NAP1L2 drives mesenchymal stem cell senescence and suppresses osteogenic differentiation

Abstract Senescence of bone marrow mesenchymal stem cells (BMSCs) impairs stemness and osteogenic differentiation, but the key regulators for senescence and the related osteogenesis are not well defined. Herein, we screened the gene expression profiles of human BMSCs from young and old donors and identified that elevation of the nucleosome assembly protein 1‐like 2 (NAP1L2) expression was correlated with BMSC senescence and impaired osteogenesis. Elevated NAP1L2 expression was observed in replicative cell senescence and induced cell senescence in vitro, and in age‐related senescent human and mouse BMSCs in vivo, concomitant with significantly augmented chromatin accessibility detected by ATAC‐seq. Loss‐ and gain‐of‐functions of NAP1L2 affected activation of NF‐κB pathway, status of histone 3 lysine 14 acetylation (H3K14ac), and chromatin accessibility on osteogenic genes in BMSCs. Mechanistic studies revealed that NAP1L2, a histone chaperone, recruited SIRT1 to deacetylate H3K14ac on promoters of osteogenic genes such as Runx2, Sp7, and Bglap and suppressed the osteogenic differentiation of BMSCs. Importantly, molecular docking analysis showed a possible bond between NAP1L2 and an anti‐aging reagent, the nicotinamide mononucleotide (NMN), and indeed, administration of NMN alleviated senescent phenotypes of BMSCs. In vivo and clinical evidence from aging mice and patients with senile osteoporosis also confirmed that elevation of NAP1L2 expression was associated with suppressed osteoblastogenesis. Taken together, our findings suggest that NAP1L2 is a regulator of both BMSC cell senescence and osteogenic differentiation, and provide a new theoretical basis for aging‐related disease.


| INTRODUC TI ON
Aging is a comprehensive process whose manifestations vary from cellular biological functions to gene transcription and protein post-translational modifications (DiLoreto & Murphy, 2015;LeBrasseur et al., 2015). The causes of bone mass loss during aging are diverse, but it is generally agreed that age-related bone loss is correlated with the imbalance of osteoblast-mediated bone formation and osteoclasts-mediated bone resorption in animals and in humans (Al-Bari & Al Mamun, 2020). Stem cell exhaustion is considered one of the promoters of aging and contributes to the progressive decrease in tissue maintenance and repair (Liu et al., 2020;Wagner et al., 2019). Despite this knowledge, key regulators of aging and a detailed understanding of the biological mechanisms underpinning aging are elusive.
Bone marrow mesenchymal stem cells (BMSCs) have the potential for self-renewal and serve as a lifelong reservoir for the generation of somatic cells and a powerful tool in bone tissue regenerative medicine (Sanghani-Kerai et al., 2018;Uccelli et al., 2008). However, it has become apparent that the osteogenic differentiation of BMSCs decreases with senescence, which limits the potential of BMSCs in bone regeneration (Abdelmagid et al., 2015). Senescence of MSCs is a dynamic process accompanied by functional alterations in metabolism, genetic and epigenetic regulation of the transcriptome, and multiple signaling pathways (Mani et al., 2020;Wong et al., 2015).
However, changes in these processes are probably a consequence of aging, and the real regulators governing BMSC senescence and osteogenesis are still unclear.
Nucleosome assembly protein 1 (NAP1) is a highly conserved gene family and is closely associated with nucleosome assembly, histone modifications, transcriptional regulation, and cell proliferation in various eukaryotes (Zlatanova et al., 2007). In humans and mice, the NAP1 gene family consists of at least five members, and previous studies have identified different roles of these molecules in regulating myocardial differentiation of induced pluripotent stem cells (iPSCs) (Gong et al., 2014), as well as homeostasis and hematopoietic differentiation of cord blood hematopoietic stem cells (Heshmati et al., 2018). Specifically, the Nap1l2 gene plays an essential role in neural tube development and neural cell differentiation in mice (Rogner et al., 2000). It is also involved in the epigenetic regulation of gene expression during neuronal differentiation (Attia et al., 2013).
Nevertheless, roles and mechanisms of NAP1L2 in regulating human BMSC senescence and osteogenesis have not yet been reported.
In the current study, we screened gene expression profiles of BMSCs derived from the young and the old donors, and identified NAP1L2 gene as one of the most upregulated genes during aging, which expression was associated with impaired osteogenic differentiation of senescent BMSCs. We aim to clarify the role and mechanism of NAP1L2 in governing BMSC senescence and osteogenic differentiation through in vitro and in vivo experiments, as well as the correlation with senile osteoporosis in clinic.

| BMSCs from elderly donors manifest senescent phenotypes and suppressed osteogenic differentiation capacity
Since the aim of this study is to identify key regulators governing BMSCs senescence and osteogenic differentiation, we firstly isolated BMSCs from the healthy young donors and the elderly donors, with the young donors aged between 17 and 45 years, and the elderly donors aged between 65 and 91 years. All these cells were examined by flow cytometry to ensure that CD90 + CD44 + CD19 − CD45 − cells were purified for further experiments according to literature (Adesida et al., 2012) (Figure S1a). At the same time, osteogenic and adipogenic differentiation were conducted to guarantee the pluripotent stemness ( Figure S1b). The positive rate of the β-galactosidase staining was significantly increased in the old group compared with that in the young BMSCs (83 ± 1.56% vs. 28 ± 2.39%) ( Figure S2a,b), and the expressions of the senescence-related p21 and p16 were significantly elevated ( Figure S2c,d). Moreover, the cell cycle of the old BMSCs was remarkably arrested at the G1 phase, indicating the proliferation capability was obviously suppressed ( Figure S2e,f).
Detection of telomere dysfunction-induced foci (TIF) revealed that accumulation of telomere DNA damage, along with γ-H2AX and 53BP1 was seen in the old BMSCs ( Figure S2g,h). As to the osteogenic differentiation capacity, we assessed the alkaline phosphatase (ALP) activity and the mineralization of the extracellular matrix of BMSCs induced with osteogenic culture media for 7 or 14 days and confirmed that the old BMSCs had considerably impaired osteogenic differentiation capacity compared with the young BMSCs ( Figure   S3a-d). Meanwhile, the expressions of osteogenesis marker genes, such as RUNX2, COL1A1, and ALP, were all significantly decreased in the old BMSCs after induction ( Figure S3e). Collectively, these data validated the stemness property of the BMSCs and confirmed the senescence phenotype of the old BMSCs.

| Highly expressed NAP1L2 is correlated with BMSCs senescence
Based on the corroboratory evidence for the BMSCs' properties and senescence, we next performed gene array sequencing and the assay for transposase accessible chromatin with highthroughput sequencing (ATAC-seq) for six biologically independent samples from the old and the young BMSC groups ( Figure S4a), respectively. Our screening analysis identified 15 upregulated and 7 downregulated genes in the old BMSCs compared with the young BMSCs (fold change ≥ 2, p ≤ 0.05) (Figure 1a), in which the expression level of the NAP1L2 gene ranked at the top of all the differentially expressed genes (DEGs) in the old BMSCs, accompanied by high expression of senescence-related secreted phenotype (SASP) factor genes (TP53, IL6, and CCXL8) (Figure 1b). Gene Ontology (GO) analysis showed that the biological processes involving these DEGs mainly involved stem cell differentiation and IL6 and IL8 secretion ( Figure 1c). Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis showed that the DEGs were primarily involved in the calcium signaling pathway and cAMP signaling pathway ( Figure   S4b). Compared with the young group, the global chromatin accessibility areas were significantly augmented in the old BMSCs ( Figure 1d). We also observed that the chromatin accessibility of the NAP1L2, IL6, IL8, CDKN2A, and TNFα genes was remarkably increased, but that of the RUNX2 gene was not obviously changed ( Figure 1e). Simultaneously, using a replicative senescence BMSCs model, after the senescence phenotype was successfully validated by telomere dysfunction and DNA damage ( Figure S5a), we further confirmed that NAP1L2 protein level was obviously upregulated ( Figure S5b,c). Thus, our data suggest that the elevation of NAP1L2 gene expression in the aging BMSCs was somehow correlated with chromatin modification and remodeling.

| NAP1L2 expression is elevated in chemotherapeutic agents-induced and oxidative stress-induced cellular senescence models
To determine the biological function of NAP1L2 in senescence, we established the chemotherapeutic agents-induced cellular senescence model with the mouse mesenchymal stem cell line C3H10T1/2 by etoposide as previously reported (Bang et al., 2019), and confirmed the senescent phenotype by β-galactosidase staining ( Figure   S6a). Telomere FISH and immunostaining for γH2AX and 53BP1 revealed significantly increased telomere dysfunction-induced foci in the etoposide treatment group ( Figure S6b), under the condition that no visible apoptosis was induced by the reagent, since the cleaved caspase-3 and PARP unchanged ( Figure S6c). Quantification by qPCR confirmed that the expression of p21 and SASP factors, such as Il1α, Il1β, Il6, and Il8, was successfully elevated along with our target gene Nap1l2 ( Figure S6d). Simultaneously, the protein levels of p53, p16, p21, and NAP1L2 were all upregulated with increasing dosage of etoposide treatment ( Figure S6e). In addition, using an oxidative stress-induced cell senescence model, whose senescent phenotype was validated by β-galactosidase staining and telomere dysfunction assay without visible cell apoptosis ( Figure S6f-i), we found that high expression of Nap1l2, p21, and SASP factors was successfully elicited by treatment of H 2 O 2 ( Figure S6j), and protein levels of NAP1L2, together with p53, p21, and p16 were all upregulated following H 2 O 2 treatment ( Figure S6k).

| Manipulation of NAP1L2 alters cellular senescence progression through NF-κB signaling pathway
Importantly, when we successfully manipulated the NAP1L2 expression with either lentivirus carrying shRNAs or vectors expressing mouse NAP1L2 in C3H10T1/2 cells ( Figure S7a concomitantly ( Figure S7f). Since the NF-κB signaling pathway is the prime signaling pathway governing aging in mammals (Osorio et al., 2012;Wang et al., 2019), we examined whether manipulation of NAP1L2 could affect the activation of the NF-κB signaling pathway. Our results showed that overexpression of NAP1L2 augmented, but knockdown of NAP1L2 suppressed the p16, p21, p53, and phosphorylation level of p65 ( Figure 3c). Moreover, the cytoplasm and nuclear fractions of p65 protein showed that majority of p65 was transported into the nucleus due to NAP1L2 overexpression, while knockdown of NAP1L2 hindered this transportation ( Figure 3d). This phenomenon was also confirmed by immunofluorescence staining of cyto-nuclear trans-localization of p65 (Figure 3e), and p65 distinctly co-localized with NAP1L2 after transported into the nucleus ( Figure 3f). Furthermore, we evaluated the alteration of the transcriptome due to NAP1L2 depletion, and the deferentially expressed genes in the heat map showed that levels of Il1a, Il1b, Il10, Il23a (P19), Cxcl15 (Il8), Trp53, Bcl2, and Mmp3 were significantly decreased ( Figure 3g). GO analysis of DEGs highlighted the involvement of the inflammatory response and immune response biological processes ( Figure   S7g), and KEGG pathway analysis showed that DEGs were mainly involved in the NF-κB signaling pathway, cellular senescence, longevity regulating pathway, and p53 signaling pathway ( Figure 3h).
These findings suggest that NAP1L2 regulates cellular senescence through the NF-κB signaling pathway and sequentially modulates the SASP factor expressions.

| NAP1L2 is a key suppressor of osteogenic differentiation
To further clarify the relationship between NAP1L2 and osteogenesis, we then evaluated the effect of loss-and gain-of-functions NAP1L2 on osteogenic differentiation in C3H10T1/2 cells. After induction with osteogenic media for 7 days, C3H10T1/2 cells with NAP1L2 knockdown showed an obvious increase in ALP staining and activity We next validated the role of NAP1L2 in regulating BMSC senescence and osteogenesis in primary cells derived from young and aged mice. We observed that loss function of NAP1L2 significantly reduced the telomere dysfunction-induced foci in BMSCs from senile mice (Figure 4g), and overexpression of NAP1L2 was capable of exacerbating telomere DNA damage in BMSCs from young mice ( Figure S8f). Compared with the BMSCs from young mice, the expression of osteogenic genes in the BMSCs from aged mice was all significantly decreased along with augmentation of NAP1L2 expression, which was consistent with our previous in vitro results (Figure S8g). However, when NAP1L2 expression was interfered in the BMSCs from aged mice, the osteogenic differentiation capacity could be obviously rescued, evidenced by the significant increase in ALP activity and mineralization of the extracellular matrix ( Figure 4h), as well as upregulated osteogenic genes after 7/14 days of induction (Figure 4i and Figure S8h). These data strongly suggest that NAP1L2 is an important suppressor for osteogenesis.

| NAP1L2 recruits SIRT1 to deacetylate H3K14ac on promoters of osteogenic genes
Since a previous study revealed that NAP1L2 regulates neuronal differentiation through histone acetylation modification in an "on the gene" its function through a similar epigenetic mechanism. We screened the global status of the most common histone modifications upon NAP1L2 depletion, including H3K27me2, H3K36me2, H3K79me3, H3K27ac, H3K14ac, and H3K9ac. We found only H3K14ac level was remarkably elevated compared with the non-target control (Figure 5a).
Meanwhile, pan-acetylation of whole histones was greatly increased after osteogenic induction in the NAP1L2 depletion C3H10T1/2 cells ( Figure 5b). On the contrary, when NAP1L2 was forcedly expressed, H3K14ac level was markedly inhibited, but H3K9ac level was not significantly affected (Figure 5c). Interestingly, after overexpression of NAP1L2, we found a direct interaction of NAP1L2 with H3K14ac and SIRT1 (Figure 5d,e). All of these findings suggested that NAP1L2 might reduce the level of H3K14 acetylation by interacting with SIRT1, thereby mediating chromatin remodeling and affecting osteogenesis.
RNA sequencing results from C3H10T1/2 cells with NAP1L2 depletion and undergoing osteogenic induction for 7 days also revealed that osteogenic genes such as Bglap, Runx2, Sp7, and Alp were all increased significantly compared with the non-target control group (Figure 5f).
GO analysis for the DEGs further suggested participation in osteoblast differentiation, mineralization, osteoblastogenesis, BMP signaling, and skeleton development (Figure 5g). Moreover, the results of H3K14ac-ChIP-seq and ChIP-qPCR showed that in the NAP1L2 knockdown group, acetylation modification of H3K14 on the promoter regions of the osteogenic genes Runx2, Sp7, and Bglap was increased significantly, which would benefit the expression of osteogenic genes and then promote osteogenic differentiation (Figure 5h,i). Thus, our present data elaborate a possible mechanism of NAP1L2 in regulating osteogenesis through H3K14 acetylation-mediated chromosomal remodeling.

| NAP1L2 expression is closely correlated with senile osteoporosis in vivo and in clinic
To verify the important role of NAP1L2 in vivo, we successfully es- femur sections of osteoporotic mice showed that the expression was significantly higher in the leptin receptor positive (LEPR + ) BMSCs than that in the control group ( Figure 6g). Notably, when we isolated BMSCs from patients with osteoporosis (Figure 6h), we also observed significantly higher NAP1L2 expression than those from agepaired healthy controls (Figure 6i). A significant negative correlation was elicited between NAP1L2 expression and bone mineral density (BMD), as well as the CT value for the L1 lumbar in osteoporosis patients (Figure 6j,k); however, no significant correlation between NAP1L2 expression and bone mineral density (BMD) or CT value in the age-paired healthy donors was observed (Figure 6l). Collectively, these in vivo and clinical data strongly demonstrate a close relationship between senile osteoporosis and NAP1L2 expression.

| DISCUSS ION
Aging is such a complicated process that the key drivers and regulators are not easy to define. In the current study, we report that NAP1L2, whose expression is significantly elevated during aging, has a close correlation with physiological and pathophysiological senescence as well as with the suppressed osteogenic capabilities of BMSCs. We also revealed a possible regulatory mechanism of NAP1L2 in activating the NF-κB pathway and impairing osteogenic potential through epigenetic regulation of histone acetylation at

H3K14.
Ample studies have revealed that aging leads to diverse changes in cellular biology and the transcriptome regulated by genomic or epigenetic mechanisms (Declerck & Vanden Berghe, 2018;Field et al., 2018). In our experiment, we firstly confirmed that BMSCs derived from elderly donors showed a significant senescent phenotype compared with those from young donors, evidenced by increasing β-galactosidase staining, upregulated age-related genes p21 and p16, cell cycle arrest at the G1 phase, limited proliferation, augmented telomere DNA damage, and suppressed osteogenic differentiation capacity. Based on these findings, we performed gene array sequencing to screen differentially expressed genes and identified 21 differently expressed genes between the two groups, with 15 being elevated and 7 being suppressed. Our focus on the Nap1l2 gene is based upon Rogner's series of work showing that NAP1L2 plays a critical role in neuronal differentiation in mice (Attia et al., 2007(Attia et al., , 2013Rogner et al., 2000). In our study, we also observed that the elevated NAP1L2 in is an epigenetic regulator of nucleosome assembly via histone acetylation modification (Attia et al., 2007). Moreover, our GO analysis showed that IL6 and IL8 secretion is one of the most changed biological processes, along with stem cell differentiation. In fact, IL6 and IL8 are important components of senescence-associated secretory phenotype (SASP) (Gorgoulis et al., 2019). In our in vitro-induced cellular senescence model and in BMSCs from the aged mice, depletion of NAP1L2 significantly alleviated the process of cellular senescence.
Concomitantly, SASP factors, such as IL6 and IL8 secretion, activation of the NF-κB signaling pathway, and activation of the p53 signaling pathway were all remarkably diminished. In sum, our data suggest that NAP1L2 regulates the cellular senescence process through the NF-κB signaling pathway and thus identifies NAP1L2 as an important regulator of the senescence of BMSCs.
It has become apparent that the osteogenic differentiation of BMSCs decreases with senescence, which limits the potential of BMSCs in bone regeneration (Abdelmagid et al., 2015;D'Ippolito et al., 1999). Our study also evaluated the impact of NAP1L2 on the osteogenesis of BMSCs, and the results indicated that deletion of Nap1l2 resulted in a significant amelioration of the osteogenic differentiation of BMSCs. Mechanistically, NAP1L2 likely participates in restraining H3K14 acetylation (H3K14ac) on promoters of osteogenic genes, such as Bglap, Runx2, and Sp7, because depletion of NAP1L2 significantly augmented the global H3K14ac and expression of these genes.
These findings were slightly different from those of a previous study, which reported that NAP1L2 coexists with chromatin in the nucleus and facilitates histone acetyltransferase activity to histones H3 and H4 (Attia et al., 2007). Interestingly, our results also suggest that NAP1L2 actually recruits the NAD + -dependent deacetylase sirtuin-1 (SIRT1), and the latter reduces the level of H3K14ac on promoters of osteogenic genes, since H3K14ac is generally linked to gene activation (Li et al., 2016). Thus, we uncovered a previously unreported role of NAP1L2 in regulating osteogenic differentiation. However, we are not sure whether NAP1L2 only affects cell senescence and osteogenic  3). (e) Representative images of 53BP1/γ-H2AX at telomeres in 24-month BMSCs (n = 3) and 6-month BMSCs (n = 3). Scale bars, 5 μm. (f) Western blot and qPCR demonstrating NAP1L2, p53, p21, and p16 expression in 24-month BMSCs (n = 3) and 6-month BMSCs (n = 3). (g) Representative micrographs of immunohistochemistry staining for LEPR and NAP1L2 in femurs of the 6-month (n = 3) and 24-month mice (n = 3). (h) CT images of age-paired healthy donors (n = 3) and patients with senile osteoporosis (n = 3). (i) qPCR showing NAP1L2 expression in osteoporosis patients' BMSCs (n = 12). (j) Correlation analysis of NAP1L2 and bone mineral density (BMD) in osteoporosis patients (n = 12). (k) Correlation analysis of NAP1L2 and CT value in osteoporosis patients (n = 12). (l) Correlation analysis of NAP1L2 and bone mineral density (BMD) or CT value in age-paired healthy donors (n = 12). p values of mean ± SD determined by Student's t test differentiation at the same time or inhibits osteogenic differentiation by inducing cell senescence.
Finally, using an in vivo aging mouse model and clinical sample from patients with senile osteoporosis, we further validated the correlation of elevated NAP1L2 expression with the development of aging and impairment of BMSCs. Similar to our above results, deletion of NAP1L2 improved the osteogenesis of the senescent BMSCs. Therefore, we verified that augmentation of NAP1L2 expression is a barrier to bone formation. Translationally, our study also suggests that nicotinamide mononucleotide (NMN), a crucial NAD + intermediate that decreases with age in mammals (Song et al., 2019), binds NAP1L2 protein, and suppresses its expression.
Other studies have verified that supplementation of NMN has preventive effects against age-associated pathophysiology and disease conditions (Yoshino et al., 2018), and also clarifies that administration of NMN reduces senescent phenotypes of BMSCs and ameliorates osteogenic differentiation.
Hence, our findings suggest that NAP1L2 serves as a new theoretical basis for aging-related disease. The translational merit of our study is to indicate that the aging antagonist reagent nicotinamide mononucleotide (NMN) has an inhibitory effect on NAP1L2 expression, thus providing a theoretical basis for the management of agerelated diseases.

| Animals and treatment procedure
We divided the mice into two groups: the old C57 mice (24 months old) and the young C57 mice (6 months old). Mice were sacrificed under 10% chloral hydrate anesthesia. Both sides of the femur were collected and fixed in 4% paraformaldehyde. The femur samples were first dehydrated using 70% ethanol for 48 h. Samples were scanned using a micro-CT scanner (Bruker microCT, Kontich, Belgium). The ROI is superiorly placed 1.0 mm from the distal growth plate of the femoral head within the femoral cortex. All images were post-operated to isolate cancellous bone from cortical bone and preserve its morphology using a threshold of 800 manually. Scans were repeated three times, bone micro-architectural parameters of each group were taken from the same region of interest (ROI). The following structural parameters of the ROI were calculated: bone volume/ tissue volume (BV/TV), trabecular number (Tb.N), trabecular thickness (Tb.Th), trabecular separation (Tb.Sp), and percentage of cortical bone area to total tissue area (B.Ar/T.Ar %).

| Cell culture
The samples were divided into groups according to age, the young group (young BMSC, 17-45 years old, n = 6), the old group (old-BMSC, 65-91 years old, n = 6). BMSCs were taken from patients undergoing fracture clean-up and requiring bone marrow examination.
Exclusion criteria include malignant tumors, family genetic diseases, and other bone diseases. All patients have signed informed consent.
Density gradient centrifugation was used to isolate adult bone marrow mononuclear cells (MNC), and BMSCs were purified by the characteristics of differential attachment. BMCSs were plated with Dulbecco's modified Eagle's medium (DMEM) including 15% fetal bovine serum (FBS) (Gibco, Life Technologies, USA), 100 U/ml penicillin, 100 μg/ml streptomycin. The medium was replaced every 3 days.
After the cell fusion degree reaches 80-90%, the cells are passaged, and passages 3 can be used for subsequent experiments. BMCSs were passaged 1:2 when cultures reached confluency. Cultures were defined as replicative senescent BMCSs when they failed to reach confluency 10 d after a 1:2 split. Primary mouse BMSCs were isolated from C57BL/6J mice at 6 or 24 months by flushing the bone marrow of femurs. Cells were cultured in minimum essential medium α (α-MEM) supplemented with 15% FBS. Human embryonic kidney 293 T cells and multipotent mouse mesenchymal stem cells (C3H10T1/2) were maintained in DMEM medium with 10% FBS.

| Plasmid construction and viral infection
Plasmids expressing shRNAs targeting Nap1l2 were purchased from Genechem (Shanghai, China). For NAP1L2 overexpression, Nap1l2 cDNA was cloned into the pITA-insert vector. Transient transfections to HEK293T cells were performed using polyethyl-

| Alizarin red Staining and calcium quantitative analysis
After cultured with osteoblast induction medium for 14 days, the cells were washed twice with PBS and fixed at room temperature for 15 min with 95% ethanol and washed with deionized water. The fixed cells were incubated with 40 mM (PH: 4.2) Alizarin red staining solution (Sigma-Aldrich, USA) for 10 min at room temperature and then washed thoroughly in deionized water. To quantitate the calcium mineral content, the stained cells were dissolved with 10% cetylpyridinium chloride for 60 min at room temperature.
Concentrations were determined by reading the absorbance at 562 nm using a standard calcium curve. The final calcium level in each group was normalized to the total protein concentrations obtained from a duplicate plate.

| Immunofluorescence-telomere FISH (IF-FISH) of telomere dysfunction-induced foci assay
Fluorescence in situ hybridization (FISH) of telomeres was performed on 2% formaldehyde fixed cells using a PNA-FISH with a (CCCTAA)3-Alexa488 telomere probe (Panagene, Daejeon, KR). 53BP1/γ-H2AX immunofluorescence-telomere FISH was carried out as described by Kasbek and colleagues(Kasbek et al., 2013). The primary antibody and the Alexa Fluor TM 555-conjugated secondary antibody were both diluted 1:500. Stained cells were photographed using an Olympus inverted microscope or confocal microscope (Olympus FV1000, JP). All antibodies, vendors were provided in the key resources table.

| Co-immunoprecipitation and Chromatin-Immunoprecipitation (ChIP)
Co-IP was conducted according to a procedure previously described (Liu et al., 2014). For the ChIP assay, 1 × 10 6 cells were crosslinked in 1% formaldehyde (Thermo Scientific, USA) for 10 min. Crosslinking was quenched with glycine, and cells were washed two times with ice PBS. The ChIP of H3K14ac (Abcam, UK) cells was lysed in lysisbuffer and chromatin was sonicated to a fragment size of 150 to 300 bp by using eight cycles. The chromatin concentration and chromatin fragments were checked using Nanodrop and 2% agarose gel. Then, samples were immunoprecipitated using an anti-H3K14ac antibody linked to magnetic beads and the DNA purified with QIAquick PCR Purification Kit (QIAGEN, USA). The normalized enrichment value was calculated as the subtraction of the IP relative value with the input relative value.

| The assay for transposase accessible chromatin with high-throughput sequencing (ATACseq)
ATAC-seq was performed as previously described (Buenrostro et al., 2015). 5 × 10 3 -5 × 10 4 cells were used for each tagmentation using Tn5 transposases. The resulting DNA was isolated, quantified, and sequenced on an Illumina NextSeq500 system. The raw reads were aligned to the human genome assembly hg19 using Bowtie40 with the default parameters, and only tags that uniquely mapped to the genome were used for further analysis.

| Statistical analysis
All quantitative data are shown as mean ± SD for at least three independent experiments. Differences between groups were determined using paired two-tailed sided Student's t test or twoway ANOVA. Pearson correlation test was used to determine the correlations between gene expressions. A p value < 0.05 was considered statistically significant compared with the controls, respectively.

ACK N OWLED G M ENTS
All authors thank the assistance from Prof. Jun Li at the Dept.

CO N FLI C T O F I NTE R E S T
All authors declare no conflict of interest.