ATM–ESCO2–SMC3 axis promotes 53BP1 recruitment in response to DNA damage and safeguards genome integrity by stabilizing cohesin complex

Abstract 53BP1 is primarily known as a key regulator in DNA double-strand break (DSB) repair. However, the mechanism of DSB-triggered cohesin modification-modulated chromatin structure on the recruitment of 53BP1 remains largely elusive. Here, we identified acetyltransferase ESCO2 as a regulator for DSB-induced cohesin-dependent chromatin structure dynamics, which promotes 53BP1 recruitment. Mechanistically, in response to DNA damage, ATM phosphorylates ESCO2 S196 and T233. MDC1 recognizes phosphorylated ESCO2 and recruits ESCO2 to DSB sites. ESCO2-mediated acetylation of SMC3 stabilizes cohesin complex conformation and regulates the chromatin structure at DSB breaks, which is essential for the recruitment of 53BP1 and the formation of 53BP1 microdomains. Furthermore, depletion of ESCO2 in both colorectal cancer cells and xenografted nude mice sensitizes cancer cells to chemotherapeutic drugs. Collectively, our results reveal a molecular mechanism for the ATM–ESCO2–SMC3 axis in DSB repair and genome integrity maintenance with a vital role in chemotherapy response in colorectal cancer.


INTRODUCTION
Double-strand breaks (DSBs) are known to be one of the most serious types of DNA lesions threatening genomic stability in mammalian cells. Failure to repair DSBs leads to loss of genetic information by mutations and chromosomal rearrangement, which contributes to the pathogenesis of cancer and di v erse inherited diseases ( 1 , 2 ). Mammalian cells employ two major DSB repair pathways, nonhomologous end joining (NHEJ) and homologous recombination (HR). NHEJ is a highly err or-pr one homologyindependent repair mechanism that is the predominant repair pathway throughout the cell cycle, whereas HR requires a homology template, such as a sister chromatid, and occurs in the S / G2 phases of the cell cycle ( 3 ). The harmonious cooperation of the different repair pathways is critical to minimize genomic damage ( 4 ).
A DSB is detected by sensor proteins that can trigger the activation of kinases such as ATM and ATR ( 5 ). ATM phosphorylates the histone variant H2AX to generate ␥ H2AX ( 6 ). 53BP1 is recruited to DSB sites by binding H4K20me2 and H2AK15ub in a manner that depends on the interaction of MDC1 and ␥ H2AX ( 7 , 8 ). 53BP1 antagonizes the resection of DSBs in G1 by recruiting downstream RIF1, REV7, and the shieldin complex and contributes to NHEJ (8)(9)(10)(11). 53BP1 undergoes liquid-liquid phase separation in response to DN A damage, w hich integrates damage detection, shielding of break sites, and checkpoint activation ( 12 ). Using 3D-SIM super-resolution microscopy, researchers have found that 53BP1 organizes DSB-flanking chromatin into circular microdomains ( 13 ). Howe v er, the regulation of 53BP1 microdomain formation and its role in the response of cancer cells to chemotherapy drugs is unclear.
Cohesin is a m ultiprotein, ring-sha ped complex, and its canonical role is to tether sister chr omatids fr om S phase to anaphase to pre v ent pr ematur e sister chromatid separation and ensure equal segregation of chromosomes ( 14 ). The cohesin complex can also r egulate thr ee dimensional chroma tin organiza tion ( 15 ) and is emerging as a key regulator in DNA damage repair by promoting homology search during recombination ( 16 ). Roberts syndrome (RBS; OMIM 268300) is a rare genetic disorder characterized by pre-and postnatal gr owth retardation, micr ocephaly, bilateral cleft lip and palate, and mesomelic symmetric limb reduction ( 17 ), and is caused by mutations in the ESCO2 gene ( 18 ). ESCO2 acetylates the cohesin subunit SMC3 at K105 / 106 (19)(20)(21)(22) and is r equir ed for the establishment of sister chromatid cohesion ( 23 , 24 ). Howe v er, the mechanism of ESCO2-mediated cohesin-dependent chromatin structure dynamics in DSB repair especially NHEJ repair remains to be elucidated.
In this study, we identify that acetyltr ansfer ase ESCO2 plays a role in regulating DSB repair. We show that ESCO2 is recruited to DSB sites in an ATM-and MDC1-dependent manner. Furthermore, ESCO2 promotes the formation of 53BP1 foci to DSB sites by stabilizing cohesin complex and is essential for resistance to chemotherapy in colorectal cancer cells (CRC).

Cell culture and transfection
HeLa, HCT116, and RKO cells were purchased from ATCC and HEK293T was acquired from National Infrastructure of Cell Line Resource. HeLa, HCT116 and HEK293T cells were cultured in DMEM medium supplemented with 10% fetal bovine serum at 37 • C with 5% CO 2 . RKO cells wer e cultur ed in RPMI 1640 medium supplemented with 10% fetal bovine serum at 37 • C with 5% CO 2 . HeLa, HCT116, RKO and HEK293T cells were transfected with PEI according to the manufacturer's instructions (Polyscience).

Plasmids
pRK5-Flag-ESCO2 and pRK5-GFP-ESCO2 plasmids were kindly provided by Huiqiang Lou at College of Biological Sciences, China Agricultural Uni v ersity. pX332-SMC3-EGFP plasmid was kindly provided by Xiong Ji at School of Life Sciences, Peking Uni v ersity. MDC1 and truncation mutants were cloned into the pCMV-HA vector. All plasmids were verified by DNA sequencing.

Laser microirradiation
Laser microirradiation was carried out following procedur es described pr eviously ( 25 ). HeLa cells wer e grown on thin glass-bottom plates and irradiated with an ultraviolet laser (16 Hz pulse, 60% laser output). Images were taken using a Dragonfly (Andor) confocal imaging system e v ery 10 s for 30 min.

Immunofluor escence microscop y
Cells wer e cultur ed on glass coverslips in six-well plates. Twenty-four hours later, cells were washed three times with pre-chilled PBS and then fixed with 1 ml of prechilled methanol for 10 min at −20 • C. After washing with pre-chilled PBS three times, the cells were blocked with 1% bovine serum albumin (BSA) for 1 h, followed by incubation with primary antibodies that had been diluted in 1% BSA for 1 h at 37 • C. After washing with pr e-chilled PBS thr ee times, the cells wer e incubated for 1 h at 37 • C with secondary antibodies that had been diluted in 1% BSA and then washed with prechilled PBS three times. Finally, 20 l of mounting solution was used to mount cells. Images were obtained using a confocal microscope (Zeiss LSM-710 NLO, Du-oScan, and Andor dragonfly microscopy) with a 63 × oil objecti v e lens. Quantification analysis was performed using ZEN 3.1 (blue edition) (Zeiss) and ImageJ software. Super-resolution 3D-SIM imaging was carried out using a DeltaVision OMX SR (GE Healthcare). 3D image analysis was carried out using QUANTEX software ( https:// figshar e.com/s/46fa39d1010d77f51d9c ). The curvatur e was defined as 'Gaussian Curvature' and was calculated according to the method described here ( https://gfx.cs.princeton. edu/pubs/Rusinkiewicz 2004 ECA/curvpaper.pdf). Curvature Points TH100 refers to the objects with a very high proportion of spiky curva ture a t a subscale > 0.75 and ≤1.0.

Clonogenic survival assay
First, 200-500 cells were seeded in six-well plates in triplicate. After 48 h, cells were cultured in medium containing a dif ferent concentra tion of b leomy cin or oxaliplatin for 24 h and washed twice with DMEM. After 12 days, cells were washed with PBS, fixed in precooled methanol for 10 min at -20 • C and stained with crystal violet (0.1% wt / vol) for 15 min. The number of clones was counted and the survival fraction was normalized to the number of untreated cells.

Mass spectrometry
HEK293T cells were lysed in modified RIPA buffer, sonicated, and pr eclear ed with protein G beads. The supernatants were incubated with anti-Flag affinity beads at 4 • C for 4 h and eluted with Flag peptide. The eluates were precipitated with TCA and subjected to mass spectrometric analysis. The MS data were aligned with the Human Reviewed Swiss-Prot database by Proteome Discoverer 2.2 softwar e. Proteins wer e consider ed to be major hits (positi v e) when matching the following criteria: (i) not found in negati v e contr ol gr oup; (ii) high Pr otein FDR Confidence (FDR < 0.01) and (iii) peptides ≥5.

NHEJ and HR assays
NHEJ r epair assays wer e performed according to a protocol as previously described ( 26 , 27 ). WT and ESCO2 KD HCT116 cells were co-transfected with HindIII-linearized pEGFP-Pem1-Ad2 and dsRED plasmids. For the HR assay, WT and ESCO2 KD HCT116 cells were co-transfected with I-SceI, DR-GFP, and dsRED plasmids for 36 h, and the fluorescence was measured using CytoFLEX S (Beckman). The repair efficiency was determined by calculating the percentage of EGFP and dsRED doub le-positi v e cells in dsRED positi v e cells. The r esults wer e normalized using the WT HCT116 cells.

Co-immunoprecipitation
Cells were cultured in 10-cm dishes and transfected with the indicated plasmids. After 48 h, cells were washed with 10 ml of pre-chilled PBS and lysed for 60 min in 1 ml NP-40 lysis buffer (50 mM Tris-HCl, 150 mM NaCl, 1 mM EDTA, 1% NP-40, pH 7.4) with 10 l of protease inhibitor cocktail. Next, 1 g of antibody or IgG was used to bind to the bait proteins for 4 h, and then incubated with 30 l protein G for an additional 3 h. Finally, the protein G w as w ashed with 1 ml of NP-40 lysis buffer three times and then heated for 10 min at 96 • C with 30 l 2 × SDS loading buffer. Samples were analyzed by SDS-PAGE and western blotting.

Generation of ESCO2 knockdown HCT116 and RKO cells by CRISPR-cas9 system
sgRNAs targeting different regions in the mRNA of the human ESCO2 gene were designed and cloned into a lentiviral sgRNA vector containing the mCherry selection marker using the Golden Gate method ( 27 ). Next, cells were cotransfected with the sgRNA and Cas9 vectors. After 48 h of transfection, mCherry-positi v e cells were selected by FACS (MOFLO, Cytomation). Single clones were obtained after 10 days of selection. The knockout efficiency was confirmed by immunoblotting. ESCO2 gene mutations were verified with PCR and sequencing.

ChIP-seq and ChIP-qPCR
HCT116 cells wer e tr eated with 4-OHT (500 nM) for 4 h. For each precipitation, 1 × 10 7 cells were crosslinked by the addition of formaldehyde directly to the growth medium to a final concentration of 1%. After 10 min, crosslinking was quenched by the addition of glycine to a final concentration of 0.25 M at room temperature. Crosslinked cells were washed with PBS, scraped, and washed again with PBS containing 1 mM EDTA, and then lysed gently on ice for 5 min using 0.5 ml of ice-cold NP-40 lysis buffer. Cell lysates were added on top of a 1.25 ml sucrose cushion (24% sucrose (wt / vol) in NP-40 lysis buffer) and centrifuged at 12 000 rpm for 10 min at 4 • C to isolate the nuclei pellet. The chromatin pellet was washed twice with 1 ml PBS / 1 mM EDTA, followed by centrifugation at 12 000 rpm for 1 min at 4 • C. The supernatant was discarded, the tube was pre-warmed at 37 • C for 2 min, and then 40 U MNase was added followed by incubation at 37 • C for 15 min with rotation at 700 rpm. Subsequently, the MNase was inactiv ated b y the addition of 20 l 0.5 M EDTA (to a final concentration of 10 mM) and 40 l 0.5 M EGTA (to a final concentration of 20 mM). The mixture was then placed on ice, mixed thoroughly by pipetting, and sonicated using pre-cooled Biorupter at 15-20 × (30 s on / 30 s off) on high position. The sonicated chromatin was spun down at 12 000 rpm for 10 min at 4 • C to collect the supernatant chromatin. Next, 1-2 g antibodies against ␥ H2AX, 53BP1, and ac-SMC3 were added to the soluble chromatin and incubated with rotation at 4 • C overnight. Protein G Dynabeads (Life Tech, 10004D) were washed three times with sonication buffer and then added to the soluble chromatin and antibodies, followed by incuba tion a t 4 • C for 2 h with rotation. The magnetic Dynabeads were pelleted by placing the tubes in a magnetic rack, washed, and then the bound DNA fragments were eluted. Sequencing libraries were prepared using 10 ng of purified DNA (average size 200-400 bp) using the NEBNext Ultra II Library Prep Kit for Illumina (E7645S, New England Biolabs) and subjected to 75-bp single-end sequencing on a Nova PE150 platform (Illumina).
For the analysis of the ChIP-seq data, the quality of each raw sequencing file (fastq) was verified with FastQC ( https://www.bioinformatics.babraham.ac.uk/ projects/fastqc/ ). All files were aligned to the reference human genome (hg19) and processed using bowtie2 ( https: //bo wtie-bio.sourcefor ge.net/ ) for mapping and samtools ( http://www.htslib.org/ ) for duplicate removal (rmdup), sorting (sort) and indexing (index). Coverage for each aligned ChIP-seq dataset (.bam) were computed with deeptools ( https://deeptools .readthedocs .io/en/latest/index. html ) and normalized using total read count for each sample. Coverage data were exported as bigwig (file format) for further processing. Averaged ChIP-seq profiles were generated using the R package ggplot2. The x axis r epr esents genomic position relati v e to DSB and the y axis r epr esents the mean coverage at each bp. Heatmaps were generated by computeMatrix tool from deepTools.
ChIP-qPCR was performed using primers shown in Supplementary Table S1. Data were analyzed using CFX Manager Software (Bio-Rad). The fold change of proteinbinding DNA was calculated by using the following steps: (i) calculate the Ct values from the ChIP (Ct 1 ) and input (Ct 2 ) C t values by using the formula Ct = Ct 1 -Ct 2 ; (ii) calculate the fold change from the corresponding Ct values by using the formula Fold change = 2 − Ct ; (iii) normalize the fold change in the ESCO2 KD cells with that of the WT cells and use Gr aphPad to gener a te the hea t map of these normalized fold change results and (iv) perform twoway ANOVA from multiple independent biological replicates to obtain and analyze the P values.

Duolink proximity ligation assay (PLA)
PLAs were performed to examine the in situ interaction between MDC1 and ESCO2 WT or the ESCO2-2A mutant in ESCO2 KD HCT116 cells transfected with Flag-ESCO2 WT and Flag-ESCO2-2A plasmids. The assay was performed using the Duolink ® In Situ PLA ® kit (DUO92101, Sigma-Aldrich) as described previously ( 28 ).

Nude mice xenograft assay
Female BALB / c nude mice at 6-8 weeks of age were purchased from Beijing Vital Ri v er Laboratory Animal Technology. HCT116 ESCO2 WT or KD cells (5 × 10 6 cells) were injected subcutaneously into both flanks, tumor size was measured e v ery 3 days using a caliper, and tumor volume was calculated using the following formula: volume = (length × width 2 ) / 2. At 24 days post-injection, tumors were dissected and weighed. The nude mice tumorigenesis assay was approved by the IACUC of the Center for Experimental Animal Research (China) and Peking Uni v ersity Laboratory Animal Center (IACUC No. LSCZhengX-2-1) and performed in accordance with the 'Guide for the Care and Use of Laboratory Animals'.

ESCO2 is involved in DSB repair and is essential for maintaining genome stability
Cells deri v ed from patients with ESCO2 mutation-induced RBS ar e mor e sensiti v e to ionizing radiation (IR)-and mitomycin C-induced DNA damage ( 29 ). To confirm whether ESCO2 is involved in the DNA damage response (DDR), we performed a laser microirradiation assay coupled with li v e imaging of protein redistribution and found that GFP-ESCO2, but not GFP-ESCO1, was ra pidl y and robustl y recruited to the micro-irradiated region ( Figure 1 A and Supplementary Figure S1A), which was detectable approximately 10-30 s after microirradiation (Figure 1 B). These observations suggest that ESCO2 is involved in the DDR.
It has been well established that overactivation of DDR pathway proteins results in resistance to chemotherapy or r adiother apy cancer treatments, and loss of DDR elements increases sensitivity to DNA damage agents ( 30 ). Analyses of TCGA and GTEx databases showed that the expression le v el of ESCO2 in colorectal cancer tissues was higher than that in normal tissues (Supplementary Figure S1B). Moreov er, the e xpression le v el of ESCO2 was correlated with those of 53BP1 and BRCA1, which are both essential for DSB repair (Supplementary Figure S1C). This analysis from cancer samples implies that ESCO2 is associated with DSB repair in colorectal cancers. Therefore, we tested ESCO2 e xpression le v els in thr ee color ectal cancer cell lines (HCT116, LO V O and RKO) and two normal colorectal epithelial cell lines (NCM460 and FHC). As expected, the abundance of ESCO2 was higher in cancer cells relati v e to normal cells (Supplementary Figure S1D). To further investigate the role of ESCO2 in the DDR in colorectal cancer, we stably knocked down ESCO2 in HCT116 cells (ESCO2 KD HCT116) and monitored ␥ H2AX foci, a biomarker of DSB damage, using an immunofluorescence assay 12 h after b leomy cin trea tment. Rela ti v e to wild-type (WT) cells, cells with depleted ESCO2 showed an increased number of ␥ H2AX foci after a 12 h recovery (Figure 1 C, D); furthermore, we achie v ed similar results in RKO cells (Supplementary Figure S2A). Next, we examined the effect of ESCO2 on the DSB repair efficiency. In addition to showing that ov ere xpression of ESCO2 promoted HR efficiency (Supplementary Figure S2C), which is consistent with a previous study ( 16 ), we also found that NHEJ repair efficiency decreased in ESCO2-depleted HCT116 cells but increased with ESCO2 ov ere xpression using 53BP1 KD cells as a negati v e control (Figur e 1 E and Supplementary Figur e S2B, C). Additionally, we performed a neutral comet assay using WT and ESCO2 KD HCT116 cells treated with or without b leomy cin. The results showed that ESCO2 deficiency increased the length of comet tails after a 12 h recovery (Figure 1 F). These results indicate that ESCO2 is involved in DSB repair and genome stability maintenance. Oxaliplatin, which can induce DNA damage, is a thirdgenera tion pla tinum drug used as a first-line chemotherapy in colorectal cancer. We examined the effect of ESCO2 depletion on cell survival following b leomy cin or oxaliplatin treatment in both HCT116 and RKO cells. Relati v e to WT HCT116 and / or WT RKO cells, knockdown of ESCO2 r ender ed both HCT116 and RKO cells more sensiti v e to b leomy cin and oxaliplatin ( Figure 1 G-I). Collecti v ely, these results suggest that ESCO2 promotes chemotherapeutic drug-induced DDR in colorectal cancer cells.

ATM r egulates the r ecruitment of ESCO2 at DSB sites by phosphorylating ESCO2 S196 and T233 residues
To determine the mechanism by which ESCO2 is recruited to DNA damage sites, we assessed whether ESCO2 recruitment is dependent on the upstream kinases, such as A TM, A TR or DNA-PKcs. Cells transfected with GFP-ESCO2 wer e tr eated with specific inhibitors targeting ATM (A TMi, KU55933), A TR (A TRi, VE821) or DNA-PKcs (DNA-PKcsi, NU7441), and the recruitment ability of GFP-ESCO2 to the laser tracks after microirradiation was examined. Interestingly, we found that ATM inhibition reduced the recruitment of ESCO2 (Figure 2 A, B and Supplementary S3A); howe v er, neither ATR nor DNA-PKcs inhibition affected ESCO2 recruitment (Supplementary Figure  S3B). Consistent with this observation, the co-localization ratio of ESCO2 and ␥ H2AX foci after b leomy cin treatment was reduced in cells treated with ATMi (Figure 2 C, D). Previous studies re v ealed the preference of PIKKs family members (including A TM, A TR and DNA-PKcs) for phosphorylating a serine or threonine followed by a glutamine (S / TQ; ( 5 )). Here, using a phospho-specific S / TQ antibod y tha t specificall y reco gnizes proteins phosphorylated on S / TQ motifs, we observed that the phosphorylation of ESCO2 responding to DNA damage significantly decreased when ATM was inhibited. Cells treated with phospha tase tha t can dephosphoryla te the phosphoryla ted serine, threonine and tyrosine residues were used as negati v e controls (Figure 2 E).
To identify the phosphorylation sites of ESCO2 by ATM, we generated a series of mutation constructs by mutating S / TQ motif residues S50, S196 and T233 to alanine, which served to mimic the non-phosphorylation status of these residues. We then tested whether these mutants could be recruited to DSB sites. Among the mutants examined, compared with GFP-ESCO2 WT, the double mutant GFP-ESCO2-S196-T233-2A (designated as 2A in figures) showed a clear reduction in its recruitment to the laser track (Figure 2 F, G and Supplementary Figure S4A, B) and had no impact on the acetyltr ansfer ase activity of ESCO2 (Supplementary Figure S4C). Moreover, the phosphorylation le v el of the GFP-ESCO2-S196-T233-2A mutant was no longer detected after b leomy cin tr eatment (Figur e 2 H). The cell survival assay results showed that over expr ession of GFP-ESCO2 WT, but not GFP-ESCO2-S196-T233-2A, rev ersed b leomy cin-hypersensiti vity in ESCO2-depleted cells (Figure 2 I). We also found a reduced number of ␥ H2AX foci in both the ESCO2-depleted cells ov ere xpressing GFP-ESCO2 WT and in the GFP-ESCO2-S196-T233-2D mutant, which mimicked the continuous phosphorylation state of ESCO2, but not in the GFP-ESCO2-S196-T233-2A mutant (Figure 2 J). These results indica te tha t ATM-media ted phosphorylation of ESCO2 on its S196 and T233 residues is essential for its recruitment to DSB sites.

MDC1 interacts with and mediates the recruitment of ESCO2 to DSB sites
To further elucidate the mechanism underlying the recruitment of ESCO2 to DNA damage sites, we performed an immunoprecipitation (IP) assa y f ollowed by mass spectrometry to identify ESCO2-interacting proteins in response to DN A damage. MDC1, w hich is a scaffold protein involved in the early steps of the DDR, was shown to be a potential partner of ESCO2 (Figure 3 A and Supplementary Table S2); this interaction between ESCO2 and MDC1 was further confirmed by co-IP assay, which re v ealed an enhanced interaction in b leomy cin-treated cells (Figure 3 B and Supplementary Figure S5A) that was not mediated by DNA (Figure 3 C). To examine whether ESCO2 regulated DSB repair through its association with MDC1, we performed laser microirradiation assays using HCT116 cells transfected with siControl or siMDC1. The results showed that ESCO2 was no longer recruited to damage sites in MDC1-depleted cells ( These results indicate that ESCO2 is recruited to DSB sites in an MDC1-dependent manner. MDC1 is composed of se v eral distinct domains and regions that can recognize and interact with its partners recruited to DSB sites. We next mapped the critical domain of MDC1 interacting with ESCO2 using an in vitro pulldown assay (Figure 3 G) and discovered that, surprisingly, the BRCT domain of MDC1 is responsible for its interaction with ESCO2. Since the BRCT domain acts as a phosphopeptide-binding domain, we needed to clarify whether ESCO2 phosphorylation is critical for its binding to MDC1; ther efor e, we tested the dir ect interaction between MDC1 and ESCO2 WT or the ESCO2-S196-T233-2A mutant. As e xpected, relati v e to ESCO2 WT, the interaction with MDC1 was largely reduced in the ESCO2-S196-T233-2A mutant (Figure 3 H). We also found that b leomy cin-induced DNA damage promoted the association  between ESCO2 and MDC1 (Figure 3 I). Furthermore, the PLA results showed that the in situ interaction between MDC1 and ESCO2 was enhanced in response to b leomy cin treatment and decreased when the phosphorylation sites wer e mutated (Figur e 3 J). Collecti v ely, these results indica te tha t the BRCT domain of MDC1 recognizes phosphorylated ESCO2 after DNA damage and recruits ESCO2 to the damage sites.

ESCO2 promotes the formation of 53BP1 foci
Since a previous study re v ealed the role of ESCO2 in HR repair ( 16 ), we then primarily investigated the mechanism by which ESCO2 promoted NHEJ repair and maintained genome stability. The immunofluorescence assay in cells treated with b leomy cin showed that ESCO2 formed damage-induced foci (Supplementary Figure S6A); additionally, the localization of ESCO2 at DSBs was devoid of 53BP1 foci, and analysis of protein intensities re v ealed that ESCO2 and 53BP1 displayed m utuall y e xclusi v e but adjacent localiza tion pa tterns (Figure 4 A). According to this observa tion, we specula ted tha t ESCO2 might regula te the formation of 53BP1 f oci; theref ore, we measured the number of 53BP1 foci in ESCO2 KD HCT116 cells. Surprisingly, relati v e to WT cells, the number of 53BP1 foci decreased significantly in ESCO2-depleted cells --regardless of bleomycin treatment --with similar results in RKO cells (Figure 4 B, C and Supplementary Figure S6B). Furthermore, depletion of ESCO2 did not reduce the number of RNF8, RNF168, H4K20me2 / 3, and FK2 foci (Supplementary Figure S6C). These results suggest that ESCO2 is important for the formation of 53BP1 foci.
A previous study has shown that 53BP1 foci consist of four to se v en nano-domains (53BP1-NDs), which assemble to form microdomains (53BP1-MDs) ( 13 ). We used 3D-SIM super-r esolution microscop y to observe the effect of ESCO2 on the formation of 53BP1-MDs and found that ESCO2 deficiency disrupted the formation of 53BP1-MDs into disordered shapes (Figure 4 D-F). QUANTEX threedimensional structure analysis of 53BP1-MDs deri v ed from ESCO2 KD HCT116 and RKO cells re v ealed an increase in mean breadth and a reduction in Curvature point TH100 (Figure 4 E, F and Supplementary Figure S6D, E), which indicated a collapse of the circularization of 53BP1-MDs. In ESCO2 KD cells, reintroduction of WT ESCO2, but not the ESCO2-S196-T233-2A mutant that could not be recruited to damage sites, could reshape the high-order organization of 53BP1-MDs (Figure 4 E, F). Moreover, Imaris image analysis was used to reconstruct the super-resolution fluorescence images of 53BP1 and GFP-ESCO2 and calculate the distance of either GFP-ESCO2 WT or 2A to the 53BP1-MDs. This analysis re v ealed that, in WT cells without ATM inhibitor treatment, GFP-ESCO2 was located either inside or at the periphery of 53BP1-MDs, but in cells treated with ATM inhibitor, it was far away from 53BP1-MDs. Additionally, GFP-ESCO2 2A showed the same pattern as that of GFP-ESCO2 in ATM inhibitor-treated cells (Supplementary Figure S6F). These results explain the exclusi v e localiza tion pa tterns seen in confocal imaging (Figure 4 A) and why ESCO2 WT, but not the ESCO2-S196-T233-2A mutant, could remold the high-order organiza-tion of 53BP1-MDs (Figure 4 E, F). In line with these observations, 53BP1 could not form repair foci in ESCO2depleted cells transfected with the ESCO2-S196-T233-2A mutant (Figures 4 G and Supplementary Figure S6G). Because 53BP1 recruits RIF1 to damage sites to pre v ent resection and channels DSB repair to the NHEJ pathway ( 9 , 11 ), we examined the number of RIF1 foci and found that the recruitment of RIF1 was restrained in ESCO2 knockdown cells (Figures 4 H and Supplementary Figure S6H). These results indicate that ESCO2 regulates the formation of 53BP1 foci and promotes NHEJ repair.

ESCO2 regulates the chromatin structure around DSB sites by acetylating SMC3 K105 / 106
The acetylation of SMC3 by ESCO2 is related to the establishment of cohesion and the increased stability of the cohesin-chroma tin associa tion ( 23 , 24 , 31 ). Furthermore, the cohesin-mediated DNA loop extrusion is involved in DSB repair ( 32 ). Our IP-MS assay that identified the ESCO2-SMC3 interaction (Figure 3 A) led us to speculate that ESCO2 promoted DSB repair by acetylating SMC3 and stabilizing cohesin structure. An ESCO2 W539G mutation has been identified in RBS patients, which results in a loss of acetyltr ansfer ase activity ( 33 ). Consistent with this study, the immunofluor escence r esults also showed that the W539G mutant could not rescue the formation of 53BP1 foci and DSB repair capability in ESCO2-depleted cells (Figure 5 A, B). Based on the abov e e xperimental results, we speculated that the ESCO2-mediated stabilization of the cohesin complex was involved in 53BP1 foci formation. To test this hypothesis, we used 3D reconstruction to monitor the high-order organization of 53BP1 and SMC3 at DSB sites. The results re v ealed that in WT cells, SMC3 localized at DSB breaks partially occupied by 53BP1, where they formed similar circular three dimensional structures; conversely, ESCO2 knockdown disrupted the threedimensional organization of SMC3 and 53BP1 (Figure 5 C, D). We mutated SMC3 K105 / 106 to glutamine (K2Q), which mimicked the hyper-acetylated state of SMC3, or to arginine (K2R), which mimicked the unacetylated state; we then analyzed the mean breadth of the 53BP1-MDs. Onl y the SMC3-K2Q m utant r escued the thr ee-dimensional organization of the 53BP1-MDs (Figure 5 D), and an increased number of 53BP1 foci was detected in ESCO2depleted cells r e-expr essed with SMC3-K2Q (Figur e 5 E, F). In keeping with this data, we observed a decrease in the number of ␥ H2AX foci after 12 h recovery in ESCO2 knockdown cells r e-expr essed with SMC3-K2Q ( Figure 5 G, H), suggesting that ESCO2-catalyzed SMC3 acetylation promoted the repair efficiency of b leomy cin-induced DSBs.
To investigate the recruitment of 53BP1 and cohesin around DSBs, we de v eloped stab le WT and ESCO2depleted HCT116 cell lines expressing AsiSI, which is a restriction enzyme that targets an 8-bp recognition sequence, fused with estrogen receptor (ER) and HA tag (HA-ER-AsiSI cells). Since treatment with 4-hydro xytamo xifen (4-OHT) triggers the nuclear translocation of HA-ER-AsiSI and generates DSBs ( 34 ), we performed chromatin immunoprecipitation with sequencing (ChIP-seq) in 4-OHTtreated ESCO2 WT and ESCO2-depleted HCT116 cells  Figure S7A, B). ChIP-qPCR analysis of pulled down DNA using primers proximal to a set of six AsiSI sites also showed that ESCO2 deficiency led to reductions of ␥ H2AX, 53BP1, and ac-SMC3 around these damage sites ( Figure 5 J and Supplementary Table S3). Consistently, SMC3 was not recruited to DSB sites in ESCO2 deficient cells ( Supplementary Figure S7C), indicating that ESCO2 stabilizes cohesin around DSB sites by acetylating SMC3. In summary, these results indica te tha t ESCO2 shapes the high-order chroma tin structure at DSB breaks by acetylating SMC3 K105 / 106, which is indispensable to the assembly of 53BP1.

Deficiency of ESCO2 leads to chemother ap y sensitivity in colorectal cancer
To further elucidate the effect of ESCO2 on genome stability in vivo , we performed a xenograft nude mouse experiment by subcutaneously injecting WT or ESCO2 KD HCT116 cells into 6-week-old female BALB / c mice. From day 14 to day 35 after inocula tion, oxalipla tin was administered intravenously at a dose of 10 mg / kg every 3 days, and the tumor volume was measur ed (Figur e 6 A). As expected, relati v e to the WT control group, ESCO2 depletion resulted in a significant decrease in tumor volume and weight in the absence of drug and r ender ed cells hypersensiti v e to oxaliplatin (Figure 6 B-D). To determine the le v els of ␥ H2AX and Ki67 --a marker of cell proliferation --in tumor specimens, tumors from mice in each group were collected and immunohistochemical assays (IHC) were performed. Consistently, ESCO2-depleted tumors treated with oxaliplatin showed that ␥ H2AX was significantly increased and Ki67 was substantially r educed (Figur e 6 E and Supplementary Figure S8). Based on TUNEL staining results, cell apoptosis also increased in ESCO2-depleted tumors after oxalipla tin trea tment (Figure 6 F). To explore the physiological function of ESCO2 phosphorylation in DSB repair in mice, we injected HCT116 WT cells, ESCO2 KD cells, and ESCO2 KD cells stably expressing either ESCO2-WT or -2A into the armpits of 6-week-old female BALB / c nude mice. Starting at day 12, we administered oxaliplatin at a dose of 10 mg / kg e v ery 3 days (Figure 6 G) and measured the tumor volumes. The results showed that re-introduction of ESCO2-WT, but not the ESCO2-2A mutant, in ESCO2 KD cells reduced the chemosensitivity to oxaliplatin. Additionally, larger tumors were observed in the ESCO2-WTexpressing group relati v e to the ESCO2 KD and ESCO2-2A groups (Figure 6 H, I). Taken together, these results suggest that ESCO2 increases the resistance of colorectal cancer cells to oxaliplatin by promoting DSB repair efficiency.

DISCUSSION
Our study demonstrates that ESCO2 facilitates the recruitment of 53BP1 to DSB sites and promotes the efficiency of 53BP1-dir ected NHEJ r epair. In r esponse to DNA damage, ESCO2 is phosphorylated by ATM kinase and recognized by MDC1, which recruits ESCO2 to DSB sites. ESCO2 acetylates SMC3 and mediates stabilization of the cohesin complex, which is essential for genome stability. Depletion of ESCO2 renders colorectal cancer cells hypersensiti v e to chemotherapeutic drugs (Figure 7 ).
A range of e xperimental e vidence supports the model that the cohesin complex and CCCTC-binding factor (CTCF) mediate DNA loop extrusion, which organizes the chromosome ( 35 ). The three-dimensional structure of the eukaryotic genome impacts and is regulated by DNA metabolism pathways such as the DNA damage response ( 36 ); for instance, chroma tin a t DSB sites compacts in a unique manner that is distinguishable from undamaged chromatin in living cells ( 37 ). Piazza et al . showed that cohesin promotes HR repair by regulating the HR pathway ( 16 ). Consistently, we also found that ov ere xpression of ESCO2 promoted HR repair and BRCA1 could not form damage-induced foci in ESCO2 KD cells (Supplementary Figure S2C, D). This finding suggests that the ESCO2stabilized cohesin complex plays an important role in the HR repair pathway. Strikingly, our results showed that ESCO2 played an indispensable role in NHEJ repair, since ESCO2 ov ere xpression increased NHEJ efficiency while ESCO2 depletion decreased NHEJ efficiency (Supplementary Figure S2C, Figure 1 E). Previous work showed that 53BP1 forms microdomains that stabilize chromatin topology at DSB sites ( 13 ). DSBs induce a genome-wide increase in cohesin to isolate damaged regions from adjacent chromatin ( 32 ). Consistently, we also found that SMC3, a major subunit of cohesin, could be recruited to the DSB site ( Figure 5 I and Supplementary Figure S7C). Furthermore, the 53BP1 foci, immunofluorescence assays were performed immediately after b leomy cin treatment ( A ). Scale bar, 2 m. To examine the ␥ H2AX foci, the cells were cultured in fresh medium for 12 h after b leomy cin treatment (5 M, 2 h) and immunofluorescence assays were then performed ( B ). Scale bar, 2 m. ( C ) 3D-SIM of 53BP1 and GFP-SMC3 foci in wild-type (WT) and ESCO2 knockdown HCT116 cells treated with b leomy cin. Scale bar, 1 m. 3D reconstruction using Imaris. ( D ) Q UANTEX anal ysis of mean-breadth of 53BP1 in WT HCT116 cells, ESCO2 knockdown HCT116 cells, and ESCO2 knockdown HCT116 cells transfected with GFP-SMC3-WT, GFP-SMC3-K2Q or GFP-SMC3-K2R plasmid. ( E , F ) ESCO2 knockdown HCT116 cells transfected with the indicated plasmids and WT HCT116 cells were treated with b leomy cin for 2 h. Immunofluorescence assays were performed to examine the 53BP1 foci. Representative images from three independent experiments are shown. Scale bar, 2 m. The graphs show mean ± SEM; n = 30 for each group. ( G , H ) ESCO2 knockdown HCT116 cells transfected with the indicated plasmids wer e tr eated with b leomy cin for 2 h, and the cells were then cultured in fresh medium for 12 h. Immunofluorescence assays were then performed to examine the ␥ H2AX foci. Representati v e images from three independent experiments are shown. Scale bar, 2 m. The graphs show mean ± SEM; n = 30 for each group. ( I ) Average profile for 53BP1 and ac-SMC3 in ESCO2 WT and ESCO2-depleted HCT116 cells. ChIP-seq analyses of WT and ESCO2-depleted HCT116 cells after 4-OHT treatment (500 nM, 4 h), using anti-53BP1 and anti-ac-SMC3 antibodies. Averaged 53BP1 and ac-SMC3 signals over a 10-kb region flanking annotated AsiSI sites are shown. ( J ) ChIP analysis was performed in AsiSI-ER-HCT116 cells after 4 h 4-OHT treatment, using anti-ac-SMC3 and anti-53BP1 antibodies as indicated. ac-SMC3 and 53BP1 enrichment was assessed by qPCR amplification using proximal primers of AsiSI sites. Statistical analysis was performed using two-way ANOVA. Statistical analysis in panels (D), (F) and (H) was performed using a Student's t test.  Model for the role of the ATM-ESCO2-SMC3 axis in the DNA damage response. When cells suffer from DSBs, ESCO2 S196 and T233 residues are phosphorylated by ATM. The phosphorylated ESCO2 is then recognized by MDC1 and recruited to DSB sites. ESCO2-mediated acetylation of SMC3 (a major subunit of cohesin) regulates the chromatin structure around DSB breaks and facilitates the formation of 53BP1 microdomains, which is important for NHEJ repair and genome stability. ESCO2-depleted colorectal cancer cells are more sensiti v e to chemotherapeutic drugs. The model was created with BioRender.com.
we identified ESCO2 as a regulator for cohesin-dependent DSB-induced chromatin structure dynamics, which is important for the recruitment of cohesin and 53BP1 to DSB sites and the formation of 53BP1-MDs. Moreover, our results suggest that DN A damage-stim ulated ESCO2 phosphoryla tion a t residues S196 and T233 by ATM is essential for the recruitment of ESCO2 to DSB sites, which is consistent with the model indicating that ATM activity is r equir ed for loop extrusion at DSBs ( 32 ). Additionally, we showed tha t phosphoryla ted ESCO2 is recognized by the phosphopeptide-binding BRCT domain of MDC1 and recruited to the damage sites in an MDC1-dependent manner. Furthermore, a previous study re v ealed that the BRCT domain of MDC1 is critical for its binding to ␥ H2AX ( 38 ). These data suggest that the BRCT domain of MDC1 proteins serves as a key functional domain to recognize dif-ferent proteins; howe v er, the detailed mechanism underlying the simultaneous binding of phosphorylated ESCO2 and ␥ H2AX to the BRCT domain of MDC1 remains to be elucidated by structural biology. Taken together, our study added an additional layer of DSB-induced chromatin dynamics through the ATM-ESCO2-SMC3 axis. Collecti v ely, our results combined with previous studies indicate that damage-induced changes in chromatin structur e ar e essential for the cellular response to DNA damage and the r ecruitment of r epair factors like 53BP1, and in turn the r ecruitment of r epair proteins at DSBs r egulates the chromatin structure around break sites and safeguards genome integrity.
Pr evious r esear ch has r e v ealed that cancer cells produce ele vated le v els of ROS ( 39 ), which has been reported to directly induce oxidati v e DNA damage, and the failure of base excision repair leads to the generation of DSBs ( 40 , 41 ). Moreover, dysfunction of Eco1 and cohesin in Sacchar om y ces cer evisiae leads to ROS overproduction ( 42 ). Here, w e show ed that depletion of ESCO2 reduced tumor growth and increased the ␥ H2AX le v el in the absence of chemotherap y r eagents, suggesting that the endogenous ROS-induced DSB damage cannot be properly r epair ed in ESCO2-depleted cells. Moreover, ESCO2-depleted cells have been reported to accumulate in S-phase ( 43 ), which at least partially contributes to a reduced prolifer ation r ate. Consistentl y, our imm unohistochemical results also showed a lower le v el of Ki67 in ESCO2-depleted tumors relati v e to WT tumors. Collecti v ely, these da ta suggest tha t depletion of ESCO2 inhibits tumor growth in the absence of chemotherapy agents by affecting endogenous DNA damage repair and cell cycle arrest.
The role of ESCO2 in colorectal cancer cells identified in this study provides a novel molecular explanation for CRC chemoresistance. As the main approach for the treatment of CRC, chemotherapy, including oxaliplatin, plays an important role in preoperati v e treatment and in reducing cancer relapse after surgery ( 44 ). Howe v er, cancer cells dev elop se v eral mechanisms to e vade oxaliplatin-induced cell a poptosis, one of w hich is the damage-induced ov ere xpression of NHEJ repair factors and improvement of NHEJ r epair efficiency, ther eby r esulting in drug r esistance ( 45 ). The upregulation of the 53BP1 expression le v el promotes NHEJ repair efficiency and leads to r adiother ap y r esistance in colorectal cancer cells ( 46 ). Our results suggest that ESCO2 regulates the formation of 53BP1 foci. Depletion of ESCO2 in CRC cells leads to the disruption of 53BP1-MDs ring-like structure and causes cancer cells to become hypersensiti v e to chemotherapeutic drugs. TCGA analysis showed that the expression level of ESCO2 positi v ely correlated with 53BP1 and BRCA1 in colorectal cancer samples (Supplementary Figure S1C). We propose that 53BP1-MDs mediated by cohesin-dependent chromatin dynamics contributes to CRC chemoresistance, and the strategies that inhibit the recruitment of ESCO2 to DBSs and / or reduce its acetyltr ansfer ase activity to disrupt the ring-like structure of 53BP1 may have therapeutic value in CRC r adiother apy and chemotherapy.
Overall, our data identify that ESCO2 plays a vital role in the DDR by regulating chromatin structure and promoting the recruitment of 53BP1. Our findings provide mechanistic insight into the relationship between the 3D genome structure and DSB repair and its role in colorectal cancer therapy.

DA T A A V AILABILITY
All data generated are included in this article and its supplementary data files.
ChIP-seq data from this study is available at the Gene Expression Omnibus ( http://www.ncbi.nlm.nih.gov/projects/ geo/ , GSE221266). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE [1] partner repository with the dataset identifier PXD039072 and 10.6019 / PXD039072. Western blot gel images can be available at https://zenodo.org/record/ 7903677#.ZFcyhi-9GTc .