1-dependent gene expression[31]. More importantly, although known to have tumor sup-pressor activity in common, BRG1 and Snf5 appear to beresponsible for preventing different types of tumors throughdistinct mechanisms. For example, Brg1 heterozygous micetend to develop mammary tumors due to haploinsufficiencyand genomic instability, whereas Snf5 heterozygote mice arepredisposed to malignant rhabdoid tumors and other sar-comas that undergo loss of heterozygosity and polyploidy[24,26].

Conflict of interest statement

None.

Acknowledgements

We are grateful to Tony Imbalzano for providing tet-VP16 andB05-1 cells. This work was supported by the grant to J.K. (R01-2007-000-10571-0) from the Korea Science and EngineeringFoundation (KOSEF) funded by the Ministry of Science & Tech-nology (MOST), and in part by the grant (R15-2007-0123-1-1)from the National Core Research Center (NCRC) program ofMOST and KOSEF through the Center for Cell Signaling & DrugDiscovery Research at Ewha Womans University.

Appendix A. Supplementary data

Supplementary data associated with this article can be found,in the online version, at doi:10.1016/j.dnarep.2008.08.011.

e f e r e n c e s

[1] J.W. Harper, S.J. Elledge, The DNA damage response: tenyears after, Mol. Cell 28 (2007) 739–745.

[2] D.C. van Gent, J.H. Hoeijmakers, R. Kanaar, Chromosomalstability and the DNA double-stranded break connection,Nat. Rev. Genet. 2 (2001) 196–206.

[3] S.P. Jackson, Sensing and repairing DNA double-strandbreaks, Carcinogenesis 23 (2002) 687–696.

[4] W.P. Roos, B. Kaina, DNA damage-induced cell death byapoptosis, Trends Mol. Med. 12 (2006)440–450.

[5] J.E. Chipuk, D.R. Green, Dissecting p53-dependent apoptosis,Cell Death Differ. 13 (2006) 994–1002.

[6] C.J. Norbury, B. Zhivotovsky, DNA damage-inducedapoptosis, Oncogene 23 (2004) 2797–2808.

[7] J.A. Downs, M.C. Nussenzweig, A. Nussenzweig, Chromatindynamics and the preservation of genetic information,Nature 447 (2007) 951–958.

[8] H. van Attikum, S.M. Gasser, The histone code at DNAbreaks: a guide to repair? Nat. Rev. Mol. Cell Biol. 6 (2005)757–765.

[9] C.L. Peterson, J. Côté, Cellular machineries for chromosomal

DNA repair, Genes Dev. 18 (2004) 602–616.

[10] C. Thiriet, J.J. Hayes, Chromatin in need of a fix:phosphorylation of H2AX connects chromatin to DNA repair,Mol. Cell 18 (2005) 617–622.

0 0 9 ) 29–39

[11] Y. Bao, X. Shen, Chromatin remodeling in DNAdouble-strand break repair, Curr. Opin. Genet. Dev. 17 (2007)126–131.

[12] A.J. Morrison, J. Highland, N.J. Krogan, A. Arbel-Eden, J.F.Greenblatt, J.E. Haber, X. Shen, INO80 and gamma-H2AXinteraction links ATP-dependent chromatin remodeling toDNA damage repair, Cell 119 (2004)767–775.

[13] H. van Attikum, O. Fritsch, B. Hohn, S.M. Gasser,Recruitment of the INO80 complex by H2A phosphorylationlinks ATP-dependent chromatin remodeling with DNAdouble-strand break repair, Cell 119 (2004)777–788.

[14] J.A. Downs, S. Allard, O. Jobin-Robitaille, A. Javaheri, A.Auger, N. Bouchard, S.J. Kron, S.P. Jackson, J. Côté, Binding ofchromatin-modifying activities to phosphorylated histoneH2A at DNA damage sites, Mol. Cell 16 (2004)979–990.

[15] T. Tsukuda, A.B. Fleming, J.A. Nickoloff, M.A. Osley,Chromatin remodelling at a DNA double-strand break site inSaccharomyces cerevisiae, Nature 438 (2005) 379–383.

[16] H. van Attikum, O. Fritsch, S.M. Gasser, Distinct roles forSWR1 and INO80 chromatin remodeling complexes atchromosomal double-strand breaks, EMBO J. 26 (2007)4113–4125.

[17] B. Chai, J. Huang, B.R. Cairns, B.C. Laurent, Distinct roles forthe RSC and Swi/Snf ATP-dependent chromatin remodelersin DNA double-strand break repair, Genes Dev. 19 (2005)1656–1661.

[18] E.Y. Shim, J.L. Ma, J.H. Oum, Y. Yanez, S.E. Lee, The yeastchromatin remodeler RSC complex facilitates end joiningrepair of DNA double-strand breaks, Mol. Cell Biol. 25 (2005)3934–3944.

[19] E.Y. Shim, S.J. Hong, J.H. Oum, Y. Yanez, Y. Zhang, S.E. Lee,RSC mobilizes nucleosomes to improve accessibility ofrepair machinery to the damaged chromatin, Mol. Cell Biol.27 (2007) 1602–1613.

[20] J.H. Park, E.J. Park, H.S. Lee, S.J. Kim, S.K. Hur, A.N.Imbalzano, J. Kwon, Mammalian SWI/SNF complexesfacilitate DNA double-strand break repair by promotinggamma-H2AX induction, EMBO J. 25 (2006)3986–3997.

[21] I.L. de La Serna, K.A. Carlson, D.A. Hill, C.J. Guidi, R.O.Stephenson, S. Sif, R.E. Kingston, A.N. Imbalzano,Mammalian SWI-SNF complexes contribute to activation ofthe hsp70 gene, Mol. Cell Biol. 20 (2000) 2839–2851.

[22] M. Castedo, J.L. Perfettini, T. Roumier, K. Andreau, R.Medema, G. Kroemer, Cell death by mitotic catastrophe: amolecular definition, Oncogene 23 (2004) 2825–2837.

[23] C.G. Takizawa, D.O. Morgan, Control of mitosis by changes inthe subcellular location of cyclin-B1-Cdk1 and Cdc25C, Curr.Opin. Cell Biol. 12 (2000) 658–665.

[24] C.W. Roberts, S.H. Orkin, The SWI/SNF complex-chromatinand cancer, Nat. Rev. Cancer. 4 (2004) 133–142.

[25] L. Mohrmann, C.P. Verrijzer, Composition and functionalspecificity of SWI2/SNF2 class chromatin remodelingcomplexes, Biochim. Biophys. Acta 1681 (2005)59–73.

[26] S.J. Bultman, J.I. Herschkowitz, V. Godfrey, T.C. Gebuhr, M.Yaniv, C.M. Perou, T. Magnuson, Characterization ofmammary tumors from Brg1 heterozygous mice, Oncogene27 (2008) 460–468.

[27] D. Lee, J.W. Kim, T. Seo, S.G. Hwang, E.J. Choi, J. Choe,SWI/SNF complex interacts with tumor suppressor p53 and

is necessary for the activation of p53-mediatedtranscription, J. Biol. Chem. 277 (2002) 22330–22337.

[28] A. Klochendler-Yeivin, E. Picarsky, M. Yaniv, Increased DNAdamage sensitivity and apoptosis in cells lacking the

( 2 0

[31] D.N. Doan, T.M. Veal, Z. Yan, W. Wang, S.N. Jones, A.N.

d n a r e p a i r 8

Snf5/Ini1 subunit of the SWI/SNF chromatin remodelingcomplex, Mol. Cell Biol. 26 (2006) 2661–2674.

[29] M.S. Isakoff, C.G. Sansam, P. Tamayo, A. Subramanian, J.A.Evans, C.M. Fillmore, X. Wang, J.A. Biegel, S.L. Pomeroy, J.P.Mesirov, C.W. Roberts, Inactivation of the Snf5 tumor

suppressor stimulates cell cycle progression and cooperateswith p53 loss in oncogenic transformation, Proc. Natl. Acad.Sci. U.S.A. 102 (2005) 17745–17750.

[30] R.G. Vries, V. Bezrookove, L.M. Zuijderduijn, S.K. Kia, A.Houweling, I. Oruetxebarria, A.K. Raap, C.P. Verrijzer,

0 9 ) 29–39 39

Cancer-associated mutations in chromatin remodeler hSNF5promote chromosomal instability by compromising themitotic checkpoint, Genes Dev. 19 (2005)665–670.

Imbalzano, Loss of the INI1 tumor suppressor does notimpair the expression of multiple BRG1-dependent genes orthe assembly of SWI/SNF enzymes, Oncogene 23 (2004)3462–3473.