ATRX cooperates with TOP2B for replication fork stability and DNA damage response through G-quadruplex regulation

Abstract

G-quadruplexes (G4s) are noncanonical DNA structures that promote genomic instability, particularly in α-thalassemia/mental retardation X-linked (ATRX)-deficient gliomas. Although TOP2B has been implicated in chromatin remodeling, its role in G4 resolution remains poorly understood. Here, we identify TOP2B as a previously unrecognized regulator of G4 homeostasis and show that it functionally cooperates with ATRX to facilitate G4 resolution during DNA replication. Disruption of this pathway by CX-5461, a small molecule originally developed as an RNA polymerase I inhibitor, leads to G4 accumulation, replication stress, and DNA damage. Mechanistically, CX-5461 acts as a TOP2B poison that selectively impairs TOP2B binding at G4 sites, alters replication fork dynamics, and induces MRE11-dependent degradation of stalled forks. These effects are strongly enhanced in ATRX-deficient glioma cells, where TOP2B plays a dominant role in G4 regulation. While etoposide similarly induces G4-related DNA damage, it does not affect the ATRX-TOP2B interaction, highlighting CX-5461’s unique mechanism. Our findings establish TOP2B as a critical player in G4 resolution, reveal CX-5461’s dual function as a TOP2B poison and G4 stabilizer, and propose G4-associated replication stress as a potential therapeutic target in ATRX-deficient gliomas.

Graphical Abstract

Graphical Abstract.

Introduction

DNA topoisomerase II (TOP2) is an essential enzyme that resolves DNA topological problems such as supercoiling and catenation [1]. Humans express two isoenzymes, TOP2α (TOP2A) and TOP2β (TOP2B), encoded by separate genes. These enzymes function as homodimers to generate DNA double-strand breaks (DSBs) with a canonical 5′ overhang [2]. Despite their catalytic and structural similarities, their functions are not redundant: TOP2A is essential for DNA replication and chromosome segregation, while TOP2B primarily regulates transcription, neural differentiation, and the expression of long neuronal genes [3–6]. Notably, TOP2B has been implicated in DSBs response and homologous recombination (HR)-mediated repair [7]. In a subset of gliomas, TOP2B is overexpressed and promotes a proliferative phenotype by modulating transcription of oncogenes [8], suggesting that targeting TOP2B could be a promising therapeutic strategy. However, the specific roles of TOP2B in DNA secondary structure resolution—particularly G-quadruplexes (G4s)—and in glioma biology remain poorly defined.

The α-thalassemia/mental retardation X-linked (ATRX) gene is a frequently mutated tumor suppressor in various human cancers, including malignant glioma, sarcoma, neuroblastoma, and pancreatic neuroendocrine tumors [9–12]. In gliomas, ATRX is one of the most commonly altered genes and serves as a critical molecular marker, particularly in isocitrate dehydrogenase (IDH)-mutant astrocytomas and pediatric gliomas [9, 13]. In pediatric glioblastomas (GBMs), mutations in the H3.3-ATRX-DAXX chromatin remodeling pathway have been identified in up to 44% of cases, underscoring their pathogenic significance [14, 15]. In adults, ATRX mutations are rare in primary IDH-wild-type GBMs but are more frequently observed in lower-grade (WHO grade II/III) gliomas and in secondary GBMs that arise from preexisting lower-grade lesions [16, 17]. ATRX mutations exhibit a strong association with IDH mutations, frequently co-occur with TP53 mutations, and are mutually exclusive with 1p/19q codeletion, thus defining a distinct molecular subclass of diffuse gliomas with important prognostic implications [18, 19]. As an SWItch/Sucrose Non-Fermentable (SWI/SNF)-like chromatin remodeler, ATRX participates in chromatin remodeling, suppression of alternative lengthening of telomeres (ALT), DNA damage repair regulation, and transcriptional control [20]. It binds to DNA G4—DNA secondary structure formed in guanine-rich regions—and interacts directly with G4 in vivo [21–24]. ATRX facilitates the replication of telomeric G4-DNA structures [25], and its absence leads to G4 accumulation at DNA replication sites [21]. Exogenous expression of ATRX reduces G4 levels in ATRX-deficient cells [26]. Moreover, ATRX supports incorporation of histone variant H3.3 at G4-containing loci through its interaction with DAXX, helping maintain these regions in a compact heterochromatic state [27]. This heterochromatinization is critical to protect cells from G4-induced replication stress. Despite well-established links among ATRX, G4 structures, and heterochromatin, the precise molecular mechanisms underlying ATRX with its enzymatic collaborator recruitment and function at G4 sites remain incompletely understood.

ATRX-deficient gliomas exhibit heightened sensitivity to G4-stabilizing agents such as CX-5461 (Pidnarulex) [26, 28], an oral ribosomal RNA (rRNA) synthesis inhibitor currently in Phase I clinical trials for BRCA1/2 (Breast Cancer Susceptibility gene 1/2)-mutant breast and ovarian cancers [29]. CX-5461 induces DNA damage, which requires both BRCA1/2-mediated HR and DNA-PK-mediated nonhomologous end-joining (NHEJ) for repair. Emerging evidence suggests that CX-5461 also functions as a TOP2 poison [30–32], raising the question of whether TOP2B interacts with G4 structures in gliomas and how CX-5461 affects this interaction.

In this study, we demonstrate that TOP2B binds to G4 structures and cooperates with ATRX to regulate G4 stability. CX-5461 alters TOP2B chromatin binding and disrupts its interaction with ATRX, reducing recruitment of TOP2B to G4 sites. This disruption leads to replication stress, transcriptional interference, DNA damage, and ultimately cell death. ATRX and TOP2B have complementary roles in resolving G4s. In the absence of ATRX, glioma cells become more dependent on TOP2B for G4 resolution, increasing their sensitivity to CX-5461. Comparatively, etoposide—a classical TOP2 poison—also induces G4-associated DNA damage but does not disrupt the ATRX-TOP2B coordinated interaction, highlighting a unique mechanism of CX-5461. These findings suggest that CX-5461 selectively targets TOP2B and represents a promising therapeutic strategy for treating ATRX-deficient gliomas.

Materials and methods

Cell lines and cell culture

U87MG and U251 cell lines were obtained from the American Type Culture Collection (ATCC, Rockville, MD, USA). These cell lines were verified with short tandem repeat (STR) profiling and were free of mycoplasma. Cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% FBS and 1:100 penicillin/streptomycin (Invitrogen, Carlsbad, CA, USA) at 37°C in 5% CO₂.

Lentivirus packaging and generation of stable cell lines

Lipofectamine 2000 reagent (Invitrogen) was incubated with Opti-MEM I Reduced Serum Medium (Gibco) and utilized for transfecting HEK293T cells with a mixture of 3 μg of the indicated plasmid, 3 μg of the pMD2.G (Addgene), and 6 μg of the psPAX2 (Addgene). Ten milliliters of fresh medium were added 6 h post-transfection. The supernatant containing the virus was gathered after 48 and 72 h and subsequently passed through a filter. 3 × 10⁵ cells were plated in a 60-mm dish (Corning) a day before the infection. Fresh medium containing 25 μl/ml lentivirus and 10 μg/ml polybrene (Sigma-Aldrich) was used to replace the original medium for infection. After 2 days, stable cell lines were chosen by exposing them to 4 μg/ml puromycin (InvivoGen) and 15 μg/ml blasticidin (InvivoGen) for a period of 14 days.

Ethics statement

This study was conducted in accordance with strict ethical guidelines and approved by the Ethics Committee of Tongji University School of Medicine. The use of patient specimens and associated clinical data was authorized by the Department of Neurosurgery, Shanghai East Hospital, Tongji University School of Medicine (Approval No. 2023tjdxsy036). Written informed consent was obtained from all participants prior to sample collection and data usage.

Patient-derived cells

Patient-derived glioma primary cells were established as follows. Glioma tissue samples were obtained from patients undergoing neurosurgical procedures at the Neurosurgery Department of Shanghai East Hospital, Tongji University School of Medicine. Fresh tumor specimens were immediately transported on ice in serum-free DMEM/F12 medium (Gibco, USA) after surgical excision. Upon arrival, surface blood was removed by washing the tissue with phosphate-buffered saline (PBS) supplemented with 5% penicillin-streptomycin. The tumor samples were then transferred to a sterile culture plate and finely minced into 1-2 mm³ fragments using sterile surgical instruments. The minced tissue was allowed to settle, and the supernatant was carefully removed. For enzymatic digestion, tissue fragments were incubated in 0.2% Type IV collagenase (YESEN, China) or Accutase (Invitrogen, USA) in a cell culture incubator for 30-45 min with gentle agitation every 5-10 min. Digestion was considered complete when no visible tumor fragments remained. To terminate enzymatic digestion, an equal volume of DMEM/F12 medium was added, and the mixture was filtered through a 70 μm cell strainer. The filtrate was centrifuged at 1500 rpm for 10-15 min, and the supernatant was discarded. To remove residual red blood cells, the cell pellet was resuspended in 500 μl of red blood cell lysis buffer and incubated at 37°C for 1-2 min. The mixture was centrifuged at 1500 rpm, and the supernatant was discarded. The resulting cell pellet was resuspended in culture medium and transferred to a cell culture incubator for further expansion. Patient-derived cells (PDCs) were maintained in DMEM/F12 medium supplemented with B27 (1:50, Gibco, USA), basic fibroblast growth factor (bFGF, PeproTech, USA) at 25 ng/ml, epidermal growth factor (EGF, PeproTech, USA) at 50 ng/ml, 10% fetal bovine serum (FBS, Gibco, USA), and 1% penicillin–streptomycin (Thermo Fisher Scientific, USA).

Proliferation assays

Cells were initially distributed into 96-well plates and subjected to appropriate compound treatments. Subsequently, Cell Counting Kit-8 (CCK-8) reagent (KeyGEN BioTECH, KGA9305) was introduced into the culture medium within the 96-well plate, achieving a final concentration of 10% (v/v). The transparent-bottomed plate was then placed in a cell culture incubator for an incubation period ranging from 1 to 4 h, after which absorbance was quantified at 450 nm.

Transwell

Add 150 μl of DMEM culture medium to each well, and incubate for 1 h in the incubator. Next, put 2 × 10⁴ cells mixed in 100 μl of DMEM culture medium into the upper compartment of a 24-well FluoroBlok insert (Corning, 8-mm pore). Prepare a 24-well plate by adding 600 μl of DMEM culture medium containing 30% FBS to each well and carefully place the Transwell chambers inside. Allow the setup to incubate in the cell culture incubator for 24 h. After incubation, carefully remove the medium from the chambers. Following this, fix the chambers with 4% paraformaldehyde for 15-30 min. Proceed to stain the chambers with 0.5%-1% Crystal Violet (Beyotime, C0121) for 20 min, then wash them three times with PBS. Finally, capture images using an Olympus CKX53 microscope.

Clonogenic assays

Initially, 500-1000 cells was seeded in 6-well plates and allowed to proliferate for 24 h. Subsequently, a drug incubation period of 5 days was implemented, after which the culture medium was aspirated, and the cells were subjected to thorough washing before being incubated with a drug-free medium for an additional 7 days. Following this incubation period, they were fixed with 4% paraformaldehyde for 30 min and subsequently stained with 0.1% (w/v) Crystal Violet (Beyotime, C0121) for 20 min. After a comprehensive wash with distilled water, the cells were left to air dry. The quantification of colonies was performed through manual counting using a stereo microscope.

Flow cytometric analyses

For cell cycle analysis utilizing the kFlour488 Click-iT EdU Flow Cytometry Kits (KeyGEN BioTECH, KGA9601-100), cells in a 6-well plate underwent labeling with 50 μM 5-ethynyl-2′-deoxyuridine (EdU) for 2 h. Subsequently, they were rinsed three times with PBS, harvested, and fixed in 70% ethanol for 30 min at room temperature. After two washes with 0.3% bovine serum albumin (BSA) in PBS, the cells were permeabilized with ice-cold 0.3% Triton X-100 in PBS for 20 min. Following the removal of the permeabilization solution, cells were subjected to a 30-min incubation at room temperature in the Click-iT reaction. Then, the cells were washed with PBS containing 0.3% Triton X-100 and underwent incubation in an RNase A reaction for 30 min at room temperature. Subsequently, the cells were incubated in a propidium iodide (PI) solution at room temperature for 15 min. Finally, the cells were assessed using the BD LSRFortessa (BD Biosciences), and the cell cycle analysis was conducted employing FlowJo software (Version 10.8.1, BD Biosciences).

Cleavage under targets & tagmentation assay

Cleavage under targets & tagmentation (CUT&Tag) was performed following previously established protocols using the CUT&Tag assay kit (Vazyme, TD904) [33]. Approximately 1 × 10⁵ PDCs were bound to ConA beads, followed by overnight incubation at 4°C with either a primary antibody or control immunoglobulin G (IgG) on a rotator. The next day, a secondary antibody was added and incubated for 60 min at room temperature, followed by pA/G-Tn5 transposome incubation and tagmentation in an activating buffer supplemented with magnesium. Details of the antibodies used are provided in Supplementary Table S1. Adaptor-ligated DNA fragments were extracted and polymerase chain reaction (PCR)-amplified using indexing primers (TD202, Vazyme). The amplified products were purified with VAHTS DNA Clean Beads (N411-01, Vazyme), and the DNA library’s quantity and quality were assessed using a Qubit^™ 4 Fluorometer and Bioanalyzer 2100 (Invitrogen). Libraries were sequenced on the Illumina platform by Novogene Bioinformatics Technology Co., Ltd. (Beijing, China), generating 150 bp paired-end reads.

Raw sequence reads underwent initial quality control using FastQC (v0.20.0) [34]. Clean reads were obtained by removing adapter sequences, poly-N stretches, and low-quality reads. Quality-filtered reads were then mapped to the reference genome (Ensembl Homo sapiens GRCh38, release 110) using BWA (v0.7.12) with the following parameters: -k 32 -T 30 -t 4 -M. Peak calling was performed using MACS2 (v2.1.0) [35] with the parameters: -q 0.05 -f AUTO –call-summits –nomodel –shift -100 –extsize 200 –keep-dup all. Peaks were adjusted to a uniform size (500 bp) centered on peak summits. Motif discovery within these loci was conducted using HOMER (v4.9.1) with the command: findMotifsGenome.pl -len 8,10,12,14 -gc -size given -homer2 -dumpFasta. Genomic annotation and identification of the nearest genes associated with peaks were performed using ChIPseeker. Functional enrichment analysis, including Gene Ontology (GO) enrichment, was subsequently conducted to determine the biological significance of peak-associated genes.

Comet assay

The comet assay was conducted using the DNA Damage Detection Kit following the manufacturer’s protocol (KeyGEN BioTECH, KGA240-100). Initially, cells cultured in a 6-well plate were exposed to 5 μM CX-5461 (Selleck, S2684) for 24 h. Subsequently, the cells were collected and suspended in 100 μl of ice-cold PBS at a concentration of 1 × 10⁶ cells/ml. The cells were combined with Comet Agarose at a ratio of 1:5 (v/v) and promptly transferred onto a frosted glass slide. Once solidified at 4°C for 30 min, the slides were placed in a pre-chilled lysis buffer for 1-2 h at 4 °C in the dark, then rinsed with PBS. Following a 30-min incubation in alkaline solution (1 mmol/L EDTA and 300 mmol/L NaOH) at room temperature, electrophoresis was performed at 25 V for 30 min. After fixation in 0.4 mM Tris-HCl (pH 7.5) three times, DNA was stained with PI. Subsequently, slides were observed and imaged using a Nikon Eclipse Ti2-E microscope at 10× magnification.

Western blot and band depletion assay

The cells were lysed, and their protein concentrations were determined using the BCA protein quantitation assay kit (KeyGEN BioTECH, KGPBCA). Subsequently, the proteins were denatured by boiling in 6× sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) loading buffer (KeyGEN BioTECH, KGP101W) and then separated using SDS-PAGE before being transferred onto polyvinylidene difluoride (PVDF) membranes. Following blocking with 5% BSA in TBS containing 0.1% Tween-20, the membranes were probed with primary antibodies, followed by horseradish-peroxidase-conjugated secondary antibodies (see Supplementary Table S1). Signal detection was achieved through enhanced chemiluminescence (ECL) utilizing the Tanon 5200CE detection system.

For the band depletion assay, cells in 6-well plates were treated with CX-5461 at a concentration that was approximately equal to their IC50 for a period of 4 h. Treated cells were then lysed in culture plates using 1 × Laemmli sample buffer (Sigma-Aldrich, 38733). Samples were boiled at 95°C for 10 min, and 20 μl of each sample was separated using 8% SDS-PAGE. Cell lysates from untreated cells were employed as a loading standard. For lanes 1-4, proteins equal to 40, 20, 10, and 5 thousand cells were loaded, respectively. For each of the rest of the lanes, protein equal to 40 thousand cells was loaded.

Co-immunoprecipitation assays

Co-immunoprecipitation (Co-IP) assays were conducted according to the manufacturer’s protocol for Protein G Sepharose^™ 4 Fast Flow (Cytiva, 17-0618-01) for immunoprecipitation. Treated cells were harvested and washed with ice-cold PBS. Cellular lysis was carried out in a suitable lysis buffer containing protease and phosphatase inhibitors, followed by incubation on ice for 30 min. The lysate was then subjected to high-speed centrifugation to eliminate cellular debris, and the resulting supernatant was transferred to a new tube. Protein A/G beads were added to the supernatant and incubated on a rotator at 4°C for 2 h. Subsequently, the sample was centrifuged to remove the beads, and the pre-cleared lysate was collected. The primary antibody (see Supplementary Table S1) was introduced to the pre-cleared lysate and incubated on a rotator at 4°C overnight. Protein A/G beads were then added to capture the antibody-protein complexes, followed by incubation on a rotator at 4°C for 2 h. Following this, the samples underwent centrifugation, and the beads were washed multiple times with ice-cold wash buffer to eliminate nonspecifically bound proteins. The eluted proteins were resolved using SDS-PAGE, transferred to a membrane, and subjected to western blotting with appropriate antibodies (see Supplementary Table S1).

Proximity ligation assay

Cells were seeded onto confocal dishes and incubated with a concentration of 5 μM CX-5461 for 24 h. Subsequently, the cells underwent a triple wash with PBS. Following the wash, cell fixation was performed using 4% paraformaldehyde for 10 min at room temperature, followed by methanol treatment for 20 min at −20°C. Permeabilization was achieved by exposing the cells to 0.5% Triton X-100 in PBS for 15 min at 4°C. After an additional triple wash with PBS, the cells were treated with 100 μg/ml RNase A (TANGEN, U8713) in 5 mM EDTA in PBS for 1 h at 37°C, followed by a triple 5-min wash in PBS. Proximity ligation assay (PLA) foci were generated following the manufacturer’s instructions for the Duolink in Situ PLA Kit (Sigma-Aldrich). Next, cells were subsequently blocked in Duolink blocking solution for 1 h at 37°C and subsequently incubated with primary antibodies (see Supplementary Table S1) diluted with Duolink antibody diluent for 1 h at 37°C. Anti-mouse PLUS and anti-rabbit MINUS PLA probes were applied to the primary antibodies for 1 h at 37°C after being washed three times with PBST (PBS with 0.05% Tween-20). Then, PLA probes were ligated for 30 min at 37°C and washed with buffer A (0.01 M Tris, 0.15 M NaCl, and 0.05% Tween-20). Next, amplification using Duolink in Situ Detection Reagents (Sigma-Aldrich) was performed at 37°C for 100 min. Following the amplification, the dishes were washed three times for 5 min each with wash buffer B (0.2 M Tris and 0.1 M NaCl) and once with PBS. Lastly, they were coated with a mounting medium containing 4′,6-diamidino-2-phenylindole (DAPI) (KeyGEN BioTECH, KGF0282). Images were captured using a Nikon Eclipse Ti2-E microscope at 60× magnification.

Rapid approach to DNA adduct recovery assay

A total of 5 × 10⁵ treated cells underwent lysis using DNAzol^® (Thermo Fisher, #10503027) in conjunction with 1% sarkosyl. Nucleic acids were precipitated by introducing 100% ethanol, which is equivalent to 1/2 the volume of the cell lysate. Following a 5-min incubation at −20°C, nucleic acids were collected through centrifugation at maximum speed for 15 min. The resulting pellet underwent two washes with 75% ethanol and was then dissolved in a freshly prepared 8 mM NaOH solution. Subsequently, the DNA concentration was determined using a NanoDrop 1000 spectrophotometer. A quantity of 1 μg DNA from each sample was transferred to a PVDF membrane utilizing the Bio-Dot SF microfiltration apparatus (Bio-Rad, #170-6542). Block the membrane with 5% nonfat milk in TBST (0.1% Tween-20), followed by probing with primary antibodies and secondary antibodies (see Supplementary Table S1). Signal detection was accomplished through ECL using the Tanon 5200CE detection system.

Immunofluorescence

The cells were cultured in confocal dishes until they reached 80% confluence. For immunofluorescence assays coupled with EdU labeling, the Fluor488 Click-iT EdU Imaging Kits (KeyGEN BioTECH, KGA9602-100) were employed, following the manufacturer’s recommended protocol. Initially, cells were incubated with 10 μM EdU for 2 h. Then, cells were fixed with 4% paraformaldehyde for 15 min at room temperature and permeabilized with ice-cold 0.5% Triton X-100 in PBS for another 15 min. After washing with 3% BSA in PBS, cells were incubated in the Click-iT reaction mix at room temperature for 30 min. Following this, the cells underwent a 3% BSA in PBS wash and were blocked with 5% BSA in PBS for 1 h at room temperature. Sequentially, cells were incubated with primary antibodies and secondary antibodies (see Supplementary Table S1) for 1 h each at room temperature within a humidified chamber. Finally, the coverslips were affixed to the confocal dishes with a DAPI-infused anti-fade mounting medium (KeyGEN BioTECH, KGF0282), all while avoiding exposure to light, for 10 min at room temperature. Images were acquired on the Nikon Eclipse Ti2-E microscope.

For quantification of G4 structures, G4 fluorescence intensity per nucleus was quantified using Fiji (ImageJ v1.54). Nuclei (≥100 per condition from three independent experiments) were defined by DAPI, and G4 intensity was measured within the nuclear mask after background subtraction. Background threshold was set at mean background + 2 × standard deviation. For phosphorylated histone H2AX (γH2AX), discrete foci per nucleus were quantified using “Analyze Particles” after thresholding, with spot counts averaged across ≥100 cells per replicate. Results were expressed as the number of foci per nucleus.

DNA fiber analysis

The cells, cultured in a 6-well plate, were treated with 5 μM CX-5461 for 24 h. Exponentially growing cells underwent a 30-min labeling period with 20 μM CldU (Sigma-Aldrich, C6891). After three washes with warm PBS, they were subsequently labeled with 100 μM IdU (Sigma-Aldrich, I7125) for 30 min, followed by three additional warm PBS washes to halt the labeling process. The labeled cells were trypsinized and then suspended in ice-cold PBS at a concentration of 1 × 10⁶ cells/ml. A 2.5 μl aliquot of this suspension was dropped onto a slide, partially air-dried, and treated with 10 μl of lysis buffer (5% SDS, 200 mM Tris-HCl pH 7.4, 50 mM EDTA) at room temperature for 15 min. Tilting the slide at a 15° angle allowed the liquid to slowly flow down. After air-drying, the slides were fixed in a methanol and acetic acid solution (3:1, v/v) for 15 min and then air-dried. DNA was denatured with 2.5 M HCl for 60 min at room temperature. Subsequently, slides were washed three times with PBS and blocked with 5% BSA/1% goat serum in PBS for 1 h at 37°C. Primary antibodies Rat anti-BrdU (1:100, Abcam, ab6323) and Mouse anti-BrdU (1:40, Becton Dickinson, 347580) were applied for 1 h at 37°C. Slides were washed with PBS containing 0.05% Tween-20 and incubated for 45 min at 37°C with Alexa Fluor 546-labeled goat anti-Rat secondary antibody (1:100, Invitrogen, A-11081) and Alexa Fluor 488-labeled goat anti-mouse secondary antibody (1:100, Invitrogen, A-11001). After washing with PBST (0.05% Tween-20 in PBS), the slides were air-dried in the dark. Coverslips (VWR, 48404-455) were mounted onto the slides using an anti-fade mounting medium (SouthernBiotech, 0100-01). Replication tracks were captured using a Nikon Eclipse Ti2-E microscope at 60× magnification and measured utilizing NIS-Elements AR software (5.42.01 64-bit). For studies involving CX-5461 and Mirin, cells were initially pulse-labeled with CldU and IdU, followed by a 3-hour treatment with 5 μM CX-5461, 100 μM Mirin (Sigma-Aldrich, M9948), or a combination of both.

RNA-seq

Total RNA of treated cells was extracted using TRIzol reagent (Invitrogen, CA, USA) according to the manufacturer’s protocol. RNA purity and quantification were evaluated using the NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, USA). RNA integrity was assessed using the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). Then, the libraries were constructed using the VAHTS Universal V6 RNA-seq Library Prep Kit according to the manufacturer’s instructions. The transcriptome sequencing and analysis were conducted by OE Biotech Co., Ltd. (Shanghai, China).

Statistical analysis

Unless otherwise specified, all experiments were conducted with 3-5 biological replicates. Statistical analyses were performed using GraphPad Prism 10. The Shapiro-Wilk test was used to assess data normality, and results are presented as mean ± standard error of the mean (SEM). For comparisons between two groups, an unpaired Student’s t-test was used. For multiple-group comparisons, one-way ANOVA was applied. If the data did not follow a normal distribution, the Mann-Whitney test was used for two-group comparisons, and the Kruskal-Wallis test was used for multiple-group comparisons. Statistical significance was defined as follows: ns = not statistically significant; *P< 0.05; **P< 0.01; ***P< 0.001.

Results

TOP2B is associated with G4 structures in glioma PDCs

To investigate the genome-wide chromatin-binding profile of TOP2B in glioma, we established patient-derived glioma cells (PDCs) from freshly resected glioma tissues (Fig. 1A). Cleavage Under Targets and Tagmentation (CUT&Tag) was then performed using a TOP2B-specific antibody, a DNA G4-specific antibody, and a protein A-Tn5 transposase fusion (pA-Tn5) to tag DNA fragments at binding sites, followed by sequencing to identify enriched regions (Fig. 1B). The CUT&Tag assay showed high reproducibility across three biological replicates, with strong correlations observed among replicates (Supplementary Fig. S1A). A total of 54 494 TOP2B peaks were identified, among which 36.5% overlapped with G4-enriched regions (Fig. 1C; Supplementary Fig. S1B and C), indicating frequent co-localization of TOP2B and G4 structures in glioma PDCs. Moreover, quantitative analysis revealed that TOP2B was significantly more enriched at G4-containing regions compared to non-G4 regions (Fig. 1D–F). Metagene analysis revealed that both TOP2B and G4 peaks were predominantly located near transcription start sites (TSS) (Fig. 1G), with the majority mapping to promoter regions (Fig. 1H), suggesting their involvement in transcriptional regulation. Motif analysis further demonstrated that TOP2B-binding sites were enriched at G4 loci containing GC-rich motifs, consistent with its function in transcriptional regulation and the resolution of DNA topological stress (Fig. 1I).

Figure 1.

TOP2B is associated with G4 structures in glioma PDCs. (A) Workflow for generating patient-derived glioma cells (PDCs) from fresh surgical samples. Tumor tissues were minced, enzymatically dissociated with collagenase IV, filtered, and cultured. (B) Schematic of the CUT&Tag assay used to map G4 regions and TOP2B-bound chromatin regions. Antibody-bound pA-Tn5 transposase inserts sequencing adaptors at target loci in situ, followed by DNA extraction and sequencing. (C) Pie chart showing the percentage of TOP2B-enriched regions that overlap with G4 peaks (19 897/54 494, 36.51%). (D) Dot plots displaying TOP2B CUT&Tag average profiles in PDCs at G4 and non-G4 sites. Error bars, mean ± SEM; ***P < 0.001, Mann-Whitney test. (E) TOP2B coverage at TSS in G4-containing versus non-G4 regions. (F) Genome browser tracks illustrating TOP2B peaks overlapping with G4 peaks in chromatin. TOP2B exhibits higher enrichment at G4 regions compared to non-G4 regions. (G) Metagene analysis of TOP2B and G4 distribution across gene bodies. Heatmap shows read counts spanning ±3 kb around gene bodies. (H) Pie chart showing the genomic annotation of TOP2B and G4-enriched regions in glioma PDCs. (I) Motif discovery analysis of TOP2B binding regions using HOMER (±250 bp from summit), revealing sequence motif characteristics. (J) PLA of TOP2B and G4 in U87MG, U251 cell lines, and glioma PDCs; scale bars: 10 μm. Green, PLA signal (TOP2B + G4); red, TOP2B staining; blue, DAPI nuclear staining. (K) Co-immunofluorescence (Co-IF) analysis of TOP2B and EdU in control and TOP2B knockdown U87MG cells; scale bars: 10 μm. TOP2B intensity per nucleus was quantified in 100 cells per condition from three independent experiments. The percentage of EdU-positive cells was calculated from five independent experiments. Error bars, mean ± SEM. ***P < 0.001, Mann-Whitney test and Student’s t-test. (L) Co-IF analysis of G4 and EdU in control and TOP2B knockdown U87MG cells; scale bars: 10 μm. Quantification of G4 intensity per nucleus was performed in 100 EdU-positive and EdU-negative cells per condition across three independent experiments. Error bars, mean ± SEM. ns, not significant; ***P < 0.001, Kruskal-Wallis test.

Given the frequent co-localization of TOP2B and G4 peaks, we next investigated whether this reflects a direct physical interaction between TOP2B and G4 structures in glioma cells. PLA, which detects close spatial proximity (<40 nm) between two target proteins or protein-DNA structures in situ, was performed using antibodies against TOP2B and G4 in U87MG and U251 glioma cell lines, as well as in glioma PDCs. Clear PLA signals were detected in all models (Fig. 1J), confirming the spatial proximity and potential interaction of TOP2B and G4 structures. Importantly, no PLA foci were observed in single antibody controls (omitting anti-G4 or anti-TOP2B), verifying the specificity of the signal (Supplementary Fig. S1D). To further clarify the functional relationship between TOP2B and G4s, we knocked down (KD) TOP2B using small interfering RNA (siRNAs) and then examined G4 accumulation (Supplementary Fig. S1E and Supplementary Table S2). Immunofluorescence staining combined with EdU incorporation assays demonstrated that TOP2B KD impaired DNA replication (Fig. 1K; Supplementary Fig. S1F and G) and led to increased G4 formation, specifically in S-phase cells (Fig. 1L and Supplementary Fig. S1H). These results suggest that in the absence of TOP2B, unresolved G4 structures accumulate at active replication forks. Collectively, these findings establish that TOP2B not only co-localizes with G4 structures at promoter regions but also plays a functional role in regulating G4 dynamics during DNA replication, likely contributing to the maintenance of replication fork progression and genome stability.

CX-5461 disrupts TOP2B chromatin binding and enhances G4 accumulation

CX-5461 has been reported to act as a TOP2B-targeting agent with selective activity in high-risk neuroblastoma models [30]. To confirm its role in gliomas, we performed the rapid approach to DNA adduct recovery (RADAR) assay to detect DNA-protein covalent complexes (DPCCs) (Fig. 2A). Rather than promoting physiological chromatin binding, CX-5461 acts similarly to other TOP2 poisons by trapping TOP2B on DNA and inducing covalent TOP2B-DNA complexes (TOP2Bcc). TOP2Bcc levels peaked at 1 hour following CX-5461 treatment before subsequently declining. Additionally, TOP2Bcc formation increased in a dose-dependent manner, reaching a maximum at 1 μM CX-5461 before slightly decreasing at higher concentrations (Fig. 2B and Supplementary Fig. S2A). To further assess TOP2Bcc stability following CX-5461 treatment, we conducted the band depletion assay, which showed a significant reduction in free TOP2B protein, indicating that TOP2B was trapped on DNA in a dose-dependent manner, mirroring classical TOP2 poisons like etoposide (Fig. 2C and D; Supplementary Fig. S2B). Notably, CX-5461 did not affect TOP3A levels (Fig. 2C), supporting the selective activity toward TOP2B.

Figure 2.

CX-5461 disrupts TOP2B chromatin binding and enhances G4 accumulation. (A) Schematic diagram of the rapid approach to DNA adduct recovery (RADAR) assay. Left: Nuclei contain DNA and proteins, some of which are covalently bound (e.g., TOP2B in DPCCs). Middle: DPCCs are isolated along with free DNA. Right: Specific DPCCs (TOP2Bcc) are detected using an antibody. Colored circles represent nuclear proteins. (B) Detection of TOP2B-DPCC (TOP2Bcc) by RADAR assay. Slot blot comparing TOP2B signal in whole-cell extract (WCE) or DPCCs isolated from U87MG cells treated with 5 μM CX-5461 for different durations (0, 0.5, 1, 2, and 3 h; left) or with increasing concentrations (0, 0.5, 1, 5, and 10 μM) of CX-5461 for an hour (right). (C) Band-depletion assay for CX-5461-treated U87MG and U251 cells. Lanes 1-4 contain proteins equivalent to 40, 20, 10, and 5 thousand cells, respectively. For all other lanes, protein equivalent to 40 thousand cells was loaded. (D) Band-depletion assay for etoposide-treated U87MG cells as a positive control. Lanes 1-4 contain proteins equivalent to 40, 20, 10, and 5 thousand cells, respectively. For all other lanes, protein equivalent to 40 thousand cells was loaded. (E) Venn diagram showing the overlap of TOP2B CUT&Tag peaks between control and CX-5461-treated glioma PDCs. Metagene analysis of TOP2B peak distribution across summits post-CX-5461 treatment. Heatmap displays read counts spanning ±3 kb around summits. (F) GO enrichment analysis of genes associated with TOP2B CUT&Tag differential peaks between control and CX-5461-treated PDCs. (Gand H) Co-IF analysis of TOP2B and G4 in U87MG cells treated with vehicle or 5 μM CX-5461 for 24 h; scale bars: 10 μm. Quantification of TOP2B and G4 intensity per nucleus in 50 cells per treatment condition across five independent experiments. Error bars, mean ± SEM. ns, not significant; ***P < 0.001, Student’s t-test. (I) Dot plots of TOP2B and G4 coverages after CX-5461 treatment, obtained from TOP2B CUT&Tag and G4 CUT&Tag in PDCs. ***P < 0.001, *P < 0.05, Mann-Whitney test. (J) Bar graph depicting the number of overlapping TOP2B and G4 CUT&Tag peaks after CX-5461 treatment in PDCs. (K) Dot plots showing TOP2B CUT&Tag peaks at G4 and non-G4 regions in control and CX-5461-treated PDCs. Error bars, mean ± SEM; ***P < 0.001, Kruskal-Wallis test. (L) Genomic tracks illustrating changes in TOP2B occupancy at G4 sites near the TSS of the Clta gene following CX-5461 treatment in PDCs.

To investigate the impact of CX-5461 on TOP2B chromatin recruitment and its relationship with G4 structures, we performed CUT&Tag assays for TOP2B and G4 in glioma PDCs. CX-5461 treatment resulted in an increased number of total TOP2B peaks but with reduced peak intensity at summits and a markedly altered genomic distribution, suggesting impaired recruitment and redistribution of TOP2B on chromatin (Fig. 2E; Supplementary Fig. S2C and D). Gene ontology (GO) enrichment analysis of genes associated with differentially enriched TOP2B peaks revealed that CX-5461 affects several DNA-related processes, including four-way junction helicase activity, DNA helicase activity, DNA recombination, DNA-binding transcription factor activity, and DNA-binding pathways (Fig. 2F). Notably, CX-5461 did not alter TOP2B protein expression but significantly increased G4 signal intensity (Fig. 2G and H; Supplementary Fig. S2E), suggesting that CX-5461 induces G4 accumulation by functionally inhibiting TOP2B rather than downregulating its expression. These findings were further supported by metagene analysis, which showed increased chromatin signals for both TOP2B and G4 structures following CX-5461 exposure (Fig. 2I), consistent with previous findings that CX-5461 stabilizes G4 structures by trapping TOP2Bcc [31].

To explore how CX-5461 affects the interaction between TOP2B and G4 structures, we analyzed the overlap between their binding peaks. CX-5461 treatment markedly reduced the number of TOP2B and G4 overlapping peaks (Fig. 2J), with a more substantial reduction at G4-enriched regions compared to non-G4 regions (Fig. 2K and Supplementary Fig. S2F). Moreover, CX-5461 altered the genomic distribution of TOP2B-G4 co-occupancy (Fig. 2L), reducing TOP2B binding on gene bodies, particularly at TSS, while G4 occupancy increased relative to TOP2B (Supplementary Fig. S2G). A shift in TOP2B occupancy from promoter regions toward intronic and intergenic regions was also observed (Supplementary Fig. S2H). Together, these results demonstrate that CX-5461 functions as a potent TOP2B poison in gliomas, trapping TOP2Bcc on chromatin, disrupting TOP2B-G4 interactions, and stabilizing G4 structures. This cascade ultimately impairs G4 resolution, perturbs genome organization, and contributes to replication stress and genomic instability.

ATRX and TOP2B co-localize with G4 structures and regulate genome stability

Given the established roles of TOP2B in chromatin regulation and ATRX in G4 resolution, we next investigated whether these proteins functionally cooperate to maintain G4 homeostasis. GeneMANIA analysis predicted a potential association between ATRX and TOP2B (Supplementary Fig. S3A). CUT&Tag analysis identified 17 534 overlapping peaks between ATRX and TOP2B, with both proteins showing preferential enrichment around TSS regions, suggesting their coordinated involvement in transcriptional regulation (Fig. 3A and B). PLA revealed distinct ATRX-TOP2B proximity signals in glioma cells (Fig. 3C), and no PLA foci were observed in single antibody controls, omitting either anti-TOP2B or anti-ATRX (Supplementary Fig. S3B). These results indicate a close spatial association and suggest a potential functional interplay between the two proteins. This interaction was further supported by Co-IP, which confirmed a specific association between ATRX and TOP2B, but not with TOP2A (Fig. 3D). Although TOP1 has also been reported to interact with G4 structures [36], its depletion modestly increased G4 levels (Supplementary Fig. S3C and D, and Supplementary Table S2). Moreover, PLA analysis in glioma PDCs demonstrated that ATRX preferentially interacts with TOP2B rather than TOP1 (Fig. 3E and Supplementary Fig. S3B). These findings suggest that ATRX and TOP2B functionally cooperate to regulate G4 dynamics, potentially by promoting G4 resolution during transcription.

Figure 3.

ATRX and TOP2B co-localize with G4 structures and regulate genome stability. (A) Venn diagram showing the overlap of ATRX and TOP2B CUT&Tag peaks in glioma PDCs. Metagene analysis of ATRX and TOP2B peak distribution across gene bodies. (B) Genome browser tracks illustrating ATRX peaks overlapping with TOP2B peaks in chromatin. (C) PLA of ATRX and TOP2B in U87MG and U251 glioma cells; scale bars: 10 μm. Red, PLA signal (TOP2B + ATRX); blue, DAPI nuclear staining. (D) Co-IP from U87MG and U251 extracts showing ATRX interaction with TOP2B but not TOP2A. Anti-ATRX was used to pull down TOP2B and TOP2A. (E) PLA of ATRX and TOP2B in glioma PDCs (top); scale bars: 10 μm. Green, PLA signal (TOP2B + ATRX); red, ATRX staining; blue, DAPI nuclear staining. The merged image demonstrates co-localization of PLA signals with ATRX in the nucleus. Insets highlight magnified regions of interest. PLA of ATRX and TOP1 in glioma PDCs (below, as negative control); scale bars: 10 μm. Green, PLA signal (TOP1 + ATRX); red, ATRX staining; blue, DAPI nuclear staining. (F) Venn diagram showing the overlap of ATRX, TOP2B, and G4 CUT&Tag peaks in glioma PDCs. (G) Metagene analysis of ATRX, TOP2, and G4 distribution across TSS in glioma PDCs. Heatmap displays read counts spanning ±3 kb around TSS. (H) Genome browser tracks illustrating ATRX peaks and TOP2B peaks overlapping with G4 peaks across the TSS of the gene STAT2. (I) PLA of ATRX and G4 in control and TOP2B knockdown U87MG cells; scale bars: 10 μm. Green, PLA signal (ATRX + G4); blue, DAPI nuclear staining. Quantification of PLA (ATRX + G4) foci per nucleus was performed in 100 cells per condition across three independent experiments. Error bars, mean ± SEM; ***P < 0.001, Mann-Whitney test. (J) PLA of TOP2B and G4 in control and ATRX knockdown U87MG cells; scale bars: 10 μm. Green, PLA signal (TOP2B + G4); blue, DAPI nuclear staining. Quantification of PLA (TOP2B + G4) foci per nucleus was performed in 100 cells per condition across three independent experiments. Error bars, mean ± SEM; ***P < 0.001, Mann-Whitney test.

To further investigate the genomic context of their cooperation at G4 structures, we performed CUT&Tag for ATRX, TOP2B, and G4 simultaneously. A total of 11 913 peaks were co-occupied by all three components, with peak enrichment primarily located from the TSS up to 3 kb upstream (Fig. 3F–H), suggesting that ATRX and TOP2B jointly localize to G4-prone regulatory regions during transcription and replication. To assess functional relevance, we evaluated G4 accumulation upon knockdown of each component. Depletion of TOP2B led to increased G4 accumulation and enhanced recruitment of ATRX to G4-enriched loci (Fig. 3I; Supplementary Fig. S3E–G). Conversely, ATRX knockdown also elevated G4 accumulation-triggered compensatory binding of TOP2B at G4 regions (Fig. 3J; Supplementary Fig. S3E and F). These reciprocal alterations indicate a compensatory mechanism, wherein ATRX and TOP2B dynamically respond to each other’s loss to preserve G4 homeostasis. Importantly, depletion of either protein induced DNA damage, and co-depletion further aggravated the damage phenotype (Supplementary Fig. S3H). GO enrichment analysis of differentially expressed genes in ATRX- and TOP2B-deficient cells showed enrichment for pathways associated with DNA replication, DSB repair via break-induced replication, and general DNA repair mechanisms (Supplementary Fig. S3I). Together, these results demonstrate that ATRX and TOP2B collaboratively safeguard replication fork stability and genome integrity through G4 regulation. Their compensatory recruitment upon loss of the other underscores a functional redundancy or backup mechanism critical for maintaining G4 homeostasis and genomic integrity.

CX-5461 stabilizes G4 by modulating ATRX and TOP2B chromatin binding

Building upon the observed cooperative role of the ATRX and TOP2B in G4 regulation, we next investigated whether CX-5461 could affect this interaction. While CX-5461 treatment did not alter the expression levels of ATRX or TOP2B, it significantly disrupted their physical association, as shown by reduced Co-IP signals (Fig. 4A and C; Supplementary Fig. S4A and B). Consistent with this, CX-5461 treatment reduced the number of ATRX-TOP2B PLA foci in both glioma cell lines and PDCs (Fig. 4B and D), indicating impaired spatial proximity between the two proteins in situ. These findings were further supported by reduced co-occupancy of ATRX and TOP2B on chromatin following CX-5461 exposure (Fig. 4E). Paradoxically, despite disruption of the ATRX-TOP2B interaction, CX-5461 enhanced the chromatin binding of both proteins (Figs 2I and 4F), suggesting that the drug alters their chromatin recruitment dynamics rather than suppressing their expression.

Figure 4.

CX-5461 stabilizes G4 structures by disrupting ATRX-TOP2B chromatin interactions. (A) Co-IP from U87MG cells showing that ATRX interacts with TOP2B, and this interaction is reduced by CX-5461. Reciprocal IPs were performed using anti-ATRX and anti-TOP2B antibodies. (B) PLA of ATRX and TOP2B in control and CX-5461-treated U87MG and U251 glioma cells. Green, PLA signal (TOP2B + ATRX); blue, DAPI nuclear staining; scale bars: 10 μm. PLA (ATRX + TOP2B) foci per nucleus were performed in 100 cells per condition across three independent experiments. Error bars, mean ± SEM; ***P < 0.001, Mann-Whitney test. (C) Co-IP from glioma PDCs showing that ATRX interacts with TOP2B, and this interaction is reduced by CX-5461 treatment. Immunoprecipitation was performed using the anti-ATRX antibody. (D) PLA of ATRX and TOP2B in control and CX-5461-treated glioma PDCs. Green, PLA signal (TOP2B + ATRX); red, ATRX staining; blue, DAPI nuclear staining; scale bars: 10 μm. PLA (ATRX + TOP2B) foci per nucleus were performed in 100 cells per condition across three independent experiments. Error bars, mean ± SEM; ***P < 0.001, Mann-Whitney test. (E) Pie chart showing the percentage of ATRX and TOP2B CUT&Tag overlapped peaks in control and CX-5461-treated glioma PDCs. (F) Dot plots displaying ATRX CUT&Tag average profiles in control and CX-5461-treated PDCs. Error bars, mean ± SEM; ***P < 0.001, Mann-Whitney test. (G) PLA of ATRX and G4 in control, TOP2B KD, and CX-5461-treated U87MG cells; scale bars: 10 μm. Green, PLA signal (ATRX + G4); blue, DAPI nuclear staining. Quantification of PLA (ATRX + G4) foci per nucleus was performed in 100 cells per condition across three independent experiments. Error bars, mean ± SEM; ***P < 0.001, Kruskal-Wallis test. (H) Dot plots displaying ATRX CUT&Tag average profiles in control and CX-5461-treated PDCs at G4 and non-G4 sites. Error bars, mean ± SEM; ***P < 0.001, Kruskal-Wallis test. (I) PLA of TOP2B and G4 in control, ATRX KD, CX-5461-treated, and the combination in U87MG cells; scale bars: 10 μm. Green, PLA signal (TOP2B + G4); blue, DAPI nuclear staining. Quantification of PLA (TOP2B + G4) foci per nucleus was performed in 100 cells per condition across three independent experiments. Error bars, mean ± SEM; ***P < 0.001, *P < 0.05, Kruskal-Wallis test. (J) Pie chart showing the percentage of ATRX and TOP2B co-enriched regions overlapping with G4 peaks in control (11 913/17 515, 68.02%) and CX-5461-treated PDCs (6556/23 126, 28.35%). Bar graph depicting the number of overlapping ATRX, TOP2B, and G4 CUT&Tag peaks after CX-5461 treatment in PDCs. (K) Metagene analysis of ATRX, TOP2B, and G4 distribution across gene bodies in control and CX-5461-treated PDCs. (L) Genome browser tracks illustrating ATRX peaks and TOP2B peaks overlapping with G4 peaks, whose locations were altered by CX-5461 treatment. (M) GO enrichment analysis of genes related to G4 CUT&Tag differential peaks between control and CX-5461-treated PDCs.

Given this, we further investigated whether CX-5461 influences ATRX and TOP2B localization at G4 regions. Similar to the phenotype observed with TOP2B knockdown, CX-5461 increased ATRX recruitment to G4 sites, as evidenced by elevated ATRX-G4 PLA foci (Fig. 4G and Supplementary Fig. S4C) and CUT&Tag signals across G4 regions (Fig. 4F and H; Supplementary Fig. S4D). In contrast, TOP2B occupancy at G4 regions was markedly reduced. While ATRX depletion alone increased the demand for TOP2B at G4 sites, this resulted in an enhanced TOP2B-G4 PLA signal. Conversely, treatment with CX-5461 alone led to TOP2B poisoning and depletion of the available TOP2B pool, thereby reducing the TOP2B-G4 PLA signal. When cells were subjected to both ATRX knockdown and CX-5461 treatment, the elevated demand for TOP2B coincided with a markedly reduced pool of functional TOP2B, creating a severe imbalance that resulted in the marked loss of TOP2B-G4 PLA signals and causing excessive G4 accumulation (Fig. 4I and Supplementary Fig. S4E). Consistently, CUT&Tag analysis revealed reduced ATRX-TOP2B co-binding at G4-enriched genomic regions (Fig. 4J and Supplementary Fig. S4F) and widespread alterations in chromatin binding patterns, including a pronounced reduction at TSS and promoter regions (Fig. 4K and L; Supplementary Fig. S4G). GO enrichment analysis of genes associated with these disrupted peaks identified significant perturbations in pathways related to DNA repair, cellular response to DNA damage, DNA polymerase activity, and transcriptional regulation (Fig. 4M).

To determine whether this effect is unique to CX-5461, we compared it with the classical TOP2 inhibitor etoposide, which is widely used in cancer treatment [37]. Similar to CX-5461, etoposide induced G4 formation, especially in ATRX-deficient cells (Supplementary Fig. S4H). However, unlike CX-5461, etoposide stabilized G4 structures without disrupting ATRX-TOP2B interactions, indicating distinct mechanisms of G4 modulation (Supplementary Fig. S4I). Instead, etoposide increased ATRX-G4 PLA foci while significantly reducing TOP2B-G4 PLA formation (Supplementary Fig. S4I), suggesting a compensatory mechanism wherein ATRX becomes the primary G4 resolver upon TOP2B inhibition.

Together, these results demonstrate that CX-5461 uniquely stabilizes G4 structures by trapping TOP2Bcc and interfering with the coordination between ATRX and TOP2B at G4 loci. This dual action impairs G4 resolution and promotes excessive G4 accumulation, especially in ATRX-deficient glioma cells where TOP2B serves as the major G4-resolving enzyme. The combined loss of ATRX and functional TOP2B activity results in unresolved G4 structures, leading to severe replication stress and genomic instability, highlighting a therapeutically actionable synthetic lethality mechanism.

CX-5461 induces G4-dependent DNA damage in ATRX-deficient glioma cells

Given that CX-5461 disrupts the association between ATRX and TOP2B and leads to aberrant G4 accumulation, we next explored whether the resulting G4 stabilization triggers replication-associated DNA damage, particularly in ATRX-deficient glioma cells. Previous research has demonstrated that chemical stabilization of G4 induces replication stress at G4-prone genomic regions, resulting in DNA damage and apoptosis [38, 39]. To evaluate this, we first performed comet assays and observed that CX-5461 treatment led to significantly increased DNA damage in ATRX-deficient glioma cells compared to wild-type (WT) cells (Fig. 5A; Supplementary Fig. S5A and B). Consistently, γH2AX immunofluorescence analysis at 0, 2, 8, and 24 h post-treatment revealed elevated and sustained DNA damage in ATRX-deficient cells, especially at later time points (8 and 24 h) (Fig. 5B and C; Supplementary Fig. S5C). To determine whether the observed DNA damage was associated with G4 accumulation, we performed G4 and γH2AX co-staining. CX-5461 markedly increased both G4 and γH2AX signals, with prominent co-localization in ATRX-deficient glioma cells (Fig. 5D), and significantly elevated numbers of G4-γH2AX double-positive foci (Fig. 5E and Supplementary Fig. S5D), supporting the notion that G4 stabilization contributes directly to DNA damage in this context.

Figure 5.

CX-5461 causes G4-associated DNA damage in ATRX-deficient glioma cells. (A) Comet assay analysis of DNA damage in control and ATRX knockdown (shATRX) U87MG cells treated with vehicle or 5 μM CX-5461 for 24 h; scale bars: 20 μm. Tail moment quantification was performed on 100 nuclei per condition using the Comet Assay Software Package (CASP). Error bars, mean ± SEM; ***P < 0.001, Kruskal-Wallis test. (Band C) IF analysis of γH2AX in control and shATRX U87MG cells at 0, 2, 8, and 24 h after 5 μM CX-5461 treatment; scale bars: 10 μm. Quantification of γH2AX foci per nucleus was performed in 100 cells per condition from three independent experiments. Error bars, mean ± SEM. ns, not significant; ***P < 0.001, two-way ANOVA. (D and E) Co-IF analysis of γH2AX and G4 in control and shATRX cells treated with vehicle or 5 μM CX-5461 for 24 h; scale bars: 10 μm. The percentage of G4 foci co-localizing with γH2AX foci was quantified in 100 nuclei per condition from three independent experiments. Error bars, mean ± SEM; *** P < 0.001, ** P < 0.01, one-way ANOVA. (F) KEGG enrichment analysis of downregulated pathways in U87MG cells following CX-5461 treatment. (G) Western blot analysis of p-ATR, ATR, p-ATM, ATM, p-CHK1, CHK1, p-CHK2, CHK2, p-RPA32, and RPA32 in control and shATRX U87MG and U251 cells treated with vehicle or 5 μM CX-5461 for 24 h.

Transcriptome profiling by RNA-seq of U87MG cells treated with 5 μM CX-5461 for 24 h further revealed a marked downregulation of genes involved in DNA replication and repair pathways (Fig. 5F and Supplementary Fig. S5E). In parallel, Western blot analysis showed activation of the ATM/ATR signaling pathway, as evidenced by increased phosphorylation of ATR (T1989), ATM (S1981), CHK1 (S345), and CHK2 (T68), along with low-level induction of RPA32 (S4/S8) phosphorylation in U87MG and U251 cells (Fig. 5G), consistent with previous findings [40]. Collectively, these results confirm that CX-5461 induces replication-associated DNA damage by stabilizing G4 structures, an effect that is markedly amplified in ATRX-deficient glioma cells. Given our earlier evidence that ATRX functionally cooperates with TOP2B at G4 loci, the absence of ATRX renders glioma cells highly vulnerable to the loss of TOP2B-mediated resolution. This highlights a synergistic role of ATRX and TOP2B in maintaining G4 homeostasis, and reveals that pharmacological disruption of this axis selectively compromises genome integrity in ATRX-deficient contexts.

CX-5461 impairs replication fork stability and induces replication stress

To explore whether the observed DNA damage is linked to replication stress, we assessed cell cycle profiles via flow cytometry. CX-5461 treatment resulted in decreased EdU incorporation (indicative of reduced active DNA replication) and a modest accumulation of G2/M-phase cells in ATRX-deficient U87MG cells (Fig. 6A). In contrast, ATRX-deficient U251 cells showed an even more pronounced S-phase reduction without G2/M arrest (Supplementary Fig. S6A), suggesting replication stress without successful checkpoint resolution. To determine whether DNA damage occurs during S-phase, we co-stained γH2AX and EdU after 24 h of CX-5461 exposure. γH2AX foci were markedly enriched in EdU-positive cells, with stronger effects in ATRX-deficient glioma cells, indicating that CX-5461-induced DNA damage is replication-dependent (Fig. 6B and Supplementary Fig. S6B).

Figure 6.

CX-5461 disrupts replication fork progression through TOP2Bcc accumulation. (A) Flow cytometry analysis of cell cycle in control and shATRX U87MG cells treated with vehicle or 5 μM CX-5461 for 24 h. (B) Co-IF analysis of γH2AX in EdU-labeled control and shATRX cells treated with vehicle or 5 μM CX-5461 for 24 h; scale bars: 10 μm. Quantification of γH2AX foci per nucleus was performed in EdU-positive and EdU-negative U87MG and U251 cells (n = 100 per condition) from three independent experiments. Error bars, mean ± SEM. ns, not significant; ***P < 0.001, Mann-Whitney test. (C) DNA fiber assay: schematic of CIdU and IdU pulse-labeling (top). Control and shATRX U87MG cells were sequentially labeled and treated with vehicle or 5 μM CX-5461 for 5 h. DNA fibers were analyzed to measure replication fork length. IdU track lengths (in μm) were converted to kilobases (1 kb = 2.59 μm). n = 100 replication tracks analyzed from three independent experiments. (D) DNA fiber analysis of U87MG cells labeled with CldU and IdU, then treated with 5 μM CX-5461, 50 mM mirin, or both for 3 h as indicated in the schematic (top). Fibers were processed and analyzed as described above. IdU and CldU track lengths were used to calculate the IdU/CldU ratio. n = 100 replication tracks analyzed from three independent experiments. (E) DNA fiber analysis of control and shATRX U87MG cells pre-treated with 5 μM CX-5461 for 24 h, washed, and sequentially labeled with CldU and IdU as indicated in the schematic (top). n = 100 replication tracks analyzed from three independent experiments. Statistical analyses (C-E) were performed using one-way ANOVA and the Kruskal-Wallis multiple comparisons test. ns, not significant; ***P < 0.001. (F) Schematic of CX-5461-induced TOP2Bcc obstructing the progression of replication forks.

We next performed DNA fiber assays to assess replication fork dynamics. Nascent replication tracks were labeled with CldU and IdU, followed by CX-5461 treatment for 5 h. ATRX-deficient glioma cells exhibited a general reduction in replication track length, indicating a decrease in replication fork speed (Fig. 6C and Supplementary Fig. S6C). Furthermore, CX-5461 treatment significantly shortened IdU track length in both ATRX-deficient and WT glioma cells, suggesting that CX-5461 promotes replication fork degradation (Fig. 6C and Supplementary Fig. S6C). To determine whether Meiotic Recombination 11 homolog A (MRE11)-mediated nuclease activity contributes to CX-5461-induced fork instability, we treated cells with Mirin, an MRE11 inhibitor [41]. Mirin treatment rescued replication fork degradation, confirming that MRE11 is involved in CX-5461-induced replication fork instability (Fig. 6D and Supplementary Fig. S6D). Next, to assess whether CX-5461 disrupts replication fork progression, we pretreated cells with CX-5461 for 24 h before pulse-labeling with CldU and IdU. Interestingly, CX-5461 pre-treatment did not alter fork length, suggesting that CX-5461 does not induce lesions that impede fork restart or progression (Fig. 6E and Supplementary Fig. S6E). Together, these findings demonstrate that CX-5461 induces replication stress by promoting MRE11-dependent fork degradation. This effect is amplified in ATRX-deficient glioma cells, which rely on TOP2B for G4 resolution. In the absence of ATRX, CX-5461-induced TOP2B trapping leads to persistent G4 accumulation, driving fork destabilization and replication-associated DNA damage (Fig. 6F). These findings highlight the cooperative function of ATRX and TOP2B in safeguarding replication fork stability under G4-induced stress.

CX-5461 inhibits glioma proliferation by impairing DNA damage response

Analysis of TCGA datasets revealed that TOP2B expression was significantly higher in ATRX-mutant gliomas compared to ATRX WT counterparts (Fig. 7A), suggesting a potential regulatory relationship between ATRX and TOP2B. Consistent with known molecular patterns, ATRX mutations were strongly enriched in IDH-mutant gliomas (96% versus 55% in ATRX-WT) and were almost exclusively associated with 1p/19q non-codeletion (98% versus 61%). Additionally, ATRX-mutant tumors exhibited a higher frequency of MGMT promoter methylation (84% versus 71%) and a marked reduction in TERT expression (4% versus 72% in ATRX-WT) (Supplementary Fig. S7A). These findings reinforce the concept that ATRX mutation defines a distinct molecular subtype of glioma, typically co-occurring with IDH mutation and characterized by features associated with a favorable prognosis. Furthermore, TOP2B expression was significantly lower in high-grade gliomas compared to low-grade gliomas, indicating a potential association between TOP2B expression and tumor aggressiveness (Supplementary Fig. S7B).

Figure 7.

CX-5461 inhibits glioma proliferation by impairing DDR. (A) Comparison of TOP2B expression between ATRX WT and mutant glioma samples in the TCGA datasets; ***P < 0.001, Student’s t-test. (B) Representative images of GSC spheroids at 0, 24, 48, and 72 h after treatment with 5 or 10 μM CX-5461; scale bar: 100 μm. Quantification of 3D cell viability was performed 72 h post CX-5461 treatment in three independent experiments. Error bars, mean ± SEM; **P < 0.01, ***P < 0.001, one-way ANOVA. (C) Colony formation assay in control and siATRX U87MG cells treated with vehicle or 5 μM CX-5461. Colony numbers were quantified from five independent experiments. Error bars, mean ± SEM. **P < 0.001, one-way ANOVA. (D) Transwell migration assay in control and siATRX U87MG cells treated with vehicle or 5 μM CX-5461. Migration was quantified from five independent experiments. Error bars, mean ± SEM; ***P < 0.001, *P < 0.05, one-way ANOVA. (E) IF analysis of EdU in control and shATRX U87MG cells treated with 5 μM CX-5461 for 0, 2, 8, and 24 h; scale bars: 100 μm. The percentage of EdU-positive cells was quantified from three independent experiments. Error bars, mean ± SEM; ***P < 0.001, **P < 0.01, two-way ANOVA. (F) Western blot analysis of γH2AX, BRCA1, RAD51, 53BP1, and RIF1 in control and shATRX U87MG and U251 cells treated with vehicle or 5 μM CX-5461 for 24 h. (G) IF analysis of BRCA1, RAD51, 53BP1, and RIF1 in control and shATRX U87MG cells treated with vehicle or 5 μM CX-5461 for 24 h; scale bars: 10 μm. Quantification of foci per nucleus was performed in 100 cells per condition from three independent experiments. The percentage of cells with ≥5 BRCA1/RAD51 or RIF1/53BP1 co-localized foci was determined. Error bars, mean ± SEM; ***P < 0.001, **P < 0.01, *P < 0.05, Kruskal-Wallis test and one-way ANOVA.

To evaluate the antiproliferative effects of CX-5461 on glioma, glioma stem cells (GSCs) were treated with 5 and 10 μM of CX-5461 for 24, 48, and 72 h. Cell viability assays showed that CX-5461 inhibited GSC proliferation in a dose- and time-dependent manner (Fig. 7B). Importantly, ATRX-deficient glioma cells displayed significantly greater sensitivity to CX-5461 compared to ATRX WT cells (Fig. 7C and D; Supplementary Fig. S7C and D). A time-dependent reduction in S-phase cells was noted following CX-5461 treatment, with ATRX-deficient cells exhibiting greater sensitivity (Fig. 7E and Supplementary Fig. S7E). To investigate the relationship between CX-5461 treatment and DNA damage response (DDR), we analyzed markers of DNA damage and DDR signaling. A marked accumulation of γH2AX was observed, especially in ATRX-deficient glioma cells, indicating heightened DNA damage (Fig. 7F and Supplementary Fig. S7F). Concurrently, expression levels of key DDR proteins, including RIF1, 53BP1, BRCA1, and RAD51, were markedly decreased (Fig. 7F), suggesting impaired activation of both HR and NHEJ pathways. Immunofluorescence analysis further confirmed a pronounced increase in γH2AX foci in ATRX-deficient glioma cells following CX-5461 treatment (Supplementary Fig. S7G and H). Interestingly, the accumulation of G4 structures mirrored the increase in γH2AX foci, implying a direct link between G4 stabilization and DNA damage induction (Supplementary Fig. S7I). CX-5461 also significantly reduced the number of RIF1, 53BP1, BRCA1, and RAD51 foci, with a corresponding decrease in BRCA1-RAD51 and 53BP1-RIF1 co-localization (Fig. 7G and Supplementary Fig. S7J), further confirming impaired DDR signaling.

To determine whether the effect is specific to CX-5461, we compared it to etoposide. Similar to CX-5461, etoposide induced G4 accumulation and γH2AX foci formation, particularly in ATRX-deficient glioma cells (Supplementary Fig. S7K). Co-localization of G4 and γH2AX exhibited a significant increase following etoposide treatment, further supporting that etoposide-induced DNA damage is associated with G4 structures (Supplementary Fig. S7L). Furthermore, a reduction in actively replicating cells was observed post-etoposide exposure, consistent with the effects of CX-5461 (Supplementary Fig. S7M). Collectively, these findings demonstrate that ATRX-deficient glioma cells exhibit heightened sensitivity to TOP2 inhibitors due to defective DDR pathways that fail to adequately resolve G4-induced DNA damage. CX-5461 inhibits glioma proliferation primarily by stabilizing G4 structures and impairing DDR activation, highlighting the critical role of ATRX in maintaining genome integrity and suggesting a therapeutic vulnerability of ATRX-mutant gliomas to TOP2-targeted agents.

Discussion

In this study, we reveal a previously unrecognized role for TOP2B in resolving G4 DNA structures and maintaining genome integrity, particularly in ATRX-deficient glioma. We demonstrate that TOP2B cooperates with ATRX to facilitate G4 resolution during DNA replication. Disruption of this coordination by CX-5461—previously characterized as an RNA Pol I inhibitor—leads to the accumulation of TOP2Bcc, altered chromatin binding of TOP2B, and impaired G4 resolution. This triggers replication stress and DSBs. Notably, in ATRX-deficient cells, where G4 resolution is already compromised, CX-5461 exacerbates G4 accumulation and DNA damage via disrupting TOP2B, ultimately inducing synthetic lethality. These findings position TOP2B as a key mediator of G4 homeostasis and suggest that G4-associated replication stress represents a therapeutic vulnerability in ATRX-deficient gliomas (Fig. 8).

Figure 8.

CX-5461 traps TOP2B and induces synthetic lethality in ATRX-deficient glioma. Schematic model showing how CX-5461 functions as a TOP2B poison to impair G4 resolution. In WT glioma cells, the ATRX and TOP2B cooperate to resolve G4 structures to ensure replication fork progression and genome stability. In ATRX-deficient cells, TOP2B becomes the primary resolver of G4s during replication. CX-5461 traps TOP2B cleavage complexes (TOP2Bcc), disrupts the ATRX-TOP2B interaction, and blocks G4 resolution, leading to G4 accumulation. This accumulation induces replication fork collapse, single-stranded DNA gaps, and DSBs, activating ATM/ATR signaling and downstream DDR pathways, including CHK1/CHK2-mediated cell cycle arrest. However, ATRX deficiency impairs key DDR components, particularly BRCA1-RAD51-mediated HR and RIF1-53BP1-mediated NHEJ, resulting in defective fork protection and repair. These vulnerabilities culminate in synthetic lethality upon CX-5461 treatment in ATRX-deficient glioma cells.

TOP2B plays crucial roles in chromosome condensation, chromatid separation, and relieving torsional stress during transcription and replication [42]. Although TOP2 enzymes have not been traditionally classified as DNA repair proteins, their association with chromatin remodeling complexes suggests a potential role in genome maintenance [43, 44]. In particular, TOP2B has been shown to interact directly with DSB sites [7, 45]. Consistent with this, our data demonstrate that TOP2B binds to G4 structures and regulates their homeostasis (Fig. 1). TOP2B KD increased G4 formation in glioma cells, which was accompanied by increased ATRX recruitment to G4 sites, and vice versa (Fig. 3I and J), suggesting functional crosstalk between the two proteins. Furthermore, we show that TOP2B and ATRX co-localize at G4 regions and cooperate in maintaining G4 stability (Fig. 3), reinforcing their joint role in mitigating replication-associated DNA damage. While previous studies have highlighted the role of ATRX in resolving G4 structures to maintain genomic stability, particularly in gliomas [46–49], our findings expand this knowledge by identifying TOP2B as a key partner of ATRX in this process.

Strikingly, analysis of glioma datasets revealed that TOP2B expression is significantly elevated in ATRX-mutant tumors. This unexpected increase likely reflects a stress-adaptive transcriptional response to unresolved G4 accumulation and persistent replication stress. Such a compensatory upregulation underscores the interdependence between chromatin remodelers and topoisomerases in maintaining genome integrity. Importantly, this adaptive rewiring also creates a therapeutic vulnerability—wherein ATRX-deficient gliomas, being overly reliant on TOP2B for G4 homeostasis, become selectively sensitive to G4-stabilizing compounds and TOP2B poisons such as CX-5461. Our findings provide new insight into the functional interplay between ATRX and TOP2B in G4 resolution and highlight a targetable synthetic vulnerability in ATRX-deficient gliomas, with potential implications for precision therapeutics.

Initially, CX-5461 was thought to exert its anticancer effects primarily through RNA polymerase I inhibition and nucleolar stress [50]; recent studies suggest it also acts as a TOP2B poison [30–32]. Consistent with this, our RADAR and band depletion assays confirmed that CX-5461 traps TOP2Bcc and induces persistent protein-DNA adducts, preventing normal TOP2B catalytic activity and altering its chromatin binding (Fig. 2). This trapping impairs TOP2B’s recruitment to G4 sites, forcing increased reliance on ATRX for G4 resolution, as evidenced by increased ATRX-G4 PLA foci and enhanced ATRX peaks at G4 sites following CX-5461 treatment (Fig. 4G and H). Additionally, CX-5461 disrupts the coordinated recruitment of ATRX and TOP2B to G4 sites, leading to the accumulation of unresolved G4 structures, replication fork stalling, and genomic instability. In ATRX-deficient glioma cells, where TOP2B is the primary mediator of G4 resolution, we observed an initial increase in TOP2B-G4 PLA foci, reflecting compensatory recruitment of TOP2B in the absence of ATRX. However, upon CX-5461 treatment, which covalently traps TOP2B on DNA and impairs its chromatin mobility, these PLA signals markedly declined, accompanied by diminished TOP2B occupancy at G4 loci. Notably, in cells with both ATRX depletion and CX-5461 exposure, TOP2B-G4 PLA foci were substantially reduced (Fig. 4I). This marked reduction is most likely explained by the dominant effect of CX-5461 on TOP2B function. ATRX loss increases the requirement for TOP2B-mediated G4 resolution, but CX-5461 sequesters TOP2B in covalent DNA complexes (TOP2Bcc), thereby depleting the pool of functional TOP2B available for recruitment. As a result, the elevated demand created by ATRX deficiency cannot be met, and the observed phenotype is driven primarily by the overriding action of CX-5461. This dual perturbation effectively disables the cell’s capacity to manage G4 structures, leading to their hyper-accumulation and exacerbated replication stress. These findings refine our understanding of CX-5461’s mechanism of action, highlighting its interference with genome maintenance machinery beyond RNA polymerase I inhibition and underscoring its selective vulnerability in ATRX-deficient contexts.

CX-5461 has been reported to induce G4-associated DNA damage in a replication-dependent manner, leading to synthetic lethality in ATRX-deficient gliomas [26, 28, 51]. Our study further explores its impact on replication fork dynamics, revealing that ATRX deficiency reduces replication fork speed in glioma cells, supporting previous findings that ATRX facilitates replication stress responses and prevents fork stalling and collapse [48, 52, 53]. Moreover, we demonstrate that CX-5461 induces MRE11-dependent fork degradation but does not generate lesions that irreversibly impede fork restart or progression (Fig. 6). Our findings suggest that ATRX and TOP2B functionally cooperate to resolve G4 structures during DNA replication. Disruption of this coordination by CX-5461 leads to G4 accumulation ahead of replication forks, causing replication fork stalling. In ATRX-deficient glioma cells, the absence of ATRX-mediated protection prevents efficient fork restart, ultimately resulting in fork collapse and the formation of DSBs upon CX-5461 treatment. Additionally, trapped TOP2Bcc at DNA ends undergoes processing, generating protein-free “naked” DSBs, which, if not efficiently repaired, become cytotoxic and further exacerbate genomic instability, contributing to the synthetic lethality observed in ATRX-deficient cells.

To determine whether CX-5461-induced disruption of ATRX-TOP2B coordination is dependent on its role as a TOP2B poison, we examined the effects of etoposide, a classical TOP2 poison, on ATRX-TOP2B interaction and G4 formation. Similar to CX-5461, etoposide induced G4 accumulation-associated DNA damage and inhibited glioma cell proliferation, particularly in ATRX-deficient gliomas. Mechanistically, etoposide trapped TOP2Bcc, preventing TOP2B recruitment to G4 sites while increasing ATRX binding at these regions. However, in contrast to CX-5461, etoposide did not affect the physical association between ATRX and TOP2B, suggesting that CX-5461 uniquely interferes with their coordinated engagement at G4 structures.

While our study establishes a functional cooperation between ATRX and TOP2B in G4 resolution and highlights the therapeutic vulnerability of ATRX-deficient glioma to CX-5461, several limitations should be acknowledged. First, although our data support a coordinated role of ATRX and TOP2B at G4 regions, the current evidence does not confirm a direct protein-protein interaction. Molecular docking suggested potential contact surfaces (data not shown); however, we did not perform site-directed mutagenesis to validate these predicted interfaces. We plan to investigate this in future work by introducing point mutations into TOP2B (R289A, E312A, K418A, and K1261A) and assessing their impact on ATRX binding and G4 resolution through co-immunoprecipitation and functional assays. These efforts aim to further clarify whether ATRX and TOP2B interact directly or operate through a larger chromatin-associated complex. Second, chromatin profiling by CUT&Tag was conducted in ATRX-WT glioma PDCs due to the lack of available ATRX-deficient PDCs. Once suitable ATRX-deficient PDCs are acquired, we will perform parallel analyses to assess how ATRX loss influences TOP2B recruitment and G4 localization. Third, the current findings are based on in vitro models; in vivo validation using ATRX-deficient patient-derived xenograft (PDX) models will be crucial to confirm the physiological and therapeutic relevance of this mechanism in a clinically relevant setting. Lastly, the broader consequences of TOP2B inhibition in normal neural cells, as well as the potential neurotoxicity of CX-5461, remain to be fully characterized. This is particularly important when considering translational applications of CX-5461 in brain tumors. Future studies should also aim to identify additional cofactors that cooperate with ATRX and TOP2B in G4 resolution. Exploring rational combinations of CX-5461 with inhibitors of DDR pathways may enhance synthetic lethality in ATRX-deficient gliomas. Moreover, given the prevalence of G4 accumulation in various cancer types, the broader applicability of CX-5461 as a G4-targeting agent warrants further investigation.

Supplementary data

Supplementary data is available at NAR online.

Conflicts of interest

The authors declare that they have no competing interests.

Funding

This study received grants from the National Natural Science Foundation of China (82172820 to J.Z.), the Natural Science Foundation of Shanghai (22ZR1466200 to J.Z.), the Clinical Research Special Funding of the Shanghai Municipal Health Commission (202340112 to J.Z.), the Research Funds for the Central Universities (22120250457 and 22120240228 to J.Z.), the Key Disciplines Group Construction Project of Shanghai Pudong New Area Health Commission (PWZxq2022-10 to C.Z.), and the Key Discipline Construction Project of Shanghai East Hospital (2024-DFZD-003S to C.Z.). Funding to pay the Open Access publication charges for this article was provided by the Key Disciplines Group Construction Project of Shanghai Pudong New Area Health Commission (PWZxq2022-10 to C.Z.).

Data availability

Data have been deposited to GEO (accession: GSE275821). Data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials.