Nuclear and genome dynamics underlying DNA double-strand break repair

Abstract

Changes in nuclear shape and in the spatial organization of chromosomes in the nucleus commonly occur in cancer, ageing and other clinical contexts that are characterized by increased DNA damage. However, the relationship between nuclear architecture, genome organization, chromosome stability and health remains poorly defined. Studies exploring the connections between the positioning and mobility of damaged DNA relative to various nuclear structures and genomic loci have revealed nuclear and cytoplasmic processes that affect chromosome stability. In this Review, we discuss the dynamic mechanisms that regulate nuclear and genome organization to promote DNA double-strand break (DSB) repair, genome stability and cell survival. Genome dynamics that support DSB repair rely on chromatin states, repair-protein condensates, nuclear or cytoplasmic microtubules and actin filaments, kinesin or myosin motor proteins, the nuclear envelope, various nuclear compartments, chromosome topology, chromatin loop extrusion and diverse signalling cues. These processes are commonly altered in cancer and during natural or premature ageing. Indeed, the reshaping of the genome in nuclear space during DSB repair points to new avenues for therapeutic interventions that may take advantage of new cancer cell vulnerabilities or aim to reverse age-associated defects.

Introduction

Eukaryotic cells are characterized by having a nucleus, which harbours the DNA and is surrounded by a nuclear envelope composed of an inner nuclear membrane (INM) and an outer nuclear membrane (ONM)^1,2 (Fig. 1a). The envelope contains transmembrane protein complexes, including the nuclear pore complex (NPC), and the linker of nucleoskeleton and cytoskeleton (LINC) complex, which is a transmembrane protein complex connecting DNA to the cytoskeleton³. The nucleus comprises functionally specialized membrane-less compartments that are considered either constitutive, such as the ribosome-manufacturing nucleolus, or assembled on-demand, such as DNA repair foci⁴. In higher eukaryotes, the nuclear envelope surrounds a meshwork of intermediate filaments constituting the nuclear lamina (Fig. 1a), which provides structural integrity to the nucleus and contacts genomic regions known as lamina-associated domains^5,6. DNA associated with core histone proteins forms nucleosomes, the fundamental organizational units of chromatin. The chromatin further folds into higher-order structures, from small-scale chromatin loops to larger-scale chromosome territories (Fig. 1a–c). Processes that include phase separation (Box 1), loop extrusion by structural maintenance of chromosome (SMC) protein complexes, chromatin attachment to nuclear landmarks (such as the nuclear envelope or the nucleolus) and histone modifications cooperate to form functional nuclear compartments.

Fig. 1 |. Principles of nuclear and genome organization relevant to DSB repair.

a, A schematic of nuclear organization principles from yeast, flies and mammals. The nuclear envelope comprises an outer nuclear membrane (ONM) and an inner nuclear membrane (INM). Fly and mammalian nuclei additionally include a network of intermediate lamin filaments constituting the nuclear lamina. The nuclear envelope harbours the nuclear pore complexes (NPCs), INM proteins, and trans-envelope protein complexes bridging nuclear structures and the cytoskeleton, such as the linker of nucleoskeleton and cytoskeleton (LINC) complex. Also shown are kinesin motor proteins moving on nuclear microtubules, detected in yeast; kinesin on cytoplasmic microtubules present from yeast to human; and myosin motor proteins moving on nuclear actin filaments (F-actin), detected in flies and mammals. The nucleus generally contains membrane-less compartments that are either constitutive such as the nucleolus or pericentromeric heterochromatin (not shown), or are assembled on-demand, such as DNA repair foci. b, The folding of the human genome. The genome is organized into large A (active) and B (inactive) chromatin compartments, which are subdivided into topologically associating domains (TADs). TADs harbour cohesin-extruded loops of chromatin, the basic unit of which is the nucleosome. Also shown is the DNA double-strand break (DSB) marker phosphorylated histone H2AX (γH2AX). c, Chromosome positioning in the nucleus can be generally characterized as chromosome territories, such as in mammalian or fly cells, or as Rabl chromosomes, such as in yeast cells. LADs, lamina-associated domains; rDNA, ribosomal DNA; SPB, spindle pole body.

Box 1 |. Compartmentalization through phase separation.

When a cluster of binding sites (for example, on chromatin) interacts with an array of reader molecules, ‘clustered binding’ occurs (see the figure, part a). Such clustered binding is typically stoichiometric and does not require the formation of networking interactions among the involved molecules. Although distinct from phase separation, clustered binding may locally enrich proteins with the capacity to self-associate, and thereby potentially promote phase separation. Phase separation describes the physical process in which a homogenous mixture (for example, a liquid) undergoes demixing and separates into distinct phases. From a cell biology perspective, liquid–liquid phase separation (LLPS) involves the formation of dynamic multivalent interactions that lack defined stoichiometry and cellular condensates with liquid droplet-like properties^252,253 (see the figure, part b). Whereas dynamic interactions between soluble molecules dominate in LLPS, interactions between associating polymers such as the chromatin fibre can lead to polymer–polymer phase separation (PPPS; also known as bridging-induced phase separation (BIPS))²⁵⁴ (see the figure, part c). Bridging between polymers in PPPS may occur through chromatin-associated proteins that can have LLPS-promoting properties on their own. Therefore, interactions occurring in LLPS and PPPS may co-occur in complex biomolecular condensates. Depending on interaction strength, viscoelastic networks can form in condensates, leading to gelation or liquid–gel phase separation (LGPS). The coupling of interaction networks and density transitions, which results in viscoelastic network fluids, is also referred to as phase separation coupled to percolation (PSCP)²⁵⁵ (see the figure, part d). Finally, liquid–solid phase separation (LSPS) or aggregation can occur, which often, but not always, reflects an irreversible pathological outcome of aberrant phase separation^64,226,252 (see the figure, part e). Although distinct material properties are associated with each of these phase separation types, they are not mutually exclusive. Instead, they may cooperate in the complex environment of the nucleus to shape DNA repair condensates and their interactions with other nuclear structures, such as the nucleoskeleton and the nuclear envelope^212,214,218, in a time-dependent and context-specific manner (Fig. 5).

All these levels of nuclear organization can be altered during DNA damage and repair. DNA double-strand breaks (DSBs) are highly toxic lesions that can pause the cell cycle and induce genome instability, cell senescence or cell death⁷. Most DSBs are repaired by non-homologous end-joining (NHEJ) or homologous recombination (HR)⁸, which are globally orchestrated by DNA damage response (DDR) kinases, including mammalian ataxia telangiectasia mutated (ATM), the HR-promoting ataxia telangiectasia and Rad3-related (ATR), and the NHEJ-driving DNA-dependent protein kinase catalytic subunit (DNA-PKcs)⁹. NHEJ is active in G1, S and G2 phases of the cell cycle. In this pathway, recruitment of the mammalian KU70–KU80 heterodimer, DNA-PKcs and TP53-binding protein 1 (53BP1) to DSBs favours the loading of additional NHEJ repair factors to DSB ends, allowing their processing and reconnection⁸. HR mostly occurs during S and G2, is substantially slower, and is typically less error-prone because it uses an intact homologous sequence (for example, the sister chromatid) as a template for repair. For HR in mammalian cells, breast cancer type 1 susceptibility protein homologue (BRCA1) antagonizes 53BP1, thereby promoting DSB end resection by factors such as the MRE11–RAD50–NBS1 (MRN) complex and CtBP-interacting protein (CtIP) to yield single-stranded DNA (ssDNA) bound by replication protein A (RPA). BRCA2-dependent substitution of RPA with the recombinase RAD51 allows ssDNA to search for homology and continue HR repair. Availability of a homologous sequence and competition between the resection-limiting 53BP1 and resection-promoting BRCA1 dictate repair pathway choice in mammals⁸. Cells also have backup DSB repair pathways, such as the error-prone alternative end-joining (altEJ) pathway⁸. Similar factors contribute to DSB repair in different organisms.

Early studies aiming to understand the dynamics of DNA repair sites in the nucleus by tracking repair proteins or damaged DNA suggested frequent clustering of DSBs in Saccharomyces cerevisiae¹⁰ (hereafter referred to as ‘yeast’, unless otherwise indicated) and human cells¹¹. Since then, the position and mobility of DSBs relative to various nuclear landmarks have emerged as key players in DNA repair and genome stability. In this Review, we discuss the mechanisms and functions of DSB mobility and compartmentalization of repair sites within the nucleus of somatic dividing cells or mitotically budding cells during interphase. We describe how these mechanisms intersect with nuclear and genome structure and with nucleus–cytoplasm communication to ensure genome stability. We also highlight demonstrated and potential connections of nuclear, genome and DSB repair dynamics to human disease and ageing. Finally, we discuss knowledge gaps and future research directions.

DSB mobility and positioning

The development of new imaging techniques and analysis methods enabled the identification of DSB mobility in different repair contexts. In this section, we discuss how these techniques helped reveal complex DSB mobility profiles and their links to DSB repair.

Evidence for DSB mobility through diffusion and directed motion

Studies using different strategies to induce and track DSBs in various organisms have revealed that DSB sites can move across distances ranging from a small area to the entire nucleus^10–19. The magnitude of this mobility differs depending on the chromosomal and nuclear context, including the chromatin state and nuclear size.

Early studies assessing the movement of DSBs in a cell population relied on mean square displacement (MSD) analyses, which revealed mobility consistent with Brownian motion or sub-diffusion within a confined space^20,21. Subsequent studies determined that these dynamics better fit an anomalous Rouse diffusion model²², which accounts for limitations to the movement due to the polymeric nature of chromatin^23,24. MSD analyses also detect directed motions when the studied particles synchronously move in one direction for a sufficiently long period of time^{18,20,21,25,26}. However, DNA repair foci typically move asynchronously, display mixed types of motion and can move in concert with dynamic nuclear structures such as intranuclear microtubule filaments. Given the complexity of these movements, more sophisticated analyses are required to identify directed motions^18,21. Directional change distribution (DCD) analysis assesses the changing direction of moving particles and provides the probability of angular changes during motion¹⁸. DCD has detected directional motion of DSBs in the yeast nucleus on both single-cell and cell population levels¹⁸. Concordantly, a modified MSD analysis named long directed motion analysis, which detects time intervals of directed motions within each trajectory, identified directed motion of heterochromatin repair sites moving to the nuclear periphery in cells of Drosophila melanogaster (hereafter also referred to as ‘fly’ or ‘flies’)^25,26. Together, these studies indicate that DSBs exhibit confined motion consistent with anomalous Rouse diffusion coupled with bursts of linear or nonlinear directed motion within the nucleus^20,21. Thus, DSB mobility can be sub-diffusive, directed or mixed, consistent with the existence of diverse mechanisms affecting DSB dynamics.

Regulation and function of DSB mobility

Several processes influence DSB mobility in different nuclear or genomic contexts. Various broadly acting or context-specific factors ultimately affect repair progression and outcome, as we discuss in this section.

Overview of DSB mobility.

The motion of repair sites is influenced by many factors, including chromatin relaxation and stiffness, nuclear filaments and motor proteins, and components of the DDR. These processes affect DSB mobility at the damaged site or more globally at the genomic scale. DSB mobility can be boosted by the release of local and global physical constraints, such as through damage site-focused chromatin relaxation or by releasing chromosomes from tethering to the nuclear envelope. Together, these features reveal a network of tightly coordinated responses to DNA damage.

The increase in local and global DSB mobility also influences repair outcomes. Higher mobility allows damaged DNA loci to escape repair-repressive nuclear domains, favour homology search, access repair proteins, avoid ectopic repair and sequester irreparable or slowly repaired breaks (Fig. 2a). Of note, rather than being linked to cell cycle phases and repair pathway choice, increased mobility of damaged DNA may be especially associated with persistent damage or slower repair. In agreement with this notion, increased chromatin mobility in yeast, fly, mouse or human cells has been associated with HR-related repair and with translocation-promoting end-joining repair mechanisms such as altEJ (generally completed within hours to over a day) rather than with the faster canonical NHEJ (generally completed within minutes to a few hours)^{2,13,14,25,27–34}. Nonetheless, movement to the nuclear periphery could also promote NHEJ, given that perinuclear-associated proteins can affect NHEJ in yeast and mammalian cells^29,35–38.

Fig. 2 |. DSB positioning, mobility and repair.

a, The mobility of DNA double-strand breaks (DSBs) affects their repair in at least four manners: by escaping repair-repressive environments; accessing repair-promoting domains, factors and homologous sequences; avoiding erroneous repair; and controlling the positioning of breaks relative to each other. b, In budding yeast (Saccharomyces cerevisiae), damaged ribosomal DNA (rDNA) sites are released from the chromosome linkage inner nuclear membrane proteins (CLIPs) and cohibin (Lrs4–Csm1) complexes, which allows damaged ribosomal RNA genes to escape the nucleolus to the nucleoplasm, where radiation sensitive 52 (Rad52) promotes homologous recombination (HR) repair (1). Irreparable or slowly repaired DSBs are targeted to monopolar spindle 3 (Mps3) (2) or to nuclear pore complex (NPC) components (3) for repair. This movement and DSB clustering are mediated by kinesins and DNA damage-inducible intranuclear microtubule (DIM) filaments (4). Upon DSB induction, phosphorylation of centromere protein 3 (Cep3) releases centromeres from the spindle pole body (SPB) (5), which, together with loss of telomere-dependent perinuclear chromosome anchoring, contributes to the increased mobility of damaged (and intact) loci (6). c, In fly (Drosophila melanogaster) cells, DSBs in pericentromeric heterochromatin relocalize towards inner nuclear membrane (INM) and NPC proteins at the nuclear periphery through damage-induced transient nuclear actin filaments (F-actin) and plus-end-directed myosins (1). Relocalization enables Rad51 recruitment and HR. In addition, DSBs cluster in the heterochromatin domain (2), and F-actin contributes to the clustering of euchromatic DSBs (3). d, In mouse (Mus musculus) cells, non-homologous end-joining (NHEJ) proteins are recruited to positionally stable pericentromeric heterochromatin DSBs in domains known as chromocenters (1), and resected DSBs exit to the surface of the chromocenters to access RAD51 and proceed with HR (2). F-actin and myosins are required for the latter movement. Forces exerted onto the nucleus by the cytoplasmic and microtubule-dependent kinesins contribute to the DSB mobility in the nucleus (3). e, In human (Homo sapiens) cells, damaged rDNA sites relocate to nucleolar caps for repair (1), whereas RAD51 can access DSBs in pericentromeric satellites (2). DSBs artificially tethered to the nuclear lamina undergo alternative end-joining (altEJ) (3), and the nuclear-envelope-embedded protein NUMEN promotes NHEJ and limits HR (4). Microtubules and plus-end-directed kinesins cooperate with nuclear envelope-embedded factors such as the linker of nucleoskeleton and cytoskeleton (LINC) complex to transiently form repair-promoting DSB-capturing nuclear envelope tubules (dsbNETs), which are reversed by microtubule minus-end-directed kinesins (5). DSB interactions with SUN1 promote HR while limiting clustering and translocations (6). By contrast, F-actin drives DSB clustering, thereby promoting translocations (7). ONM, outer nuclear membrane.

Both short-range and long-range movements can be energy-consuming, requiring chromatin relaxation at a minimum. Longer-range dynamics also rely on molecular motors such as myosins or kinesins, which cooperate with nuclear actin filaments (F-actin) or nuclear or cytoplasmic microtubules to directionally move DSBs^18,25,28,38 (Fig. 2b–e). Despite these differences, many of the components that are required to mobilize repair sites appear to be conserved across organisms and repair pathways (Figs. 2b–e and 3), revealing a remarkable ability of core factors (Supplementary Table 1) to adapt to different repair challenges.

Fig. 3 |. Molecular mechanisms underlying DSB mobility for repair.

Factors required for the movement of DNA double-strand break (DSB)-repair sites in different contexts, with outcomes indicated as homologous recombination (HR, including break-induced replication (HR-BIR)), non-homologous end-joining (NHEJ), or alternative end-joining (altEJ). a, In yeast (Saccharomyces cerevisiae), persistent or irreparable DSBs move to the nuclear pore complex (NPC) or the inner nuclear membrane protein monopolar spindle 3 (Mps3) in a manner requiring checkpoint kinases, the structural maintenance of chromosome (Smc)5–Smc6 complex, Siz2, the Slx5–Slx8 SUMO-targeted ubiquitin ligase (STUbL), Swi2/Snf2-related (Swr1)-dependent H2A.Z incorporation, H2A.Z sumoylation and damage-inducible intranuclear microtubules (DIMs). Radiation sensitive (Rad)51 and Ino80 mediate targeting specifically to Mps3 in the S and G2 cell cycle phases. H2A.Z, DNA end resection, Rad51, checkpoint factors and Ino80 are also required for homology search (not shown). b, Yeast eroded telomeres relocalize to NPCs for HR, which depends on Siz1 and Siz2-mediated sumoylation of telomeric proteins, including replication protein A (RPA). NPC anchoring and HR require the SUMO protease ubiquitin-like specific protease 1 (Ulp1) and the STUbL Slx5–Slx8. Mouse (Mus musculus) uncapped telomeres are mobilized through a pathway requiring TP53-binding protein 1 (53BP1) phosphorylation and cytoplasmic microtubules, which transfer forces to the nucleus through kinesins and the LINC complex. Repair of eroded telomeres through the alternative lengthening of telomeres (ALT) pathway in human (Homo sapiens) cancer cells occurs by RAD51-dependent homology search, requires the meiotic homologous chromosome synapsis complex homologous-pairing protein 2 (HOP2)–meiotic nuclear division protein 1 (MND1) and induces telomere coalescence into ALT-associated promyelocytic leukaemia bodies (APBs) for BIR repair. Human telomeres damaged during replication relocalize to NPCs through ataxia telangiectasia and Rad3-related protein (yeast Mec1) (ATR)-induced nuclear actin filaments (F-actin). c, Yeast subtelomeric DSBs relocalize to NPCs for HR-BIR repair in a pathway requiring Rad9 (homologous to human 53BP1), Swr1-dependent H2A.Z incorporation, cohibin, Rad52 condensates and kinesin-14 (Cik1–Kar3) travelling along DIMs. Yeast nucleoporin (Nup)84 is homologous to fly and human Nup107. d, Damaged yeast ribosomal DNA (rDNA) is released from perinuclear tethering through phosphorylation and Siz2-dependent sumoylation of chromosome linkage inner nuclear membrane proteins (CLIPs) and cohibin subunits followed by their Cdc48/p97-dependent and Ufd1-dependent disassembly. Exclusion of Rad52 foci from the nucleolus requires Rad52 sumoylation and Smc5–Smc6. At least some rDNA breaks are repaired at NPCs. In mammals, damaged rDNA is transcriptionally repressed by the kinase MST2 (STK3) and cohesin–human silencing hub (HUSH) complexes, thereby promoting DSB movement to nucleolar caps. Silencing and movement also depend on the damage checkpoint proteins ataxia telangiectasia mutated (ATM; yeast Tel1), checkpoint kinase (CHK)2, ATR and CHK1, the resection complex MRE11–RAD50–NBS1 (MRN), and DNA topoisomerase II-binding protein 1 (TOPBP1)–TCOF1. Movement also depends on the histone demethylase jumonji domain-containing 6 (JMJD6). LINC, actin-related protein (ARP)3 and UNC45 contribute to this relocalization, suggesting movements between nucleolar caps and the nuclear envelope. e, In fly (Drosophila melanogaster) cells, DSBs in pericentromeric heterochromatin relocalize to the NPCs and inner nuclear membrane (INM) proteins through the STUbL Dgrn, the histone demethylase Kdm4A, the acetyltransferase Gcn5, checkpoint activation and resection. Nuclear myosins promote this movement by walking along F-actin induced by Scar and Wash. The Smc5–Smc6 complex is targeted to the heterochromatin domain by heterochromatin protein 1 (HP1)a, contributing to this movement by interacting with myosins, recruiting the myosin activator Unc45 and establishing Nup98 condensates within heterochromatin. Nup98 recruitment to DSBs is also mediated by nucleoplasmic Sec13 or Nup88. Sumoylation by dPIAS also halts HR inside the heterochromatin domain upstream of Nup98 condensates. Dgrn STUbL enables HR restart at the nuclear periphery together with the Rad60–Esc2–NIP45 (RENi) protein Rad60. In mouse cells, relocalization relies on Smc5–Smc6, myosins, F-actin, and might be facilitated by chromatin relaxation through HP1β phosphorylation. f, In yeast, short microtubule-dependent oscillations promote clustering of Rad52 repair condensates. In fly cells, DSB clustering requires Arp2–Arp3-dependent F-actin in euchromatin and Nup98 condensates in heterochromatin. In human cells, DSB clustering of transcribed genes relies on formin-2-dependent F-actin, MRN, ATM and LINC in G1, or on resection, Wiskott–Aldrich syndrome family protein (WASP) and ARP2–ARP3-mediated nuclear F-actin in S/G2. g, In human cells, DSB-capturing nuclear envelope tubules (dsbNETs) are formed by cytoplasmic microtubules through plus-end directed kinesins (kinesin family member (KIF)5B, KIF13B), and reversed by KIFC3. Checkpoint kinases, especially ATM and DNA-PK, promote α-tubulin acetyltransferase 1 (ATAT1)-dependent acetylation of α-tubulin, thereby increasing kinesin recruitment. INM proteins and NPCs mediate dsbNET formation. dsbNET assembly partly requires the nuclear periphery-tethering protein period circadian regulator (PER)1 and F-actin, which can connect dsbNETs to DSBs. h, PER2 contributes to repair of transcription-coupled DSBs (TC-DSBs) by targeting them to the NPC and the LINC subunit SUN1. F-actin structures are illustrated when directly detected in the nucleus. LINC is shown when its direct link to the nuclear periphery has been demonstrated. BRCA1, breast cancer type 1 susceptibility protein homologue; FMN2, Formin-2; Mlp1 and Mlp2, Myosin-like protein 1 and 2; ONM, outer nuclear membrane.

Release of global physical constraints.

In budding yeast, global constraints on chromosome movement are imposed by the attachment of ribosomal DNA (rDNA) and telomeres to the nuclear envelope and the anchoring of centromeres to the nuclear-envelope-embedded spindle pole body (SPB)^2,39–41 (Fig. 1c). Intriguingly, DSB induction can release these physical restraints and contribute to the increased mobility of both DSB-harbouring and undamaged DNA^2,40,42,43 (Fig. 2b). Additionally, the chromatin remodeller Ino80 affects chromatin mobility at least in part through proteasome-dependent histone degradation, thereby increasing chromatin accessibility throughout the genome^41,44–47. The checkpoint kinase Mec1 (human ATR) facilitates Ino80-mediated histone degradation⁴⁶. Although it is unclear whether similar processes occur in organisms other than yeast, increased mobilization of undamaged chromatin has been observed in Arabidopsis thaliana, fly, mouse and human cells^{25,29,38,48,49}.

Release of local constraints and homology search.

Early studies in yeast revealed that a single, resected DSB causes increased mobility of the damaged site and of chromatin more globally, which is likely to facilitate the identification of homologous sequences for strand invasion and HR^15,16 (Fig. 2b). The local increase in dynamics is promoted by releasing local physical constraints through chromatin remodelling that is activated by checkpoint kinases. Specifically, upon DSB induction, activation of Mec1 recruits the Swi/Snf Swr1 and Ino80 chromatin remodellers to the damaged site, loosening the surrounding chromatin⁵⁰ and promoting DSB movement^15,28,44,51 (Fig. 3a). In the fission yeast Schizosaccharomyces pombe, Ino80 is also required for the juxtaposition and protective translocation of rDNA to eroded telomeres in the absence of telomerase^52,53, suggesting a broad role for this chromatin remodeller in genome dynamics and stability. In budding yeast, despite becoming more accessible, DSB-flanking chromatin also exhibits features suggestive of increased stiffness, which can facilitate the movement of repair sites through the chromatin meshwork^23,54. Such stiffening could reflect negative charges introduced by histone H2A phosphorylation, the formation of Rad51 nucleofilaments or the action of motors pulling the chromatin onto nuclear microtubules during directed transport^{18,20,23,54–56}. Consistently, DNA end resection or Rad51 facilitate the relocalization of repair sites in yeast^15,16,55,57 and homology search is associated with the directed motion of recombining telomeres undergoing alternative lengthening of telomeres in mammals^58,59 (Fig. 3b).

Controlled escape from the repair-restrictive nucleolus.

Mobile DSBs can escape nuclear neighbourhoods that are generally predisposed to aberrant repair or that can limit certain repair steps. Such neighbourhoods include nucleoli, telomere clusters and pericentromeric heterochromatin domains, which all typically harbour repetitive DNA sequences prone to ectopic recombination^39,60–62. Studies in yeast and mammalian cells have established that escape from the nucleolus, a major compartment associated with RNA-dependent and protein-dependent phase separation^63–67 (Box 1), is a prerequisite to complete HR and avoid ectopic recombination (Fig. 3d).

Replicating yeast sister chromatids harbouring rDNA repeats are tethered to each other and to the nuclear envelope within the crescent-shaped perinuclear nucleolus through the chromosome linkage inner nuclear membrane protein (CLIP) complex (Heh1–Nur1) and cohibin (Lrs4–Csm1), and are further stabilized by the cohesin complex^2,42,68–70 (Figs. 2b and 3d). Perinuclear tethering of rDNA repeats prevents sister chromatid misalignment and deleterious overexposure to radiation sensitive 52 (Rad52; equivalent of human BRCA2 (refs. 71,72)) recombination proteins in the nucleoplasm^2,13,42,73. CLIP-dependent tethering preserves the stability of rDNA repeats independently of chromatin silencing^2,74. The exclusion of Rad52 repair foci from the nucleolus is mediated by the Smc5–Smc6 complex and by Rad52 sumoylation^13,75. Mutations abrogating this exclusion result in Rad52 foci inside the nucleolus, rDNA hyper-recombination and extrachromosomal rDNA circles^13,74. In wild type cells, rDNA damage also induces sumoylation of Lrs4 and Nur1, and phosphorylation of Nur1, thereby releasing cohibin from CLIP and allowing the damaged rDNA to escape the nucleolus for controlled HR^2,42 (Fig. 3d). HR-mediated repair of released rDNA may also be completed at nuclear pores⁷⁶.

Similarly, in mammalian cells, DSBs in rDNA induce transcriptional silencing and the relocation of the damaged sequences into newly formed nucleolar ‘caps’ at the edge of the nucleolus^77–85 (Figs. 2e and 3d). Silencing and repositioning depend on ATM, ATR, the MRN complex, the human silencing hub (HUSH) complex, cohesin and the nucleolar-specific protein TCOF1 (also known as Treacle)^80–83. Silencing is mediated by MST2 (also known as STK3)-dependent phosphorylation of histone H2B⁸⁶ and HUSH-complex-dependent trimethylation of histone H3 Lys9 (H3K9me3)⁸¹. The histone demethylase Jumonji domain-containing 6 (JMJD6) induces relocalization of damaged DNA to nucleolar caps without affecting transcription, thereby uncoupling relocalization from transcriptional silencing⁸⁷. The LINC complex, actin nucleator actin-related protein 3 (ARP3) and myosin activator UNC45 are also required for DSB relocalisation to nucleolar caps, suggesting the involvement of movement to and contacts with the nuclear envelope⁸¹. Relocalization to nucleolar caps generally occurs after end resection but before the recruitment of additional HR proteins, and repositioning preserves rDNA stability^79–82.

Mobilization of persistent DSBs for repair at the nuclear envelope.

In yeast, relocalization to the nuclear periphery might reflect the sequestration of irreparable DSBs lacking donor sequences^14,57,88 or repair by slower HR-linked repair processes such as break-induced replication (BIR)^28,30 (Figs. 2b and 3a–c). Such DSBs are commonly targeted to the nucleoporin Nup84 at NPCs directly or through the SUN domain protein monopolar spindle 3 (Mps3)^{14,28,51,57,88,89}. Mps3, which may exist as an INM protein alone, or bind the KASH-like domain-containing ONM protein Mps2 as part of a yeast LINC-like complex^90–92, can also mediate DSB repair independently of Nup84 (refs. 51,88) (Figs. 2b and 3a). DSB relocation to NPCs and subsequent repair depend on the Nup84-interacting SUMO-targeted ubiquitin ligase (STUbL) Slx5–Slx8 complex, which may facilitate the proteasomal degradation or removal of HR-pausing factors^{14,28,93–95} (Fig. 3a). Checkpoint activation, Swr1-dependent incorporation of histone H2A.Z (also known as Htz1) and H2A.Z sumoylation also promote relocation of persistent breaks, probably through chromatin loosening^51,57. In yeast, a similar relocalization pathway mediates the often slower repair of eroded telomeres^96,97 (Fig. 3b), subtelomeric DSBs^28,35 (Fig. 3c), damaged replication forks at CAG repeats^93,98,99 and stalled forks^94,95. Of note, yeast Nup84 can also contribute to DNA repair through NHEJ^35,36 and altEJ^30,35, and Slx5–Slx8 can partly promote NHEJ²⁸, although it remains unclear whether physical relocation of damaged DNA occurs in these contexts.

Studies in yeast have shown a higher importance of mobility for repair the closer DSBs are to telomeres^18,28,35,41. Also, remodelling of chromatin flanking DSBs by Swr1 promotes the recruitment of Kinesin-14 (Cik1–Kar3) for directed mobility on DNA damage-inducible intranuclear microtubule (DIM) filaments towards Nup84 for BIR repair of subtelomeric or internal DSBs²⁸ (Figs. 2b and 3c). Subtelomeric DSBs additionally require the telomere-associated factor cohibin to engage this pathway, indicating that breaks within different chromosomal contexts require distinct factors to use similar repair mechanisms at the nuclear periphery²⁸. Similar to the release of perinuclear rDNA tethers for repair, damage has been reported to result in the phosphorylation of the kinetochore-associated centromere protein 3 (Cep3), leading to the release of centromeres from the SPB⁴⁰ (Fig. 2b). Centromere release from the SPB appears to be less important for repair the further away the DSBs are from centromeres⁴⁰, may not be a universal feature of increased chromatin mobility during DNA repair^43,47 and can accelerate but is not essential for repair⁴¹. Instead, global nucleosome depletion⁴⁷ and microtubule-dependent forces⁴³ may have broader roles in promoting chromatin mobility during repair. For instance, artificial centromere release is sufficient to induce DIM formation from the SPB, highlighting the potential coordination of removing global centromere restraint with the emergence of nuclear microtubules on which DSBs can be transported¹⁸ (Fig. 2b). Similarly, as reported in a preprint, the targeting of collapsed replication forks at CAG repeats to NPCs depends on Cep3 phosphorylation and DIMs⁹⁹. Unlike in yeast, there is still no direct evidence for the existence of nuclear microtubules inside mammalian nuclei. However, their potential existence is indirectly suggested by the observation that DSBs recruit the nuclear kinesin family member 18B (KIF18B), whose microtubule-dependent motor activity is required for NHEJ and altEJ¹⁰⁰. In addition, the DSB-enriched kinesin KIF2C and its microtubule depolymerase activity promote the mobility of human DSBs, and along with it NHEJ or HR¹⁰¹.

Escaping heterochromatin domains for repair at the envelope.

In flies, DSBs in pericentromeric heterochromatin move from inside the ‘heterochromatin domain’ to the nuclear periphery for HR^17,25,27,62 (Figs. 2c and 3e). Pericentromeric heterochromatin is typically in a silent and compact state, and has phase separation properties (Box 1). During DSB relocalization in fly and mouse cells, the pericentromeric heterochromatin domain ‘expands’, probably reflecting a large-scale decompaction of heterochromatin, which can contribute to DSB mobility by releasing physical constraints^27,102,103. In fly cells, the histone demethylase Kdm4A mediates H3K9me3 and H3K56 demethylation and the histone acetyltransferase Gcn5 drives H3K9 acetylation around the DSB^104,105, resulting in Smc5–Smc6 recruitment and DSB relocalization¹⁰⁶. These activities suggest that additional, local chromatin changes help mobilize the repair sites. Relocalization of heterochromatic DSBs to the nuclear periphery also relies on a dynamic and transient network of nuclear actin filaments that connect DSBs with the nuclear periphery¹⁰⁷, which are assembled by the Arp2–Arp3 complex and its activators suppressor of cAR (Scar) and Wiskott–Aldrich syndrome protein and scar homologue (Wash) at repair sites²⁵ (Fig. 3e). Nuclear myosins are specifically responsible for moving repair sites from the edge of the heterochromatin domain to the nuclear periphery through directed motion along these filaments²⁵.

At the nuclear periphery, DSBs associate with the Nup107 complex (homologous to yeast Nup84) of the NPCs or with the INM proteins Koi and Spag4 (refs. 17,108). Here, Rad51 is likely to be recruited through the action of Dgrn STUbL associated with Rad60–Esc2–NIP45 (RENi) proteins, which are enriched at these locations^17,109 (Figs. 2c and 3e). The Smc5–Smc6 complex enriched at DSBs is a central regulator of this pathway²⁷. Smc5–Smc6 associates with nuclear myosins in response to damage and recruits the myosin activator Unc45 to DSBs²⁵. Smc5–Smc6 also blocks HR inside the heterochromatin domain and promotes relocalization through sumoylation of unknown targets in concert with the fly SUMO E3 ligase dPIAS^17,27,109. Sumoylated proteins are presumably removed at the nuclear periphery by STUbL to enable repair¹⁷. Excluding Rad51 from the heterochromatin domain is central for preventing ectopic recombination among repeat sequences^{17,25,27,109–111} (Fig. 3e).

As reported in a recent preprint, in addition to these functions, Smc5–Smc6 recruits the nucleoporin Nup98 to DSBs inside the heterochromatin domain, where Nup98-dependent phase separation excludes Rad51 from, while facilitating the early mobilization of repair sites inside, the heterochromatin domain¹¹⁰ (Fig. 3e,f). These early diffusive motions are likely to be promoted by capillary forces generated by surface tension between immiscible condensates, namely the heterochromatin domain on one hand and repair sites on the other hand, drawing new connections between condensate properties and nuclear dynamics¹¹⁰. Thus, off-pore nucleoporins may serve as a long-distance communication tool to initiate processes that enable the eventual directed movement towards the nuclear periphery. Elements of this pathway are conserved in mouse cells^25,112–114, where repair sites exit heterochromatic chromocenters through SMC5–SMC6, ARP2–ARP3, F-actin and myosins²⁵ (Figs. 2d and 3e). In human cells, DNA-damage-dependent actin polymerization is induced by different sources of DSBs and replication stress^{32,38,115,116}. In addition to heterochromatin repair, ARP2–ARP3-induced F-actin has been linked to clustering of repair foci for HR repair of euchromatin in human and fly cells^25,32 (Fig. 3f). In human cells, F-actin also mediates the myosin-driven relocalization of damaged replication forks to the nuclear periphery^117,118, the protection and remodelling of stressed forks^116,118,119 and the repair of telomeric damage induced by replication defects¹²⁰ (Fig. 3b), thereby revealing additional roles of nuclear actin filaments in DNA damage responses.

Positional stability and mobility of mammalian DSBs.

Mouse and human cells show similarities and differences in DSB mobility and its regulation. Human pericentromeric DSBs that are repaired by HR are more positionally stable than in mouse and fly cells, with RAD51 being recruited within large heterochromatin satellites (Fig. 2e). By contrast, NHEJ occurs at the periphery of the satellites¹²¹. Conversely, mouse pericentromeric DSBs localize to the periphery of heterochromatin domains known as chromocenters to recruit RAD51 during HR^25,113, whereas NHEJ factors access positionally stable DSBs inside the chromocenters¹¹³ (Fig. 2d).

Although these and other studies are consistent with more limited movement of DSBs in the mammalian nucleus relative to yeast and fly cells^{29,38,113,121,122}, at least some DSBs, damaged replication forks and fragile telomeres exhibit features consistent with mid-range to long-range mobility in mammalian cells^{31,33,34,38,58,117,123–125}. When interpreting these findings, it should be considered that most adherent cultured human cells are relatively flat; thus, small movements along the Z-axis sufficient to reach the nuclear periphery may go undetected¹¹⁸.

DSBs induced in transcriptionally competent regions and in rDNA were also found to contact the nuclear envelope in human cells^38,81,125 (Fig. 2e). Tethering of DSBs to the nuclear envelope mostly requires SUN1 and the nucleoporin NUP153 (refs. 38,125) (Fig. 3d,g,h). In this context, the nuclear periphery can promote repair protein assembly, fostering repair^38,125. In addition, as reported in a preprint, nuclear envelope–DSB contacts for HR repair depend on PERIOD proteins, which are involved in the regulation of the circadian clock, consistent with a link between DSB mobility and circadian rhythms¹²⁵ (Fig. 3h).

Nuclear envelope mechanics and nuclear–cytoplasmic DNA damage responses

While increased chromatin targeting to the nuclear envelope can promote DSB repair, the extent to which mammalian DSBs exhibit such repositioning remains unclear. Recent findings suggest that DSB–nuclear envelope interactions can be promoted also by membrane remodelling, at least in mammalian cells.

Nuclear envelope mobilization towards DSBs.

Upon DSB induction in human cells, checkpoint kinases promote the remodelling of cytoplasmic microtubules³⁸ (Figs. 2e and 3g). These filaments work in concert with the nuclear envelope-embedded LINC and NPCs and with the microtubule plus-end-directed cytoplasmic kinesins KIF5B and KIF13B, to push the nuclear envelope inwards³⁸ (Figs. 2e and 3g). This process forms a transient network of tubular nuclear-envelope invaginations dubbed DSB-capturing nuclear envelope tubules (dsbNETs), which encompass the INM, ONM and nuclear lamina^38,126. dsbNETs deliver the envelope and its proteins to DSBs, which helps reconnect DSB ends within DNA repair centres, thereby promoting NHEJ and HR, before the reversal of tubules by the microtubule minus-end-directed kinesin KIFC3 (ref. 38). dsbNETs also shorten the distance a DSB must travel to access the nuclear envelope for repair and their formation can alter global chromatin organization, with chromatin decompressing as the tubules reverse³⁸. These observations are also consistent with the role of LINC, kinesins and microtubules in mobilizing uncapped telomeres and DSBs for NHEJ in mouse cells²⁹ (Fig. 3b).

Of note, in human cells, DSBs tethered to the nuclear lamina favour altEJ repair¹²⁷. This preference is consistent with the observation that the human nuclear envelope at the edge of the nucleus is enriched in the protein nuclear membrane endonuclease-exonuclease (NUMEN; also known as ENDOD1), which promotes NHEJ and limits HR by generating short 5′ overhangs at damage sites within transcriptionally silenced lamina-associated domains, deprotected telomeres and euchromatin DSBs³⁷ (Fig. 2e). Additional localization of NUMEN in lamin-B1-positive, non-peripheral regions of the nucleus is reminiscent of the tubular invaginations of dsbNETs^37,38. Thus, nuclear-envelope-embedded transmembrane proteins that promote repair may be able to access peripheral and non-peripheral DSBs. As reported in a preprint, increased contacts of transcription-associated and HR-repairable DSBs with the nuclear envelope have also been observed¹²⁵ (Fig. 3h). Taken together, dsbNETs or other damaged DNA-capturing nuclear envelope tubules may deliver specific factors to activate or repress different repair pathways at diverse chromosomal lesions throughout the nucleus^37,38,125. Intriguingly, dsbNETs contact F-actin, and actin filaments are at least partly required for dsbNET formation, suggesting that F-actin can help in guiding the interaction of dsbNETs with DSBs³⁸ (Fig. 3g).

dsbNETs change our perception of the nuclear envelope from a destination to a dynamic structure that actively localizes with DSB sites to regulate repair. It is unknown whether dsbNETs are conserved, but cooperation between KIF5B, LINC and dynamic microtubules promotes the small-scale mobility of uncapped telomeres in mouse cells²⁹ (Fig. 3b). In addition, lipid metabolism and dynamics are intimately linked with genome stability in yeast¹²⁸. Overall, these studies reveal different mechanisms through which nuclear envelope–DSB contacts can be established and highlight an under-appreciated role of cytoplasmic factors and processes in the dynamic reorganization of the nucleus and genome during DSB repair.

The mechano-responsiveness of DDR kinases.

Considering that the DDR kinases may regulate both nuclear and cytoplasmic responses that drive DSB mobility and DSB–nuclear-envelope contacts, it will be crucial to decipher whether these kinase functions are part of the canonical DDR or are non-canonical. For example, the DDR kinases control cytoplasmic mechanical forces that reshape the nuclear envelope to promote DNA repair in humans³⁸. Upon DNA damage induction, DNA-PKcs, ATM and ATR promote the α-tubulin acetyltransferase 1 (ATAT1)-dependent acetylation of Lys40 of α-tubulin within the microtubule lumen³⁸. This post-translational modification can promote the ability of microtubules to recruit plus-end-directed kinesins, which push onto the nuclear envelope, thereby forming dsbNETs and promoting DSB repair^38,129–131 (Figs. 2e and 3g). ATAT1 activation may also boost the DNA damage checkpoint^38,132.

Notably, DDR kinases can exhibit mechano-responsiveness also in the absence of DNA damage^133–135. For instance, temporary mechanical stretching of cells activates ATR at the nuclear envelope independently of DNA damage, altering the plasticity of the envelope and its association with chromatin^133,135. This work further highlights a role for ATR in coupling cytoskeletal forces exerted onto the nuclear envelope with the chromatin inside the nucleus¹³⁵. Mechanical stress also activates ATM through a reactive oxygen species-linked mechanism, driving nucleoskeleton, cytoskeleton and chromatin remodeling¹³⁴.

Additionally, DDR kinases can alter the building blocks of the human nuclear lamina during DSB repair^136,137. DSB induction has been reported to trigger ATR-dependent phosphorylation of Ser282 of the LMNA gene products lamin A and lamin C (laminA/C), promoting rupture of the nuclear envelope¹³⁶. This rupture exposes nuclear DNA to the cytoplasm, thereby activating the cytosolic DNA sensor cyclic GMP–AMP synthase (cGAS)¹³⁶, which can mediate innate immunity responses^138,139. Furthermore, mechanical-confinement-induced nuclear envelope rupture in p53-negative cells depends on the ATR-mediated phosphorylation of lamin A¹³⁶. Of note, in the absence of external mechanical stress, nuclear envelope rupture is transient and affects a small percentage of wild type cells, though the cellular fraction affected is higher in BRCA2-deficient cells¹³⁶. ATR-dependent nuclear envelope rupture may also have a protective function, such as by aiding in the clearance of cells with an excessively damaged nucleus. Concordantly, ATR-dependent phosphorylation of laminA/C at Ser395 and further phosphorylation of Ser392 by CDK1 was reported to promote the rupture of lamin B-deficient micronuclei. This potentially led to cytoplasmic DNA and cGAS activation and the clearance of affected cells through autophagy, senescence or natural killer cells¹³⁷. However, the rupture of micronuclei-bearing chromatin without naked double-stranded DNA (dsDNA) may not effectively activate cGAS¹⁴⁰, possibly because cGAS is directly inhibited by histones^141–146, and supporting the possibility that not all micronuclei are the same¹³⁷. Nevertheless, induction of interferon-stimulated genes can be observed in the absence of cGAS activation and conditions yielding naked dsDNA from nucleosome-depleted regions or annealed telomeric ssDNA repeats may still activate cGAS¹⁴⁰. There is also crosstalk between micronuclei and mitochondria^147,148, and it is possible that the rupture of micronuclei compromises mitochondria, causing them to release some of their nucleosome-free dsDNA in the cytoplasm and ultimately activating cGAS or similar effectors.

Thus, the DDR kinases can promote and respond to mechanical forces to drive DNA repair or promote the clearance of excessively damaged cells. DNA-damage-induced post-translational modifications may also reinforce the cytoskeleton and loosen the nuclear lamina to facilitate repair-promoting processes, such as the formation of dsbNETs. It remains unclear whether the disruptive effect of DDR kinases on the integrity of the nuclear lamina and nuclear envelope reflects mechanisms that evolved to protect the nucleus, clear cells with an excessively damaged nucleus, eliminate micronuclei-bearing cells or a combination of these functions. Nonetheless, the above findings are consistent with DDR kinases exerting roles within an integrated mechano-response that is separable from their canonical DDR functions.

Emerging mechanisms of dynamic compartmentalization of DSB repair

The insights into DSB and chromatin mobility discussed so far emphasize the importance of subnuclear environments for repair efficiency and outcome. In addition, emerging mechanisms are revealing how repair compartments built around DNA lesions are remodelled and fused to form larger repair centres within the context of a broader genome and nuclear organization.

General principles of higher-order chromatin organisation

In general terms, two complementary mechanisms regulate chromosome folding and spatial organization within the nucleus, namely chromatin loop extrusion and compartmentalization through phase separation.

Loop extrusion is a dynamic, energy-consuming process allowing the rapid formation and growth of chromatin loops by ATP-dependent translocase activities of SMC complexes such as cohesin, SMC5–SMC6 and condensin¹⁴⁹. Although the imaging of loop extrusion in live single nuclei has recently become more feasible¹⁵⁰, this process has been extensively visualized using assays such as Hi-C, which measures contact frequencies between different DNA loci on a genomic scale at the cell population level. Loop extrusion can transiently pause upon encountering genetic boundary elements (Fig. 4a). In the mammalian genome, cohesin-mediated loop extrusion mainly halts at genomic loci that are occupied by the insulator protein CCCTC-binding factor (CTCF) in a convergent orientation, which was proposed to be the main mechanism by which topologically associating domains (TADs) are formed¹⁵¹. The list of loop-extrusion barriers has recently expanded to include DSBs^152,153, origins of DNA replication¹⁵⁴, stalled replication forks¹⁵⁵ and the transcription machinery^155–158, supporting the concept of loop extrusion barriers being dynamic. Given its potential to scan the genome and bring distant genomic loci into close proximity, SMC-mediated DNA loop extrusion is proposed to perform a broad range of biological functions, including enabling enhancer–promoter communication¹⁵⁹, chromatin condensation, sister chromatid disentanglement and segregation^160,161, VDJ and class switch recombination¹⁶², histone post-translational modification, and DSB repair^152,153,163, as well as the regulation of DNA replication^154,164 and chromosome topology¹⁶⁵.

Fig. 4 |. Roles of genome organization and compartmentalization in DSB repair.

Two main processes dictate chromosome architecture and genome organization in the nucleus: loop extrusion and phase separation. a, Cohesin-mediated loop extrusion allows the formation of topologically associating domains (TADs), within which loci display enhanced interaction. When encountering a DNA double-strand break (DSB), cohesin-mediated loop extrusion halts, triggering one-sided loop extrusion on each side of the break, likely to be fostered by nipped-B-like (NIPBL), which also accumulates at DSBs. Ataxia telangiectasia mutated (ATM)-mediated phosphorylation of H2AX-containing nucleosomes during the loop extrusion process allows γH2AX spreading on both sides of the DSB until CCCTC-binding factor (CTCF)-bound loci are reached, resulting in γH2AX accumulation within the entire TAD. Cohesin-mediated loop extrusion, possibly with or without DNA entrapment, at the DSB also contributes to the synapsis of break ends and homology search. The structural maintenance of chromosomes protein (SMC)5–SMC6 complex, which also mediates loop extrusion, is present at DSBs, but its potential role in DSB-induced loop extrusion is unknown. b, Phase separation triggers self-segregation of heterochromatin (B-compartment) and euchromatin (A-compartment). Following DSB induction, γH2AX-marked TADs display a potential to self-segregate from the rest of the genome, forming a DSB-induced chromatin compartment (D-compartment). Current models for DSB clustering and D-compartment formation in human cells involve the phase separation properties of repair proteins accumulating across the entire damaged TADs, such as TP53-binding protein 1 (53BP1), and the ATP-dependent motion of DSBs along nuclear actin filaments (not shown). In addition to γH2AX–53BP1 covered TADs, the D-compartment also contains a subset of DNA damage response (DDR) and R-loop-enriched genes. The targeting of DDR genes to the D-compartment contributes to their optimal activation during DNA repair.

In addition to loop extrusion, chromatin regions can self-associate and segregate from other regions within the nucleus, thereby forming chromatin compartments and microcompartments (Fig. 4b). Chromatin is broadly organized into two main compartments, corresponding to euchromatin (‘A’ compartment) and silent heterochromatic sequences (‘B’ compartment), which can be visualized cytologically and by Hi-C-derived chromatin contact frequency maps^166,167. Chromatin compartmentalization has been proposed to involve weak and multivalent molecular interactions between distant genomic loci, promoting phase separation^168–170. Although different forms of phase separation have been described (Box 1), the general principle is that a mixture prone to undergo phase separation is characterized by energetically favourable interactions among molecules of the same type (type A with A molecules; type B with B) and by less favourable interactions between types of molecules (type A with type B). When applied to chromatin polymers, phase separation allows the physical compartmentalization of similar chromatin types and the selective concentration of associated factors. Phase separation can underly large-scale and heterochromatin protein 1 (HP1)-dependent compartmentalization of pericentromeric heterochromatin^171,172. In addition, small-scale phase separation (‘microphase separation’) may account for the formation of many other smaller subnuclear structures, including enhancer–promoter contacts and super-enhancers, Cajal bodies, promyelocytic leukaemia bodies (PML bodies), and Polycomb bodies¹⁷³, and contribute to TAD formation^174,175.

These two complementary mechanisms driving chromosome architecture and spatial genome organization were recently shown to cooperate in the formation of DNA repair centres and contribute to the activation of the DDR and to repair^34,152,153. Indeed, the rapid generation of DSBs (within a few minutes) gives rise to subnuclear structures of high complexity, originally observed by microscopy and termed ‘irradiation-induced foci’. Long considered to be repair foci or repair ‘compartments,’ recent methodological advances, including in chromosome conformation capture and super-resolution microscopy, have provided new and unique insight into the formation, protein and nucleic acid composition, and function of these compartments.

Chromosome folding through loop extrusion upon DSB induction

One of the first steps of DNA repair foci formation in human cells is the ATM-mediated phosphorylation of Ser139 of the histone variant H2AX (γH2AX), which spreads within large chromatin domains (up to about one megabase) surrounding the DSB¹⁷⁶. γH2AX spreading is contained within the DSB-containing TAD^152,177,178, underlining a key role of chromosome architecture in establishing γH2AX domains (Fig. 4a). Moreover, SMC complexes, including SMC5–SMC6 and cohesin (reviewed elsewhere^179–181) and their regulators (for example, nipped-B-like (NIPBL; Scc2 in yeast) or establishment of sister chromatid cohesion N-acetyltransferase 2 (ESCO2))^182,183 accumulate at DSB sites from yeast to mammals, suggesting that DSBs can stall loop extrusion factors. Indeed, Hi-C experiments in human¹⁵² and yeast¹⁵³ have revealed the potential of DSBs to stall loop extrusion, resulting in unidirectional cohesin-mediated loop extrusion. Whether other SMC complexes also affect DSB-induced loop extrusion is unknown, but it is probable given the involvement of SMC5–SMC6 in DSB mobility and repair in different organisms^{13,17,25,27,30,109,110,113,184}. Notably, in human cells, the cohesin complex is subjected to post-translational modifications following DSBs, including the ATM-mediated phosphorylation of SMC1 Ser966 and Ser957 and of SMC3 Ser1083 (ref. 185), and the acetylation of SMC3 by ESCO2 (ref. 182). SMC3 acetylation modulates loop extrusion¹⁸⁶ and cohesin phosphorylation regulates cohesin residency on chromatin¹⁸⁷. As damaged TADs are enriched in modified cohesin subunits^152,182, it is tempting to speculate that cohesin modification affects DSB-flanking loop extrusion dynamics, resulting in increased loop strength in damaged TADs, as has been revealed by Hi-C experiments^{33,152,188,189}.

DSB-induced loop extrusion has crucial functions in the initiation of the DDR and in DSB repair. First, although not formally demonstrated, the fact that DSBs can act as barriers for loop extrusion raises the possibility that SMC complexes sense DNA lesions while translocating along chromosomes. Second, the cohesin complex was found to regulate the joining of distant DSB ends¹⁹⁰ and computational, experimentally constrained simulations suggested that DNA loop extrusion favours the synapsis of DNA ends, thereby promoting end-joining¹⁶³. In addition, loop extrusion dictates the use of the homologous donor in yeast, favouring intra-chromosomal over inter-chromosomal donors¹⁵³, in agreement with the observation that intra-chromosomal recombination correlates with contact frequencies determined by chromosome conformation capture^191,192. Finally, loop extrusion at DSBs promotes γH2AX spreading across nucleosomes brought into physical proximity to the DSB-associated kinase (ATM) during the extrusion process¹⁵². The surrounding CTCF-bound loop boundary elements further halt loop extrusion, therefore confining γH2AX spreading to the damaged TAD and, to a lesser extent, to the surrounding TADs depending on the strength of TAD boundaries¹⁵² (Fig. 4a). Additional DSB-induced chromatin modifications can take place across the damaged TAD, including eviction of linker histone H1, histone polyubiquitylation, and recruitment of mediator of DNA damage checkpoint 1 (MDC1) and 53BP1 (ref. 193). Whether loop extrusion also contributes to these chromatin modifications is currently unknown. Together, these studies suggest that cohesin-mediated loop extrusion at DSBs ensures the spreading of γH2AX to rapidly barcode damaged TADs, thereby initiating the chromatin response to DSBs (Fig. 4a).

The damaged TADs further fold and adopt a particular structure in the nucleus, as shown by super-resolution imaging. 3D-structured illumination microscopy has revealed the fine structure of γH2AX and 53BP1 (γH2AX–53BP1) condensates in human cells, comprising 53BP1 nanodomains (approximately 60–180 nm in diameter) separated by CTCF and further assembled into a circular domain (up to a few micrometres in size) surrounding BRCA1 and RPA foci^194–197. Interestingly, in mammalian cells, this arrangement of 53BP1 nanodomains depends on Schlafen 5 (ref. 198) and on RIF1 (ref. 196), which regulates chromosome architecture. Although the exact mechanism by which RIF1 functions is not fully understood, its intrinsically disordered region¹⁹⁹ could promote weak multivalent interactions between nanodomains, thereby forming an organized micro-compartment. The significance of this multilayered organization of γH2AX–53BP1 condensates is unclear, but it suggests that the DSB is buried inside the circular domain, which also agrees with Hi-C data showing a decreased interaction of the DSB with neighbouring TADs³⁴. This organization is consistent with the role of DSB mobility in lesion sequestration, escape from repair-repressive domains or avoidance of ectopic repair (Fig. 2). Such local reorganization might also explain why long-range dynamics of repair sites are less-frequently observed in mammalian cells, where the γH2AX–53BP1 local domain might similarly mediate the spatiotemporal separation of repair sites from surrounding chromatin. Besides these TAD-sized chromatin modifications, Hi-C experiments in yeast have identified a cohesin-independent, local chromatin folding at the site of damage termed local interaction pattern (LIP)¹⁵³. The LIP was suggested to reflect DSB-end tethering¹⁵³, but a recent study indicated that end tethering involves cohesin²⁰⁰, so the contribution of the LIP requires further investigation.

Compartmentalization of damaged chromatin in the nucleus

Human genome A-compartments and B-compartments are maintained following DSB induction, though DSBs induced in the B-compartment can occasionally shift to the A-compartment^33,34. This phenomenon may be analogous to DSBs escaping repair-repressive domains or relocating to repair-promoting regions (Fig. 2a). Hi-C analyses have revealed that damaged TADs marked with γH2AX–53BP1 can self-segregate³⁴, consistent with the clustering of DSB foci observed from yeast to human cells^{11,12,20,25,27,31,32,34,58,110,124,201–205} (Fig. 3f). Recent Hi-C data obtained in human cells has also indicated that entire damaged TADs can co-segregate, forming a new chromatin compartment referred to as the ‘D-compartment’ (for DSB-induced)³⁴ (Fig. 4b). The formation of this compartment, as detected by Hi-C in human cells³⁴, is consistent with the fusion of DNA repair foci in live cells in different organisms^{10,12,20,25,32,201,206}. Notably, DSB clustering commonly involves difficult-to-repair lesions at either transcriptionally active loci^25,31,32,34 or pericentromeric heterochromatin^25,27,110 in mammalian or fly cells. Moreover, inhibition of DNA-PKcs in human cells, which strongly impairs DSB repair, boosts DSB clustering and D-compartment formation^33,34,207. These findings suggest that compartmentalization of γH2AX–53BP1-associated chromatin occurs especially when DSBs persist long enough to sustain the γH2AX–53BP1 signal, thereby permitting the segregation and clustering of difficult-to-repair foci.

Although the formation of the D-compartment awaits additional investigation, recent evidence suggests two potentially complementary mechanisms. On one hand, formation of this compartment in human cells depends on ATM activity and on 53BP1 (ref. 34), which also mediates the movement and clustering of dysfunctional telomeres in mouse cells^29,208 (Fig. 3b). Interestingly, human 53BP1 has the ability to undergo phase separation and form droplets in vitro, and 53BP1 foci exhibit the typical behaviour of phase-separated nuclear bodies in mammalian cells^201,206. Recent work has additionally indicated that 53BP1 can bridge chromatin at different DSBs, consistent with bridging-induced phase separation (also known as polymer–polymer phase separation) or phase separation coupled to percolation^34,209 (Box 1). Interestingly, DNA-damage-induced phase separation also depends on R-loops and RNAs that have been found to contribute to HR^34,206. However, the exact contribution of RNA to long-distance chromatin contacts and γH2AX–53BP1 domain formation awaits further investigation. Of note, as reported in a preprint, impairing the anchoring of DSBs to the nuclear envelope enhances D-compartment formation¹²⁵. This effect is reminiscent of the behaviour of H3K9me3 heterochromatin domains, which, when detached from the nuclear lamina, coalesce into large internal heterochromatic foci that can trigger the formation of ‘inverted’ nuclei in extreme cases²¹⁰. Therefore, a growing body of data argues for the role of multivalent interactions involved in phase separation in the compartmentalization of damaged, γH2AX–53BP1-marked TADs.

On the other hand, DSB clustering in human and fly cells depends on the Wiskott–Aldrich syndrome family proteins (WASP), ARP2–ARP3 or Formin-2, which promote polymerization of nuclear actin^25,31–33 (Fig. 3f). The dependence on nuclear F-actin suggests that the directed motion of DSBs could contribute to DSB clustering and D-compartment formation. Notably, neuronal WASP (N-WASP) can form a membrane-associated biomolecular condensate through liquid–liquid phase separation and promote ARP2–ARP3-dependent actin polymerization in the cytoplasm²¹¹. Thus, the increased concentration of monomeric actin, WASP and ARP2–ARP3 within a biomolecular condensate assembled at DSB sites could stimulate actin polymerization. This notion is supported by the observation that short microtubule-dependent forces promote the clustering of Rad52 liquid-like droplets in yeast²¹². The resulting larger droplets then concentrate tubulin and project long microtubule filaments, thereby promoting the directed transport of DSBs to the nuclear periphery²¹². Similarly, a recent preprint reported that human RAD52 forms liquid-like droplets that project RAD51–ssDNA fibrillar structures²¹³, though it is unclear whether they contribute to DSB mobility during HR repair. How F-actin-driven movement intersects with the phase separation of repair domains requires additional investigation²¹⁴. Interestingly, as reported in a recent preprint and a prior study, DSB clustering in flies specifically requires Arp2–Arp3 in euchromatin and Nup98 in heterochromatin, revealing distinct mechanisms of clustering in different chromatin environments^25,110 (Fig. 3f).

The D-compartment may fulfil different biological functions. For instance, in addition to gathering damaged TADs, this compartment attracts some DDR genes³⁴ (Fig. 4b). Relocalization of DDR genes within the D-compartment may explain why DNA repair foci in human cells are enriched with different transcription factors²¹⁵, including the tumour suppressor p53 (ref. 201). Although the mechanisms mediating the segregation of DDR genes are unknown, D-compartment-associated genes display more R-loops than other DDR-responsive genes³⁴, suggesting a potential function of R-loops in this pathway. Of note, targeting to the D-compartment contributes to the transcriptional activation of these genes following DSB induction. Accordingly, 53BP1 was identified as crucial for the full activation of p53 target genes in mammals^216,217, and disrupting the formation of 53BP1 droplets suppressed p53 signalling and checkpoint activation^201,217. Thus, formation of the D-compartment may set the magnitude of the DDR depending on DSB load and persistence. However, DSB clustering also increases the risk of illegitimate rejoining of break ends and chromosomal translocations. Indeed, a translocation between two genetic loci can arise from their increased DSB-induced proximity¹²⁴, and D-compartment formation also favours translocation^33,34. DSB clustering might also promote repair progression (Fig. 3f), for example by enhancing end resection and HR³².

Formation of DNA repair condensates

Once assembled, DSB repair foci resemble biomolecular condensates, which are membrane-less organelles that selectively concentrate molecules involved in DNA repair around damaged areas of the genome while excluding non-repair factors^62,197,218. Concentrating repair factors around DNA lesions favours biochemical reactions required for repair, whereas exclusion of proteins that are not involved in DNA repair limits conflicts between repair and transcription or replication and prevents inappropriate biochemical reactions. Several mechanisms have been proposed to contribute to the formation of DNA repair condensates (Box 1). Although distinct material properties are associated with each of these models, they are not mutually exclusive and may cooperate in the complex environment of the nucleus to shape DNA repair in a time-dependent and context-dependent manner.

First, histone modifications behave as ensembles or clusters of molecular binding sites on chromatin surrounding DNA lesions²¹⁹. They can selectively engage with and enrich repair factors through their chromatin reader domains, such as MDC1’s BRCA1 C-terminal (BRCT) domain, which recognizes the phosphorylated residue of γH2AX²²⁰. DNA-damage-induced spreading of some of these repair factors along the chromatin fibre can occur through mechanisms of feed-forward amplification. For example, damage-induced ubiquitylation by the E3 ubiquitin ligase ring finger protein 168 (RNF168) spreads along chromatin owing to the ability of RNF168 to bind to its own catalytic product²¹⁹. Clustered binding, which typically involves fixed stoichiometries, can selectively enrich molecules without requiring multivalency, cooperativity or protein network interactions¹⁹⁷. Emerging evidence suggests that DNA repair compartments combine clustered binding and phase separation in its various forms (Box 1). The composition and physical properties of DNA repair condensates is time-dependent and context-dependent, enabling transition between and mixing of states. For example, such states may include chromatin-associated clustered binding coupled with phase separation in the surrounding environment, eventually leading to increased network interactions and gelation^{197,201,206,221}.

One of the first responses to DNA damage is the activation of PARP enzymes, which generate poly(ADP-ribose) (PAR), a highly anionic nucleic acid-like biopolymer, at sites of DNA lesions^222,223 (Fig. 5a). Within seconds of DNA damage induction in mammalian cells, PARylation leads to liquid demixing and assembly of a multitude of PAR-binding proteins, several of which can drive phase separation²²⁴. Among them is fused in sarcoma (FUS), an intrinsically disordered protein with a high propensity to undergo physiological phase separation and aberrant aggregation in disease^224–226. The PAR-dependent protein assembly at DNA break sites is fast, yet transient, and kept in check by poly(ADP-ribose) glycohydrolase-mediated degradation of PAR chains. With somewhat slower kinetics, γH2AX formation is induced by ATM around DNA lesions, a phenomenon that is required for the formation of the γH2AX–53BP1 repair compartment (Fig. 5b). This repair compartment has liquid-like properties, but it gradually acquires features associated with a viscoelastic network fluid^201,206 (Box 1).

Fig. 5 |. Chromosome topology and phase separation cooperate to stabilize DNA repair compartments.

a, An immediate response to DNA double-strand breaks (DSBs) is poly(ADP-ribose) (PAR) formation by PAR polymerase 1 (PARP1) and PARP2, which leads to the assembly around DSBs of multiple PAR-binding, RNA-binding and DNA repair factors in human cells (for example, fused in sarcoma (FUS), EWSR1, TAF15 (collectively known as FET proteins) and related heterogeneous nuclear ribonucleoproteins (hnRNPs)), primarily through multivalent interactions with the PAR scaffold. Whereas single-strand breaks (SSBs) are dealt with by PAR-dependent SSB repair (SSBR), carried out by XRCC1, DNA polymerase-β (Polβ) and DNA ligase 3 (LIG3), clustered SSBs and DSBs are dealt with by PAR-facilitated non-homologous end-joining (NHEJ), involving KU70–KU80, XRCC4, XLF and LIG4, or by microhomology-mediated end-joining (MMEJ), depending on the cell cycle phase and the complexity of the lesion. b, If fast repair by SSBR, NHEJ or MMEJ is not possible, or if a sister chromatid with undamaged homologous sequences is available in S or G2 for faithful repair by homologous recombination (HR), then ATM-mediated phosphorylation of H2AX (γH2AX), RING finger protein (RNF)8-dependent and RNF168-mediated ubiquitylation of H2A and H2AX, and poly(ADP-ribose) glycohydrolase-dependent PAR degradation (not shown) result in remodelling of the DNA damage compartment and formation of the γH2AX–TP53-binding protein 1 (53BP1) domain. 53BP1 limits DNA end resection and the formation of single-stranded DNA (ssDNA), thus reducing the recruitment of ssDNA-binding proteins such as replication protein A (RPA), radiation sensitive (RAD)51 and RAD52. The antagonism between 53BP1 and the HR-promoting breast cancer type 1 susceptibility protein (BRCA1)–BRCA1-associated RING domain protein 1 (BARD1) complex, which is regulated by the absence or presence of an undamaged sister chromatid sequence, determines the choice of DSB repair pathway. c, Persistent breaks, which cannot be easily repaired by NHEJ or by HR, or which are actively halted from repair to prevent aberrant recombination (for example, at ribosomal DNA (rDNA) or pericentromeric heterochromatin), may form a more stable DNA repair compartment, which exhibits increased mobility to enable (micro)homology search (for example, to search for homologous sequences if the sister chromatid sequence is not available) or to escape nuclear environments that are unfavourable for repair. This movement can be facilitated by nuclear actin (F-actin) and, in yeast, by nuclear microtubules, thereby enabling DSB clustering or formation of the D-compartment in transcribed regions. The clustering of DSBs within repair centres can be beneficial for repair, but is also associated with an increased risk of chromosomal translocations. SAFA, scaffold-attachment factor A; NONO, non-POU domain-containing octamer-binding protein; SFPQ, splicing factor, proline- and glutamine-rich; DDX3X, DEAD box protein 3, X-chromosomal; MRN, MRE11–RAD50–NBS1 complex; BLM, Bloom syndrome protein; TOPBP1, DNA topoisomerase II-binding protein 1.

Owing to their physical properties and related ability to attract or repel repair proteins, molecular condensates and their multilayered organization may regulate the choice of DNA repair pathway. For instance, 53BP1 accumulation protects DNA ends from excessive resection and can thereby antagonize HR. Interestingly, ssDNA-binding proteins and several mediators of homology-directed repair also form condensates, from bacteria to human cells^{20,110,221,227–237}, many of which promote key repair events and some also form within a larger repair domain, such as the 53BP1 microdomain in mammalian cells¹⁹⁶. This suggests a broader involvement of phase separation in DNA repair than previously appreciated and also indicates the existence of spatial segregation of HR processes within 53BP1-marked DNA repair centres.

PAR-induced FUS assembly also facilitates repair in part by preventing the separation of broken DNA ends^238–243. Components of the mammalian NHEJ machinery, which are needed to re-ligate tethered DNA ends, form condensates through networks of multivalent interactions to stimulate DNA end ligation²⁴⁴. PARP and PAR-dependent single-strand break repair involves enhanced nucleosome dynamics and the condensation of the repair factors DNA polymerase-β, XRCC1 and DNA ligase 3 to stimulate repair^245–247. Condensates can also selectively exclude repair factors, as reported in a preprint for Nup98-induced condensates that prevent Rad51 recruitment to repair foci in heterochromatin¹¹⁰ (Fig. 3e).

Chromatin context and the cell cycle strongly influence the spatiotemporal dynamics of DSB repair condensates. In replicated parts of the human genome in S and G2, where the sister chromatid sequence can be used as a template for HR, the initial assembly of 53BP1 around DSBs is antagonized by BRCA1 and its partner protein BARD1, which bind to unmethylated H4K20 in the chromatin of newly synthesized DNA^248,249. BRCA1–BARD1 binding generates a resection-permissive compartment that is surrounded by 53BP1, which in turn provides a barrier against excessive, mutagenic resection^194,250 (Fig. 5b). Conversely, in non-replicated parts of the genome, and particularly in the G1 phase when sister chromatids are absent and DNA end resection is inhibited, 53BP1 assembly is dominant over BRCA1 (ref. 251). In such conditions, complex or persistent DSBs, which are prevented by 53BP1 from undergoing HR and which cannot be easily re-ligated by classical NHEJ, further stabilize the 53BP1 compartment and also promote DSB clustering, thereby fostering DSB repair at the expense of an increased translocation frequency (Fig. 5c).

In summary, active DSB-induced processes such as H2AX phosphorylation by ATM and cohesin-mediated loop extrusion cooperate with self-associating properties of DSB repair factors to dynamically compartmentalize damaged chromatin, and sequentially remodel repair condensates to facilitate and coordinate repair in a context-dependent manner.

Summary and future directions

Emerging evidence suggests that a set of nuclear and genome organization hallmarks underly the dynamics of DSB repair (Fig. 6). These hallmarks operate at different but interconnected spatial organization scales and implicate numerous proteins, many of which are linked to human disease and ageing (Box 2 and Supplementary Table 1). However, key gaps remain in our knowledge. For instance, the detailed molecular processes through which cells temporally and spatially coordinate such a diverse set of hallmarks will require further research. On that front, a key question is whether cells use common DSB-sensing effectors, such as the DDR kinases, to simultaneously drive the nuclear and cytoplasmic processes underlying repair; conversely, it remains unclear which factors are specific for different DSB-relocalization pathways. Furthermore, considering the recent identification of the intersection of DDR kinase activities with tubulin modifications in driving nuclear envelope remodelling for repair, future studies should address whether a more elaborate ‘code’ of post-translational modifications affecting tubulin, actin and motor proteins influences nuclear remodelling and DSB mobility for repair. Similarly, how various chromatin modifications and non-chromatin-associated post-translational modifications are coordinated to modulate nuclear and genome organization for DSB repair remains under-explored.

Fig. 6 |. Integrated view of nuclear and genome organization hallmarks underlying DSB repair.

The proposed hallmarks are arranged into three categories, defined by the relative scale or size at which the underlying mechanisms operate to drive DNA double-strand break (DSB) repair. The distinction between the different mechanisms within each scale category and the transitions between scale categories should not be viewed as a strict barrier, but as a gradual transition. This consideration is especially important as certain hallmarks can operate at different scales. For instance, phase separation and chromatin loop extrusion can function and cooperate at the interface of the small and medium scales. Furthermore, processes operating across the different scales can exhibit two-way communication or coordination (bi-directional arrows), highlighting the interconnectedness of these hallmarks.

Box 2 |. Nuclear and genome organization factors in disease and ageing.

Mutations in DNA repair factors are linked with disorders characterized by chromosome instability, cancer predisposition, growth deficiencies, and neurological and immune defects²⁵⁶. Such disorders include ataxia telangiectasia (caused by ATM mutations), Nijmegen breakage syndrome (NBS1 mutation), Bloom syndrome (BLM mutations) and Werner syndrome (WRN mutations). These diseases also exhibit molecular and clinical similarities to natural or premature ageing.

Laminopathies and envelopathies — disorders caused by mutations in genes encoding lamins or their interacting nuclear-envelope-embedded proteins, respectively — are intimately linked to ageing²⁵⁷. LMNA mutations lead to compromising of the nuclear lamina through the expression of a protein called progerin, thereby causing Hutchinson-Gilford progeria syndrome (HGPS), which is characterized by severe premature ageing²⁵⁸. Cells of individuals with HGPS show anomalies in the nuclear lamina, lamina associated domains, nuclear envelope and their associated proteome^257,258. This phenotype is mirrored by gene expression changes and DNA damage build-up²⁵⁷. HGPS is characterized by decreased repair through non-homologous end-joining owing to delayed recruitment of TP53-binding protein 1 (53BP1) and reduced function of DNA-dependent protein kinase catalytic subunit (DNA-PKcs) and KU70 (refs. 259,260); disruption of KU70 or KU80 also causes early ageing in mice^261,262. RAD51 recruitment and repair through homologous recombination are also delayed, and poly(ADP-ribose) polymerase 1 (PARP1) function is decreased in HGPS cells^259,263. Considering the contribution of nuclear lamina–DNA double-strand break (DSB) contacts to DNA repair, it will be crucial to assess whether altering such contacts contributes to disease and whether restoring optimal contacts is a viable longevity-promoting therapeutic strategy. Concordantly, inhibiting progerin farsenylation, which typically allows this mutant protein to engage the inner nuclear membrane and disrupt the lamina, using the FDA-approved drug lanofarnib, improves clinical outcomes and decreases the chance of death from HGPS^{257,264–266}.

Like ageing cells, cancer cells have high baseline levels of DNA damage. Cancer cells exhibit increased dependence on DSB-capturing nuclear envelope tubules (dsbNETs) for DNA repair and cell survival³⁸. Disrupting dsbNETs increases chromosome instability and decreases the ability of breast cancer type 1 susceptibility protein homologue (BRCA1)-mutant triple-negative breast cancer cells to form tumours in mice³⁸. Similarly, the higher propensity of cancer-cell nuclei for rupture makes them potentially hyper-sensitive to interventions such as farnesyl transferase inhibition, which can partly and transiently disrupt the nuclear lamina¹³⁶. Furthermore, given the role of nuclear reshaping through mechano-transduction in DSB repair^{38,133,134,136}, it will be important to assess how much of the observed nuclear deformations in cancer or ageing represents bona fide nuclear defects, beneficial DNA repair or damaged cell clearance.

It will also be informative to evaluate how disease-causing mutations in genes encoding cohesin, tubulin, actin or their associated factors afect their roles in repair and how such disruptions contribute to disease. For instance, cohesinopathies such as Cornelia de Lange syndrome, are caused by mutations in genes encoding cohesin subunits or their regulatory factors, including nipped-B-like (NIPBL), structural maintenance of chromosome (SMC)1A or SMC3 (refs. 267–271). Cohesinopathies are characterized by developmental abnormalities and increased cancer risk²⁶⁷. In addition, tubulinopathies are caused by mutations in tubulin genes such as in TUBA1A in lissencephaly, and actinopathies are linked to actin gene mutations such as in ACTG1 in Baraitser-Winter syndrome^272,273. Kinesinopathies and dyneinopathies are linked to mutations in genes encoding kinesin or dynein motor proteins, respectively^274–278. Tubulinopathies, actinopathies, kinesinopathies and dyneinopathies can be characterized by developmental delays, intellectual disability and increased cancer incidence. The pathobiological mechanisms underlying these diseases are unclear and their characterization should benefit from exploring connections to DSB dynamics and repair.

Nuclear, genomic and DSB dynamics during DSB repair may not be exclusively beneficial and could instead induce chromosome instability. For example, the controlled release of ribosomal DNA (rDNA) repeat units (ribosomal RNA genes) carrying DSBs from the human or yeast nucleolus can promote DSB repair^2,13,42,77. By contrast, the excessive mobility of rDNA DSBs and their release from nucleoli induces aberrant recombination, which compromises chromosome stability and shortens cellular lifespan^2,73. Similarly, increased DSB mobility and D-compartment formation may promote repair, but at an elevated risk of chromosomal translocations^31–34. Such increases in translocation can be exploited clinically. For example, BRCA1-deficient cancer cells are more dependent on PARP for DNA repair and survival, and thus the PARP inhibitor olaparib can restrain these cancer cells by further inhibiting DSB repair, instead driving chromosome fusion events in a manner dependent on dynamic microtubules, the LINC complex and kinesins^29,38. This process eventually triggers cell senescence and restrains cancer cell growth in vitro and in vivo³⁸.

On another front, the threshold of DNA-damage load necessary to trigger formation of the D-compartment and the potential impact on DNA repair pathway choice should be explored further. It may reveal a proportional relationship between DNA-damage load and the size of the D-compartment, and the number and types of DDR genes the compartment attracts. In addition, a more detailed investigation is required of the molecular processes that coordinate loop extrusion, phase separation and other chromosomal events that control γH2AX spreading and repair-centre formation. Future work should also decipher the precise events and signalling cues that are responsible for the sequential restructuring of the composition of repair centre condensates and their ability to ensure the ordered progression of DNA repair. How condensates affect nuclear dynamics of repair sites in different contexts also needs further exploration. Finally, considering the robust links of cancer, ageing and age-related diseases with the hallmarks of nuclear and genome organization underlying DNA repair, it will be crucial to assess whether the modulation of factors controlling these hallmarks uncovers avenues for the development of new therapeutics.

Supplementary Material

Additional information

Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41580-025-00828-1.

Glossary

Anomalous Rouse diffusion

A type of sub-diffusive motion where a single particle, such as a segment of a DNA polymer, moves slower than expected for normal diffusion owing to physical constraints, including those imposed by polymer entanglement

Brownian motion

Random movement of particles suspended in a medium, such as a liquid, with no preferential directionality

Cajal bodies

Nuclear bodies first reported by Santiago Ramón y Cajal in 1903 often associating with the nucleolus and containing RNA processing factors and involved in the biogenesis of small nuclear ribonucleoprotein particles

DNA repair foci

Microscopically discernible accumulation of DNA repair proteins at sites of DNA damage. Also referred to as ionizing radiation-induced foci (or IRIF) when induced by ionizing radiation

Loop extrusion

An energy-dependent process carried out by structural maintenance of chromosomes complexes, wherein chromatin is reeled in by a molecular motor and extruded as a loop. Loop extrusion contributes to genome organization and stability

Nucleolus

Large membrane-less nuclear compartment, where ribosomal (rDNA) is transcribed and ribosome biogenesis occurs. A hotspot of genomic instability owing to the highly repetitive nature of rDNA, which also harbours replication-fork-blocking and double-strand break-inducing elements

Polycomb bodies

Microscopically discernible accumulation of polycomb group (PcG) proteins in the nucleus, associated with PcG-dependent gene repression.

Promyelocytic leukaemia bodies

Nuclear bodies characterized by the promyelocytic leukaemia (PML) protein and multivalent interactions between SUMOylated proteins and proteins containing SUMO-interacting motifs. PMLs are involved in multiple cellular processes, including in telomere repair

Super-enhancer

Nuclear cluster of gene enhancers; associated with the accumulation and clustering of transcription factors and coactivators and active histone modifications such as H3K27ac

Topologically associating domain

Genomic region of ~1 Mb in human cells, in which DNA sequences interact more frequently with each other than with sequences at other genomic regions. Topologically associating domain borders are enriched in CCCTC-binding factor and cohesin binding