Dual epigenetic and nuclear export inhibition by chidamide and selinexor in high grade B-cell lymphomas via survivin and PI3K/AKT inhibition

Abstract

Backgound

High-grade B-cell lymphoma with concurrent MYC and BCL2/BCL6 rearrangements (HGBL-DHL) is a highly aggressive disease that is resistant to conventional first-line immunochemotherapeutic regimens. This resistance necessitates the exploration of innovative therapeutic strategies.

Result

In this study, the combination of chidamide and selinexor showed significant synergistic antilymphoma effects in the treatment of HGBL-DHL. The synergistic effects were evidenced by the inhibition of cell proliferation, induction of apoptosis, and perturbation of the cell cycle in cell lines, as assessed by Cell Counting Kit-8, Annexin V/PI staining, and PI staining assays. Furthermore, in a xenograft mouse model of HGBL-DHL, this combination therapy markedly reduced the tumor burden without causing lethal toxicity. At the mechanistic level, the combination of chidamide and selinexor resulted in the synergistic downregulation of survivin and the PI3K/AKT signaling pathway. This dual inhibition was attributed to the interactive effects of the two drugs. The downregulation of key downstream targets of the PI3K/AKT pathway, including c-Myc, MCL1, BCL-XL, cyclin A2, and survivin, was synergistic and aligned with the phenotypic outcomes. Notably, survivin, an anti-apoptotic gene, underwent transcriptional repression by FOXO1 at the level of epigenetic regulation. Chidamide combined with selinexor synergistically down-regulated survivin in both the nucleus, cytoplasm and total protein levels via HDAC/FOXO1/survivin, HDAC3/PI3K/AKT/XPO1/survivin, XPO1/FOXO1/survivin, and XPO1/survivin axes.

Conclusion

Our preclinical data highlighted the potential synergistic efficacy of chidamide and selinexor in targeting HGBL-DHL, providing a rationale for further clinical investigation of this therapeutic combination for the treatment of this refractory disease.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13148-026-02056-6.

Introduction

HGBL-DHL, a highly aggressive subtype of high-grade B-cell lymphoma, is associated with a poor prognosis. It is defined by concurrent rearrangements of the MYC and BCL2 genes [1], which drive the malignant behavior of the disease. These genetic alterations impart HGBL-DHL with features that integrate Burkitt lymphoma (BL) and diffuse large B-cell lymphoma (DLBCL) in terms of morphology, immunophenotype, and genetic landscape. HGBL-DHL poses significant therapeutic challenges because of its inherent resistance to conventional chemoimmunotherapy regimens. Patients with HGBL-DHL often exhibit poor responses to first-line treatments and have a high likelihood of relapse [2–4]. The prognosis for HGBL-DHL patients remains poor even with intensive therapies such as stem cell transplantation, with a median overall survival of less than 2 years [5]. This underscores the unmet need for novel therapeutic approaches to improve outcomes in this population.

Chidamide, a selective inhibitor of HDAC1/2/3/10, has been approved for use in combination with rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) for the treatment of MYC- and BCL2-positive DLBCL. Preclinical studies have demonstrated its efficacy in HGBL-DHL cell lines and xenograft mouse models. In clinical settings, patients with triple-hit lymphoma (THL), characterized by concurrent MYC, BCL2, and BCL6 rearrangements, have shown favorable responses to a preparative regimen of carmustine, etoposide, cytarabine, cyclophosphamide, chidamide, and high-dose rituximab [6]. However, the complexity of this multidrug regimen has raised concerns regarding medication adherence and patient tolerability. Selinexor, a novel nuclear export inhibitor targeted on exportin 1 (XPO1), is approved for relapsed/refractory DLBCL (R/R DLBCL) patients who have received at least two prior systemic therapies. However, its wide application is limited by significant gastrointestinal and neurological toxicity. Combining chidamide with selinexor may enhance the antitumor effects of selinexor, potentially allowing for dose reduction and lower toxicity. Additionally, selinexor may enhance the efficacy of chidamide while simplifying the medication regimen. This study aimed to explore the efficacy and mechanisms of chidamide combined with selinexor in HGBL-DHL.

The expression of survivin, a novel antiapoptotic gene, is elevated in DLBCL [7, 8]. Patients with survivin expression had a significantly lower 5-year overall survival rate compared to those without survivin expression. Survivin expression may serve as a novel adverse prognostic factor for DLBCL. Moreover, in various cancers, its high expression is indicative of unfavorable clinical outcomes, contributes to resistance against chemotherapy [9–11], and is associated with relapse [12–18]. By suppressing survivin activity, tumors can become more sensitive to conventional therapies, and this approach may assist in overcoming resistance to multiple drugs [19–22]. In DLBCL-xenograft models, the survivin inhibitor YM155 was found to eliminate tumors and markedly enhance survival when compared to rituximab monotherapy [23]. Mechanistically, survivin inhibits apoptosis in the cytoplasm by blocking caspases and promotes cell division in the nucleus [24]. On the basis of these findings, downregulation of survivin has emerged as a potential therapeutic strategy for DLBCL.

Histone Deacetylase (HDAC) inhibitor increases the expression of the transcription factor Forkhead box O1 (FOXO1) [25], and can also promote its nuclear translocation [26], both of which can promote the transcription suppression of survivin by FOXO1 [27, 28]. This mechanism is testified in lymphoma, where HDAC inhibitors can downregulate survivin mRNA [29] and protein expression [30]. In nasal natural killer/T-cell lymphoma, survivin expression is induced through the phosphatidylinositol 3-kinases /protein Kinase B (PI3K/Akt) pathway both in vitro and in vivo [31]. The overexpression of PI3K or Akt significantly increases survivin mRNA expression, whereas the overexpression of wild-type phosphatase and tensin homolog (PTEN, a negative regulator of PI3K) decreases survivin mRNA levels [32]. Additionally, PI3K inhibitors can also downregulate survivin protein expression [33, 34]. Moreover, abnormal PI3K/AKT activation can increase the nuclear export of tumor suppressor proteins (TSPs) such as FOXO1 and survivin, with XPO1 being a key nuclear export mediator. Therefore, inhibiting the PI3K/AKT pathway may also inhibit the nuclear export of FOXO1 and survivin by inhibiting XPO1, leading to the transcriptional repression, accumulation and faster degradation of survivin in the nucleus, where survivin is known to be less stable than it is in the cytoplasm [35]. This hypothesis is also consistent with the literature reporting that selinexor treatment decreases survivin protein levels [36]. Our previous study also revealed that chidamide can suppress the PI3K/AKT pathway in HGBL-DHL by downregulating HDAC3 [37]. Thus, chidamide combined with selinexor may synergistically reduce survivin transcription at the epigenetic regulation level, through inhibiting HDAC/FOXO1, HDAC3/PI3K/AKT/XPO1/FOXO1 and XPO1/FOXO1 pathway

In view of the multiple inhibitory effects of chidamide and selinexor on survivin, we investigated their combined effects and molecular mechanisms in HGBL-DHL. Our study revealed that chidamide significantly enhanced the cytotoxicity of selinexor both in vitro and in vivo. This synergistic effect was evident in human cell lines and a cell-derived xenograft (CDX) mouse model of HGBL-DHL, driven by marked suppression of survivin expression.

Materials and methods

Reagents and cell lines

Chidamide (HBI-8000) was obtained from Shenzhen Chipscreen Biosciences (Shenzhen, China), and selinexor was obtained from Shelleck Chemicals (Houston, Texas, USA). For in vitro studies, each drug was dissolved in sterile DMSO (Sigma, MO, USA) at 50 mM to prepare stock solutions, which were stored at -20 °C. For oral administration, chidamide was suspended in 0.5% (w/v) CMC-Na. In accordance with the manufacturer’s instructions, the selinexor solution used for the mouse experiments consisted of 2% DMSO, 40% PEG300, 5% Tween 80, and 53% double-distilled water.

The HGBL-DHL cell lines TMD8, Toledo, MCA, and LR were acquired from the MD Anderson Cancer Center Laboratory. The cells were cultured in an incubator at 37 °C with 5% CO₂. All HGBL-DHL cells were cultured in RPMI-1640 (HyClone, Thermo Scientific, Logan, UT, USA) supplemented with 10% fetal bovine serum (Gibco, Thermo Scientific, Grand Island, NY, USA). The culture media were fortified with 100 units/ml penicillin and 100 mg/ml streptomycin.

Assessment of cell viability

Cell viability was assessed via the Cell Counting Kit-8 (CCK-8) from APExBIO (Texas, USA). TMD8, Toledo, MCA, and LR cells were seeded in 96-well plates at a density of 2 × 10⁵ cells/ml in 100 µl of medium and treated with DMSO, chidamide, selinexor, or their combination. For all the cells mentioned above, the selinexor concentrations used were 0.0625 µM, 0.125 µM, 0.25 µM, 0.5 µM and 1 µM, with chidamide fixed at 2 µM. After drug treatment for 24–48 h, 10 µl of CCK-8 reagent was added to each well and the mixture was incubated for an additional 2 h. The absorbance at 450 nm was measured via a Bio-Rad microplate reader (Bio-Rad, CA, USA). The inhibition rate was calculated according to the following formula (1):

1

Using CompuSyn software which was based on Chou-Talalay method, we determined the fraction affected (Fa) and combination index (CI) for each treatment group and produced the corresponding Fa-CI curves.

Apoptosis analysis via flow cytometry

To evaluate apoptosis, TMD8 and Toledo cells were treated with DMSO, chidamide, selinexor, or a combination of selinexor and chidamide for 24–48 h. The cells were then harvested, washed with ice - cold phosphate-buffered saline (Gibco BRL, Rockville, MD, USA), and stained with the APC Annexin V (556454, BD Bioscience, CA, USA) and FITC Annexin V Apoptosis Detection Kit (4201822 ,BD Bioscience, CA, USA) for 15 min in the dark at room temperature following the manufacturer’s protocol. Flow cytometry (Quanteon, ACEA Biosciences, USA) was performed within 1 h of staining. In the dot plot, the cells located in the right quadrant and exhibiting Annexin V positivity indicated apoptosis.

Cell cycle analysis via flow cytometry

TMD8 cells at a density of 2 × 10⁵ cells per well were first starved in FBS-free medium for 24 h, and then treated with DMSO, 3 µM chidamide, 3 µM selinexor, or a combination of 3 µM chidamide and 3 µM selinexor in complete medium for an additional 24 h. Similarly, 2 × 10⁵ Toledo cells were subjected to the same starvation and treatment protocol, with the exception that the chidamide concentration was reduced to 2.5 µM and the selinexor concentration was 0.5 µM. Following treatment, the cells were harvested, washed with ice-cold PBS (Gibco BRL, Rockville, MD, USA), and fixed with 70% ethanol overnight at 4 °C. The fixed cells were resuspended in PBS containing RNase and stained with propidium iodide (PI) for 30 min in the dark. Cell cycle analysis was performed via flow cytometry (Quanteon, ACEA Biosciences, USA).

Network Pharmacological analysis

Firstly, the 3D structures of chidamide and selinexor were retrieved from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/). PharmMapper (https://lilabecust.cn/pharmmapper/) was utilized to predict potential targets for these drugs. The corresponding gene sets were identified via UniProt (https://www.uniprot.org/). Initial searches of the OMIM, Phenopedia, and TTD databases via “Diffuse large B-cell lymphoma” retrieved relevant disease-related targets. Both disease-related and drug-related gene sets were subsequently used to construct a protein-protein Interaction (PPI) network via STRING, with a confidence threshold of 0.4. The vital network was analyzed via Cytoscape 3.9.1. Finally, GO and KEGG enrichment analyses were performed via ClusterProfler GO and KEGG in R 4.1.2, which were obtained from Bioconductor (https://www.bioconductor.org/), with an adjusted P-value < 0.05 and a Q-value < 0.05 as significance thresholds.

RNA sequencing

TMD8 cells were exposed to 2 µM chidamide, 0.125 µM selinexor, or a combination of 2 µM chidamide and 0.125 µM selinexor for 24 h. The cells were subsequently lysed via TRIzol reagent (Invitrogen, NY, USA), and RNA was extracted in accordance with the manufacturer’s instructions. Next-generation sequencing libraries were constructed through a series of steps, including sample quality control, mRNA isolation, mRNA fragmentation, cDNA synthesis, end repair, adenine addition, adaptor ligation, PCR amplification, library quality control, and circularization. Sequencing was performed via combinatorial probe-anchor synthesis (cPAS) following the manufacturer’s protocol. The Dr.TOM II online platform was used for data analysis, which involved data filtering, RNA identification, gene quantification, and differential expression analysis to facilitate subsequent gene annotation. To investigate phenotypic changes, we utilized the Phyper tool to conduct Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses on the annotated genes on the basis of the hypergeometric test. Terms and pathways with a Q value of < 0.05 were considered significantly enriched. A detailed protocol for each step of the RNA sequencing process is provided in Supplementary Document 1. The RNA-seq experiment was conducted with three biological replicates.

Western blot analysis

Approximately 1 × 10⁶ TMD8 and Toledo cells were each inoculated onto a 6 cm cell culture dish. Both cell lines were subjected to treatment with DMSO, 1.8 µM chidamide, 1 µM selinexor, or a combination of 1.8 µM chidamide and 1 µM selinexor for a duration of 24 h. Both cell types were subsequently harvested through centrifugation at 300 × g for 5 min at 4 °C. Each sample was lysed with 200 µL of RIPA lysis buffer (Thermo, USA) on ice for 10 min. The whole-cell extracts were obtained by collecting the supernatant after a subsequent centrifugation at 10,000 rpm for 10 min at 4℃. These extracts were then subjected to electrophoresis via 8% or 10% gels and transferred onto PVDF membranes (Millipore, Billerica, MA, USA). The membranes were blocked with 5% nonfat milk in TBST solution and incubated with specific primary antibodies, followed by incubation with HRP-conjugated goat secondary antibodies (HuaAn Biotechnology, Zhejiang, China) or HRP Goat anti-mouse secondary antibodies (ABclonal Technology, Hubei, China).

For the detection of various proteins, a range of primary antibodies were used. Survivin (2808), FOXO1(2880), FOXO3a(12829) and CyclinA2 (4656T) antibodies were obtained from Cell Signaling Technology (Danvers, Massachusetts, USA). ABclonal Technology (Hubei, China) provided the antibodies against HDAC3 (A19537), PI3K85α (A4992), phospho-PI3KP85α/P55γ/P85β-Y467/Y199/Y464 (AP0854), and c-Myc (A9032). Additional antibodies used included those against PI3K110δ (ET1705-46), AKT1 (ET1609-47), phospho-AKT1 (Ser473) (ET1607-73), BCL-XL (ET1603-28), and MCL1 (ET1606-14) were obtained from HuaAn Biotechnology (Zhejiang, China). An antibody against Ki67 (Ab16667) was purchased from Abcam (Cambridge, UK). For detection of total protein levels, an antibody against GAPDH (HA721136, 1:10000, HuaAn Biotechnology, China) was used as a loading control. For the nuclear and cytoplasmic fractionation experiments, antibodies against LaminB1 (13435) from Cell Signaling Technology and GAPDH (HA721136) were used as internal reference antibodies for the nuclear and cytoplasmic proteins, respectively. The visualization of proteins was achieved via an enhanced chemiluminescence (ECL) western blotting detection kit (Gene-Flow, Staffordshire, UK) in conjunction with the Bio-Rad ChemiDoc XRS + detection system (Bio-Rad, CA, USA). All experiments were performed in triplicate under identical conditions to ensure reproducibility.

Coimmunoprecipitation (Co-IP) experiment

In brief, 2 × 10⁷ TMD8 cells were cultivated in a 100 mm cell culture dish and exposed to DMSO, 1.8 µM chidamide, 1 µM selinexor, or a combination of 1.8 µM chidamide and 1 µM selinexor for 24 h. Following treatment, the cells were harvested and lysed with 1 mL of cell lysis buffer (RM00022, ABclonal Technology). The protein extracts were then incubated with survivin antibodies and protein A/G agarose (Millipore Corp., Burlington, MA, USA) overnight at 4 °C. After incubation, the beads underwent a minimum of three washes with cell lysis buffer and were subsequently eluted in 1x SDS-PAGE loading buffer. Further analysis of the samples was conducted via Western blotting using antibodies against survivin, HDAC3, and XPO1.

Nuclear and cytoplasmic protein extraction assay

These experiments were performed via a nuclear and cytoplasmic protein extraction kit (P0027, Beyotime Biotechnology, Shanghai, China). Briefly, 5 × 10⁶ TMD8 cells were cultivated in a 600 mm cell culture dish and exposed to DMSO, 1.8 µM chidamide, 1 µM selinexor, or a combination of 1.8 µM chidamide and 1 µM selinexor for 24 h. After treatment, the cells were harvested and supplemented with cytoplasmic protein extraction reagent A containing phenylmethanesulfonyl fluoride (PMSF). They were vortexed at maximum speed for 5 s to fully suspend and disperse the cell pellet and then placed in an ice bath for 10–15 min. Next, 10 µL of cytoplasmic protein extraction reagent B was added. After vortexing again at maximum speed for 5 s and incubating in an ice bath for 1 min, another vortexing step at maximum speed for 5 s was performed. The mixture was then centrifuged at 4 °C for 15 min at 12,000–16,000 × g. The supernatant was collected as the cytoplasmic protein extract. For the nuclear protein extraction, the pellet was resuspended in 50 µL of nuclear protein extraction reagent containing PMSF. The mixture was vortexed at maximum speed at 4 ℃ for a total of 30 min. Finally, the mixture was centrifuged at 4 °C for 10 min at 12,000–16,000 × g, and the supernatant was collected as the nuclear protein extract. The obtained cytoplasmic and nuclear protein samples were subjected to Western blot analysis with FOXO1,FOXO3a and survivin antibody.

Metabolomics analysis

A total of 2 × 10⁵ TMD8 cells were cultivated in a 600 mm cell culture dish and exposed to DMSO, 2 µM chidamide, 0.125 µM selinexor, or a combination of 2 µM chidamide and 0.125 µM selinexor for 24 h. After centrifugation, 100 µL of the culture medium was removed and mixed with 400 µL of methanol/acetonitrile (1/1, v/v) by vortexing for 30 s.The mixture was subsequently sonicated for 10 min at 4 °C. After 1 h of incubation at -20 °C, the samples were centrifuged at 13,000 rpm for 15 min at 4 ℃. The supernatant was transferred to a new tube and evaporated to dryness at 4 °C via a vacuum concentrator. The samples were subsequently reconstituted with 100 uL of acetonitrile/H2O solution (1/1, v/v) for injection analysis. The injection volume was 3 µL. Analysis of metabolites was performed on a Shimadzu Prominence UPLC system (Nexera UHPLC LC-30 A, Japan) interfaced with a TripleTOF 5600 + system (SCIEX Framingham, MA, U.S.A.) equipped with an ESI source. Metabolites were separated through a HILIC column (ZIC-pHILIC, 5 μm, 2.1 × 100 mm, PN: 1.50462.0001, Millipore) with the column temperature maintained at 40 °C. The mobile phases consisted of 25 mM ammonium acetate in 25 mM ammonia water (mobile phase A) and acetonitrile (90/10, v/v) (mobile phase B) and were run at a flow rate of 0.3 ml/min. The gradient was as follows: 95% B for 1 min, 65% B for 14 min, 40% B within 16 min and held for 18 min, then increased to 95% B for 18.1 min, then returned to 9% B and held for another 23 min. The flow rate for the mobile phases was set at 0.3 ml/min. The mass spectrometer was run in positive, information-dependent acquisition (IDA) mode, with a source temperature of 550 °C, the ion source gases 1 and 2 at 55 psi, the curtain gas at 35 psi, the collision energy at + 30 eV or −30 eV, the ion spray voltage floating at 5.5 kV or −4.5 kV, and the mass range at 60–1250 m/z. The accumulation time for the full scan was set at 150 ms, and the accumulation time for each IDA scan was 45 ms. Peaks of metabolites with intensities greater than 100 c.p.s. after adding up the signals from 10 rounds of IDA scans were chosen for further analysis. Data were collected via Analyst TF 1.6 software (SCIEX), and analysed via KEGG (https://www.kegg.jp/) and MetaboAnalyst6.0 (https://www.metaboanalyst.ca/).

In vivo efficacy of Selinexor combined with Chidamide in HGBL-DHL cell-derived xenograft mouse models

The CB17-SCID mice (8–10 weeks of age) utilized in this study were obtained from the Animal Care Facility of Xiamen University and maintained under germ-free conditions. All animal-based investigations were approved by the Animal Care and Use Committee and Ethics Committee of Xiamen University (reference number: 231007). Additionally, all the animal experiments were conducted in strict compliance with the Directive 2010/63/EU in Europe. To establish HGBL-DHL cell-derived xenograft models, mice were subjected to subcutaneous injection of 2 × 10⁷ TMD8 cells following exposure to a 1.5-Gy total body irradiation dose. On day 9 post-injection, the mice were randomly allocated into four groups (n = 10). The treatment were administered for eight consecutive days as follows: (1) control group (0.5% carboxymethylcellulose sodium); (2) chidamide-treated group (oral gavage at 15 mg/kg/day); (3) selinexor-treated group (oral gavage at 7.5 mg/kg, three times a week); and (4) combination-treated group (chidamide at the same dose as group 2 plus selinexor at the same dose as group 3). Tumor size and body weight were measured daily, and tumor volume was calculated via Eq. (2) as shown below. Specifically, “L” indicates the maximum longitudinal diameter, and “W” represents the maximum transverse diameter, with both parameters measured via a caliper. On day 17, five mice from each group were euthanized with CO₂, and the tumors were harvested, weighed, and divided into two parts. One part was fixed in 4% paraformaldehyde for immunohistochemistry (IHC), hematoxylin and eosin (HE) staining, and terminal deoxynucleotidyl transferase-Mediated dUTP nick end labeling (TUNEL) assays. The other part was frozen at −80 °C for subsequent western blot analysis. Primary antibodies against survivin and ki-67 were applied to the tumor sections, which were subsequently incubated overnight at 4 °C. DAB (ZLI − 9018, ZSGB - Biotechnologies, Beijing, China) was then applied at room temperature until a color change was observed on the microscope slide. IHC and HE analyses were conducted via a fluorescence microscope (CX23, OLYMPUS, Tokyo, Japan). TUNEL assays were performed via the Pannoramic MIDI (3DHISTECH, Hungary). The remaining five mice per group continued to receive their respective treatments and were monitored for survival analysis. A tumor volume exceeding 1500 mm³ was designated as the endpoint.

2

Statistical analyses

Each experiment was independently repeated three times in triplicate, and the data are presented as the means ± SDs. Statistical analysis was performed via GraphPad Prism v6.0 (GraphPad Software, La Jolla, CA, USA). Comparisons between two groups were made via Student’s t-test. For multigroup comparisons, two way ANOVA was employed. A P value of less than 0.05 was considered statistically significant. In the manuscript, the significance levels are represented as *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Results

Synergistic effects of selinexor and chidamide on HGBL-DHL cells through proliferation inhibition, apoptosis induction, and cell cycle arrest in vitro

In this study, we investigated how the combination of chidamide and selinexor affects the viability, apoptosis, and cell cycle progression of HGBL-DHL cell lines (TMD8, Toledo, MCA, and LR). Selinexor alone reduced cell viability in a dose-dependent manner. Notably, the addition of chidamide significantly strengthened the anti-proliferative effects of selinexor across all the cell lines after 24 h (Fig. 1a and b, Supplementary Fig. 1a) and 48 h (Supplementary Fig. 1b). The combination index (CI) values for TMD8 and Toledo cells at 24 h were all less than 1.0 (Fig. 1a and b), which indicated the synergy of these two drugs. Selinexor also induced apoptosis in a concentration-dependent manner. Chidamide significantly enhanced this effect (Fig. 1c and d, Supplementary Fig. 1c). For example, after 24 h of treatment in TMD8 cells, the addition of 2 µM chidamide increased the apoptosis rate of 8 µM selinexor from 34.85% to 50.08%, marking a 1.43 - fold increase. In Toledo cells, the percentage of apoptotic cells in the combination group (2 µM chidamide and 8 µM selinexor) was 139% greater than that in the chidamide alone group.

Fig. 1.

The synergistic antilymphoma effect of selinexor in combination with chidamide on HGBL-DHL in vitro. a, b Proliferation and combination index (CI) values of human TMD8 and Toledo cells treated with chidamide and selinexor for 24 h were evaluated. c, d Typical flow cytometric images for Annexin V/PI staining and the corresponding percentages of apoptotic TMD8 and Toledo cells are presented. e, f The percentages of TMD8 e and Toledo cells f in the G1, S, and G2 phases after treatment with chidamide and selinexor for 24 h were determined. The values are presented as the means ± SDs (n = 3). Two-way ANOVA was applied for statistical analysis, with significance levels set at P < 0.05. Significance levels are marked as *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001

As HDAC dysregulation can affect cell cycle progression [38, 39], chidamide may induce cell cycle arrest, thus contributing to its cytotoxicity in HGBL-DHL cells. We then explored whether chidamide could enhance selinexor-induced cell cycle arrest. In TMD8 cells, the combination increased the number of cells in the G1 phase population and reduced the number of cells in the S and G2 phases (Fig. 1e). Compared with the single agent treatments, the combination notably decreased the population in the G2 phase in Toledo cells (Fig. 1f). In Toledo cells, compared with the selinexor monotherapy group, the proportion of cells arrested in the G1 phase was significantly increased in the combination therapy group.

Selinexor and chidamide modulate targets associated with the PI3K/AKT signaling pathway in silico and at the transcriptional level

A network pharmacology approach was used to explore the potential targets and shared pathways associated with chidamide and selinexor. We identified 65 drug targets and 1,474 disease targets for DLBCL via PharmMapper. The intersection of these targets revealed 26 common targets (Supplementary Fig. 2a). The interaction network was visualized via Cytoscape 3.9.1(Supplementary Fig. 2b,2c).

Considering that both chidamide and selinexor can interact with the PI3K/AKT pathway, we investigated the regulatory interactions between the 26 target genes and this pathway. Fourteen genes including SPEN, NOTCH1, ERBB2, CTNNB1, CCNE1, CDK2, AXIN1, MMP2, CD38, ZEB1, BTK, ANXA1, GF2, and S100B, had direct regulatory interactions with the PI3K/AKT pathway, whereas three genes including SMARCA4, PARP1, and CUL1, had indirect regulatory interactions. ClusterProfiler GO identified 20 enriched GO terms. Specifically, the GO enrichment results revealed that RNA polymerase II-specific DNA-binding transcription factor binding, DNA-binding transcription factor binding, transcription coregulator activity, and transcription coregulator binding were enriched biological processes (Fig. 2a), which coincided with the epigenetic regulatory function of chidamide.

Fig. 2.

In silico and transcriptional analyses revealed combined effects of chidamide and selinexor on PI3K/AKT-related pathways, as well as their effects on epigenetic regulation and nuclear export inhibition. a Bubble chart displaying the top 20 GO terms by significance. The count reflects the number of enriched genes, whereas the gene ratio represents the proportion of enriched genes relative to all targets. b Fold changes in the mRNA levels of PI3K/AKT - associated genes across treatment groups (S: selinexor, C: chidamide, Com: combination). c, d, e, f GO and KEGG pathway enrichment analyses highlighted differential pathways and enriched genes among the treatment groups. The mRNA levels are presented as the means ± SDs from three biological replicates. Statistical analysis was performed via two-way ANOVA with significance set at P < 0.05. Significance levels are marked as *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001

The RNA-seq analysis of TMD8 cells verified the results of network pharmacology, as the combination of selinexor and chidamide downregulated transcripts of genes related to the PI3K/AKT axis. The combination downregulated transcripts of AKT1, the proliferation-associated gene MYC, the antiapoptotic genes BCL-XL, and MCL1, and the cell cycle-associated gene cyclin A2 (Fig. 2b). Compared with the control group, the levels of the transcription factor FOXO1 mRNA were upregulated in both the chidamide and combination groups, while the mRNA levels of its downstream target survivin were significantly downregulated in the chidamide group, suggesting that FOXO1 may inhibit survivin transcription (Fig. 2b, Supplementary Table 1).

Moreover, among the 17 joint targets of DLBCL and drugs that associated with the PI3K/AKT pathway obtained from network pharmacology analysis, ERBB2, CCNE1, and CUL1 mRNA expression was significantly lower in the combination group than in the monotherapy groups. SPEN and AXIN1 mRNA expression was significantly lower in the combination group than in the chidamide and/or selinexor monotherapy groups (Fig. 2b, Supplementary Table 1). GO biological process enrichment analysis revealed that differential genes between the control and combination groups were involved mainly in protein transport, the cell cycle, and cell division (Fig. 2c). In addition to these three processes, genes whose expression differed between the control and selinexor groups were also enriched in the protein stabilization and the mitotic cell cycle (Fig. 2d). GO molecular function enrichment analysis revealed that differential genes between the control and combination groups were enriched mainly in transcription coactivator activity (Fig. 2e), and the differential genes between the control and chidamdie group were enriched in genes related to transcription coactivator activity (Supplementary Fig. 2d). KEGG pathway enrichment analysis revealed that differential genes between the control and combination groups (Fig. 2f), as well as between the chidamide group and the control group (Supplementary Fig. 2e), were enriched mainly in the pathways related to nucleocytoplasmic transport and the cell cycle. Since chidamide loosens chromatin via promoter acetylation, easing access for transcription factors and RNA polymerase II, thereby boosting gene expression, all the functional enrichment pathways mentioned above are consistent with the epigenetic regulatory mechanisms of chidamide and the nuclear export inhibition effects of selinexor.

Selinexor combined with Chidamide inhibits PI3K/AKT signaling pathway at the translation level in vitro

The protein levels of effectors in the PI3K/AKT signaling pathway were evaluated to confirm the inhibitory impact of selinexor plus chidamide in TMD8 and Toledo cells. We focused on downstream targets of the PI3K/AKT pathway including c-Myc, survivin, MCL1, BCL-XL, and cyclin A2, which are linked to cell proliferation, anti-apoptosis, and the cell cycle. The western blotting results revealed that the drug combination reduced the levels of PI3K110δ, PI3K p85α, phospho-PI3K p85α/P55γ/P85β-Y467/Y199/Y464, AKT1, and phospho-AKT1 (Ser473), and downstream genes such as -Myc, MCL1, BCL-XL, survivin, and cyclin A2 (Fig. 3a). Time-course analysis revealed distinct effects of selinexor and chidamide on survivin protein levels (Fig. 3b). After 6 h, monotherapy with either drug upregulated survivin, while the combination treatment downregulated it. At 12 h, selinexor alone downregulated survivin, and combination therapy still had a downregulatory downregulation effect. By 24 h, both monotherapies downregulated survivin, with the combination resulting in increased synergistic downregulation.

Fig. 3.

Chidamide and selinexor mediate antilymphoma effects via PI3K/AKT and survivin inhibition. a Differential protein levels of PI3K/AKT effectors in TMD8 and Toledo cells after 24 h of chidamide or selinexor treatment were assessed by Western blotting. b The effect of drug treatment duration on survivin protein levels was examined. c Interaction between XPO1, HDAC3 and survivin was assessed via co-IP. d The subcellular localization of FOXO1, FOXO3a and survivin in the nucleus and cytoplasm following drug treatment was determined

Coi

mmunoprecipitation experiments demonstrated that survivin interacted with both HDAC3 and XPO1 (Fig. 3c). We propose that in addition to the PI3K/AKT pathway, the effects of drugs on HDAC3 and XPO1 may also contribute to survivin downregulation. As survivin can be exported from the nucleus via XPO1-mediated transport, we performed nuclear and cytoplasmic extraction experiments to assess survivin distribution in these compartments. The results showed that after 48 h of treatment, survivin levels synergistically decreased in both compartments (Fig. 3d). This dual reduction likely enhances the anticancer effect, as survivin in the cytoplasm inhibits caspase 3/7/9 activation, and survivin in the nucleus is crucial for cell division [24]. Both chidamide and selinexor can increase FOXO1 accumulation in nucleus, thus exerting its transcriptional suppressive effect on its downstream target survivin (Fig. 3d).

Selinexor combined with Chidamide influence metabolic reprogramming associated with the PI3K/AKT signaling pathway

According to previous research [40], as a key precursor for purine synthesis, serine provides one - carbon units through the mitochondrial tetrahydrofolate cycle. By regulating MTHFD2 and ATF4, the mTORC1 signaling pathway enhances the conversion of serine to purine. When PI3K/AKT/mTOR is inhibited, serine cannot be converted to purines, thus depriving tumor cells of the purines required for proliferation and affecting their growth. KEGG classification analysis revealed that the differentially abundant metabolites between the control group and the combination group were enriched in “Glycine, serine and threonine metabolism” (Fig. 4a). The level of serine in the combination group was significantly greater than that in the control group (Supplementary Table 2), indicating reduced serine metabolism and utilization. This is consistent with the increased serine content caused by the inhibition of PI3K/AKT/mTOR. Previous research [41] demonstrated that the PI3K/AKT activator artesunate can promote glycerophosphocholine metabolism. In the comparison with the selinexor group, “choline metabolism in cancer pathway” is up - regulated in the combination group (Fig. 4b), with the glycerophosphocholine content significantly increased (Supplementary Table 2). This finding is in line with the trend reported in the literature [41], which showed that the glycerophosphocholine concentration may increase when PI3K/AKT is inhibited. When the chidamide group was compared with the combination group, differentially abundant metabolites were enriched in “Cysteine and methionine metabolism” (Fig. 4c). In the combination group, the contents of methionine and serine in this pathway are significantly increased (Supplementary Table 2). As shown in a previous study [42], methionine deficiency is closely correlated with poor prognosis in newly - diagnosed DLBCL patients. The increase in methionine in the combination group may imply a better prognosis. Pathway analysis revealed that the differentially abundant metabolites between the control group and the selinexor group were enriched in “Glycerophospholipid metabolism” (Fig. 4d). The PI3K/AKT activator artesunate can promote glycerophospholipid metabolism. Choline is a key precursor in glycerophospholipid metabolism. The increased choline content in the selinexor group (Supplementary Table 2) indicates impaired glycerophospholipid metabolism, which is consistent with the trend of PI3K/AKT inhibition that hinders glycerophospholipid metabolism.

Fig. 4.

Chidamide and selinexor affected metabolic pathways associated with the PI3K/AKT pathway. a KEGG classification analysis of the control vs. combination groups. b KEGG classification analysis of selinexor vs. combination. c KEGG classification analysis of chidamide vs. combination. d KEGG pathway analysis of the control vs. selinexor groups

Selinexor in combination with Chidamide synergistically suppresses tumor growth in a HGBL-DHL xenograft mouse model

Encouraged by these promising in vitro results, we further investigated the synergistic antilymphoma effects of selinexor in combination with chidamide in a TMD8 cell-based xenograft mouse model. Initially, the mice were subjected to sublethal radiation at 1.5 Gy, followed by subcutaneous injection of TMD8 cells (2 × 10⁷) in the right flank. After nine days, the tumor-bearing mice were randomly divided into four groups: vehicle, chidamide, selinexor, and combination therapy. The vehicle or drugs were administered orally for nine days according to the protocol outlined in Fig. 5a. Throughout the treatment period, tumor volumes were measured daily to evaluate therapeutic efficacy. Upon completion of nine days of treatment, 20 mice in total were euthanized, with tumors collected from five mice per group. The tumor volume, representative photographs, and tumor weights are presented in Fig. 5b, c, and d, respectively.

Fig. 5.

The combination of chidamide and selinexor synergistically inhibits tumor growth in xenograft mouse models. a Cell transplantation protocol for the HGBL-DHL tumor xenograft model. b, c Tumor volume (b) and representative images (c) of the mice were recorded daily. The tumor volumes and weights are expressed as the means ± S.D.s to evaluate the synergistic effects of chidamide and selinexor. d Tumor weight analysis of TMD8 - bearing xenograft mice following various treatments (n = 5). e Kaplan-Meier survival analysis of mice with tumor xenografts. f Body weight changes of in the mice (mean ± S.D.) used to assess the toxicity of chidamide and selinexor. g IHC, TUNEL, and HE staining of tumor tissue sections. Tumor samples from the vehicle, chidamide, selinexor, and combination groups were processed and analyzed. h Protein levels of PI3K/AKT cascade of tumor tissues were determined by western blot. Images were captured via an OLYMPUS fluorescence microscope (original magnification, ×200). Two - way ANOVA was used for statistical analysis, with significance set at P < 0.05. Significance levels are indicated as *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001

Our observations revealed that both chidamide and selinexor monotherapies significantly inhibited the in vivo proliferation of TMD8 cells. Notably, the combination therapy demonstrated an enhanced inhibitory effect compared with either monotherapy alone (Fig. 5b). Furthermore, tumor weights were markedly lower in the combination therapy group than in the vehicle control and monotherapy groups (Fig. 5d). Survival analysis indicated that both chidamide and selinexor monotherapies substantially prolonged survival, with the combination therapy further enhancing this survival benefit (Fig. 5e). Importantly, no significant differences in body weight were observed among the four groups (Fig. 5f).

To further elucidate the in vivo effects of chidamide and selinexor on the protein expression of PI3K/AKT signaling effectors, immunohistochemical (IHC) staining (Fig. 5g) and western blot analysis (Fig. 5h) were performed on tumor tissues from CDX mice. Additionally, to assess cellular proliferation within tumor tissues, IHC analysis was conducted to evaluate the expression of the proliferation marker Ki-67. Compared with the vehicle control and monotherapy groups, the combination therapy group presented significantly fewer survivin- and Ki-67-positive tumor cells (Fig. 5g). Western blot analysis of tumor tissues from CDX mice was also performed to examine the protein levels of HDAC3, PI3K p110δ, PI3K p85α, phosphor-PI3K p85α/P55γ/P85β-Y467/Y199/Y464, AKT1, phospho-AKT1 (Ser473), and downstream targets regulated by the PI3K/AKT pathway, including c-Myc, Cyclin A2, MCL1, BCL-XL, and survivin. As depicted in Fig. 5h, the combination of selinexor and chidamide significantly downregulated the expression of these proteins compared with that in the other three groups. TUNEL assays conducted on tumor tissue sections from CDX mice revealed that the combination therapy group exhibited increased apoptosis (Fig. 5g) in vivo. HE staining of tumor tissues from the vehicle control group revealed abundant tumor cells and active tumor growth. In contrast, tumor tissues from the chidamide and selinexor monotherapy groups presented reduced tumor cell density and increased necrosis. Most notably, the combination therapy group presented minimal tumor cell density and extensive necrosis (Fig. 5g). Collectively, these results strongly support the notion that the combination of chidamide and selinexor exhibits remarkable synergistic activity and acceptable tolerability for the treatment of HGBL-DHL in vivo.

Discussion

HGBL-DHL is a notably intractable subtype that is frequently resistant to front-line immunochemotherapy regimens, highlighting the urgent need to investigate new therapeutic approaches. Our study revealed that the combination of chidamide and selinexor had synergistic effects on HGBL-DHL, which was associated with the comprehensive downregulation of the PI3K/AKT signaling pathway, and especially via the multifaceted reduction in survivin expression. In particular, the protein levels of several downstream targets of the PI3K/AKT pathway, such as c-Myc, MCL1, BCL-2, and cyclin A2, were decreased both in vitro and in vivo, culminating in inhibited cell proliferation, increased apoptosis, and cell cycle arrest (Fig. 6).

Fig. 6.

A hypothetical model illustrating the cooperative pharmacological mechanism of chidamide and selinexor, showing how their combined action may curb HGBL-DHL progression

As a critical antiapoptotic and mitophagy-related protein, survivin is present in both the cytoplasm and nucleus. Within the cytoplasm, it forms a survivin-XIAP (X-linked inhibitor of apoptosis protein) complex to inhibit caspase-3/7/9, thereby preventing apoptosis. In the nucleus, it plays a critical role in cell division. Specifically, survivin is a nonenzymatic regulatory component of the chromosomal passenger complex (CPC), together with borealin and inner centromere protein (INCENP). This complex plays a key role in mitosis. Survivin helps direct the CPC to centromeres during prometaphase, ensuring correct chromosome alignment before segregation at anaphase. It is phosphorylated by Aurora B kinase to maintain dynamic associations with centromeres. As cells progress through mitosis, survivin moves from centromeres to the spindle midzone during the metaphase–anaphase transition, and then to midbodies during telophase, where it is degraded after cytokinesis. Additionally, survivin can bind to mitotic spindle microtubules, centrosomes, and kinetochores, thus to regulate microtubule dynamics and nucleation. Conversely, its depletion increases microtubule nucleation and catastrophe events (Fig. 6). A reduction in survivin in the cytoplasm induces apoptosis, whereas a reduction in survivin in the nucleus disrupts cell division. Both mechanisms contribute to anticancer effects.

To clarify how chidamide downregulates survivin, we analyzed its transcription, post-translational modification, cellular localization, and protein degradation through a review of the relevant literatures. At the transcriptional level, prior studies have indicated that siHDAC2 and siHDAC3 can decrease survivin mRNA levels [35]. Our RNA-seq data also revealed that chidamide reduces survivin mRNA. Mechanistically, Chidamide may enhance FOXO1 transcription by promoting histone acetylation within the FOXO1 promoter region through its epigenetic regulatory effects, and also facilitate the nuclear translocation of FOXO1. Additionally, Chidamide increases the level of histone acetylation at the survivin promoter, thereby promoting the binding of FOXO1 to the survivin promoter region. All these actions ultimately lead to the suppression of survivin transcription [27, 28]. Additionally, siHDAC3 increases survivin acetylation and promotes its nuclear translocation. Since nuclear survivin is less stable than cytoplasmic survivin [35], the nuclear translocation process accelerates its degradation within the nucleus. This process requires the interaction between HDAC3 and survivin, which is confirmed by co-immunoprecipitation experiments in our study. Finally, siHDAC3 upregulates 26 S proteasome expression, which can also enhance survivin degradation [35].

To explore the mechanism by which selinexor downregulates survivin, we concentrated on its cellular localization and protein degradation by examining prior studies. XPO1 can bind to the nuclear export sequence (NES) of survivin, preventing its nuclear export. Our coimmunoprecipitation experiments confirmed the interaction between survivin and XPO1. As reported in previous research [43], selinexor inhibits XPO1-mediated nuclear export of survivin in breast cancer cells within 4 h, leading to increased nuclear survivin and decreased cytoplasmic survivin. After 12 h, survivin levels in both compartments start to decrease, which is consistent with the trend observed after 48 h of drug treatment in the present study.When selinexor is combined with caspase inhibitors, total survivin levels partially recover, indicating that caspase-9 may promote survivin degradation and relieve the inhibitory effect of survivin on caspase-9, thereby inducing apoptosis. Selinexor can also increase the accumulation of FOXO1 in the nucleus, which helps FOXO1 to inhibit the transcription of survivin at the level of epigenetic regulation, since chidamide may also facilitate the binding of FOXO1 to the survivin promoter by enhancing local histone acetylation within this region.

Abnormal activation of the PI3K/AKT pathway can increase the nuclear export of TSPs medicated by XPO1. Active PI3K/AKT phosphorylates TSPs such as FOXO1, strengthening its binding to XPO1 and promoting its nuclear export, which undermines its tumor-suppressive functions [44]. Conversely, downregulation of PI3K/AKT may suppress XPO1-mediated FOXO1 export and increase FOXO1 accumulation in the nucleus, thereby inhibiting survivin transcription. Our study indicated that chidamide downregulates HDAC3 (Fig. 3a), leading to inhibition of the PI3K/AKT axis [37]. It is possible that chidamide-induced PI3K/AKT suppression and the resulting inhibition of XPO1 function, combined with the direct inhibition of XPO1 by selinexor, might synergistically cause survivin accumulation and its subsequent rapid degradation in the nucleus.

In conclusion, chidamide and selinexor may work synergistically to downregulate survivin levels by simultaneously targeting the HDAC/FOXO1/surviving, HDAC3/PI3K/AKT/XPO1/survivin, XPO1/FOXO1/survivin, and XPO1/survivin axes.

In R/R DLBCL, chidamide resistance is linked to increased CREB-binding protein (cAMP response element binding protein) expression, which in turn stabilizes Aurora kinase A (AURKA) [45]. XPO1 inhibition reduces survivin transcription by preventing CREB-binding protein-driven STAT3 acetylation and STAT3 attachment to the survivin promoter [43]. Selinexor lowers AURKA levels, and AURKA inhibitors inhibit liposarcoma cell growth [46]. Hence, selinexor might increase the effectiveness of chidamide by targeting resistance targets related to chidamide. When treated with selinexor, nonresponder/resistant acute myeloid leukemia (AML) cells show increased FOXO3 S253 phosphorylation, which is a key inhibitory site. Given that FOXO3 serves as a key transcription factor capable of upregulating several apoptosis-associated genes, the inhibition of apoptosis may constitute a primary resistance mechanism in AML [47]. However, HDAC1 and HDAC2 siRNAs increase FOXO3a expression in human neuroblastoma cells [48]. Pan-HDAC inhibitors can drive FOXO3a nuclear translocation [26]. Our results also found that chidamide can increase FOXO3a expression and nuclear translocation (Fig. 3d), which can increase the proapoptotic role of FOXO3a by transcriptional regulation within the nucleus. ATP-binding cassette subfamily C member 4 (ABCC4), also known as multidrug resistance-associated protein 4 (MRP4), reduces drug effectiveness by expelling drugs out of cells. ABCC4 may be a key resistance mechanism for selinexor in multiple myeloma [49]. However, HDAC inhibitors can lower ABCC4 mRNA expression [50]. The mutual regulatory effects of chidamide and selinexor on each other’s resistance targets likely play a part in their synergistic effects against HGBL-DHL.

According to the drug instructions, for MYC and BCL2 positive DLBCL patients, chidamide is dosed at 20 mg on days 1, 4, 8, and 11 of a 3-week cycle. In R/R DLBCL patients who have undergone at least two treatment lines, selinexor is recommended at 60 mg orally on days 1 and 3 weekly. On the basis of their pharmacokinetic properties, we suggest exploring these agents in HGBL-DHL according to the instructed regimens. Both selinexor and chidamide peak approximately 4 h after a single dose. Selinexor is cleared continuously with a mean terminal half-life (T_1/2) of 6–8 h, whereas chidamide was subjected to sustained exposure with a T_1/2 of 17 h. On days 11–21, chidamide was paused, but selinexor was continued on days 15 and 17. Thus, chidamide compensates for the declining exposure of selinexor, and selinexor fills the dosing break of chidamide. Moreover, both drugs bind highly to plasma proteins (≈ 90%). When used together, these two compounds may compete for plasma protein binding, raising each other’s free concentration and boosting their efficacy.

Given these dosing regimens, synergistic pharmacological effects and complementary pharmacokinetic benefits of the drug combination can maximize clinical efficacy. However, several unresolved issues necessitate further exploration in future research. In the present study, the mechanisms by which HDAC3 modulate PI3K/AKT axis, as well as the regulatory relationship between PI3K/AKT and XPO1 in HGBL-DHL, were not been fully clarified. Additionally, the degradation process of survivin require further verification. To date, we have demonstrated only the preclinical efficacy and safety of the combination of chidamide and selinexor. The effects of this combination on primary HGBL-DHL cells and in clinical settings remain to be determined. Therefore, further cellular and molecular studies, along with clinical trials, are needed to address these open questions.

In conclusion, this study provides compelling evidence that survivin and the PI3K/AKT signaling pathway, may be promising therapeutic targets for HGBL-DHL. The multiple inhibitory effects of the combination of chidamide and selinexor on survivin and the PI3K/AKT axis effectively target HGBL-DHL without causing fatal toxicity, thus representing a potential antilymphoma strategy. Specifically, given that survivin is an anti-apoptotic protein, the synergistic downregulation of survivin by the combination of chidamide and selinexor in HGBL-DHL cell lines is consistent with the phenotypic result of increased apoptosis rate by the combination treatment.The promising preclinical efficacy and safety profile of this combination regimen highlight the necessity for further evaluation of its clinical potential in HGBL-DHL management.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

JZL, JZ, BX: conception and design. JZL, HZ, XGZ : research performance. SMJ, DMQ, BZC, MXZ : provision of study thought and technology. JZ, JZL, HZ : data analysis and interpretation. JZ, BX, JZL, XGZ : funding support. JZ, HZ: manuscript writing and review. JZL, JZ, BX: project administration and supervision.

Funding

This research was funded by the National Natural Science Foundation of China (U22A2090 and 82170180 granted for Bing Xu; 82470187 granted for Jie Zha), the Natural Science Foundation of Fujian Province (2023J06054 granted for Jie Zha), Xiamen Municipal Bureau of Science and Technology (3502Z20234001 granted for Jie Zha; 3502Z20244015 granted for Bing Xu), Young Investigator Research Program of Xiang’an Hospital of Xiamen University (XM02070002 granted for Jiazhen Lin), Open Innovation Fund for undergraduate students of Xiamen University (KFJJ-202410 granted for Bozhang Chen), XMU Training Program of Innovation and Enterpreneurship for Undergraduates (S202510384225 granted for Mingxin Zhuang),

Xiamen Medical and Health Guidance Project (3502Z20224ZD1014 granted for Xinguo Zhuang), the Natural Scientific Foundation of Xiamen (3502Z20227340 granted for Xinguo Zhuang), and the Fujian Natural Science Foundation of China (2022J011372 granted for Xinguo Zhuang).

Data availability

Data will be made available upon request to Bing Xu. The datasets generated and/or analysed during the current study are available in the GEO repository: Series GSE298302, https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE298302.

Declarations

Ethics approval and consent to participate

The animal study was performed in line with Directive 2010/63/EU in Europe. Approval of animal experiment was granted by the Animal Care and Use Committee and Ethics Committee of Xiamen University.

Consent to publish

Not applicable.

Competing interests

The authors declare no competing interests.