DNA double-strand breaks promote methylation ofhistone H3 on lysine 9 and transient formation ofrepressive chromatinMarina K. Ayrapetova,1, Ozge Gursoy-Yuzugullua,1, Chang Xua,b, Ye Xua, and Brendan D. Pricea,2

aDepartment of Radiation Oncology, Dana–Farber Cancer Institute, Harvard Medical School, Boston, MA 02215; and bInstitute of Radiation Medicine, TianjinKey Laboratory of Molecular Nuclear Medicine, Chinese Academy of Medical Sciences and Peking Union Medical College, Tianjin 300192, People’s Republicof China

Edited by Steven Henikoff, Fred Hutchinson Cancer Research Center, Seattle, WA, and approved May 14, 2014 (received for review February 26, 2014)

Dynamic changes in histone modification are critical for regulatingDNA double-strand break (DSB) repair. Activation of the Tip60acetyltransferase by DSBs requires interaction of Tip60 with his-tone H3 methylated on lysine 9 (H3K9me3). However, how H3K9methylation is regulated during DSB repair is not known. Here, wedemonstrate that a complex containing kap-1, HP1, and the H3K9methyltransferase suv39h1 is rapidly loaded onto the chromatin atDSBs. Suv39h1 methylates H3K9, facilitating loading of additionalkap-1/HP1/suv39h1 through binding of HP1’s chromodomain tothe nascent H3K9me3. This process initiates cycles of kap-1/HP1/suv39h1 loading and H3K9 methylation that facilitate spreading ofH3K9me3 and kap-1/HP1/suv39h1 complexes for tens of kilobasesaway from the DSB. These domains of H3K9me3 function to acti-vate the Tip60 acetyltransferase, allowing Tip60 to acetylate bothataxia telangiectasia-mutated (ATM) kinase and histone H4. Conse-quently, cells lacking suv39h1 display defective activation of Tip60and ATM, decreased DSB repair, and increased radiosensitivity. Im-portantly, activated ATM rapidly phosphorylates kap-1, leading torelease of the repressive kap-1/HP1/suv39h1 complex from thechromatin. ATM activation therefore functions as a negative feed-back loop to remove repressive suv39h1 complexes at DSBs, whichmay limit DSB repair. Recruitment of kap-1/HP1/suv39h1 to DSBstherefore provides a mechanism for transiently increasing the levelsof H3K9me3 in open chromatin domains that lack H3K9me3 andthereby promoting efficient activation of Tip60 and ATM in theseregions. Further, transient formation of repressive chromatin maybe critical for stabilizing the damaged chromatin and for remodel-ing the chromatin to create an efficient template for the DNArepair machinery.

histone methylation | homologous recombination

DNA double-strand breaks (DSBs) are toxic and must berepaired to maintain genomic stability. Detection of DSBs

requires recruitment of the mre11–rad50–nbs1 (MRN) complexto the DNA ends (1). MRN then recruits and activates the ataxiatelangiectasia-mutated (ATM) kinase (2, 3) through a mecha-nism that also requires the Tip60 acetyltransferase (3). Tip60directly acetylates and activates ATM’s kinase activity (4–6) andfunctions, in combination with MRN, to promote ATM-dependentphosphorylation of DSB repair proteins (3), including histoneH2AX. This process creates domains of phosphorylated H2AX(γH2AX) extending for hundreds of kilobases along the chro-matin (7, 8). Mdc1 then binds to γH2AX, providing a landingpad for other DSB repair proteins, including the RNF8/RNF168ubiquitin ligases (1, 3, 9, 10). Tip60 also plays a critical role inregulating chromatin structure at DSBs as part of the NuA4–Tip60 complex (11). NuA4-Tip60 catalyzes histone exchange (viathe p400 ATPase subunit) and acetylation of histone H4 (by Tip60)at DSBs (12–15), leading to the formation of open, flexible chro-matin domains adjacent to the break (12, 13). These open chro-matin structures then facilitate histone ubiquitination, the loadingof brca1 and 53BP1, and repair of the DSB (13, 16). The ordered

acetylation and ubiquitination of the chromatin and loading ofDNA repair proteins is therefore critical for DSB repair.Activation of Tip60’s acetyltransferase activity requires in-

teraction between Tip60’s chromodomain and histone H3 meth-ylated on lysine 9 (H3K9me3) on nucleosomes at the break (4, 6).This interaction, in combination with tyrosine phosphorylation ofTip60 (17), increases Tip60’s acetyltransferase activity and pro-motes acetylation of both the ATM kinase and histone H4 (4–6,17). Consequently, loss of H3K9me2/3 leads to failure to activatethe ATM signaling pathway, loss of H4 acetylation during DSBrepair, disruption of heterochromatin, genomic instability, anddefective DSB repair (4, 17–19). H3K9me3s therefore play a crit-ical role in linking chromatin structure at DSBs to the activation ofDSB signaling proteins such as Tip60 and ATM.How Tip60 gains access to H3K9me3 and how H3K9me3 levels

at DSBs are regulated is not known. H3K9me3 is concentratedin heterochromatin domains, where it recruits HP1, kap-1, andH3K9 methyltransferases (20, 21) to maintain the silent, compactconformation of heterochromatin (20). This implies that Tip60’sacetyltransferase activity can only be activated at DSBs in regionsof high H3K9me3 density, such as heterochromatin. Alterna-tively, H3K9 methylation may be actively increased at DSBs inregions of low H3K9me3 density to allow for Tip60 activation and

Significance

Double-strand break (DSB) repair initiates dynamic changes inhistone modifications that are required to maintain genomestability. Methylation of histone H3 lysine-9 (H3K9me3) iscritical for activating ataxia telangiectasia-mutated (ATM) ki-nase, but how H3K9 methylation is regulated at DSBs is un-known. We show that a complex containing the suv39h1methyltransferase is rapidly recruited to DSBs, where it directsH3K9 methylation on large chromatin domains adjacent to theDSB. This process results in transient formation of repressivechromatin and serves to both stabilize the chromatin structureand promote activation of DSB-signaling proteins, including ATMkinase. Dynamic changes in H3K9 modification in euchro-matin by suv39h1 are therefore one of the earliest signalingevents required for processing and remodeling of the damagedchromatin template.

Author contributions: O.G.-Y. and B.D.P. designed research; B.D.P. conceived and plannedthe study, with input from M.K.A. and O.G.-Y.; M.K.A., O.G.-Y., C.X., and Y.X. performedresearch; M.K.A. performed laser microirradiation, plasmid construction, and cell-basedanalysis; O.G.-Y. performed the ChIP experiments with contributions from C.X.; Y.X. con-tributed to assay development; M.K.A., O.G.-Y., and B.D.P. analyzed data; and M.K.A.,O.G.-Y., and B.D.P. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.1M.K.A. and O.G.-Y. contributed equally to this work.2To whom correspondence should be addressed. E-mail: brendan_price@dfci.harvard.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1403565111/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1403565111 PNAS | June 24, 2014 | vol. 111 | no. 25 | 9169–9174

CELL

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Mar

ch 1

2, 2

021

efficient DSB repair in euchromatin. Understanding the dy-namics of H3K9 methylation at DSBs is therefore critical to un-derstanding how Tip60 activity is regulated by the local chromatinarchitecture. Here, we show that the suv39h1 methyltransferase isrecruited to DSBs in euchromatin as part of a larger kap-1/HP1/suv39h1 complex. Suv39h1 increases H3K9me3 at DSBs, acti-vating Tip60’s acetyltransferase activity and promoting the sub-sequent acetylation and activation of ATM. Further, loss ofinducible H3K9me3 at DSBs leads to defective repair and increasedradiosensitivity. Finally, loading of the kap-1/HP1/suv39h1 com-plex is transient, and the complex is rapidly released from thechromatin through a negative feedback loop driven by ATM-dependent phosphorylation of the kap-1 protein.

ResultsInitially, we determined whether H3K9me3 participates in DSBrepair in chromatin domains that lack endogenous H3K9me2/3.The p84–zinc finger nuclease (p84-ZFN) creates a DSB in intron 1of the PPP1R12C gene (12, 13). PPP1R12C lacks significantH3K9me2/3 but is rich in marks associated with transcription(ENCODE database, http://encodeproject.org/ENCODE/). Chro-matin IP (ChIP) demonstrated increased phosphorylation ofH2AX (γH2AX) at the p84-ZFN DSB (Fig. 1A). ChIP withH3K9me3 antibody (Fig. S1A) demonstrated increased H3K9me3on either side of the DSB (±1.5 kb), with lower levels of H3K9me3extending >200 kb away from the DSB (Fig. 1A). A small in-crease in H3K9me2 was also detected (Fig. 1B). Histone H3levels were unchanged (Fig. 1B), indicating increased methylationrather than changes in H3 content. Further, no change in eitherH3K36me3 or H4K20me2 was seen at the DSB (Fig. S1B). Fi-nally, H3K9me3 was not increased at a distal chromosome site(Fig. S1C), indicating that the increased H3K9me3 is re-stricted to the chromatin domain adjacent to the DSB.The H3K9 methyltransferase suv39h1 has been implicated in

DSB repair (4, 18, 19). Depletion of suv39h1 with siRNA (Fig.S1D) significantly reduced H3K9me3 at the p84-ZFN DSB (Fig.1C). Furthermore, ChIP demonstrated that suv39h1 was loadedonto the chromatin at the DSB (Fig. 1D and Fig. S1E). Finally,suv39h1 was rapidly (within 5 min) recruited to regions of DNAdamage created with laser microirradiation (Fig. 1E). Thesuv39h1 methyltransferase is therefore recruited to DSBs andincreases local H3K9me3 in response to DSBs.Interaction between Tip60 and H3K9me3 (4) promotes acet-

ylation of ATM (5, 6, 17) and histone H4 (12–14) by Tip60.Depletion of suv39h1 reduced inducible H3K9me3 (Fig. 1C) andinhibited acetylation of histone H4 by Tip60 (Fig. 2A). Fur-thermore, depletion of suv39h1 attenuated ATM activation(Fig. S2A) and reduced phosphorylation of kap-1 (Fig. S2B).This finding is consistent with methylation of H3K9 by suv39h1playing an essential role in activation of Tip60’s acetyltransferaseactivity and the subsequent acetylation and activation of theATM kinase. Furthermore, cells lacking suv39h1 have increasedradiation sensitivity (Fig. 2B) and reduced homologous recombi-nation (HR)-mediated repair (Fig. 2C). Finally, recruitment ofbrca1 and RPA32 to DSBs (Fig. 2D) were reduced following lossof suv39h1, consistent with the decrease in HR-mediated repairin suv39h1-deficient cells (Fig. 2C). However, nonhomologousend-joining (NHEJ) activity was not significantly altered by lossof suv39h1 (Fig. S2C), implying that regulation of the Ku70/80complex and DNA–PKcs activity does not require H3K9 methyla-tion. These results are consistent with previous studies demonstratingincreased genomic instability in mice and other experimental systems(4, 18, 19) in which suv39h1 was inactivated. Methylation ofH3K9 by suv39h1 at DSBs therefore plays a critical role in acti-vating Tip60, controlling ATM signaling, and in directing DSBrepair and maintaining genomic stability.Two heterochromatin-binding proteins, kap-1 and HP1, are

corecruited to sites of DNA damage (22, 23), although how theyimpact DSB repair is not clear. However, suv39h1 can interact withkap-1/HP1 repressor complexes (20, 21, 24), suggesting that kap-1and HP1 may recruit suv39h1 to DSBs. Initially, we confirmed that

kap-1 interacts with the HP1α, HP1β, and suv39h1 and that thisinteraction was not altered by DNA damage (Fig. S3 B and C).When laser “striping” was used to create focused regions of DNAdamage, suv39h1 (Fig. 3A) and kap-1 (Fig. 3B) were recruited tosites of DNA damage with similar kinetics, such that ∼90% ofγH2AX stripes were colocalized with suv39h1 and kap-1 (Fig.3C). Furthermore, ChIP demonstrated that HP1β was alsorecruited to sites of DNA damage (Fig. 3D). Importantly,

Fig. 1. Suv39h1 promotes H3K9 methylation at DSBs. (A) 293T cells trans-fected with p84-ZFN were processed for ChIP by using γH2AX orH3K9me3 antibodies, followed by quantitative RT-PCR (RT-qPCR) withprimer pairs located at the indicated positions. Results are fold enrichmentrelative to uncut DNA, which is assigned a value of 1 (solid black line; Uncut).Results are ±SD (n = 3). (B) 293T cells transfected with p84-ZFN (ZFN) orvector (Vec) were processed for ChIP by using antibodies to H3, γH2AX,H3K9me3, or H3K9me2 and primer pairs located 1.5 kb to the right of theDSB. Results are expressed as fold enrichment relative to uncut DNA. Resultsare ±SD (n = 3). (C) 293T cells were transfected with nonspecific (−) or siRNAtargeting suv39h1 (+). Forty-eight hours later, cells were transfected withvector (Vec) or p84-ZFN (ZFN) and processed for ChIP by using antibodyagainst H3K9me3 and primers located 1.5 kb to either side of the DSB.Results are ±SD (n = 3). (D) 293T cells were transfected with vector (Vec) orp84-ZFN (ZFN) and processed for ChIP with suv39h1 antibody and primerslocated 1.5 kb to either side of the DSB. Results are ±SD (n = 3). (E ) U2OScells were transfected with nonspecific (control) or suv39h1-specific(siSuv39h1) siRNA. Forty-eight hours later, focused regions of DNA damagewere produced by using a scanning laser system. Cells were fixed for im-munofluorescent staining by using antibodies to γH2AX (green) or suv39h1(red). Nuclei were stained with DAPI (blue).

9170 | www.pnas.org/cgi/doi/10.1073/pnas.1403565111 Ayrapetov et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

2, 2

021

suv39h1, kap-1, and HP1 were only transiently retained at damagesites (Fig. 3 A and B), indicating that they function during theinitial few minutes following DSB production. Furthermore, de-pletion of kap-1 (Fig. S3A) blocked recruitment of both suv39h1(Fig. 3 A and C) and HP1β (Fig. 3D) to DSBs. Similarly, de-pletion of suv39h1 blocked recruitment of kap-1 (Fig. 3 B and Cand Fig. S3D) and HP1β (Fig. 3D) to DSBs. Loss of any one ofsuv39h1, kap-1, or HP1 therefore prevents loading of all threeproteins at the DSB. Finally, depletion of kap-1, which blocksrecruitment of both suv39h1 and HP1 to DSBs, abolished theincrease in H3K9me3 at DSBs (Fig. 3E). Combining these datawith previous work demonstrating direct interaction betweenkap-1, HP1 family members, and suv39h1 (24–28) implies thatkap-1, HP1, and suv39h1 are recruited to DSBs as a single kap-1/HP1/suv39h1 complex and that it is this complex that directsH3K9 methylation at DSBs.Next, we examined suv39h1’s methyltransferase activity.

Suv39h1MD, containing a point mutation in the catalytic domain,was efficiently incorporated into the kap-1/HP1/suv39h1 complex(Fig. S4A). However, suv39h1MD was not recruited to regions of

laser damage, and, furthermore, it blocked recruitment of kap-1(Fig. 4A), such that <5% of γH2AX stripes in suv39h1MD cellscontained either kap-1 or suv39h1 (see Fig. S4E for quantita-tion). Dimethyloxalylglycine (DMOG), a pan-specific inhibitorof H3K9 demethylases (29), was then used to increase globalH3K9me3 levels (Fig. S4 B and D). DMOG rescued recruitmentof both suv39h1MD and kap-1 to regions of laser damage (Fig.4A), such that >80% of γH2AX stripes in suv39h1MD cells thencolocalized with suv39h1 and kap-1 (Fig. S4E). Further, DMOGrescued recruitment of kap-1 and HP1α, even in cells lackingexpression of suv39h1 (Fig. 4B). This finding demonstrates thatH3K9me3, rather than suv39h1, mediates retention of kap-1,HP1, and suv39h1 on the chromatin. Because the chromodomainof HP1α binds to H3K9me3, we deleted HP1α’s chromodomain(Fig. S4D). Loss of HP1α’s chromodomain did not alter in-teraction of HP1α with either suv39h1 or kap-1 (Fig. S4 F and G).Importantly, HP1αCD was not recruited to sites of laser micro-irradiation and inhibited corecruitment of suv39h1 (Fig. 4C).Furthermore, increasing H3K9me3 with DMOG did not restoreloading of HP1 or suv39h1 in HP1αCD cells (Fig. 4C), demon-strating that both HP1α’s chromodomain and H3K9me3 are re-quired to retain kap-1/HP1/suv39h1 at DSBs. This finding suggestsa model in which the initial positioning of kap-1/HP1/suv39h1 atDSBs promotes H3K9 methylation by suv39h1. This “priming”event then recruits additional kap-1/HP1/suv39h1 complexes,which are retained through interaction between HP1’s chromo-domain and newly synthesized H3K9me3. This kap-1/HP1/suv39h1then bridges to more distal nucleosomes, promoting additionalcycles of H3K9 methylation and kap-1/HP1/suv39h1 binding,which spreads H3K9me3 along the chromatin away from the DSB.Spreading therefore requires both suv39h1 (to create H3K9me3)and HP1 (which binds to H3K9me3). Consequently, loss of HP1,suv39h1, or kap-1 prevents spreading of H3K9me3 and thereforeloading of the complex onto the chromatin.Kap-1, HP1, and suv39h1 are only retained at the DSB during

the first few minutes after damage (Figs. 3 and 4). Because theATM-dependent phosphorylation of kap-1 weakens kap-1’s in-teraction with chromatin (30, 31), we examined whether ATMwas required to release kap-1/HP1/suv39h1 from DSBs. TheATM inhibitor Ku-55933 (ATMi) did not alter interaction betweenkap-1 and suv39h1 (Fig. S5A). Suv39h1 (Fig. 5A) and kap-1 (Fig.5B) were recruited to DSBs in the presence of ATMi. However,ATMi blocked release of suv39h1 and kap-1 (Fig. 5 A and B) andincreased retention of kap-1 at p84-ZFN–generated DSBs (Fig.S5B). Similarly, inactivation of the MRN complex, which is re-quired for ATM activation (2), blocked phosphorylation of kap-1by ATM (Fig. S5D) and prevented release of suv39h1 from sitesof DNA damage (Fig. S5C). Phosphorylation of serine 824 ofkap-1 by ATM weakens kap-1 interaction with chromatin (30–32).Kap-1wt, kap-1S824A, and kap-1S824D (containing a phospho-mimic)were expressed in U2OS cells (Fig. S5E). Kap-1wt was transientlyrecruited to and released from regions of DNA damage, whereasthe nonphosphorylatable kap-1S824A was recruited to DSBs butretained at DSBs for an extended time (Fig. 5C and Fig. S5F).This result is consistent with kap-1 phosphorylation promotingrelease of kap-1/HP1/suv39h1 from DSBs. Intriguingly, thephospho-mimic kap-1S824D was poorly retained at both sites oflaser damage (Fig. 5C) and at DSBs created by the p84-ZFN(Fig. S5F). We conclude that the rapid release of kap-1/HP1/suv39h1 from DSBs requires the ATM-dependent phosphoryla-tion of serine 824 of kap-1.Because many early responses to DNA damage require poly

(ADP-ribose) polymerase (PARP) family members, which cata-lyze synthesis of PAR chains on the chromatin at DSBs (33), weexamined whether the rapid recruitment of kap-1/HP1/suv39h1to DSBs required PARP activity. The parp inhibitor olaparib(PARPi) did not alter interaction between kap-1, HP1α, andsuv39h1 (Fig. S5A). However, inhibition of PARP blocked therapid recruitment of suv39h1 (Fig. 6A) and kap-1 (Fig. S6B)to sites of DNA damage. Chromatin PARylation after DNAdamage was not altered in cells lacking suv39h1 (Fig. S6A),

Fig. 2. Suv39h1 regulates Tip60 activity, genomic stability, and homologousrecombination (HR)-mediated repair. (A) 293T cells expressing nonspecificshRNA (control) or shRNA targeting suv39h1 were transfected with vector(Vec) or p84-ZFN (ZFN), followed by ChIP with antibody to H4Ac. RT-qPCRwas carried out using primer pairs located to 1.5 kb to the right of the DSB.Results are ±SD (n = 3). (B) 293T cells expressing nonspecific shRNA (●) orshRNA targeting suv39h1 (○) were irradiated, and clonogenic cell survivalassays were carried out. Results are ±SD (n = 3). (C) U2OS cells containinga GFP–HR reporter were stably transfected with nonspecific (NS) or shRNAtargeting suv39h1, followed by transient transfection with vector (Vec) orthe I-Sce1 endonuclease. GFP-positive cells were counted by FACS. Resultsare ±SD (n = 4 biological replicates). (D) U2OS cells expressing nonspecific (−)or shRNA targeting suv39h1 (+) were irradiated (2 Gy), allowed to recoverfor 30 min, and stained with antibodies to γH2AX, brca1, or RPA32. Cellswith more than five foci were scored as positive. Results are ±SD (n = 3). Pvalue was determined by using a t test.

Ayrapetov et al. PNAS | June 24, 2014 | vol. 111 | no. 25 | 9171

CELL

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Mar

ch 1

2, 2

021

demonstrating that PARylation is upstream of the kap-1/HP1α/suv39h1 complex. Furthermore, siRNA to parp1 also blockedrecruitment of suv39h1 (Fig. S6C), implying that parp1, ratherthan other parp family members, is required for recruitment ofkap-1/HP1/suv39h1 to damaged chromatin. PARPi led to a signifi-cant reduction in H3K9me3 at DSBs (Fig. 6C), consistent with thefailure to recruit suv39h1 to DSBs in the presence of PARPi(Fig. 6A) or parp1 siRNA (Fig. S6C). In addition, becauseH3K9me3 is important for full activation of ATM (4), PARPialso inhibited phosphorylation of kap-1 by ATM after DNAdamage (Fig. 6B). Finally, depletion of suv39h1 increased cellsensitivity to PARPi (Fig. 6D), consistent with previous reportsthat cells lacking ATM are sensitive to PARP inhibition (34, 35).

DiscussionWe have shown that the suv39h1 methyltransferase is rapidlyrecruited to DSBs, where it functions to create domains ofH3K9me3 adjacent to DSBs. Previous work demonstrated that

the repressive HP1 and kap-1 proteins were also recruited toDSBs (4, 22, 23, 32, 36), although how these proteins contributedto DSB repair was not clear. Kap-1, HP1 family members, andsuv39h1 can form large repressive complexes (24–28). Here,we show that loading of suv39h1, kap-1, and HP1 at DSBs wasinterdependent, with loss of any one protein inhibiting re-cruitment of the other two. This finding is consistent with theidea that kap-1, HP1, and suv39h1 are recruited to DSBs as asingle kap-1/HP1/suv39h1 complex. We propose a model in whichloading of the kap-1/HP1/suv39h1 complex at DSBs increasesH3K9me3 methylation on nucleosomes on either side of theDSB. This promotes recruitment of additional kap-1/HP1/suv39h1(through interaction of HP1’s chromodomain with H3K9me3),which then methylates H3K9 on nucleosomes further from theDSB. This process leads to cycles of kap-1/HP1/suv39h1 loadingand H3K9 methylation, which catalyzes the spreading ofH3K9me3 (and kap-1/HP1/suv39h1) along the chromatin. Thismodel is similar to the spreading of heterochromatin, in which an

Fig. 3. Recruitment of suv39h1 to DSBs requires kap-1 and HP1. (A) U2OS cells were transfected with control siRNA or siRNA to kap-1. DNA damage wascreated by using a laser, and cells were allowed to recover for 0, 15, or 30 min. Cells were costained with antibody to suv39h1 (red) or γH2AX (green). (B) U2OScells were transfected with control siRNA or siRNA to suv39h1. DNA damage was created by using a laser, and cells were allowed to recover for 0, 15, or 30min. Cells were costained with antibody to kap1 (green) or γH2AX (red). (C) Quantitation of results in A and B. The percentage of γH2AX laser stripes thatcolocalized with either suv39h1 or kap-1 stripes were noted. Results are ±SD (n = 25–60 cells). (D) 293T cells expressing nonspecific shRNA (control) or shRNAto kap-1 or suv39h1 were transfected with vector (Vec) or p84-ZFN (ZFN), followed by ChIP using HP1β antibody and primers 1.5 kb to the right of the DSB.Results are ±SD (n = 3). (E) 293T cells expressing nonspecific shRNA (control) or shRNA against kap-1 (shKap1) were transfected with p84-ZFN (ZFN) or vector(Vec) followed by ChIP using H3K9me3 antibody and primers located 1.5 kb to the right of the DSB. Results are ±SD (n = 3).

Fig. 4. HP1’s chromodomain is required to recruit kap-1, HP1, and suv39h1 to DSBs. (A) U2OS cells expressing myc-suv39h1 (SuvWT) or catalytically inactivemyc-suv39h1 (SuvMD) were exposed to laser damage and stained with antibody to myc and γH2AX or kap-1 and γH2AX. Some cells were preincubated indimethyloxalylglycine (DMOG) (1 mM/1 h). (B) U2OS cells expressing nonspecific shRNA (shVec) or shRNA to suv39h1 (shSuv) were exposed to laser damageand costained with antibodies to HP1α and γH2AX or kap-1 and γH2AX. Some cells were preincubated in DMOG (1 mM/1 h). (C) U2OS cells expressing vector,HP1α (HA-HP1α), or HP1α with the chromodomain deleted (HA-HP1αCD) were exposed to laser damage and costained with antibodies for HA and γH2AX orsuv39h1 and γH2AX. Some cells were preincubated in DMOG (1 mM/1 h).

9172 | www.pnas.org/cgi/doi/10.1073/pnas.1403565111 Ayrapetov et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

2, 2

021

initial nucleation event positions HP1 complexes containingH3K9 methyltransferases on the chromatin (20, 37). Subsequentcycles of H3K9 methylation and loading of HP1 complexes resultin the spreading of heterochromatin along the chromatin (26, 38,39). In this way, an initial nucleation event at DSBs can spreadH3K9me3 and kap-1/HP1/suv39h1 along the chromatin domainsflanking the DSB, leading to the rapid formation of repressivechromatin at the DSB.The initial nucleation event that recruits kap-1/HP1/suv39h1 to

DSBs required parp1. Several PARP family members arerecruited to DSBs where they rapidly create PAR chains. PARprovides docking sites for several proteins implicated in DSBrepair, including macroH2A and the ALC1 and NuRD remod-eling complexes (40–44). However, neither kap-1 nor HP1 norsuv39h1 contain conserved PAR binding motifs or undergochanges in PARylation after DNA damage. Thus, whether thekap-1/HP1α/suv39h1 complex binds directly to PAR chains on thechromatin, or whether the complex contains additional PAR-bind-ing subunits, remains to be determined. Alternatively, PARylationmay alter nucleosome structure at DSBs, thereby facilitating H3K9methylation by the kap-1/HP1/suv39h1 complex. In eithercase, both parp1 activity and H3K9me3 spreading are re-quired to stably, but transiently, load kap-1/HP1α/suv39h1onto the chromatin.H3K9me3 is required for activation of the Tip60 acetyl-

transferase (4, 17). However, because H3K9me3 is primarilylocated within silent, heterochromatic regions (20, 37, 38), thisrequirement suggests that Tip60 activity during DNA repair maybe restricted to chromatin domains with a high density ofH3K9me2/3. Here, we demonstrate that transient loading ofkap-1/HP1/suv39h1 at DSBs provides a mechanism for rapidlyincreasing H3K9me3 in open (euchromatin) domains that lackpreexisting H3K9me3. Furthermore, reducing H3K9me3 by tar-geting either suv39h1 or PARP activity inhibited Tip60’s acetyl-transferase activity and attenuated activation of ATM by DSBs.This finding is consistent with published studies in which loss ofMRN (2), Tip60 (4, 6), or acute PARP inhibition (35) led todefective activation of ATM’s kinase activity. Increased H3K9methylation by the kap-1/HP1/suv39h1 complex is thereforecritical for full activation of Tip60 and ATM and for the repair ofDSBs within open, euchromatic regions of the genome.The recruitment of kap-1/HP1/suv39h1 and the increase in

H3K9me3 may reflect a need to temporarily stabilize and “heter-ochromatinize”DSBs in open regions. Other repressive complexes,such as NuRD and histone deacetylases (HDACs) (43–46), also

transiently accumulate at DSBs, supporting this idea. The for-mation of repressive structures at DSBs in open chromatin hasparallels with DSB repair in heterochromatin. DSB repair inheterochromatin requires the ATM-dependent phosphorylationof kap-1 (30, 31), which releases the repressive CHD3 chromatinremodeler (47) and promotes chromatin relaxation and facili-tates DSB repair (31, 32, 47). Thus, immediately after DNAdamage, both euchromatin and heterochromatin domains havesimilar structural organization, including high density of H3K9me3and the presence of repressive complexes, such as kap-1, HP1α,methyltransferases, HDACs, and CHD3/CHD4 remodelingATPases (31, 32, 43–47). This rapid, but temporary, formation ofrepressive chromatin may inhibit local transcription, compact thelocal chromatin structure, and rewrite the local epigeneticlandscape, stabilizing open chromatin structures and limitingDSB mobility during the initial moments following DSB pro-duction. However, because repressive chromatin inhibits DSBrepair (30, 31, 47), it is important that these repressive structuresare rapidly dismantled. As the damage response unfolds, H3K9methylation increases, leading to Tip60 activation and increasedATM kinase activity. ATM then phosphorylates kap-1, releasingthe repressive kap-1/HP1/suv39h1 from the chromatin and therebyproviding a negative feedback loop that regulates both H3K9methylation and loading of the kap-1/HP1/suv39h1 complex atDSBs. In heterochromatin, kap-1 phosphorylation releases theCHD3 complex (47), leading to relaxation of the chromatin struc-ture (31), although kap-1 remains associated with the heterochro-matin. The retention of phosphorylated kap-1 in heterochromatinmay result from the presence of Kruppel-associated box zinc fin-ger proteins (21), which anchor kap-1 to heterochromatin, butwhich are absent from open, euchromatin regions. In this way, thecompact structure of heterochromatin and the transient estab-lishment of repressive chromatin at DSBs in open regions can bereversed through a common mechanism dependent on the ATMkinase. This structure then allows further chromatin processing to

Fig. 5. Phosphorylation of kap-1 by ATM releases suv39h, kap-1, and HP1from DSBs. (A and B) U2OS cells were preincubated with ATMi (100 μM) orsolvent for 60 min. After laser microirradiation, cells were immediately fixed(0 min) or allowed to recover for 15 min and then costained with antibodiesto suv39h1 and γH2AX (A) or kap-1 and γH2AX (B). (C) U2OS cells expressingwild-type kap-1 (mycKap1wt), kap-1 with an alanine mutation in the ATMphosphorylation site (mycKap1S824D), or kap-1 with a phospho-mimic in thesame site (mycKap1S824D) were exposed to laser microirradiation and mycKap1and γH2AX detected with myc or γH2AX antibody.

Fig. 6. Chromatin PARylation recruits kap-1, HP1, and suv39h1 to DSBs. (A)U2OS cells were preincubated with PARPi (olaparib; 1 h/20 μM), followed bylaser microirradiation. Cells were either fixed (0 min) or allowed to recoverfor 15 min, and then costained with antibody to suv39h1 and γH2AX. (B)U2OS cells were preincubated in PARPi (olaparib; 20 μM) for 1 h, and thenexposed to bleomycin (5 μM) for the indicated times. Kap1, pkap1, and tu-bulin were monitored by Western blot analysis. (C) 293T cells were trans-fected with vector (Vec) or p84-ZFN (ZFN), followed by solvent (control) orPARPi (olaparib; 20μM). ChIP was carried out by using H3K9me3 antibodyand primers located 1.5 kb to the right of the DSB. Results are ±SD (n = 3).(D) 293T cells expressing nonspecific shRNA (○) or shRNA targeting suv39h1(●) were incubated with PARPi for 24 h, and clonogenic cell survival assayswere carried out. Results are ±SE (n = 3).

Ayrapetov et al. PNAS | June 24, 2014 | vol. 111 | no. 25 | 9173

CELL

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Mar

ch 1

2, 2

021

create open, flexible chromatin structures, which are essential forDSB repair (11).Dynamic methylation of histones and phosphorylation of kap-

1 during DSB repair therefore provide a regulated mechanismfor increasing compaction of open chromatin and decreasing thecompaction of heterochromatin, so that DSBs in both regionsbegin to have similar epigenetic and structural organization. Thisprocess is critical for establishing H3K9me3 for activation ofTip60 and the ATM kinase, as well as for contributing to the earlyprocessing of the chromatin at DSBs. These dynamic changes inchromatin organization are therefore critical for creating a com-mon chromatin template that is an efficient substrate for the HRor NHEJ DSB repair pathways.

Materials and MethodsDetails on cell growth, HRassays, transfection, antibodies, plasmid construction,Western blot, mRNA analysis, and shRNA/siRNA are in SI Materialsand Methods.

ChIP. For ChIP assays (12, 13), cross-linked chromatin was sonicated, andequivalent amounts were incubated with primary antibody and protein G

agarose beads precoated with sperm DNA. After washing, immunopurifiedchromatin was eluted and digested with proteinase K, and purified DNA wasquantified by quantitative RT-PCR (RT-qPCR) using the Step One Plus real-time PCR system (Applied Biosystems). Results are expressed as fold increasein signal relative to uncut sample. Detailed protocols, primer pairs, and ChIPgrade antibodies are described in SI Materials and Methods.

Laser Microirradiation and Immunofluorescence. Laser damage was producedby using a 30-mW, 405-nm diode laser focused through the 40×-Plan Apochro-mat/1.25-N.A. oil objective (Leica TCS SP5; Leica Microsystems) in combinationwith Hoechst 33258 (12). The time between the initial laser exposure and ter-mination by fixation was 5 min, which is referred to as time 0. At least 50 nucleiwere microirradiated per slide. For recruitment of brca1 and RPA32 to ionizingradiation induced foci, cells were cultured on coverslips and irradiated in a Cs137

irradiator. Cells were fixed and incubated with primary and secondary antibodiesas described in SI Materials and Methods and imaged by using a Zeiss AxioIm-ager Z1 microscope.

ACKNOWLEDGMENTS. We thank Sangamo Biosciences for p84-ZFN andDipanjan Chowdhury for the kap1 constructs. This work was supported byNational Institutes of Health Grants CA64585, CA93602, and CA177884 (to B.D.P.).

1. Ciccia A, Elledge SJ (2010) The DNA damage response: Making it safe to play with

knives. Mol Cell 40(2):179–204.2. Shiloh Y, Ziv Y (2013) The ATM protein kinase: Regulating the cellular response to

genotoxic stress, and more. Nat Rev Mol Cell Biol 14(4):197–210.3. Sun Y, Jiang X, Price BD (2010) Tip60: Connecting chromatin to DNA damage sig-

naling. Cell Cycle 9(5):930–936.4. Sun Y, et al. (2009) Histone H3 methylation links DNA damage detection to activation

of the tumour suppressor Tip60. Nat Cell Biol 11(11):1376–1382.5. Sun Y, Xu Y, Roy K, Price BD (2007) DNA damage-induced acetylation of lysine 3016 of

ATM activates ATM kinase activity. Mol Cell Biol 27(24):8502–8509.6. Sun Y, Jiang X, Chen S, Fernandes N, Price BD (2005) A role for the Tip60 histone

acetyltransferase in the acetylation and activation of ATM. Proc Natl Acad Sci USA

102(37):13182–13187.7. Rogakou EP, Boon C, Redon C, Bonner WM (1999) Megabase chromatin domains

involved in DNA double-strand breaks in vivo. J Cell Biol 146(5):905–916.8. Iacovoni JS, et al. (2010) High-resolution profiling of gammaH2AX around DNA

double strand breaks in the mammalian genome. EMBO J 29(8):1446–1457.9. Jackson SP, Durocher D (2013) Regulation of DNA damage responses by ubiquitin and

SUMO. Mol Cell 49(5):795–807.10. Fradet-Turcotte A, et al. (2013) 53BP1 is a reader of the DNA-damage-induced H2A

Lys 15 ubiquitin mark. Nature 499(7456):50–54.11. Price BD, D’Andrea AD (2013) Chromatin remodeling at DNA double-strand breaks.

Cell 152(6):1344–1354.12. Xu Y, et al. (2012) Histone H2A.Z controls a critical chromatin remodeling step re-

quired for DNA double-strand break repair. Mol Cell 48(5):723–733.13. Xu Y, et al. (2010) The p400 ATPase regulates nucleosome stability and chromatin

ubiquitination during DNA repair. J Cell Biol 191(1):31–43.14. Murr R, et al. (2006) Histone acetylation by Trrap-Tip60 modulates loading of repair

proteins and repair of DNA double-strand breaks. Nat Cell Biol 8(1):91–99.15. Soria G, Polo SE, Almouzni G (2012) Prime, repair, restore: The active role of chro-

matin in the DNA damage response. Mol Cell 46(6):722–734.16. Tang J, et al. (2013) Acetylation limits 53BP1 association with damaged chromatin to

promote homologous recombination. Nat Struct Mol Biol 20(3):317–325.17. Kaidi A, Jackson SP (2013) KAT5 tyrosine phosphorylation couples chromatin sensing

to ATM signalling. Nature 498(7452):70–74.18. Peters AH, et al. (2001) Loss of the Suv39h histone methyltransferases impairs

mammalian heterochromatin and genome stability. Cell 107(3):323–337.19. Peng JC, Karpen GH (2009) Heterochromatic genome stability requires regulators of

histone H3 K9 methylation. PLoS Genet 5(3):e1000435.20. Grewal SI, Jia S (2007) Heterochromatin revisited. Nat Rev Genet 8(1):35–46.21. Iyengar S, Farnham PJ (2011) KAP1 protein: An enigmatic master regulator of the

genome. J Biol Chem 286(30):26267–26276.22. Baldeyron C, Soria G, Roche D, Cook AJ, Almouzni G (2011) HP1alpha recruitment to

DNA damage by p150CAF-1 promotes homologous recombination repair. J Cell Biol

193(1):81–95.23. Luijsterburg MS, et al. (2009) Heterochromatin protein 1 is recruited to various types

of DNA damage. J Cell Biol 185(4):577–586.24. Fritsch L, et al. (2010) A subset of the histone H3 lysine 9 methyltransferases Suv39h1,

G9a, GLP, and SETDB1 participate in a multimeric complex. Mol Cell 37(1):46–56.25. Lechner MS, Begg GE, Speicher DW, Rauscher FJ, 3rd (2000) Molecular determinants

for targeting heterochromatin protein 1-mediated gene silencing: Direct chromoshadow

domain-KAP-1 corepressor interaction is essential. Mol Cell Biol 20(17):6449–6465.

26. Nielsen AL, et al. (1999) Interaction with members of the heterochromatin protein 1(HP1) family and histone deacetylation are differentially involved in transcriptionalsilencing by members of the TIF1 family. EMBO J 18(22):6385–6395.

27. Li X, et al. (2010) SUMOylation of the transcriptional co-repressor KAP1 is regulatedby the serine and threonine phosphatase PP1. Sci Signal 3(119):ra32.

28. Ivanov AV, et al. (2007) PHD domain-mediated E3 ligase activity directs intramolecularsumoylation of an adjacent bromodomain required for gene silencing. Mol Cell 28(5):823–837.

29. Ayrapetov MK, et al. (2011) Activation of Hif1α by the prolylhydroxylase inhibitordimethyoxalyglycine decreases radiosensitivity. PLoS ONE 6(10):e26064.

30. Noon AT, et al. (2010) 53BP1-dependent robust localized KAP-1 phosphorylation isessential for heterochromatic DNA double-strand break repair. Nat Cell Biol 12(2):177–184.

31. Goodarzi AA, et al. (2008) ATM signaling facilitates repair of DNA double-strandbreaks associated with heterochromatin. Mol Cell 31(2):167–177.

32. Ziv Y, et al. (2006) Chromatin relaxation in response to DNA double-strand breaks ismodulated by a novel ATM- and KAP-1 dependent pathway. Nat Cell Biol 8(8):870–876.

33. Gibson BA, Kraus WL (2012) New insights into the molecular and cellular functions ofpoly(ADP-ribose) and PARPs. Nat Rev Mol Cell Biol 13(7):411–424.

34. Williamson CT, et al. (2010) ATM deficiency sensitizes mantle cell lymphoma cells topoly(ADP-ribose) polymerase-1 inhibitors. Mol Cancer Ther 9(2):347–357.

35. Haince JF, et al. (2007) Ataxia telangiectasia mutated (ATM) signaling network ismodulated by a novel poly(ADP-ribose)-dependent pathway in the early response toDNA-damaging agents. J Biol Chem 282(22):16441–16453.

36. Ayoub N, Jeyasekharan AD, Bernal JA, Venkitaraman AR (2008) HP1-beta mobiliza-tion promotes chromatin changes that initiate the DNA damage response. Nature453(7195):682–686.

37. Tamaru H (2010) Confining euchromatin/heterochromatin territory: Jumonji crossesthe line. Genes Dev 24(14):1465–1478.

38. Hathaway NA, et al. (2012) Dynamics and memory of heterochromatin in living cells.Cell 149(7):1447–1460.

39. Cheutin T, et al. (2003) Maintenance of stable heterochromatin domains by dynamicHP1 binding. Science 299(5607):721–725.

40. Ahel D, et al. (2009) Poly(ADP-ribose)-dependent regulation of DNA repair by thechromatin remodeling enzyme ALC1. Science 325(5945):1240–1243.

41. Xu C, Xu Y, Gursoy-Yuzugullu O, Price BD (2012) The histone variant macroH2A1.1is recruited to DSBs through a mechanism involving PARP1. FEBS Lett 586(21):3920–3925.

42. Timinszky G, et al. (2009) A macrodomain-containing histone rearranges chromatinupon sensing PARP1 activation. Nat Struct Mol Biol 16(9):923–929.

43. Polo SE, Kaidi A, Baskcomb L, Galanty Y, Jackson SP (2010) Regulation of DNA-damage responses and cell-cycle progression by the chromatin remodelling factorCHD4. EMBO J 29(18):3130–3139.

44. Chou DM, et al. (2010) A chromatin localization screen reveals poly (ADP ribose)-regulated recruitment of the repressive polycomb and NuRD complexes to sites ofDNA damage. Proc Natl Acad Sci USA 107(43):18475–18480.

45. Smeenk G, et al. (2010) The NuRD chromatin-remodeling complex regulates signalingand repair of DNA damage. J Cell Biol 190(5):741–749.

46. Miller KM, et al. (2010) Human HDAC1 and HDAC2 function in the DNA-damage responseto promote DNA nonhomologous end-joining. Nat Struct Mol Biol 17(9):1144–1151.

47. Goodarzi AA, Kurka T, Jeggo PA (2011) KAP-1 phosphorylation regulates CHD3 nu-cleosome remodeling during the DNA double-strand break response. Nat StructMol Biol 18(7):831–839.

9174 | www.pnas.org/cgi/doi/10.1073/pnas.1403565111 Ayrapetov et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

2, 2

021