Incorporation of 53BP1 into phase-separated bodies in cancer cells during aberrant mitosis

The p53-binding protein 1 (53BP1, also known as TP53BP1) is a large multifunctional protein best known as a key mediator of the non-homologous end joining (NHEJ) DNA repair pathway (Aparicio et al., 2014; Shibata, 2017). Recruitment of 53BP1 to DNA damage sites requires its phosphorylation by ATM and ATR (ATM/ATR) kinases and interaction with modified chromatin at double-strand breaks (DSBs). 53BP1 clusters at DNA repair foci in a complex architecture with RIF1, the shieldin complex and other components (Ochs et al., 2019). These foci have liquid–liquid phase separation (LLPS) properties, which likely serve to retain repair components close to the break site (Kilic et al., 2019; Piccinno et al., 2019). In these foci, 53BP1 alters the topological arrangement of the chromatin and serves as an assembly platform for effectors, such as RIF1, PTIP (also known as PAXIP1) and others. The assembled complex prevents binding of BRCA1, thus shielding the broken DNA ends from excessive nucleolytic digestion. This blocks steps toward homology-directed repair (HDR) and channels DNA repair toward NHEJ.

53BP1 can be found in cells throughout the cell cycle. However, NHEJ is most active in G1 phase, before sister chromatids are available for HDR. In addition, 53BP1 forms nuclear bodies around unresolved replication intermediates arising during S phase that are transmitted through mitosis into G1 phase (Harrigan et al., 2011; Lukas et al., 2011). These 53BP1 nuclear bodies are thought to shield problematic DNA sites until appropriate DNA repair pathways are reactivated following the completion of mitosis.

The functional capabilities of 53BP1 are owed to its complex structure (Panier and Boulton, 2014; Mirman and de Lange, 2020). 53BP1 comprises three main regions: (1) an N-terminal region containing multiple Ser/Thr-Gln (S/T-Q) sites that can be phosphorylated by ATM/ATR kinases, which is a prerequisite for 53BP1 recruitment to DSBs; (2) a central region that includes an oligomerization domain (OD), two tandem Tudor domains (TDs) and additional motifs, which together make the minimal focus-forming region (MFFR) essential for recognition and binding to methylated and ubiquitylated chromatin at DSBs; and (3) a C-terminal tail that includes two tandem BRCA1 C-terminal (BRCT) domains involved in amplification and transmission of DNA damage signaling, mainly through the MRN complex–ATM axis and p53 (also known as TP53). The N- and C-termini of 53BP1 contain intrinsically disordered regions that confer high conformational flexibility and may serve as hubs for transient multiprotein interactions and their compartmentalization (Kilic et al., 2019). The OD domain has been shown to be critically involved in 53BP1 phase separation at sites of DNA damage, with the BRCT domains also positively contributing to this process (Kilic et al., 2019). Interestingly, the largely disordered N terminus appears to be dispensable for 53BP1 LLPS at sites of DNA damage during interphase (Kilic et al., 2019).

Although 53BP1 has well established roles in the DNA damage response during interphase (Mirza-Aghazadeh-Attari et al., 2019; Zhang and Gong, 2021), less is known about its functions during mitosis, when DNA repair pathways are suppressed. In mitotic cells, 53BP1 is detected at kinetochores, where it facilitates the resolution of merotelic attachments, and its depletion increases chromosome lagging and missegregation (Jullien et al., 2002; Wang et al., 2017). 53BP1 has also been shown to limit sister chromatid intertwinements formed in S phase to prevent ultrafine bridges during cytokinesis (Tiwari et al., 2018). However, 53BP1 activity in mitotic cells must be strictly regulated, since aberrant activation of NHEJ during mitosis can lead to telomere fusion and the induction of lethal chromosome breakage–fusion cycles, particularly in cells with HDR defects (Cesare, 2014; Orthwein et al., 2014). The regulation of 53BP1 might be critically important in response to mitotic stress, as delayed mitosis may increase frequency of DNA defects, chromatin bridging and telomere fusion.

In previous work, we have found that most colon cancer cells released from mitotic arrest divide successfully. However, some cells undergo apoptosis, with the extent of apoptosis reflective of the DNA damage response during mitosis (Chopra et al., 2016). Cells arrested in mitosis typically activate a DNA damage response, although the cause and role of this DNA damage response are not completely understood (Tu et al., 2013; Martín et al., 2014). In our previous work, we have observed that AK3-derived mitotic inhibitors that target the mitotic spindle generate a strong H2AX phosphorylation (γH2AX) response (Chopra et al., 2016; Bond et al., 2018). Given the importance of 53BP1 in DNA damage responses and apoptosis, we investigated 53BP1 regulation in colon cancer cells arrested in mitosis by AK306 and other arrest agents. We found that in cells entering mitosis, 53BP1 localizes to centromeres, as previously reported (Jullien et al., 2002; Wang et al., 2017). However, in arrested cells, we observed the formation of large, roughly spherical 53BP1 foci, referred to as mitotic stress bodies (MSBs). These bodies form regardless of the arrest agent employed and are also observed in a small percentage of normally dividing cancer cells. Here, we report the properties of 53BP1 MSBs, the molecular determinants that underlie their formation, and their potential role in regulating 53BP1 activity and cell fate after mitosis.

In agreement with published data, we detected small 53BP1 foci at centromeres from early prophase to metaphase in normally dividing cells (Fig. 1A,B, left panels). However, in mitotically arrested HCT116 colon cancer cells, we observed large 53BP1 bodies close to the mitotic chromatin. Each mitotic cell had, on average, 5–6 53BP1 bodies (Fig. 1B, middle panel). These bodies included 53BP1 phosphorylated at N-terminal ATM/ATR kinase target residues (Fig. 1C). When the arrest agent was withdrawn and cells were allowed to divide, these structures decreased in number, moved away from the chromatin and disappeared upon completion of cytokinesis (Fig. 1B, right panels).

To ensure that our staining was specific, we performed an siRNA-mediated knockdown experiment. Transfection of HCT116 cells with 53BP1-targeting siRNA, followed by mitotic arrest, almost completely abolished formation of the 53BP1 MSBs. Cells transfected with control non-targeting siRNA were not affected (Fig. 1D). We found these structures in cells treated with different mitotic arrest agents; treatment with AK306, microtubule destabilizers (nocodazole and Colcemid), a microtubule stabilizer (paclitaxel) or a topoisomerase II inhibitor (ICRF-193) all resulted in generation of a similar number of bodies that had a similar size (Fig. 1E,F).

Finally, formation of 53BP1 MSBs was observed in both chromosomally stable HCT116 cells and chromosomally unstable HT29 cells (Fig. 1G, left and middle panels); however, 53BP1 MSBs were not observed in arrested retinal epithelium cells (ARPE-19; Fig. 1G, right panel).

As a large protein with extensive intrinsically disordered regions, 53BP1 has been reported to form LLPS assemblies in interphase cells at sites of DNA damage (Kilic et al., 2019; Piccinno et al., 2019). Stimulated emission depletion (STED) imaging revealed that 53BP1 MSBs were spherical structures with a diameter of ∼1 μm (Fig. 2A), which is consistent with these mitotic bodies having liquid-like droplet properties. We further tested this possibility by determining their sensitivity to high salt concentration and osmotic stress. As shown in Fig. 2B, 53BP1 MSBs were disrupted by treatment with either NaCl or sorbitol. The disruption affected all cells with MSBs and was evident by the residual presence of dispersed 53BP1 in place of the discrete rounded structures characteristic of MSBs. Finally, 1,6-hexanediol, which is frequently used to disrupt LLPS structures, completely dissolved the 53BP1 MSBs (Fig. 2C, left panels). Washout of 1,6-hexanediol allowed the bodies to reform, further supporting their dynamic nature (Fig. 2C, right panels).

Since 53BP1 is positioned near mitotic centromeres in normally dividing cells, we investigated whether the 53BP1 MSBs were similarly positioned. Analysis of cell images in 3D space showed that most 53BP1 MSBs were positioned near centromeres (Fig. 3A). Time course analysis showed that formation of the 53BP1 MSBs began within 1 h of arrest with the accumulation of 53BP1 in discrete foci at a subset of centromeres (Fig. 3B, left panel). Over time, the number of these bodies per cell decreased, while their area increased (Fig. 3C). Some off-chromatin bodies were observed during prolonged mitotic arrest, accounting for ∼13% of all bodies after 22 h of arrest (Fig. S1). Although we don't yet know which chromosomes associate with the 53BP1 bodies, staining for UBF (also known as UBTF) showed that these bodies did not specifically associate with the five acrocentric chromosomes with nucleolus organizer regions (Fig. S2).

After release from mitotic arrest, the majority of 53BP1 MSBs dissociated from the chromatin (Fig. 3D). This process began once the cells resumed mitosis. At the advanced mitotic stages, cells typically had a decreased number of MSBs, with almost all of them not associated with chromatin; however, some cells displayed a few small chromatin-associated 53BP1 foci (Fig. 3D). It is not clear whether these were residual or newly formed entities.

A series of co-staining experiments were performed to identify other components in the 53BP1 MSBs. Probing for PML, coilin and nucleophosmin showed no colocalization of these proteins with 53BP1 MSBs (data not shown). We then examined colocalization with lamin A/C. Lamin A/C interacts with 53BP1 and helps retain 53BP1 in the nucleus of interphase cells (Gibbs-Seymour et al., 2015; Mayca Pozo et al., 2017). Following nuclear envelope breakdown in mitosis, some of the lamin A/C associates with the chromatin in proximity to the centromere, where it is important for proper spatial positioning of centromeres and kinetochore assembly (Taimen et al., 2009; Eisch et al., 2016). As shown in Fig. 4A, much of the lamin A/C was found dispersed in the cytoplasm of mitotically arrested cells. However, areas where lamin A/C was concentrated overlapped with 53BP1 MSBs (Fig. 4A, top panels). As described previously, some interphase cells possess multiple 53BP1 foci, which might be related to S-phase DNA breakage (Andreassen et al., 2006). However, unlike the mitotic 53BP1 bodies, interphase 53BP1 foci did not include lamin A/C (Fig. 4A, bottom panels). STED microscopy confirmed localization of lamin A/C to 53BP1 MSBs, where it was found in areas of low 53BP1 density within the bodies (Fig. 4B). 53BP1 knockdown abolished the mitotic lamin A/C foci, which is consistent with 53BP1 driving lamin A/C accumulation (Fig. 4C). Like the 53BP1 MSBs, the lamin A/C foci were disrupted by treatment with 1,6-hexanediol (Fig. 4D, top panels) and could reform after 1,6-hexanediol washout (Fig. 4D, bottom panels).

Another protein found to colocalize with the 53BP1 MSBs was RIF1 (Fig. 4E). RIF1 is a protein involved in DSB repair that binds to phosphorylated 53BP1 (Chapman et al., 2013; Escribano-Díaz et al., 2013; Zimmermann et al., 2013). It also helps resolve ultrafine DNA bridges at the centromere during mitosis (Hengeveld et al., 2015).

We have previously shown that treatment with AK3 compounds leads to an elevated DNA damage response marked by induction of histone H2AX phosphorylation (Chopra et al., 2016). Thus, we determined whether 53BP1 MSBs localize to sites of phosphorylated H2AX. Co-staining of AK306-arrested cells for 53BP1 and γH2AX showed little overlap between 53BP1 and γH2AX foci, (Fig. 5, left panels), whereas γH2AX-positive interphase cells from the same experiments showed almost complete colocalization of 53BP1 and γH2AX staining (Fig. 5, right panels). These results suggest that 53BP1 MSBs do not localize to DNA DSBs.

We next sought to determine the regions of 53BP1 involved in the formation of MSBs. For these experiments we generated GFP–53BP1 fusion constructs and transiently transfected them into HCT116 cells. The choice of the domains was based on the reported structural domains of 53BP1. These domains include the N-terminal domain, with S/T-Q repeats phosphorylated by ATM/ATR kinases; the OD, which facilitates DNA repair foci formation; the chromatin-binding TDs; and the BRCT domains, which are integral to the DNA damage response. The constructs generated and tested are shown in Fig. 6A.

The results with the GFP–53BP1 constructs are shown in Fig. 6B. The full-length GFP–53BP1 construct formed bodies in mitotically arrested cells. Analysis of the shortened constructs showed that the N-terminal, TD and BRCT regions alone were not capable of incorporation into 53BP1 MSBs. However, the N terminus with the addition of the OD (N–OD⁺ construct) showed efficient incorporation into the bodies, as indicated by large foci. The OD region combined with the more C-terminal part of the protein (OD⁺–TD⁺–BRCT construct) showed very limited incorporation, as indicated by typically very small GFP foci (see arrowheads in Fig. 6B). The regions that formed large, small or no MSB foci are summarized in Fig. 6C and suggest that the OD region of 53BP1 is required to initiate MSBs, while the N terminus provides capacity for their growth.

Introduction of exogenous 53BP1 into HCT116 cells did not increase MSB size or number, but it did increase the level of MSB-free 53BP1 severalfold (Fig. 6D). We asked how the increase in MSB-free 53BP1 affects arrested and released cells. We found that cells transfected with GFP–53BP1 had slower mitotic progression than control cells transfected with GFP alone. GFP–53BP1 cells displayed a significant increase in the percentage of cells in prophase or metaphase and a decrease in the percentage of cells in anaphase and telophase compared to control cells (Fig. 7A). Cells with elevated free 53BP1 were also more prone to apoptosis, based on staining with annexin V (Fig. 7B,C). However, increased apoptosis was not observed in non-arrested cells transfected with GFP–53BP1. These results indicate that sequestering of 53BP1 into MSBs might allow cells to re-enter mitosis more readily and suppress apoptosis. Since apoptosis following mitotic arrest is often accompanied by increased DNA damage (Chopra et al., 2016), we determined whether exogenous GFP–53BP1 increased the formation of γH2AX foci. As shown in Fig. 7D,E, cells expressing GFP–53BP1 had an increased average number of γH2AX foci per cell, compared to cells expressing GFP alone (Fig. 7D, top panels and Fig. 7E), and also more frequently displayed extensive γH2AX staining (Fig. 7D, bottom panels).

We next looked at the effects of 53BP1 knockdown in cells released from mitotic arrest using an RNAi approach. Cells transfected with 53BP1-targeting siRNA were slightly less likely to divide after 2 h (Fig. 7F; Fig. S3A) and did not show a significant change in apoptosis (based on annexin V staining; Fig. 7G; Fig. S3B). Formation of γH2AX foci was also not affected by 53BP1 knockdown (Fig. 7H).

Cancer cells can spontaneously experience mitotic stress resulting from mutations that affect spindle formation (for example, mutation of APC) or from debilitated cell cycle checkpoints that cause premature entrance into M phase (for example, CHFR silencing) (Green et al., 2005; Kashima et al., 2012). We therefore determined whether 53BP1 MSBs also formed under normal growth conditions. Both HCT116 and HT29 cells had subpopulations of mitotic cells that possessed 53BP1 MSBs that co-stained for both lamin A/C and RIF1 (Fig. 8). We observed that ∼20% of mitotic HCT116 cells and HT29 cells showed multiple, small 53BP1 MSBs (<1 μm diameter), and ∼2% of cells featured MSBs comparable in size to those found in arrested cells (∼1 μm diameter).

Most cancer cells emerging from mitotic arrest divide successfully, whereas a small percentage undergo apoptosis. However, the variables determining cell fate under these conditions are not entirely clear. In this study we examined the regulation of 53BP1 during mitosis, a protein whose unrestricted activity can promote fusion of mitotic chromosomes and apoptosis (Cesare, 2014; Orthwein et al., 2014; Rybanska-Spaeder et al., 2014). While studying components of the DNA damage response during mitotic arrest, we found that 53BP1 in colon cancer cells formed large bodies in close proximity to a subset of centromeres. Following release from arrest, these 53BP1 bodies move away from the mitotic chromosomes, diminish in number and then disappear upon cell division. These bodies appear to be distinct from DNA repair foci and 53BP1 G1-phase nuclear bodies in that they form during mitosis and include lamin A/C. Since lamin A/C may prevent 53BP1 DNA repair activity by binding to its TDs (Gibbs-Seymour et al., 2015), we envision these bodies restraining 53BP1 repair activity in mitotic cells. RIF1, a 53BP1 effector in the NHEJ pathway, was also found in these bodies, which is consistent with a broader role of these bodies in sequestering multiple components of the DNA repair machinery.

A time course analysis of mitotically arrested colon cancer cells showed that 53BP1 MSBs start forming within 1 h of arrest, grow to a size of about 1 μm as arrest persists and display LLPS features. Although they are most apparent in arrested cells, we also found bodies containing 53BP1, lamin A/C and RIF1 in mitotic cells in normally dividing cultures, suggesting that they form when cells encounter stochastic mitotic delays. 53BP1 MSBs might represent part of an important regulatory mechanism that restrains 53BP1 activity and enables cells with mitotic defects to survive and complete mitosis.

Mitotic stress can arise more frequently in cancer cells with checkpoint defects, as cells with damaged DNA are allowed to progress into mitosis. The perpetuation of this damage is supported by studies showing elevated 53BP1 and γH2AX expression in colon cancer tissues compared to the expression in normal tissues (Kim et al., 2013; Yang et al., 2021). This may also account for why 53BP1 MSBs weren't typically observed in the non-transformed cell line tested. On a few occasions, we found similar bodies in ARPE-19 cells, but only after release from mitotic arrest and never during mitotic arrest (data not shown).

The presence of lamin A/C in the 53BP1 MSBs is of interest, since lamin A/C helps to regulate 53BP1 during interphase (Gonzalez-Suarez et al., 2009; Gibbs-Seymour et al., 2015; Mayca Pozo et al., 2017). Specifically, lamin A/C retains 53BP1 in the nucleus during interphase through its interaction with the 53BP1 TDs. Lamin A/C also stabilizes 53BP1, suppressing its proteasomal degradation during interphase. This interaction therefore ensures that adequate levels of 53BP1 are available when needed. However, for an effective DNA damage response, 53BP1 must be released from lamin A/C so that it can interact with the chromatin at DNA damage sites. We therefore envision lamin A/C also playing a regulatory role in 53BP1 MSBs.

The structural details of the 53BP1 MSBs are not yet clear. For example, the 53BP1 TDs are not required for their formation, suggesting a different mode of interaction in mitotic cells versus interphase cells. It is also notable that 53BP1 is present in the mitotic bodies together with both lamin A/C and RIF1, which is not the case with DNA repair foci. The 53BP1 is released from lamin A/C at the onset of a DNA damage response, prior to formation of DNA repair foci, but its interaction with RIF1 occurs later in the pathway, after 53BP1 is phosphorylated by ATM/ATR kinases (Chapman et al., 2013; Escribano-Díaz et al., 2013). Although 53BP1 MSBs show N-terminal phosphorylation at ATM/ATR target sites (Ser25/Ser29), this is not necessary for the formation of the bodies – they form even in the presence of the ATM/ATR inhibitors caffeine and AZD0156 (Fig. S4). One possibility is that 53BP1, RIF1 and lamin A/C don't assemble into a highly structured complex in the MSBs but are instead trapped within a LLPS condensate formed by their disordered domains.

Recruitment of 53BP1 to telomeres during mitosis initiates NHEJ that can cause telomere fusion and formation of dicentric chromosomes (Orthwein et al., 2014). This can lead to breakage–fusion–bridge cycles complicating cell division and impacting survival. Prolonged mitosis increases the risk of telomere deprotection and telomere-dependent activation of a DNA damage response (Hayashi et al., 2012; Hain et al., 2016; Bernal and Tusell, 2018). Sequestering 53BP1 into MSBs might suppress aberrant activation of NHEJ and chromosome fusion. Consistent with this possibility, we found that transgenic expression of 53BP1 in mitotic cells increased the amount of ‘free’ 53BP1 while also increasing γH2AX foci and apoptosis. The increased γH2AX foci might indicate aberrant chromosome fusion and breakage. Alternatively, the γH2AX foci might form as a result of apoptosis-associated DNA fragmentation, thereby being more a result of the role of ‘free’ 53BP1 in apoptosis. Regardless of the exact mechanism, our data indicate that sequestering 53BP1 during mitotic stress is likely to increase cell survival. Interestingly, MSBs are similarly formed in HCT116 cells, which express wild-type p53, and in HT29 cells, which express mutant p53, suggesting that these bodies protect cells from apoptosis regardless of p53 status.

A distinctive feature of the 53BP1 MSBs is their formation near centromeres. As reported previously, 53BP1 associates with each centromere at the beginning of mitosis where it helps ensure proper spindle attachment (Wang et al., 2017; Petsalaki and Zachos, 2020). During a prolonged mitosis, we found that 53BP1 vacated some of the centromeres and accumulated at others in the MSBs. Since the number of bodies found on average was close to the number of human acrocentric chromosomes, we performed co-staining experiments with UBF, a transcription factor that binds to the nucleolar organizing region on acrocentric chromosomes. However, we did not observe specific association of 53BP1 MSBs with acrocentric chromosomes. At present it is unclear whether 53BP1 MSBs form on particular chromosomes, or whether the association is random.

The localization of 53BP1 near the centromere is intriguing since studies with colon cancer cell lines have shown that up to 60% of chromatin breaks occur in the centromeric and pericentromeric regions (Knutsen et al., 2010). Mitotic stress may expose centromeric defects due to malfunction or degradation of protective components. It is possible that 53BP1 recognizes chromosomes based on centromere abnormalities and/or defects in kinetochore formation. This possibility is consistent with the formation of 53BP1 MSBs in cells arrested with ICRF-193, an agent that inhibits topoisomerase IIα (Fig. 1E, right panel, and Fig. 1F). In mitotic cells, topoisomerase IIα accumulates at centromeres and is required for sister-chromatid decatenation, which is largely completed by the end of prophase, and for resolving chromatin intertwinements and chromatin bridges (Spence et al., 2007; Nagasaka et al., 2016; Nielsen et al., 2020). Depletion of topoisomerase IIα at prometaphase causes severe chromatin entanglements (Nielsen et al., 2020). It is conceivable that 53BP1 MSBs are formed at entangled centromeric regions, particularly when such defects become more exposed during mitotic arrest. Entangled and distorted centromeric heterochromatin may also reveal previously hidden breaks, which would make the prevention of 53BP1 activity and its segregation crucial for cell survival.

The results of our experiments are consistent with 53BP1 being tightly regulated in cancer cells during mitosis. Based on our data and published reports, we speculate that 53BP1 MSBs suppress 53BP1 apoptotic signaling and potentially prevent aberrant NHEJ of mitotic chromosomes. It is also possible that these bodies help deal with centromeric defects. We propose that cancer cells with debilitated mitotic checkpoints or a damaged mitotic apparatus may have a higher dependence on this 53BP1 regulatory mechanism. The HCT116 cell line used in this study has a defective mitotic checkpoint resulting from CHFR silencing, which increases the frequency of mitotic errors (Kashima et al., 2012). The HT29 cell line carries an APC truncation mutation that complicates spindle assembly and attachment (Green et al., 2005). These defects might cause cells to stall in mitosis more frequently and increase the risk of aberrant 53BP1-mediated NHEJ and apoptosis. The role of 53BP1 in cancer cell biology and cancer therapy is well established (Mirza-Aghazadeh-Attari et al., 2019; Trenner and Sartori, 2019). Our findings point to a novel mode of regulation during mitosis that might ultimately be targeted for therapeutic purposes.

HCT116 and HT29 colon cancer cell lines and the ARPE-19 normal retinal epithelium cell line were obtained from the American Type Culture Collection (Manassas, VA). All cell lines were frozen in aliquots with fresh cultures started every few months. HT29 and HCT116 cells were cultured in McCoy's 5A medium. ARPE19 cells were cultured in DMEM/F12 medium. Media were supplemented with 10% fetal bovine serum, non-essential amino acids and antibiotic-antimycotic. All media and supplements were purchased from Thermo Fisher Scientific (Gibco brand; Life Technologies, Eugene, OR, USA). Cells were regularly checked for contamination.

The AK306 compound was synthesized as described previously (Bond et al., 2018) and was used at a concentration of 100 nM for 12 h, unless otherwise indicated. Release from mitotic arrest was for 2.5 h, unless otherwise indicated. Nocodazole, topoisomerase IIα inhibitor ICRF-193, caffeine and D-sorbitol were obtained from Sigma-Aldrich (St. Louis, MO, USA). 1,6-hexanediol was purchased from Acros Organics (Morris Plains, NJ, USA). AZD0156 was purchased from Selleck Chemicals (Houston, TX, USA). Colcemid was from Enzo Life Sciences (Farmingdale, NY, USA), and paclitaxel was from Cytoskeleton Inc. (Denver, CO, USA). Nocodazole (200 nM), ICRF-193 (10 μM), Colcemid (50 ng/ml) and paclitaxel (10 nM) were used for 5 h. 1,6-hexanediol (3%) was used for 25 min. Caffeine (8 mM) and AZD0156 (250 nM) were used for 5 h 20 min. D-sorbitol and NaCl were applied at 0.4 M and 0.3 M concentrations, respectively, for 1 h. ProGold and ProLong Glass antifade mounting reagents and DNA dye DAPI were purchased from Thermo Fisher Scientific (Life Technologies, Carlsbad, CA, USA).

The following primary antibodies were purchased from Thermo Fisher Scientific (Invitrogen): anti-53BP1 (PA1-16565; 1:1100 for confocal microscopy, 1:4000 for STED microscopy), anti-phospho-53BP1 (PA5-101035; 1:250), and anti-CENP-A clone-3-19 (MA1-20832; 1:200). The anti-RIF1 clone B-3 (sc-515573; 1:100 for confocal microscopy, 1:380 for STED microscopy), anti-phospho-histone H3 (sc-12927; 1:320), anti-topoisomerase IIα clone G-6 (sc-166934; 1:350) and anti-UBF clone F-9 (sc-13125; 1:200) primary antibodies were purchased from Santa Cruz Biotechnology. Anti-Aurora kinase B (ab2254; 1:500) was from Abcam. Anti-γH2AX clone JBW301 (05-636; 1:1000) was from EMD Millipore. The anti-lamin A/C (MANLAC1 clone 4A7; deposited by G. E. Morris; 1:60 for confocal microscopy, 1:180 for STED microscopy) and anti-β-tubulin clone E7 (deposited by M. Klymkowsky; 1:160) primary antibodies were obtained from Developmental Studies Hybridoma Bank, IA, USA. All antibodies were validated by western blotting and immunofluorescence. The antibody to γH2AX was additionally validated by ChIP, and antibodies to 53BP1 and CENP-A received advanced verification (information from EMD Millipore and Invitrogen). The anti-53BP1 antibody was also confirmed to be specific through RNAi experiments. Secondary antibodies for confocal microscopy were obtained from Thermo Fisher Scientific (Eugene, OR, USA) and Jackson ImmunoResearch (West Grove, PA, USA) and were used at a dilution of 1:570. Secondary antibodies for STED microscopy were obtained from Abberior GmbH (Gottingen, Germany) – namely STAR 635P, STAR RED and STAR ORANGE – and were used at a dilution of 1:170.

Cells cultured on coverslips were fixed with 3% paraformaldehyde at room temperature and permeabilized with 0.3% Triton X-100 (Sigma-Aldrich, St. Louis, MO, USA) in Dulbecco's phosphate-buffered saline (DPBS). Cells were blocked in 5% normal donkey serum (Jackson ImmunoResearch, West Grove, PA, USA) in DPBS, incubated with primary antibody in 5% serum on a shaker for 1 h at room temperature, and then incubated with secondary antibody for 45 min. DNA was visualized using DAPI (5 μg/ml in DPBS). Coverslips were mounted on slides using ProLong Gold. ProLong Glass was used to mount cells expressing GFP. Confocal images were acquired using a Nikon A1R confocal microscope (version 2.11, Nikon Instruments Inc.) with a 60× oil immersion objective and NIS-Elements Advanced Research Software (version 4.4, Nikon Instruments Inc.) Images were viewed and analyzed using Fiji software (Schindelin et al., 2012). The ‘Max Intensity’ projection algorithm was used for image stacking. Background subtraction was accomplished by setting the minimum displayed brightness and contrast value to the value of a region of known background. Brightness and contrast were uniformly modified when appropriate. Quantifications of 53BP1 foci were done using ‘RenyiEntropy’ threshold and the menu command ‘Analyze particles’. Generated masks were applied to images in the ‘Sum Slices’ projection. Off-chromatin positions of 53BP1 foci were confirmed using 2D and 3D stacks.

STED microscopy was performed using an Abberior Instruments Expert Line system (Abberior Instruments America LLC) with the inverted stand Olympus IX83 and Physik Instrumente piezo z stage insert. Images were acquired using a 100×/1.4 NA oil immersion objective. The following combinations of excitation/depletion lasers were used: 640 nm/775 nm for samples labeled with STAR 635P; 640 nm and 561 nm/775 nm for samples co-labeled with STAR RED and STAR ORANGE. Abberior Imspector software was used for acquisition. Pixel size was 20×20 nm². Images were visualized using Fiji software.

ON-TARGET plus SMART pool siRNA duplexes against 53BP1 and non-targeting siRNA were obtained from GE Dharmacon (Lafayette, CO, USA). Double reverse siRNA transfection was performed in a 24-well plate using Lipofectamine RNAiMAX (Invitrogen 13778-100, Thermo Fisher Scientific). Briefly, siRNA duplexes (45 nM final concentration) were diluted in 100 μl Opti-MEM medium (Gibco 31985-062, Thermo Fisher Scientific) and added to each well. Next, 1.2 μl Lipofectamine was mixed in, and the mixture was incubated for 20 min at room temperature. HCT116 cells were diluted (to give ∼50% confluency) in 500 μl complete growth medium without antibiotic-antimycotic and added to each well containing the siRNA–lipofectamine mixture. Plates were placed in cell culture incubator for 48 h. Then, cells were lifted, re-transfected using the same procedure, and incubated for an additional 36–48 h.

pEGFP-h53BP1 plasmid was from Addgene (110301; deposited by Chris Kok-Lung Chan). The plasmid was expressed in 5-alpha competent E. coli (C2987; New England Biolabs, NEB) using the NEB transformation protocol and was purified using a CompactPrep Plasmid Midi kit (12843, Qiagen) according to the Qiagen protocol. Transient plasmid transfections were performed in 24-well plates using Lipofectamine LTX with Plus Reagent kit (Invitrogen A12621, Thermo Fisher Scientific). Briefly, HCT116 cells were seeded at ∼70% confluency and incubated for ∼3.5–4 h to allow for attachment. Two solutions were prepared in Opti-MEM medium (Gibco 31985-062, Thermo Fisher Scientific) in the indicated amounts per well: (1) 500 ng plasmid DNA, 0.5 μl Plus Reagent and 25 μl Opti-MEM; and (2) 1 μl Lipofectamine and 25 μl Opti-MEM. The solutions were mixed, and the mixture was incubated for 5 min before being added to each well. Cells were incubated for ∼33–34 h. Transfection efficiency was 50–70%. The ranges and average fluorescence intensities were similar for all constructs, except for the BRCT construct, where the average intensity was ∼15% higher.

GFP–53BP1 constructs were generated using Q5 site-directed mutagenesis kit (E0554, NEB) and NEB cloning protocol. NEBaseChanger (http://nebasechanger.neb.com/) and NEB Tm calculator (https://tmcalculator.neb.com), were used for primer design. The following primer pairs (forward/reverse) were used with pEGFP-h53BP1 plasmid: 5′-TAGCGGGATCAATTCCGC-3′/5′-CTGGCTATGGAGCGACTC-3′ for the N construct (amino acids 1–1220), 5′-TAGCGGGATCAATTCCGC-3′/5′-TTCACCGGTGTTGTCTCC-3′ for the N–OD⁺–TD⁺ construct (amino acids 1-1711), 5′-GGAGAAGAAGAGTTTGATATG-3′/5′-CATCTCGAGATCTGAGTAC-3′ for the OD⁺–TD⁺–BRCT construct (amino acids 1221-1972) and 5′-CCCTCTGCCCTGGAAGAG-3′/5′-CATCTCGAGATCTGAGTACTTGTAC-3′ for the BRCT construct (amino acids 1712-1972). The N–OD⁺ construct (amino acids 1-1466) was found among N–OD⁺–TD⁺ clones.

The N–OD⁺–TD⁺ construct was utilized to generate TD⁺ (amino acids 1480–1711) and N–TD⁺ (amino acids 1–1220 and amino acids 1480–1711) constructs using 5′-TCCTCTCCAGGAAATAGC-3′/5′-CATCTCGAGATCTGAGTAC-3′ and 5′-TCCTCTCCAGGAAATAGC-3′/5′-CTGGCTATGGAGCGACTC-3′ primer pairs, respectively. Sanger sequencing of the constructs was performed by Eurofins Genomics (Louisville, KY, USA).

A GFP-certified Apoptosis/Necrosis detection kit and supplied protocol (Enzo Life Sciences, Farmingdale, NY, USA) were used for this assay. Cells grown on coverslips were washed twice with DPBS (with CaCl₂ and MgCl₂; Gibco, Thermo Fisher Scientific), then incubated with annexin V–EnzoGold in binding buffer (1:100) for 14 min at room temperature on rocker in the dark. Cells were fixed with 2% paraformaldehyde for 15 min followed by immunofluorescence staining and imaging. Annexin V-positive cells were counted manually, with chromatin condensation and fragmentation used to confirm apoptosis. Calculations were conducted in Microsoft Excel.

Statistical analysis was performed with Prism GraphPad, and all data were tested for the assumptions of the tests. The data were analyzed using 95% confidence intervals. A two-sided Fisher's exact test was used for comparing two treatment groups with sample sizes between 50 and 1100 cells in the experiments assaying mitotic progression and annexin V labeling. The data are presented as percentages. An ordinary one-way ANOVA and Tukey's post-hoc tests were used to compare number of foci per cell between three or more groups in the time course experiments and experiments involving different mitotic agents. Kruskal–Wallis and Dunn's post-hoc tests were used to compare area per focus between three groups in the time course experiment since the assumption of normality was not met. Analysis of γH2AX foci was performed using ordinary two-way ANOVA and Holm–Šídák post-hoc tests. P≤0.05 was considered to be significant. All images for quantifications were taken from random fields of view. Cells were scored as mitotic based on the appearance of condensed mitotic chromosomes, increased fluorescence intensity of DAPI staining and a rounded morphology. Assessments of mitotic phases and annexin V-positive cell counts were performed by an expert who was not familiar with the project and was not informed of the group allocation.

We thank the National Institutes of Health and Dr Christopher O'Connell, Director of the Advanced Light Microscopy Facility of the University of Connecticut, for access to the STED microscope. The Abberior Instruments STED microscope was purchased with funds from a National Institutes of Health shared instrumentation grant, S10OD023618, awarded to Christopher O'Connell. We thank Dr O'Connell for performing STED imaging and his expert assistance with microscopy questions.