Developmental propagation of V(D)J recombination-associated DNA breaks and translocationsin mature B cells via dicentric chromosomesJiazhi Hu1, Suprawee Tepsuporn1, Robin M. Meyers, Monica Gostissa2,3, and Frederick W. Alt2

Howard Hughes Medical Institute, Program in Cellular and Molecular Medicine, Boston Children’s Hospital, and Department of Genetics, Harvard MedicalSchool, Boston, MA 02115

Contributed by Frederick W. Alt, June 2, 2014 (sent for review May 15, 2014)

Mature IgM+ B-cell lymphomas that arise in certain ataxia telangi-ectasia-mutated (ATM)-deficient compound mutant mice harbortranslocations that fuse V(D)J recombination-initiated IgH double-strand breaks (DSBs) on chromosome 12 to sequences downstreamof c-myc on chromosome 15, generating dicentric chromosomesand c-myc amplification via a breakage-fusion-bridge mechanism.As V(D)J recombination DSBs occur in developing progenitor B cellsin the bone marrow, we sought to elucidate a mechanism by whichsuch DSBs contribute to oncogenic translocations/amplificationsin mature B cells. For this purpose, we applied high-throughputgenome-wide translocation sequencing to study the fate of intro-duced c-myc DSBs in splenic IgM+ B cells stimulated for activation-induced cytidine deaminase (AID)-dependent IgH class switch re-combination (CSR). We found frequent translocations of c-mycDSBs to AID-initiated DSBs in IgH switch regions in wild-type andATM-deficient B cells. However, c-myc also translocated frequentlyto newly generated DSBs within a 35-Mb region downstream of IgHin ATM-deficient, but not wild-type, CSR-activated B cells. Moreover,we found such DSBs and translocations in activated B cells that did notexpress AID or undergo CSR. Our findings indicate that ATM deficiencyleads to formation of chromosome 12 dicentrics via recombination-activating gene-initiated IgH DSBs in progenitor B cells and that thesedicentrics can be propagated developmentally into mature B cellswhere they generate new DSBs downstream of IgH via breakage-fusion-bridge cycles. We propose that dicentrics formed by joiningV(D)J recombination–associated IgH DSBs to DSBs downstream ofc-myc in ATM-deficient B lineage cells similarly contribute to c-mycamplification and mature B-cell lymphomas.

B-lineage lymphomas are frequently characterized by recur-rent chromosomal translocations (1–3), such as the T8;14

translocation in human Burkitt’s lymphoma that results in jux-taposition of the c-myc oncogene on chromosome 8 to the Ig heavychain (IgH) locus on chromosome 14, leading to c-myc deregulationby putting it under the control of the powerful IgH transcriptionalenhancers (1, 4). Translocations generally result from the joiningof double-strand breaks (DSBs) on two separate chromosomes(5, 6). Both in humans and in mice, translocations that are foundin tumors arising from early developing B cells usually involveIgH DSBs initiated by recombination-activating gene (RAG) 1and 2 endonuclease during the V(D)J recombination process thatassembles IgH variable region exons from component VH, DH, andJH gene segments (1, 5, 7). Likewise, most oncogenic translocationsin human and mouse mature B-cell tumors involve activation-in-duced cytidine deaminase (AID)-initiated DSBs that occur withinlong repetitive switch (S) regions in the CH portion of IgH duringIgH class switch recombination (CSR) in mature, peripheral B cells(1, 5, 8). RAG-initiated DSBs have also been implicated in IgHlocus translocations found in peripheral B-cell lymphomas (9).DSBs at partner oncogene loci that participate in translocationwith IgH can also be caused by AID and RAG (8, 10, 11), as well asby other, more general mechanisms (1, 5, 11).DNA DSBs, including those initiated by RAG and AID, lead to

activation of the ataxia telangiectasia-mutated (ATM)–dependent

DNA DSB response (12). To effect the DSB response, the ATMkinase phosphorylates downstream factors that contribute to acti-vation of cellular checkpoints and to DSB repair (13, 14). In thiscontext, ATM phosphorylation of the p53 tumor suppressor acti-vates the G1 checkpoint, allowing cells to repair DSBs beforeprogressing into the replicative stage of the cell cycle or, alterna-tively, signaling apoptotic death of cells with persistent DSBs (14).ATM also activates a series of downstream substrates that con-tribute to the actual repair of the DSB, likely in part by tetheringthe broken DNA ends (5, 7, 15). ATM is required for stabilizingRAG-initiated DSB intermediates during V(D)J recombination(16). In the absence of ATM, V(D)J recombination is not abro-gated, but a fraction of RAG-dependent DSBs are not properlyjoined and can persist and lead to chromosomal translocations(16, 17). Similarly, IgH CSR is not abrogated by ATM defi-ciency but is moderately impaired, with a fraction of AID-initiatedS-region DSBs not being joined and leading again to chromosomalbreaks and translocations in ATM-deficient B cells activated forCSR (18, 19).

Significance

Antibody production depends on a cut-and-paste genomicrearrangement termed “V(D)J recombination” that takes placeduring early B-lymphocyte development. Mistakes in V(D)J re-combination can lead to chromosomal translocations that acti-vate oncogenes. Such mistakes usually lead to immature B-cellcancers. However, in the absence of the ATM kinase, mice candevelop mature B-cell tumors with translocations resulting fromV(D)J recombination-associated breaks. Normally persistentchromosome breaks activate cellular checkpoints that eliminatecells harboring such dangerous lesions. The current studies re-veal that, in the absence of ATM, V(D)J recombination-generatedbreaks are cycled into aberrant chromosomes, termed “di-centrics,” that avoid checkpoints and are propagated throughdevelopment, generating new breaks and translocations inmature B cells.

Author contributions: J.H., S.T., M.G., and F.W.A. designed research; J.H., S.T., and M.G.performed research; J.H., R.M.M., and M.G. contributed new reagents/analytic tools; J.H.,S.T., R.M.M., M.G., and F.W.A. analyzed data; and J.H., S.T., M.G., and F.W.A. wrotethe paper.

The authors declare no conflict of interest.

Freely available online through the PNAS open access option.

Data deposition: The sequence reported in this paper has been deposited in the GeneExpression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no.GSE58599).1J.H. and S.T. contributed equally to this article.2To whom correspondence may be addressed. E-mail: alt@enders.tch.harvard.edu ormonica.gostissa2@childrens.harvard.edu.

3Present address: 121 Bio LLC, Cambridge, MA 02139.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1410112111/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1410112111 PNAS | July 15, 2014 | vol. 111 | no. 28 | 10269–10274

GEN

ETICS

Dow

nloa

ded

by g

uest

on

Oct

ober

1, 2

020

Both V(D)J recombination- and IgH CSR-associated DSBs,which occur in the G1 cell-cycle phase, are joined by the classicalnonhomologous DNA end-joining pathway (C-NHEJ), one ofthe major forms of cellular DSB repair (20, 21). In the absence ofC-NHEJ, V(D)J recombination DSBs are not joined in pro-genitor (pro)-B cells, and a substantial number of CSR DSBs arenot joined in CSR-activated mature B cells, leading to chromo-some breaks and translocations (5). However, C-NHEJ–deficientmice are not prone to lymphoma development because pro-liferation of cells bearing persistent DSBs or oncogenic trans-locations is prevented by the p53-dependent cell-cycle checkpoint(22, 23). Indeed, C-NHEJ/p53 dual deficiency in mice leads toinevitable pro–B-cell lymphomas carrying translocations that joinRAG-initiated DSBs at the IgH JH locus to DSBs downstream ofc-myc, leading to c-myc amplification via a breakage-fusion-bridge(BFB) mechanism (24, 25). Mechanistically, it appears that C-NHEJdeficiency leads to persistent RAG-initiated JH breaks that, inthe absence of p53 are replicated, and thereby can be convertedinto dicentric chromosomes, which promote the downstreamtranslocation/amplification process (24, 25). Notably, C-NHEJ/p53double deficiency in murine mature B cells leads to peripheralB-cell lymphomas that harbor IgH/c-myc translocations in whichS-region DSBs are fused to sequences upstream of c-myc, lead-ing to c-myc overexpression by juxtaposition to the IgH 3′ regu-latory region (26, 27).In humans, ATM deficiency leads to both T- and B-cell lym-

phoma. In mice, however, ATM deficiency alone does not leadto B-cell lymphoma but does promote recurrent thymic lym-phomas (28, 29). These tumors frequently harbor complex chro-mosome 14 translocations that are generated via dicentrics thatarise during TCRδ locus V(D)J recombination, often from DδDSBs, and that lead to amplification of genes upstream of TCRδ(30). ATM-deficient thymic lymphomas also harbor mechanisti-cally related T(12;14) translocations that fuse TCRδ breaks tochromosome 12 sequences over a large region downstream ofIgH, leading to deletion of potential tumor suppressors, includingBcl11b (30, 31). Human T-ALLs also frequently harbor trans-locations initiated from Dδ (10). Notably, the translocations inATM-deficient mouse thymic lymphomas appear to arise viamechanisms reminiscent of those found for C-NHEJ/p53–deficientpro–B-cell lymphomas. This mechanistic overlap likely occurs be-cause ATM deficiency alone produces defects similar to thosecaused by combined C-NHEJ and p53 deficiency, including aDSB repair defect, albeit not as severe as that of C-NHEJ de-ficiency, and a G1 checkpoint defect (16, 17). In this regard, inATM-deficient mice, RAG-dependent IgH locus breaks that aregenerated in pro-B cells can persist through development and bemanifested in mature B cells as chromosome 12 centromericfragments that end within IgH or within downstream regions andthat lack upstream (telomeric) sequences (17). These chromo-some fragments, which can participate in translocations in ma-ture B cells, have been proposed to often lose IgH and variableamounts of downstream sequences due to nuclease-mediatedend erosion in the absence of protective telomeres (17).Given the potential for both unrepaired V(D)J recombination-

associated and CSR-associated DSBs, along with G1 checkpointdefects in ATM-deficient B cells, an intriguing question is whyATM-deficient mice do not develop peripheral B-cell lympho-mas. In this regard, we recently generated ATM-deficient mousemodels that routinely develop peripheral mature B-cell lympho-mas and found that they routinely harbor translocations mecha-nistically related to those of C-NHEJ/p53 pro–B-cell lymphomas,in which RAG-initiated IgH DSBs are joined to sequences down-stream of c-myc, leading to dicentrics and c-myc amplification viaa BFB mechanism (32). To elucidate how RAG-initiated DSBscould lead to dicentrics and c-myc amplification in peripheral B-celltumors, we used our high-throughput genome-wide translocation se-quencing (HTGTS) approach to assay for sequences that translocate

to DSBs within the c-myc gene in wild-type (WT) and ATM-deficient B cells. Our findings point to a mechanism in whichRAG-initiated DSBs in ATM-deficient pro-B cells are propagatedvia dicentric chromosomes and through BFB cycles give rise tonew DSBs downstream of IgH in mature B cells.

ResultsATM Suppresses Genome-Wide Translocations in Stimulated B Cells.To elucidate potential mechanisms by which ATM deficiencypromotes peripheral IgM+ B-cell lymphomas with c-myc ampli-fication that results from dicentric translocations involving ap-parent V(D)J recombination-associated IgH DSBs (32), we usedHTGTS to assess genome-wide translocation patterns from DSBsintroduced into the c-myc gene in WT and ATM-deficient matureIgM+ B cells activated in culture by treatment with α-CD40 plusIL4 to induce IgH class switching from IgM to IgG1 and IgE. Forthis purpose, we used B cells that harbor a cassette with 25 con-secutive I-SceI recognition sites (referred to as the 25xI-SceI cas-sette) within c-myc intron 1 (33). Upon expression of I-SceI duringB-cell activation, I-SceI–specific DSBs generated in the 25xI-SceIc-myc cassette can be used as HTGTS “bait” to capture endoge-nous prey DSBs to which they translocate genome-wide (33). Forthese studies, we used primers located about 100 bp centromericto the 25xI-SceI cassette to isolate translocation junctions in-volving the 5′ broken end (BE) of the 25xI-SceI cassette DSBs (5′I-SceI-BEs). Junctions of 5′ BEs are considered in the (+) ori-entation if prey sequences are read from the junction in a cen-tromere-to-telomere direction and the (−) orientation if read inthe opposite direction (33). In all HTGTS experiments described,we performed at least three independent repeats for each geno-type under each given experimental condition. As all repeats yieldedthe same major findings, we pooled results from each set of ex-perimental repeats for presentation. For ready visualization of ge-nome-wide translocation patterns on different chromosomes, wepresent HTGTS junctions as a dot plot in which each chromosomeis divided into 2-Mb bins and the number of junctions within eachbin is represented by various numbers of color-coded dots (33)(Fig. 1 and Figs. S1A, S2, and S3A).Consistent with prior results (33), a large fraction (72%) of 5′

I-SceI-BE HTGTS junctions from CSR-activated WT B cellsmaps within 20 kb of the I-SceI break-site within c-myc onchromosome 15, with the majority of these representing rejoin-ing of single I-SceI DSBs following resection (Fig. 1A and Fig. S1A and B). HTGTS libraries from ATM-deficient B cells had alower fraction of junctions than WT within this immediate break-site region (51%) and, correspondingly, a higher percentage ofjunctions in the form of translocations along the length of thebreak-site chromosome (17 vs. 11% in WT) and to other chro-mosomes (31 vs. 17% in WT) (Fig. S1B). The increased levelsof intrachromosomal and interchromosomal translocation ofbait DSBs in ATM-deficient B cells is consistent with aberrantresolution and/or prolonged persistence of DSBs promoting theirtranslocation.Translocations to known AID-dependent hotspots (33, 34)

and cryptic I-SceI sites (33) were only modestly increased or, insome cases, decreased in ATM-deficient vs. WT cells (Fig. 2),perhaps due to competition for joining from increased levels ofother persistent DSBs within given chromosomes (5, 35). In thiscontext, the major translocation hotspots in both WT and ATM-deficient–activated B cells were the Sμ, Sγ1, and Se regions (Fig.2 A and B), which are the primary targets for AID-initiated DSBsin α-CD40/IL4–stimulated B cells. Notably, however, the relativelevel of the Sμ and Se region translocation hotspots was slightlydiminished by ATM deficiency (Fig. 2B), again potentially be-cause of increased levels of other unresolved DSBs that competefor translocation of c-myc bait DSBs in ATM-deficient cells.Strikingly, although in WT B cells translocations dropped tobackground levels just downstream of IgH S region hotspots, in

10270 | www.pnas.org/cgi/doi/10.1073/pnas.1410112111 Hu et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

1, 2

020

ATM-deficient B cells a substantial number of translocations, farabove background levels, were observed over a 35-Mb regiondownstream (centromeric) of IgH on chromosome 12 (Fig. 1Aand Fig. S1A, indicated by red boxes).

De Novo DSBs Over a 35-Mb Region Downstream of IgH Provide c-mycTranslocation Targets in Activated Splenic B Cells. To determinewhether the high level of translocation junctions found down-stream of the IgH locus in ATM-deficient vs. WT-activated Bcells resulted from generation of new DSBs vs. potential erosionof developmentally persistent chromosome 12 centromeric frag-ments resulting from IgH DSBs as proposed previously (17), weassayed the frequency of junctions in the plus and minus ori-entations. The erosion model of developmental DSB persistence,in its most simple form (Discussion), predicts occurrence of DSBsonly in the minus orientation (17) (Fig. 3A) whereas newly gen-erated DSBs provide both ends for joining and thus result injunctions occurring at similar frequency over the region in bothplus and minus orientations (33) (Fig. 3B). Strikingly, we foundthat the dramatically increased junctions across this large regiondownstream of IgH in activated ATM-deficient B cells vs. WT Bcells were evenly distributed in the plus and minus orientations(Fig. 3C, Top and Middle), demonstrating that they involved denovo prey DSBs introduced during CSR activation of the ATM-deficient B cells. Notably, although DSBs were greatly increasedacross the region, we did not identify any focal hotspots.

There is no obvious mechanism by which off-target activities ofAID or RAG could be imagined to generate DSBs over sucha large and specific chromosomal region in the absence of ATM.However, we previously found that ATM-deficient B cells activatedfor CSR generated substantial numbers of chromosome 12 dicentrictranslocations (18), suggesting that some newly generated DSBsdownstream of IgH in ATM-deficient B cells could arise de novoduring activation and cell division via breakage of chromosome12-based dicentrics generated via aberrant CSR. To test whetherthe downstream IgH DSBs could occur secondary to dicentricsthat arise during CSR activation, we prepared HTGTS librariesfrom WT B cells treated with the KU55933 ATM inhibitor (ATMi)during activation with α-CD40 plus IL4. This ATMi treatment in-deed resulted in increased accumulation of plus and minus HTGTSjunctions in the 35-Mb region downstream of IgH, showing that theDSBs that led to them did not arise from earlier developmentalevents, but rather arose from events that occurred in the course ofCSR activation. However, relative levels of these downstreamHTGTS junctions (DSBs) were lower than those of similarly ac-tivated ATM-deficient B cells, consistent with the possibility thatsome of the downstream DSBs were either generated or predis-posed by events that occurred before CSR activation of ATM-deficient B cells (Fig. 3C, Bottom, and Fig. 3D).

Frequent DSBs Downstream of IgH in Splenic ATM-Deficient B Cells DoNot Require CSR Activation or AID Induction. To determine whetheractivated ATM-deficient splenic B cells also were predisposed to

ATM

-/- +

-C

D40

/IL4

WT

+ -C

D40

/IL4

Cen.

Tel.

ATM

-/- +

RP

105

WT

+ R

P10

5

Cen.

Tel.

A

B

AID-dependent hotspot

cryptic I-SceI site

AID-dependent hotspot

cryptic I-SceI site

Fig. 1. ATM suppresses widespread translocations downstream of IgH in CSR-activated B cells. (A and B) Profiles of unique translocation junctions on selectedchromosomes in HTGTS libraries from α-CD40/IL4–stimulated (A) or RP105-stimulated (B) B cells from WT (shown on left of each chromosome diagram) orATM-deficient (shown on right of each chromosome diagram) mice, as indicated. Genome-wide translocation patterns (for all chromosomes) are shown inFigs. S1 and S2. Plots represent pooled data from at least three independent experiments, all of which gave similar results. Chromosomes are drawn withcentromere (Cen.) on top and telomere (Tel.) on the bottom. The blue box on chromosome 15 indicates the break-site. The red box on chromosome 12indicates the 35-Mb region centromeric to IgH that contains frequent translocation junctions in ATM-deficient activated B cells. Triangles indicate known AID-dependent hotspots (blue) (33 and 34) or cryptic I-SceI sites (green) (33). Numbers of translocation hits within each 2-Mb bin are represented by colored dots,as keyed in the box.

Hu et al. PNAS | July 15, 2014 | vol. 111 | no. 28 | 10271

GEN

ETICS

Dow

nloa

ded

by g

uest

on

Oct

ober

1, 2

020

generating AID-independent DSBs downstream of IgH that couldcontribute to dicentric translocations, we performed HTGTS onWT and ATM-deficient B cells during expansion in the presenceof RP105 (Fig. 1B and Fig. S2), which allows B cells to proliferateand survive in culture without expressing AID or undergoing CSR(17). Consistent with lack of AID expression, HTGTS librariesfrom RP105-treated WT and ATM-deficient B cells had no de-tectable genome-wide AID hotspots, with the few junctionsfound in S regions occurring at the very low levels observed inAID-deficient cells (33) (Fig. 2 A and B). However, HTGTS li-braries from RP105-treated WT and ATM-deficient cells didretain similar levels of cryptic I-SceI HTGTS hotspots as foundin libraries from α-CD40/IL4–activated cells (Fig. 2C), showingthat RP105 stimulation did not affect translocations more gen-erally. Notably, although libraries from RP105-stimulated WT Bcells had only background levels of HTGTS junctions in the 35-Mb region downstream of IgH, similar to those observed inα-CD40/IL4–stimulated WT B cells (Fig. 1B, Fig. 4A, and Fig.S2), RP105-stimulated, ATM-deficient B cells retained sub-stantial levels of plus-and-minus-orientation HTGTS junctionsover this 35-Mb region (Fig. 4A), with a distribution similar tothat of the DSBs that occurred during the course of CSR(compare Fig. 3C and 4A). To further verify increased HTGTSjunctions downstream of IgH derived from DSBs generated inATM-deficient B cells during early development rather thanduring activation in culture, we generated HTGTS libraries fromATMi-treated WT B cells stimulated with RP105. Under theseculture conditions, we observed significant levels of cell deathand a higher level of HTGTS junctions genome-wide, the latterlikely due to increased random DSBs and translocations in dyingcells (Fig. S3A). This “background” made it hard to normalize toother libraries; however, it still was quite clear that the 35-Mbregion downstream of IgH was not enriched in translocationjunctions under these stimulation conditions (Fig. S3B). Takentogether, these results indicate that a substantial proportionof DSBs in the IgH downstream region in activated ATM−/−

peripheral B cells originate from an AID-independent mechanism.

DiscussionA genetic alteration that increases DSB frequency around c-myccombined with an Eμ-Bcl2 transgene that enhances B-cell sur-vival leads to the development of mature B-cell lymphomas withnearly 100% penetrance in the context of ATM deficiency (32).Remarkably, all of these ATM-deficient mature B-cell lymphomasharbor oncogenic c-myc amplifications generated by a BFBmechanism from dicentric chromosomes formed through fusionof V(D)J recombination-associated IgH DSBs on chromosome12 to sequences downstream of c-myc on chromosome 15 (32).The finding of recurrent IgH/c-myc translocations/amplificationsin mature B-cell lymphomas via an apparently RAG-associatedmechanism was unanticipated and led to our current use ofHTGTS to identify potential mechanisms. Our HTGTS studiesrevealed that mature ATM-deficient B cells, when activated toproliferate in culture, frequently generate translocations betweenc-myc breaks and sequences spread over a 35-Mb region down-stream of IgH. Because these translocations are equally distributedin both plus and minus chromosomal orientations on chromosome12, they must arise from newly generated DSBs across this largeregion. As this phenomenon also is observed in B cells activated toproliferate in the absence of AID induction and CSR, we suggestthat ATM deficiency allows IgH V(D)J recombination DSBs indeveloping B cells to generate chromosome 12 dicentrics that, viaBFB cycles, persist to generate new DSBs and c-myc–activatingoncogenic translocations/amplifications that can contribute to thedevelopment of mature B-cell lymphomas (Fig. 3B and Fig. S4).The clustering of the BFB breaks closer to the initial dicentric breakpoint is notable and could have a number of conceivable explan-ations. To our knowledge, our system provides the first experi-mental approach to elucidate the basis of the central DSB featureof the long-known BFB mechanism (36).Prior cytogenetic studies of ATM-deficient B cells detected

large chromosome 12 centromeric fragments, in some cases con-taining downstream IgH sequences and in other cases lacking IgHsequences, and showed further that presence of such chromosome12 fragments was AID-independent but RAG2-dependent (17).The model proposed to explain these observations was that, in the

A

B C

Fig. 2. HTGTS junctions at cryptic I-SceI sites and AID-dependent hotspots α-CD40/IL4– and RP105–stimulated splenic B cells. (A) Linear plots showing HTGTSjunctions at IgH locus S regions in libraries from WT (Left) or ATM−/− (Right) B cells activated with either α-CD40/IL4 (blue) or RP105 (red). The y axis shows thenumber of hits in each 2.5-kb bin; note that different scales are used in the two plots. A diagram of the IgH locus with the position of S regions is shown on thebottom; S regions that are AID targets upon α-CD40/IL4 stimulation are indicated. (B and C) Average number of translocation junctions at AID-dependenthotspots (B) (33, 34) or cryptic I-SceI sites (C) (33) in HTGTS libraries from α-CD40/IL4–stimulated B cells (WT, blue triangles; ATM−/−, red triangles) or RP105-stimulated B cells (WT, blue squares; ATM−/−, red squares); data are shown as mean ± SD of three or four independent experiments.

10272 | www.pnas.org/cgi/doi/10.1073/pnas.1410112111 Hu et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

1, 2

020

absence of ATM, unrepaired RAG-generated IgH DSBs in pro-Bcells persist as large centromeric chromosome fragments to themature B-cell stage, but extensive erosion of the broken DNAends due to lack of telomeres over time leads to loss of sequencescentromeric to the IgH locus (17). We note, however, that ourcurrently proposed dicentric model is consistent with all of theearlier cytogenetic findings that led to the erosion model fordevelopmental DSB persistence in the absence of ATM. In ad-dition, although the ATM/p53 pathway is involved in checkpointactivation throughout the cell cycle (37, 38), cells lacking ATMmostly accumulate chromosome breaks, indicating a more prom-inent role of ATM in detection and repair of prereplicative DSBs(39, 40), with unrepaired DSBs and unprotected chromosomeends in S- and G2/M-activating ATM-independent postreplicativecheckpoints (36, 37). In this context, chromosomes containingtranslocation/amplifications generated via BFB are routinely“stabilized” by acquiring a new telomere (25, 41). Therefore,in ATM-deficient cells, which lack the G1 checkpoint, the BFB

mechanism that we propose would prevent activation of post-replicative checkpoints and thereby propagate DSBs during celldivision that contribute to translocations and gene amplifica-tions, as has been observed in C-NHEJ/p53 doubly deficient pro-B cells (24, 25). In light of our current findings, and given theabove considerations, we propose that a dicentric/BFB mechanismlikely contributes majorly to developmental propagation of RAG-generated DSBs.Our findings also may help explain the origin of recurrent chro-

mosomal aberrations in ATM-deficient mouse thymic lymphomas.These tumors frequently harbor complex chromosome 14 trans-locations involving dicentrics generated during TCRδ V(D)J re-combination that amplify genes upstream of TCRδ, as well as T(12;14) translocations that fuse TCRδ breaks on chromosome 14 tochromosome 12 sequences over a large region downstream of IgH,which may result in the deletion of putative tumor suppressors (27).RAG-initiated DSBs in the JH locus occur in developing T cells,and, in the absence of ATM, have been proposed to persist indeveloping and peripheral T cells as chromosome 12 centromericfragments generated via exonucleolytic processing of broken ends,similar to what is observed in ATM-deficient B cells (42). Again,however, we note that these findings are consistent with persistenceof JH breaks via the dicentric/BFB mechanism, which would providesubstrates for translocation to persistent RAG-initiated TCRδ breaksalso propagated via BFB cycles. Our findings may also be relevantto the mechanisms of translocations and amplifications observedin mantle cell lymphomas that have somatic ATM inactivation(32). In theory, in permissive backgrounds, DSBs that are prop-agated via dicentrics through development in other lineages maycontribute to cancers beyond those of the immune system.

Experimental ProceduresMouse Strains Used. Previously generated c-myc25xI-SceI mice (33) were bredinto the ATM-deficient background (43). All mice used were heterozygousfor modified alleles containing I-SceI cassettes. All animal experiments were

Fig. 3. Frequent DSBs within the 35-Mb region downstream of IgH in CSR-activated ATM-deficient B cells. (A and B) Two models to explain the accu-mulation of translocation junctions downstream of IgH. (A) Erosion of per-sistent IgH DSB ends results in c-myc/IgH junctions only in minus orientation(adapted from ref. 17). (B) c-myc 5′ BEs joining to de novo breaks within ordownstream of IgH form translocation junctions in both plus and minusorientations. Junction orientations are defined in the text. Chromosome 15is in green and chromosome 12 in blue. Centromere (Cen.), telomere (Tel.),and gene names are indicated. Red arrows indicate the position and di-rection of the HTGTS sequencing primer. (C) Distribution of translocationjunctions on the 45-Mb telomeric region of chromosome 12 in α-CD40/IL4–activated WT (Top), ATM-deficient (Middle), or ATMi-treated (Bottom) Bcells. Libraries were normalized based on the numbers of junctions at crypticI-SceI sites (Experimental Procedures). Schematic of the chromosome 12 re-gion is shown at the top. Red indicates plus-orientation junctions, and blueindicates minus-orientation junctions. The y axis shows the number of hits ineach 1.2-Mb bin. Numbers above the telomeric peaks indicate the totalHTGTS junctions within IgH S region. (D) Statistical analysis of the accumu-lation of junctions in the 35-Mb hotspot downstream of IgH from α-CD40/IL4–stimulated B cells of indicated backgrounds. Nlocal represents the numberof junctions in the 35-Mb hotspot; Nglobal hotspots represents the number ofjunctions within cryptic I-SceI sites in each library. Bar graphs show themean± SDof three or four independent experiments. *P < 0.05 and ***P < 0.001 by Stu-dent’s two-tailed t test.

A B

Fig. 4. Frequent DSBs downstream of IgH in activated ATM-deficient B cellsactivated under conditions that do not Induce CSR. (A) Distribution oftranslocation junctions on the 45-Mb telomeric region of chromosome 12 inRP105-stimulated WT (Upper) or ATM-deficient (Lower) B cells, depicted asdescribed in the legend to Fig. 3. Libraries were normalized based on thenumbers of junctions at cryptic I-SceI sites. Schematic of the chromosome 12region is shown at the top. (B) Statistical analysis of the accumulation ofjunctions in the 35-Mb hotspot downstream of IgH from RP105-stimulated Bcells of indicated backgrounds. Nlocal represents the number of junctions inthe 35-Mb hotspot; Nglobal hotposts represents the number of junctions withincryptic I-SceI sites in each library. Bar graphs show the mean ± SD of threeindependent experiments. **P < 0.01 by Student’s two-tailed t test.

Hu et al. PNAS | July 15, 2014 | vol. 111 | no. 28 | 10273

GEN

ETICS

Dow

nloa

ded

by g

uest

on

Oct

ober

1, 2

020

performed under protocols approved by the Institutional Animal Care andUse Committee of Boston Children’s Hospital.

Splenic B-Cell Purification, Activation in Culture, and Retroviral Transfection.Splenic B cells were isolated as previously described (9) and cultured in 15%(vol/vol) FBS-containing RPMI medium supplemented with α-CD40 (1 μg/mL;eBioscience) plus IL4 (20 ng/mL; R&D Systems) or RP105 (2 μg/mL; eBioscience).Retroviral infection with I-SceI-GFP–expressing retrovirus (33) was performedafter 24 h following α-CD40/IL4 stimulation, and cells were harvested at day 4.For RP105 stimulation, infection was performed after 48 h and cells were har-vested at day 5. For indicated experiments, ATM inhibitor KU55933 (10 μM;Tocris) was added at time of infection and maintained for the whole durationof the stimulation.

Generation of HTGTS Libraries. HTGTS libraries were prepared as previouslydescribed (33). Nucleotide sequences of junctions within some libraries weregenerated by 454 (Life Sciences) sequencing, and others were generated byMi-Seq (Illumina) sequencing. For each genotype and condition analyzed,libraries sequenced by different methods showed comparable patterns. Atleast three independent libraries were generated and analyzed for eachexperimental condition shown and shown to give very similar results foreach genotype and experimental condition. Thus, for presentation, theselibraries were pooled. Details of the library pools are as follows: α-CD40/IL4–stimulated WT B cells—four independent libraries sequenced via 454 (totaljunctions = 6,897/5,897/8,378/4,673); α-CD40/IL4–stimulated ATM-deficient Bcells—three independent libraries sequenced with 454 (total junctions =7,725/8,446/5,675); α-CD40/IL4–stimulated and ATMi-treated WT B cells—three independent libraries sequenced with 454 (total junctions = 11,528/

17,035/18,264); RP105-stimulated WT B cells—three independent librariessequenced with Mi-seq (total junctions = 6,304/7,342/10,658); and RP105-stimulated ATM-deficient B cells—three independent libraries sequencedwith Mi-seq (total junctions = 7,278/4,484/11,627). Five additional librariesfrom α-CD40/IL4–stimulated WT or ATM-deficient B cells were sequencedwith Mi-seq and were used only to compare the genome-wide patternsbetween the two sequencing methods, which were very similar.

Data Analysis and Normalization. Sequence data were aligned and analyzed asdescribed (33). For Mi-seq libraries, we further filtered the reads by removingjunctions with a >30-bp gap between bait and prey sequences. RP105stimulation resulted in loss of AID-dependent hotspots (Fig. 2B), but DSBs(hotspots) at cryptic I-SceI sites (33) were found to occur independently ofbackground and stimulation (Fig. 2C). Thus, for comparison, libraries gen-erated under different stimulation conditions were normalized based on thenumbers of junctions at cryptic I-SceI sites in each given library.

ACKNOWLEDGMENTS. We thank David Schatz, Barry Sleckman, MichaelLieber, Zhengfei Lu, and Andre Nussenzweig for helpful comments on thismanuscript. This work was supported by Grant P01CA109901 from theNational Cancer Institute of the National Institutes of Health, Grant AI076210from the National Institutes of Health, and a Leukemia and Lymphoma Societyof America Specialized Center of Research grant (to F.W.A.). F.W.A. is an In-vestigator of the Howard Hughes Medical Institute. M.G. was supported by anEleanor and Miles Shore/Boston Children’s Hospital Career Development Fellow-ship Award. J.H. is supported by the Robertson Foundation/Cancer Research In-stitute Irvington Fellowship.

1. Gostissa M, Alt FW, Chiarle R (2011) Mechanisms that promote and suppress chro-mosomal translocations in lymphocytes. Annu Rev Immunol 29:319–350.

2. Küppers R, Dalla-Favera R (2001) Mechanisms of chromosomal translocations in B celllymphomas. Oncogene 20(40):5580–5594.

3. Mitelman F, Johansson B, Mertens F (2007) The impact of translocations and genefusions on cancer causation. Nat Rev Cancer 7(4):233–245.

4. Janz S (2006) Myc translocations in B cell and plasma cell neoplasms. DNA Repair(Amst) 5(9-10):1213–1224.

5. Alt FW, Zhang Y, Meng F-LL, Guo C, Schwer B (2013) Mechanisms of programmedDNA lesions and genomic instability in the immune system. Cell 152(3):417–429.

6. Zhang Y, et al. (2010) The role of mechanistic factors in promoting chromosomaltranslocations found in lymphoid and other cancers. Adv Immunol 106:93–133.

7. Nussenzweig A, Nussenzweig MC (2010) Origin of chromosomal translocations inlymphoid cancer. Cell 141(1):27–38.

8. Robbiani DF, Nussenzweig MC (2013) Chromosome translocation, B cell lymphoma,and activation-induced cytidine deaminase. Annu Rev Pathol 8:79–103.

9. Wang JH, et al. (2009) Mechanisms promoting translocations in editing and switchingperipheral B cells. Nature 460(7252):231–236.

10. Larmonie NS, et al. (2013) Breakpoint sites disclose the role of the V(D)J re-combination machinery in the formation of T-cell receptor (TCR) and non-TCR asso-ciated aberrations in T-cell acute lymphoblastic leukemia. Haematologica 98(8):1173–1184.

11. Tsai AG, Lieber MR (2010) Mechanisms of chromosomal rearrangement in the humangenome. BMC Genomics 11(Suppl 1):S1.

12. Helmink BA, Sleckman BP (2012) The response to and repair of RAG-mediated DNAdouble-strand breaks. Annu Rev Immunol 30:175–202.

13. Bednarski JJ, Sleckman BP (2012) Lymphocyte development: Integration of DNAdamage response signaling. Adv Immunol 116:175–204.

14. Shiloh Y, Ziv Y (2013) The ATM protein kinase: Regulating the cellular response togenotoxic stress, and more. Nat Rev Mol Cell Biol 14(4):197–210.

15. Bassing CH, Alt FW (2004) H2AX may function as an anchor to hold broken chro-mosomal DNA ends in close proximity. Cell Cycle 3(2):149–153.

16. Bredemeyer AL, et al. (2006) ATM stabilizes DNA double-strand-break complexesduring V(D)J recombination. Nature 442(7101):466–470.

17. Callén E, et al. (2007) ATM prevents the persistence and propagation of chromosomebreaks in lymphocytes. Cell 130(1):63–75.

18. Franco S, et al. (2006) H2AX prevents DNA breaks from progressing to chromosomebreaks and translocations. Mol Cell 21(2):201–214.

19. Ramiro AR, et al. (2006) Role of genomic instability and p53 in AID-induced c-myc-Ightranslocations. Nature 440(7080):105–109.

20. Boboila C, Alt FW, Schwer B (2012) Classical and alternative end-joining pathways forrepair of lymphocyte-specific and general DNA double-strand breaks. Adv Immunol116:1–49.

21. Lieber MR (2010) The mechanism of double-strand DNA break repair by the non-homologous DNA end-joining pathway. Annu Rev Biochem 79:181–211.

22. Gao Y, et al. (2000) Interplay of p53 and DNA-repair protein XRCC4 in tumorigenesis,genomic stability and development. Nature 404(6780):897–900.

23. Difilippantonio MJ, et al. (2000) DNA repair protein Ku80 suppresses chromosomalaberrations and malignant transformation. Nature 404(6777):510–514.

24. Difilippantonio MJ, et al. (2002) Evidence for replicative repair of DNA double-strandbreaks leading to oncogenic translocation and gene amplification. J Exp Med 196(4):469–480.

25. Zhu C, et al. (2002) Unrepaired DNA breaks in p53-deficient cells lead to oncogenicgene amplification subsequent to translocations. Cell 109(7):811–821.

26. Gostissa M, et al. (2009) Long-range oncogenic activation of Igh-c-myc translocationsby the Igh 3′ regulatory region. Nature 462(7274):803–807.

27. Wang JH, et al. (2008) Oncogenic transformation in the absence of Xrcc4 targetsperipheral B cells that have undergone editing and switching. J Exp Med 205(13):3079–3090.

28. Lavin MF (2013) The appropriateness of the mouse model for ataxia-telangiectasia:Neurological defects but no neurodegeneration. DNA Repair (Amst) 12(8):612–619.

29. Perlman S, Becker-Catania S, Gatti RA (2003) Ataxia-telangiectasia: Diagnosis andtreatment. Semin Pediatr Neurol 10(3):173–182.

30. Zha S, et al. (2010) ATM-deficient thymic lymphoma is associated with aberrant tcrdrearrangement and gene amplification. J Exp Med 207(7):1369–1380.

31. Gutierrez A, et al. (2011) The BCL11B tumor suppressor is mutated across the majormolecular subtypes of T-cell acute lymphoblastic leukemia. Blood 118(15):4169–4173.

32. Tepsuporn S, et al. (2014) Mechanisms that can promote peripheral B cell lymphomain ATM-deficient mice. Cancer Immunol Res, 10.1158/2326-6066.CIR-14-0090.

33. Chiarle R, et al. (2011) Genome-wide translocation sequencing reveals mechanisms ofchromosome breaks and rearrangements in B cells. Cell 147(1):107–119.

34. Klein IA, et al. (2011) Translocation-capture sequencing reveals the extent and natureof chromosomal rearrangements in B lymphocytes. Cell 147(1):95–106.

35. Zhang Y, et al. (2012) Spatial organization of the mouse genome and its role in re-current chromosomal translocations. Cell 148(5):908–921.

36. McClintock B (1941) The stability of broken ends of chromosomes in Zea mays. Ge-netics 26(2):234–282.

37. Deckbar D, Jeggo PA, Löbrich M (2011) Understanding the limitations of radiation-induced cell cycle checkpoints. Crit Rev Biochem Mol Biol 46(4):271–283.

38. De Lange T (2005) Telomere-related genome instability in cancer. Cold Spring HarbSymp Quant Biol 70:197–204.

39. Franco S, Alt FW, Manis JP (2006) Pathways that suppress programmed DNA breaks fromprogressing to chromosomal breaks and translocations. DNA Repair (Amst) 5(9-10):1030–1041.

40. Zha S, Sekiguchi J, Brush JW, Bassing CH, Alt FW (2008) Complementary functions ofATM and H2AX in development and suppression of genomic instability. Proc NatlAcad Sci USA 105(27):9302–9306.

41. Murnane JP (2012) Telomere dysfunction and chromosome instability. Mutat Res730(1-2):28–36.

42. Callén E, et al. (2009) Chimeric IgH-TCRalpha/delta translocations in T lymphocytesmediated by RAG. Cell Cycle 8(15):2408–2412.

43. Borghesani PR, et al. (2000) Abnormal development of Purkinje cells and lymphocytesin Atm mutant mice. Proc Natl Acad Sci USA 97(7):3336–3341.

10274 | www.pnas.org/cgi/doi/10.1073/pnas.1410112111 Hu et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

1, 2

020