Double-strand breaks in faculta3ve heterochroma3n require speciﬁc movements and chroma3n changes for eﬃcient repair

DNA double-strand breaks (DSBs) must be properly repaired within diverse chromaUn domains to maintain genome stability. Whereas euchromaUn has an open structure and is associated with acUve transcripUon, facultaUve heterochromaUn is essenUal to silence developmental genes and forms compact nuclear condensates, called polycomb bodies. Whether the speciﬁc chromaUn properUes of facultaUve heterochromaUn require disUnct DSB repair mechanisms remains unknown. Here, we integrate single DSB systems in euchromaUn and facultaUve heterochromaUn in Drosophila melanogaster and ﬁnd that facultaUve heterochromaUc DSBs rapidly move outside polycomb bodies. These DSB movements coincide with a break-proximal reducUon in the canonical heterochromaUn mark histone H3 Lysine 27 trimethylaUon (H3K27me3). We demonstrate that DSB movement and loss of H3K27me3 at heterochromaUc DSBs both depend on the histone demethylase dUtx. Moreover, loss of dUtx speciﬁcally disrupts compleUon of homologous recombinaUon at heterochromaUc DSBs. We conclude that DSBs in facultaUve heterochromaUn require dUtx-mediated loss of H3K27me3 to promote DSB movement and repair.


INTRODUCTION
EukaryoUc cells are conUnuously exposed to factors that break or chemically alter DNA.One parUcularly dangerous type of DNA damage is a double-strand break (DSB), in which both strands of the DNA helix are severed.Improper repair of DSBs can directly lead to inserUons, deleUons and chromosomal rearrangements associated with disease development including cancer 1 .To overcome these detrimental outcomes, cells have evolved mechanisms to repair DSBs of which the two main pathways are Non-Homologous End Joining (NHEJ) and Homologous RecombinaUon (HR).During NHEJ, the severed DNA ends undergo limited end processing and are directly ligated, which can result in small inserUons or deleUons (indels) at the break site and is therefore considered error-prone 2 .HR, on the other hand, is usually more precise since it relies on a homologous template to repair the DSB.During HR, the broken DNA ends undergo 5' to 3' end resecUon, generaUng 3' single-stranded DNA (ssDNA) overhangs.This overhang invades a homologous sequence on the sister chromaUd or the homologous chromosome, which serves as a repair template.The choice of DSB repair pathway depends on mulUple aspects, including cell cycle phase, the sequence context of the surrounding DNA, as well as the pre-exisUng chromaUn state [2][3][4] .
The eukaryoUc nucleus consists of a variety of chromaUn domains each characterized by specific molecular and biophysical properUes.Whereas euchromaUn has an open chromaUn structure with acUvely transcribed genes, heterochromaUn is more condensed and transcripUonally inacUve.One type of heterochromaUn is facultaUve heterochromaUn, which is essenUal to silence specific developmental genes.FacultaUve heterochromaUn can cover large genomic distances (e.g.developmental genes such as Hox genes) 5 , or regulatory regions (e.g.promoters) 6 .This type of heterochromaUn is enriched for Histone H3 Lysine 27 trimethylaUon (H3K27me3) and polycomb group (PcG) proteins, and accumulates in nuclear foci, called polycomb bodies [7][8][9] .Polycomb bodies cluster PcG-bound transcripUonally repressed genomic regions to maintain correct silencing of developmental genes [9][10][11][12][13] .Although the DSB response in open, euchromaUc regions has been extensively studied, the DSB repair response in facultaUve heterochromaUn remains largely unknown.
In the past decade, it has become clear that the pre-exisUng chromaUn state can directly influence the DSB repair response.For example, DSBs in acUvely transcribed regions are prone to clustering and repair by HR 14,15 .Moreover, DSBs in centromeres 16 , nucleoli [17][18][19] and consUtuUve heterochromaUn domains 16,20- 22 have been found to move outside the respecUve domains to facilitate repair.Previous evidence suggests that DSB movements can also occur within the inacUve X chromosome 23 , which is a specific type of facultaUve heterochromaUn enriched for both H3K27me3 and H3K9me3 24 .IrradiaUon of female human fibroblasts resulted in the specific exclusion of DSB repair proteins outside the inacUve X chromosome, suggesUve of DSB movement 23 .Moreover, decompacUon of the inacUve X upon laser irradiaUon has also been observed 25 .Nevertheless, live imaging of individual DSBs to precisely monitor their dynamics within facultaUve heterochromaUn has never been performed.More importantly, the response of polycomb bodies to DSBs in a physiological, in vivo sefng, and whether this chromaUn environment facilitates parUcular movements of DSBs, remains unknown.
Various histone modificaUons have been idenUfied to play a role in DSB repair in euchromaUn 3,26 .Silencing histone modificaUons, including H3K27me3 27,28 and H3K9me2/3 29,30 have been described to be deposited at DSBs in euchromaUn, resulUng in local, transient heterochromaUnizaUon and transcripUonal silencing.To restart transcripUon aier DSB repair in euchromaUn, acUve removal of H3K27me3 by the mammalian histone demethylase UTX (Ubiquitously transcribed TetratricopepUde repeat on X chromosome) has been suggested to occur specifically in cancer cells, not healthy fibroblasts 31 .In contrast to the accumulaUon of silencing marks at euchromaUc DSBs, we previously idenUfied a loss of H3K9me2/3 at DSBs within Drosophila consUtuUve heterochromaUn 32,33 .These findings suggest that eu-and heterochromaUn regions require differenUal changes in silencing histone modificaUons to repair their DSBs.Whether specific H3K27-modifying acUviUes are needed to repair DSBs in H3K27me3-enriched facultaUve heterochromaUn domains remains untested.
Here, we study the dynamic DSB response in facultaUve heterochromaUn in vivo by integraUng inducible single DSB systems 22,34 in euchromaUn and facultaUve heterochromaUn regions in the fruit fly Drosophila melanogaster.Using high-resoluUon live imaging, we find that the majority of DSBs in polycomb bodies rapidly move outside these domains.Moreover, we find that facultaUve heterochromaUc DSBs specifically undergo a local decrease in the canonical heterochromaUn histone mark H3K27me3, which is mediated by the histone demethylase dUtx.Early steps of HR can occur efficiently within polycomb bodies and are independent of dUtx, while dUtx is required for subsequent DSB movement and compleUon of HR.Together, our results reveal that DSBs in facultaUve heterochromaUn move outside the compacted polycomb bodies to promote Umely repair by HR.

Development of a single double-strand break system in faculta8ve heterochroma8n
To study DSB repair in facultaUve heterochromaUn in detail in animal Ussue, and directly compare responses to euchromaUn, we generated a set of inducible single DSB systems in Drosophila.We integrated our previously established in vivo DR-white reporters 22,34 into three facultaUve heterochromaUn regions and two euchromaUn regions (Fig. 1A, B).Upon expression of I-SceI, DSBs are induced in the upstream white gene.DSB repair pathway usage can subsequently be determined by sequencing the resulUng repair products; HR with the downstream iwhite sequence will generate an intact upstream white gene, while NHEJ will generate small indels at the cut site (Fig. 1A).We performed ChIP-qPCR (ChromaUn Immuno-PrecipitaUon followed by quanUtaUve PCR (qPCR)) for the canonical facultaUve heterochromaUn histone modificaUon H3K27me3 and confirmed enrichment of H3K27me3 at the three DR-white integraUons in heterochromaUn when compared to the two euchromaUc inserUons (Fig. 1C).Moreover, internal controls revealed strong specificity of the anUbody used for H3K27me3 ChIP analysis (Fig. S1A).
To allow Umed DSB inducUon, we combined our DR-white systems with either a heat-shock inducible I-SceI transgene (hsp70.I-SceI, Fig. 1D) or an ecDHFR-I-SceI transgene, which depends on the ligand trimethoprim to stabilize the ecDHFR-I-SceI protein (Fig. 1E) 22 .We first tested the efficiency and inducibility of our DR-white systems by performing ChIP-qPCR for phosphorylated H2Av on Serine 137 (gH2Av, gH2AX in mammals), one of the earliest chromaUn markers of DSB inducUon 35 .Heat-shock inducible expression of I-SceI resulted in a local increase in gH2Av levels within six hours at both euchromaUc and heterochromaUc DR-white loci (Fig. 1F, G).Moreover, we find the appearance of single gH2Av foci in nuclei of imaginal discs six hours aier heat-shock inducible I-SceI expression in larvae containing either a eu-or heterochromaUc DR-white inserUon (16-18% of cells contained a single gH2Av focus compared to 3-5% in control cells) (Fig. S1B).These results suggest efficient DSB inducUon in both eu-and heterochromaUc loci.
To directly determine which DSB repair pathways play a role in facultaUve heterochromaUn, we performed Sanger sequencing followed by TIDE analysis 36 on the DR-white reporters upon I-SceI expression.Feeding trimethoprim throughout development (3-4 days) to DR-white larvae expressing ecDHFR-I-SceI (Fig. 1E) results in the appearance of repair products in both eu-and heterochromaUn (Fig. 1H).Repair rates vary slightly between integraUon sites (55-70%), possibly reflecUng differenUal cufng efficiency or repair Uming at the different sites.However, no apparent differences in the number of idenUfied repair products were found between heterochromaUn and euchromaUn, indicaUng that both domains undergo efficient DSB inducUon and repair.More importantly, repair pathway analyses revealed that I-SceI-induced DSBs in both eu-and heterochromaUn regions employ HR (17-26%) and NHEJ (74-83%) to a similar extent (Fig. 1I).In line with this, expression of hsp.I-SceI also resulted in the appearance of DSB repair products (7-19%) and similar HR and NHEJ repair percentages in eu-and heterochromaUn DR-white integraUons (Fig. S1C, D).Our heat-shock inducible system resulted in an overall lower number of repair products when compared to ecDHFR-I-SceI, likely reflecUng the shorter duraUon of I-SceI expression (24 hours in hsp.I-SceI compared to 3-4 days in ecDHFR-I-SceI system).To confirm the role of HR repair at facultaUve heterochromaUc DSBs, we depleted the end resecUon HR protein CtIP using RNAi (Fig. S1E) and analyzed DR-white repair products following hsp.I-SceI inducUon (Fig. S1F, G).As expected, loss of CtIP resulted in a decrease in HR repair products (3-11% compared to 23-33% in control) in both eu-and heterochromaUn, indicaUng that the idenUfied HR repair products indeed reflect end resecUon dependent HR repair.
Together, these data reveal that our systems efficiently induce DSBs with liole variaUon between euchromaUn and facultaUve heterochromaUn regions in repair products, suggesUng that both chromaUn regions undergo similar DSB repair efficiency and pathway choice.Our inducible single DSB system in vivo therefore allows us to perform detailed analyses of DSB repair in facultaUve heterochromaUn, and directly compare it to the DSB response in euchromaUn.

DSBs rapidly move outside polycomb bodies
Specific DSB spaUotemporal dynamics are associated with a variety of chromaUn domains, such as centromeres 16 , nucleoli [17][18][19] and consUtuUve heterochromaUn 16,20,22 .These dynamics include the movement of DSBs to the periphery of the respecUve domain 19,21 .FacultaUve heterochromaUn forms disUnct domains in the fly and mammalian nucleus, termed polycomb bodies [7][8][9] .We wished to determine whether the disUnct molecular-and biophysical-properUes of polycomb bodies [37][38][39][40] could impact DSB dynamics and promote movements similar to those previously idenUfied in other nuclear domains.To this end, we employed our DR-white systems to perform in vivo live imaging of single DSBs in facultaUve heterochromaUn (Fig. 2A) using fluorescently tagged Mu2 to visualize DSBs, and fluorescently tagged polyhomeoUc-proximal (ph-p) to visualize polycomb bodies (Fig. 2B).Mu2 is the Drosophila ortholog of mammalian MDC1, and directly binds gH2Av 41,42 , while ph-p is one of the four core subunits of the Drosophila PRC1 (Polycomb Repressive Complex 1) and is enriched in fly polycomb bodies 43,44 (Fig. 2B).Strikingly, our live imaging analyses revealed that the majority of Mu2 foci (60%) that appear within polycomb bodies move outside the domain within ten minutes (Fig. 2C, D, Fig. S2A), while euchromaUc DSBs that appear outside polycomb bodies remain outside unUl their disappearance (Fig. S2B).Importantly, these movements are not specific for I-SceI-induced DSBs, since inducing DSBs in larval Ussue using 5Gy gamma-radiaUon resulted in similar DSB dynamics in ph-p marked polycomb bodies (Fig. 2E, F).The majority of radiaUon-induced Mu2 foci that appeared within polycomb bodies moved outside this domain within ten minutes aier appearance, whereas euchromaUc DSBs are resolved outside polycomb bodies (Fig. S2C).Taken together, our live imaging data demonstrate for the first Ume that DSBs move outside polycomb bodies to conUnue repair.

Reduc8on in H3K27me3 at faculta8ve heterochroma8c DSBs is mediated by the histone demethylase dUtx
Considering the compact and silent state of facultaUve heterochromaUn, we hypothesized that local chromaUn changes could coincide with the specific DSB movements in polycomb bodies.To this end, we assessed the levels of the canonical facultaUve heterochromaUn histone modificaUon H3K27me3 by ChIP-qPCR at DSB sites (Fig. 1F, 3A).We observed a decrease in H3K27me3 (loss of 22-34%) at two of the three heterochromaUc DSB sites aier I-SceI inducUon, while H3K27me3 at euchromaUc DSB sites remained unchanged (Fig. 3A).Since 16-18% of cells obtain a single DSB six hours aier hsp.I-SceI inducUon (Fig. S1B), this decrease in H3K27me3 levels suggests a significant loss of H3K27me3 at individual heterochromaUc DSBs.To exclude that the observed reducUon in H3K27me3 is due to histone loss at the break site, we performed ChIP-qPCR for histone H3, which did not reveal any significant differences in histone H3 levels at euchromaUc and heterochromaUc DSB sites (Fig. 3B, Fig. S3A).The reducUon in H3K27me3 was observed in two of the three heterochromaUc integraUons, suggesUng that not all heterochromaUc DSBs induce evident loss of H3K27me3.A possibility is that our ChIP assay may not be sensiUve enough to pick up on subtle differences in H3K27me3 levels.Nevertheless, our data reveal that DSBs in facultaUve heterochromaUn are frequently accompanied by a local reducUon in H3K27me3.
We hypothesized that the reducUon in H3K27me3 levels at heterochromaUc DSBs could be mediated by a histone demethylase that acUvely removes the methyl groups from H3K27.In Drosophila, dUtx is the only protein described to demethylate H3K27me3 45 .To determine whether dUtx removes the methyl group at heterochromaUc DSB, we depleted dUtx using RNAi in 3 rd instar larvae (Fig. 3C) and assessed H3K27me3 levels at the DSB sites using ChIP-qPCR.Indeed, dUtx depleUon leads to retenUon of H3K27me3 at DSBs in heterochromaUn, whereas the H3K27me3 levels at euchromaUc DSBs remain unaffected upon DSB inducUon in the presence or absence of dUtx (Fig. 3D).Loss of dUtx did not alter the levels of the DSB marker gH2Av, indicaUng that the retenUon of H3K27me3 at DR-white sites upon dUtx depleUon is not due to inefficient cufng by I-SceI (Fig. S3B).These data suggest that dUtx mediates the removal of H3K27me3 at DSB sites specifically in facultaUve heterochromaUn.

DSB movement and HR repair in faculta8ve heterochroma8n depend on dUtx
We next hypothesized that the local loss of H3K27me3 at heterochromaUc DSBs could be required to promote DSB movement outside polycomb bodies.To test this, we turned to Drosophila cells in culture which allow for in-depth visualizaUon of repair processes in combinaUon with RNAi-mediated depleUons.We visualized the dynamics of the early HR protein ATR InteracUng Protein (ATRIP) within polycomb bodies (ph-p domains) in the presence or absence of dUtx using live imaging of 5Gy gamma-radiated Drosophila Kc cells (Fig. 4A).ATRIP binds to RPA-coated ssDNA overhangs, which are produced early in HR during 5' to 3' end resecUon of the DSB 46 .We find that upon irradiaUon of control cells, ATRIP foci appear inside polycomb bodies and move outside these domains within ten minutes (Fig. 4B, C, Fig. S4A), which recapitulates our findings on Mu2 foci dynamics at heterochromaUc DSBs in larval Ussues (Fig. 2C, D, F).InteresUngly, ATRIP gets recruited inside polycomb bodies upon irradiaUon in the absence of dUtx, suggesUng that end resecUon steps are independent of dUtx (Fig. 4D, E, Fig. S4B).However, dUtx depleUon does lead to a delay in the movement of ATRIP-coated DSBs outside polycomb bodies.In control cells, 77% of ATRIP foci move outside the polycomb bodies within 10 minutes, while in dUtx depleted cells only 53% move out within this Umeframe (Fig. 4C).In line with this, we find an increased accumulaUon of ATRIP foci within the ph-p domains (21% in control cells during the course of our imaging experiment, compared to 33% in dUtx-depleted cells) indicaUve of defects in DSB movement upon dUtx depleUon (Fig. 4E).Together, these results suggest that the early steps of HR (end resecUon, ATRIP recruitment) occur efficiently within polycomb bodies and are independent of H3K27me3 demethylaUon at the DSB site.However, dUtxmediated demethylaUon of H3K27me3 is required for the subsequent DSB movement outside the polycomb body.
To test whether this dUtx-dependent DSB movement is required for later repair steps in facultaUve heterochromaUn, we employed our in vivo reporter system, which allows the direct assessment of DSB repair pathway choice (HR/NHEJ) by sequencing repair products (Fig. 1A).Strikingly, loss of dUtx in larvae revealed a 39-52% relaUve reducUon in the proporUon of HR repair products at two of the three heterochromaUc DSB sites, while euchromaUc DSB repair products remained unchanged (Fig. 4F).This reducUon in HR repair was accompanied by an increase in NHEJ repair (Fig. S4C).These changes in DSB repair pathway choice are not due to indirect cell-cycle effects, since dUtx depleUon did not significantly affect cell cycle progression in the Fly-FUCCI system 47 (Fig. S4D, E).
Although dUtx depleUon did not affect relaUve HR levels at the fHet3 integraUon (Fig. 4F), we do find that loss of dUtx results in a reducUon in the total number of idenUfied repair products at the fHet3 integraUon, as well as at the fHet2 integraUon site (Fig. S4F).A reducUon in idenUfied repair products is indicaUve of defects or delays in DSB repair.InteresUngly, this reducUon in total idenUfied repair products is more evident when strictly assessing repair of the fHet3 region in wing disc Ussues, which have strong silencing of genes nearby the fHet3 integraUon site (e.g.hmx) 48 .This indicates that DSB sites with high H3K27me3 levels depend more heavily on dUtx for repair (Fig. S4G).Indeed, brain Ussues with high gene expression levels (low H3K27me3) nearby fHet3 do not show this reducUon in repair efficiency upon dUtx depleUon.Together, these results suggest that DSB repair regulaUon is defecUve at all heterochromaUc sites in the absence of dUtx.
Finally, although DSBs at the fHet2 locus did not induce H3K27me3 loss (Fig. 3A) this region showed a clear defect in HR in the absence of dUtx (Fig. 4F).This suggests that dUtx plays an important role in DSB repair at this integraUon site and we may not have been able to idenUfy dUtx-mediated loss of H3K27me3 levels at this site due to compensatory histone methylaUon acUviUes or relaUvely low sensiUvity of our ChIP assay.Altogether, our results reveal that early HR steps in facultaUve heterochromaUn can be performed in the absence of dUtx, and that dUtx is specifically required for DSB movement and compleUon of repair.
To assess the physiological role of dUtx in DSB repair in facultaUve heterochromaUn, we wished to determine whether development of flies mutant for dUtx depends on the presence of intact DNA damage signaling.To do so, we crossed heterozygous dUtx mutant flies with flies that contain a truncaUon mutaUon in Ataxia telangiectasia and Rad3 related (ATR, mei41).ATR is one of the earliest kinases that rapidly respond to DSB events 49 .Combining dUtx mutant flies with an ATR mutaUon reduces relaUve viability by 16%, indeed suggesUng that dUtx mutant flies depend on correct DNA damage repair signaling for their development (Fig. S4H).This result reveals the physiological role of dUtx and suggests that in the absence of proper facultaUve heterochromaUc DSB repair, flies depend on acUve DNA damage checkpoint signaling to maintain viability.

DISCUSSION
ChromaUn forms dynamic domains in the nucleus, each characterized by specific molecular properUes, which can directly influence the DSB response.However, how transcripUonally inacUve facultaUve heterochromaUn (i.e.polycomb chromaUn) influences DSB repair remains poorly understood.To address this quesUon and understand how eukaryoUc cells maintain the integrity of silenced developmental genes, we here integrated inducible single DSB systems in euchromaUn and facultaUve heterochromaUn in Drosophila Melanogaster.This allowed us to comprehensively study DSB repair in facultaUve heterochromaUn in animal Ussue for the first Ume.We find that DSBs in facultaUve heterochromaUn rapidly move outside polycomb bodies within minutes aier their appearance.This movement depends on the H3K27me3 demethylase dUtx.In line with this, we find evidence for dUtx-mediated loss of the silencing mark H3K27me3 near DSBs in facultaUve heterochromaUn.Our data further reveal that early steps of HR (i.e.end resecUon) can occur efficiently within polycomb bodies and are independent of dUtx, whereas dUtx is required to promote subsequent DSB movement and the compleUon of HR.Together, we propose a model in which resected DSBs in polycomb bodies are subjected to dUtx-mediated loss of the silencing mark H3K27me3, which in turn promotes DSB movement and Umely repair by homologous recombinaUon (Fig. 5).
Specific movements of DSBs have been idenUfied to occur in a variety of chromaUn compartments including centromeres 16 , nucleoli [17][18][19] and pericentromeric consUtuUve heterochromaUn 16,[20][21][22] .Movement of these DSBs has been suggested to promote binding of HR proteins 17,18 , as well as prevent aberrant recombinaUon between repeUUve sequences 20,21 .Here, we find for the first Ume that DSBs move outside polycomb bodies in vivo (Fig. 2).In contrast to centromeres, nucleoli and consUtuUve heterochromaUn, facultaUve heterochromaUn mainly contains unique sequences and is deprived of repeUUve sequences.We therefore propose that DSB movement in facultaUve heterochromaUn did not evolve to prevent aberrant recombinaUon, but rather reflects the necessity to create a repair-competent state, facilitaUng access to the DSB repair machinery (Fig. 5).
Our data suggest that the movement of DSBs outside polycomb bodies is directly regulated by the dUtx-mediated removal of H3K27me3 at the break site (Fig. 3, 4).H3K27me3 is required to recruit PcG proteins to enhance compacUon and maintain a silenced state [9][10][11][12][13] .Therefore, acUve removal of H3K27me3 at the DSB site by dUtx could directly lead to a local loss of PcG proteins.This can subsequently lead to changes in the molecular-and biophysical-properUes of the DSB locus, creaUng an environment disUnct from the surrounding polycomb body and the acUve expulsion or passive separaUon of the DSB from the polycomb body.In line with this hypothesis, we find that in the absence of dUtx, and the subsequent retenUon of H3K27me3, DSBs remain longer within the polycomb body (Fig. 4).These results reveal analogies with our previous findings at DSBs in consUtuUve heterochromaUn, where we found that loss of the silencing mark H3K9me3 at DSBs by the histone demethylase dKDM4A ensures DSB movement outside the consUtuUve heterochromaUn domain 32,33 .
Since we did not observe a complete inhibiUon of facultaUve heterochromaUc DSB movement upon dUtx depleUon, addiUonal processes are likely involved, such as DSB end processing or the recruitment of specific chromaUn-or repair-proteins.End resecUon as well as chromaUn proteins, including the cohesinand SMC5/6-complexes, drive DSB movement in other chromaUn domains 16,18,20 , suggesUng that addiUonal components could be driving movements in polycomb bodies.
Our findings indicate that dUtx-mediated demethylaUon of H3K27me3 at facultaUve heterochromaUc DSBs is important for repair pathway choice, since dUtx loss shiis the choice towards NHEJ, resulUng in decreased HR (Fig. 4).Considering that dUtx depleUon only affects DSB repair pathway choice in facultaUve heterochromaUn, not euchromaUn, these repair pathway changes are unlikely to be driven by indirect general defects in cell cycle progression (Fig. S4) or transcripUonal regulaUon of repair genes 50 .Therefore, we hypothesize that the HR/NHEJ repair pathway choice at facultaUve heterochromaUn could be directly regulated by dUtx through two non-mutually exclusive mechanisms: (1) defects in DSB movement, and (2) direct impact on binding of HR-or NHEJ-proteins at DSB sites.
In the first model, dUtx is promoUng HR by moving the DSB to a more HR-prone chromaUn state depleted of silencing marks.The movement might therefore specifically facilitate the access to 'late' HR repair proteins (e.g.Rad51 loading or helicases to resolve D-loops), usually excluded from the compact polycomb state.Moreover, moving an HR-proficient DSB away from the compact facultaUve heterochromaUn might provide the required chromaUn mobility necessary to perform homology search 51 .Indeed, we observe that loss of dUtx has no impact on the iniUal stages of HR, such as end resecUon and ATRIP loading, in polycomb bodies.However, dUtx loss does impede DSB movement, and subsequent later HR steps as evidenced by the decreased number of HR repair products idenUfied at facultaUve heterochromaUc DSB sites (Fig. 4).
In our second model, we propose that the decreased frequency of HR repair in the absence of dUtx is caused by changes in binding of repair proteins to histone modificaUons at the break site.It is possible that H3K27me3 is able to directly bind specific NHEJ proteins within the polycomb body.AlternaUvely, H3K7me1 or unmethylated H3K27 residues, as a result of dUtx-mediated H3K27me3 demethylaUon, could directly recruit proteins important for HR.In line with this hypothesis, previous work idenUfied that the HR-promoUng TONSL-MMS2L complex has a higher affinity for unmodified histones, generated following DNA replicaUon 52 .
Despite differences in dUtx-dependency for repair pathway choice, we find the frequencies of HR and NHEJ repair pathway usage in facultaUve heterochromaUn and euchromaUn to be similar in wild type Drosophila (Fig. 1, Fig. S1).These results are consistent with previous findings in which DSBs in H3K27me3enriched imprinted loci in mice did not differ in repair pathway usage when compared to the corresponding acUve allele 53 .In addiUon, both HR and NHEJ components are recruited to laser-damaged inacUve X chromosomes in female human cells 25 .In contrast, a recent study that used a sequencing-based reporter system in cancer cells to invesUgate the impact of chromaUn on CRISPR-Cas9-induced DSBs did reveal differences in repair pathway usage in H3K27me3-enriched regions 54 .The authors found a relaUve decrease in NHEJ and concurrently an increase in usage of Microhomology-Mediated End-Joining (MMEJ) within H3K27me3-enriched regions.MMEJ is an error-prone mechanism that relies on short-range end resecUon and uses homologous sequences to align the broken ends.In contrast to our results, these findings suggest that end resecUon-based repair pathways are preferred at facultaUve heterochromaUc DSBs.These different outcomes could indicate differences in repair pathway usage between species or could be due to differences in the approach used to induce DSB inducUon (CRISPR-Cas9 versus I-SceI).Moreover, heterochromaUn properUes in vivo may vary from that observed in cultured tumor cells, potenUally leading to disparate outcomes.
In conclusion, our work demonstrates that DSBs in facultaUve heterochromaUn require specific local chromaUn changes and DSB movements for their faithful repair in animal Ussue.Our results emphasize the importance of understanding how different chromaUn components influence DSB repair pathway choice and maintain genome stability across diverse chromaUn domains.FacultaUve heterochromaUn regions are oien associated with high mutaUonal loads in cancer 55 , indicaUng that these domains are parUcularly vulnerable to aberrant DNA damage repair.Moreover, the human homolog of dUtx, UTX, is oien mutated in cancer 56 .In the long-term, research into DNA damage repair in heterochromaUn will give insights into how misregulaUon of chromaUn proteins, such as UTX, could result in increased genome instability and specific mutaUonal signatures in cancer, ulUmately contribuUng to disease development.

Constructs
The DR-white construct was created previously 22 .ph-p was N-terminally tagged with mCherry and cloned in a pCasper5 vector for random p-element transformaUon in flies.For cell culture experiments, ph-p and ATRIP were cloned into pCopia vectors containing N-terminal mCherry or C-terminal eYFP epitope tags respecUvely.ph-p was cloned from the pFastBac plasmid (Addgene #1925), whereas ATRIP was cloned from cDNA generated from RNA extracted from wild type flies.

Fly lines
Flies were grown at room temperature on standard medium, except otherwise specified.Embryo injecUon and generaUon of new DR-white and ph-p-mCherry fly lines were performed by BestGene, Inc. DR-white aoB containing plasmids were integrated in Minos-mediated integraUon casseoe (MiMIC) integraUon sites as described previously 22 .FacultaUve heterochromaUn integraUon sites were selected based on high H3K27me3 levels in OregonR flies (modEncode) and near a gene known to be regulated by H3K27me3 and/or PcG proteins 57 .EuchromaUn integraUon sites were selected based on low H3K27me3 and low H3K9me2/3 levels.To create ph-p-mCherry fly lines, pCasper5-ph-p-mCherry plasmid containing the copia promoter and P-element transposons was injected in embryos of w1118 flies by Bestgene (Chino Hills, CA, USA).To induce knockdown of either CtIP or dUtx, flies containing Gal4 driven by an AcUn5C promoter were crossed with UAS-CtIP RNAi or UAS-dUtx RNAi.A list of MiMIC integraUon sites to generate the DRwhite fly lines as well as all fly lines used can be found in Supplemental Table S1.

DR-white repair product analysis
QuanUficaUon of somaUc repair products in DR-white, I-SceI larvae was performed as described previously 22 .In short, either hsp.I-SceI was expressed using heat-shock in third instar DR-white/hsp.I-SceI larvae, 24 hours before single larvae were collected.AlternaUvely, ecDHFR-I-SceI was expressed by feeding DRwhite/ecDHFR-I-SceI 80μM trimethoprim (Sigma) throughout development (3-4 days), thereby stabilizing the ecDHFR-I-SceI protein.To prepare trimethoprim containing food, 1.67g instant Drosophila medium (Formula 4-24, Carolina Biological Supply) was mixed with 5 mL non disUlled water containing 5.3 uL of 100mM trimethoprim while vortexing.To analyze repair products, the upstream white gene was PCR amplified and the PCR product was treated with ExoSAP-IT to enzymaUcally remove excess primers and unincorporated nucleoUdes, followed by Sanger sequencing.Analysis of Sanger sequences was performed using the TIDE (tracking of indels by decomposiUon) algorithm 36 .HR repair products were idenUfied by loss of the I-SceI cleavage site and appearance of the wildtype white gene, which is essenUally a 23-nucleoUde deleUon at the I-SceI cut site.NHEJ products were idenUfied as inserUons and deleUons up to 25bp, except for the 23-nucleoUde deleUon.For PCR and sequencing primers, see Supplemental Table S1.

Immunofluorescence staining
Wing discs were dissected from third instar DR-white larvae and fixed on slides as described earlier 59 .Slides were stored in 96% ethanol at -20°C unUl staining procedure.Slides were thawed at room temperature and washed in PBS for 20 minutes.Tissues were blocked using 0.4% Triton in PBS and 5% milk for 1 hour at room temperature.Primary anUbody incubaUons were performed overnight at 4°C in block buffer.Primary anUbody used for imaging was mouse anU-gH2Av (1:250, Developmental Studies Hybridoma Bank, UNC93-5.2.1).Slides were washed 3 Umes with block buffer.Secondary anUbody incubaUon was performed at room temperature in PBS 0.4% Triton for 2 hours.Secondary anUbody used was Alexa 568 goat anU-mouse (1:600; Invitrogen).Slides were subsequently washed 3 Umes with 0.4% Triton in PBS, incubated with 3μg/ml DAPI for 30 minutes, washed with PBS and mounted using Prolong Diamond AnUfade Mountant and a 20x20 mm #1.5 coverslip.

Cell culture
Kc cells were cultured in CCM3 medium (Avantor) supplemented with AnUbioUc AnUmycoUc SoluUon (Sigma) at 27°C.To induce expression of ph-p-mCherry and ATRIP-eYFP, Kc cells were transiently transfected with 400ng of each plasmid using the TransIT-2020 reagent (Mirus).Live imaging was performed 72 hours aier transfecUon.To induce RNAi-mediated depleUon of dUtx, cells were transfected with 5ug dsRNA (TransIT-2020 reagent (Mirus)) and harvested or subjected to imaging 3 days later.dsRNA was generated using a MEGAScript T7 transcripUon kit (Life Technologies).PCR products containing a T7 promoter sequence and the target regions were used as templates (Supplemental Table S1).

RT-qPCR
RNA was isolated by homogenizing either single larvae or Kc cells in 200μL Trizol (Invitrogen) using an electrical douncer (VWR).Aier addiUon of 40μL of chloroform and centrifugaUon, RNA from the aqueous phase was precipitated using isopropanol and further purified using an ethanol wash step.cDNA was synthesized using iScript following standard cDNA synthesis protocol (Bio-Rad).qPCR was subsequently performed on cDNA with gene-specific primers (Supplemental Table S1).

Imaging
Images of fixed Ussue, Fly-FUCCI wing discs and transfected Kc cells were acquired using a 60x oil immersion objecUve (NA 1.42) on a DeltaVision microscope (DeltaVision Spectris; Applied Precision, LLC).For all DSB tracking experiments in wing discs, Ume-lapse images were acquired using a LD C-Apochromat 63x/1.15W Korr M27 objecUve on a LSM880 microscope with Airyscan (Zeiss), and images were processed using the Zeiss ZEN soiware.Time-lapse images were acquired once every 5-10 minutes.Image analysis and focus tracking were performed manually using the Fiji image analysis soiware.For live imaging of wing discs, third instar larvae were dissected and wing discs were placed on a slide in 10 μL of Schneider S1 medium supplemented with 10% FBS and covered with a 20×20 mm #1.5 coverslip as described previously 60 .To induce ecDHFR-I-SceI in larval Ussue, 400 μM trimethoprim (Sigma) was added 15 minutes prior to imaging.

Figure 1 .
Figure 1.DR-white system to induce single DSBs in euchromaEn and facultaEve heterochromaEn.A) SchemaUc of the DR-white system.Inducible expression of I-SceI creates a single DSB in the upstream white gene.HR with the downstream (5' and 3') truncated iwhite sequence results in loss of the I-SceI cut site (-23bp), and thereby an intact upstream white gene.NHEJ can result in indels at the DSB site.Perfect NHEJ or HR using the sister chromaUd can lead to recreaUon of the I-SceI cut site.B) SchemaUc of DRwhite integraUon sites in the fly genome (2x in euchromaUn [DR eu1, DR eu2] and 3x in facultaUve heterochromaUn [DR fHet1, DR fHet2, DR fHet3]).C) ChIP-qPCR analysis for H3K27me3 at the DR-white locus using qPCR primers that bind 1.4kb downstream of the I-SceI cut site (indicated in Fig 1A).H3K27me3 levels were normalized using ubx gene qPCR primers as a posiUve internal control (ubx has consistently high H3K27me3 levels).Averages are shown for ≥4 independent experiments +SEM.D) SchemaUc of the hsp.I-SceI construct.The hsp70 promoter upstream of I-SceI can be acUvated by shiiing larvae for one hour to 37°C.RT = room temperature.E) SchemaUc of the ecDHFR-I-SceI system.Proteasomal degradaUon of ecDHFR-I-SceI can be blocked by adding the stabilizing compound trimethoprim (TMP).F) Experimental set up for ChIP-qPCR experiments upon DSB inducUon using hsp.I-SceI.3 rd instar DR-white larvae with and