A dual-mechanism model of Selinexor in DLBCL: p53 reactivation and metabolic reprogramming.

Introduction

Diffuse large B-cell lymphoma (DLBCL) is the most common type of non-Hodgkin’s lymphoma (NHL), accounting for 25–50% of cases [ 1 ]. The disease demonstrates significant clinical and therapeutic diversity, indicating its complex genetic and molecular underpinnings. Increasing evidence across hematologic malignancies suggests that molecular biomarkers critically influence disease progression and clinical outcomes. Recent meta-analytic data have further highlighted the prognostic significance of angiogenesis-related factors such as vascular endothelial growth factor-A (VEGF-A), reinforcing the importance of biology-driven risk stratification strategies in lymphoma and leukemia [ 2 ]. Although the standard R-CHOP regimen (rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone) yields high initial response rates, approximately 30%-40% of patients experience relapse or develop refractory disease [ 3 ]. Recent advancements in chemotherapy and immunotherapy combined with autologous stem cell transplantation have improved outcomes for many patients. However, managing relapsed or refractory DLBCL remains a major challenge, particularly in cases that are resistant to conventional therapies. These persistent challenges highlight the urgent need for innovative therapeutic strategies and have spurred the development of numerous new drugs in recent years to enhance salvage therapy efficacy [ 4 ].

Exportin 1 (XPO1, also known as CRM1), is a nuclear export receptor that regulates the transport of proteins and RNAs with nuclear export signals (NES) from the nucleus to the cytoplasm [ 5 ]. Dysregulation of XPO1-mediated export can mislocalize key tumor suppressor proteins, such as p53, to the cytoplasm. This disables their tumor-suppressive functions and promotes oncogenic signaling. Aberrant XPO1 activity has been implicated in the pathogenesis of DLBCL and other cancers [ 6 , 7 ]. Accordingly, pharmacologic inhibition of XPO1, has emerged as an attractive therapeutic approach. By blocking XPO1, selective inhibitors prevent the nuclear export of tumor suppressors like p53, leading to their nuclear retention and reactivation of downstream cell-cycle arrest and apoptotic pathways. Consistent with this mechanism, XPO1 inhibitors have demonstrated potent antitumor effects in preclinical studies by restoring tumor suppressor function. This underscores XPO1 as a promising therapeutic target with a mechanism of action distinct from that of conventional chemotherapies.

Selinexor (KPT-330) is a first-in-class selective inhibitor of nuclear export (SINE) that targets XPO1. It has demonstrated broad anticancer activity and has been approved for certain malignancies. In relapsed or refractory DLBCL, single-agent Selinexor demonstrated meaningful clinical efficacy in the phase II SADAL trial, leading to its FDA approval for DLBCL patients who had received at least two prior systemic therapies [ 8 ]. Despite this clinical progress, therapeutic responses to targeted agents in hematologic malignancies are increasingly recognized to be influenced by underlying genetic and molecular heterogeneity. Pharmacogenomic studies in multiple myeloma have demonstrated that genetic polymorphisms significantly modulate treatment outcomes of proteasome inhibitors and immunomodulatory drugs, emphasizing the need for mechanism-oriented and biologically tailored therapeutic strategies [ 9 ]. Nevertheless, the precise molecular mechanisms by which Selinexor exerts its anti-lymphoma effects remain unclear. This gap in understanding necessitates integrated investigations of Selinexor’s action in DLBCL at both the clinical and molecular levels.

Given the growing recognition that molecular context critically determines therapeutic response, in the present study we aimed to elucidate the molecular mechanisms of Selinexor in DLBCL and evaluate its therapeutic impact in a real-world patient cohort. We hypothesized that Selinexor would exert a dual anti-lymphoma mechanism of action: activation of the p53 tumor suppressor pathway via enforced nuclear retention of p53 and reprogramming of tumor metabolic pathways, especially those involved in amino acid uptake and nucleotide synthesis. To test this hypothesis, we performed an integrative analysis combining clinical outcome assessment with transcriptomic and proteomic profiling in DLBCL models. This dual mechanistic approach aimed to elucidate the antitumor mechanisms of Selinexor and to provide a robust scientific rationale for optimizing the clinical use of XPO1 inhibitors in DLBCL.

Materials and methods

Data collection of patients

This retrospective study analyzed data from lymphoma patients treated with Selinexor at the Hematology Department of the Fourth Hospital of Hebei Medical University from June 2021 to December 2023. Eligible patients included those with a confirmed diagnosis of lymphoma who received Selinexor as part of their treatment regimen.

Clinical data collection

Comprehensive clinical data were collected from medical records, including demographic information (age and sex), clinical characteristics at diagnosis (histological type, Ann Arbor stage, presence of B symptoms), treatment history, and outcomes. Laboratory and diagnostic parameters such as International Prognostic Index (IPI) score, Eastern Cooperative Oncology Group (ECOG) performance status, levels of lactate dehydrogenase (LDH), and β2-microglobulin were also recorded. Immunohistochemical findings and genetic tests, including assessments for TP53 mutations and MYC rearrangements, were gathered to ascertain the molecular profile of the tumors.

Treatment and response evaluation

Treatment details such as Selinexor dosage, treatment combinations, and response evaluations based on the 2014 Lugano Classification were documented. Survival outcomes, including overall survival (OS) and progression-free survival (PFS), were recorded through December 2023.

Univariate and multivariate analysis of clinical data

Univariate analysis was performed using the Kaplan-Meier method to evaluate the impact of various clinical variables on OS and PFS among lymphoma patients treated with Selinexor. Significant factors from the Kaplan-Meier analysis were subsequently examined in a multivariate Cox proportional hazards model. This approach allowed for the assessment of independent prognostic factors influencing patient outcomes.

Cell culture

Human DLBCL-GCB SU-DHL6 (DHL6) cells, authenticated by Short Tandem Repeat (STR) analysis, were cultured in RPMI 1640 medium (Gibco, Grand Island, NY, USA) supplemented with 10% fetal bovine serum (FBS) and 100 U/mL penicillin and 100 mg/mL streptomycin. The cells were maintained at 37 °C in a humidified atmosphere with 5% CO₂.

Human DLBCL-GCB OCI-LY7 (LY7) cells, generously provided by Dr. Wang from Tianjin Medical University Affiliated Cancer Hospital, were grown in IMDM (Gibco, Grand Island, NY, USA) with 10% FBS and the same antibiotic supplementation, under identical incubation conditions.

Human DLBCL-ABC OCI-LY3 (LY3) cells, also verified by STR analysis, were cultured in RPMI 1640 medium supplemented with 20% FBS, 100 U/mL penicillin and 100 mg/mL streptomycin. These cells were incubated at 37 °C in a humidified 5% CO₂ atmosphere.

All cell lines were tested for Mycoplasma contamination, and confirmed to be free of Mycoplasma. The DLBCL cell lines DHL6, LY3, and LY7 were selected as established experimental models representing distinct molecular subtypes of DLBCL. DHL6 and LY7 are classified as GCB models, whereas LY3 represents non-GCB subtype according to previously reported transcriptional profiling studies. These cell lines collectively encompass key biological contexts of DLBCL and have been widely used in investigations of subtype-associated signaling and therapeutic vulnerabilities. The inclusion of multiple cell lines allowed evaluation of conserved versus context-dependent responses to XPO1 inhibition across biologically distinct DLBCL backgrounds.

Selinexor was prepared as a 100 mM stock solution in DMSO.

Cell counting kit-8 (CCK-8) assay

Cell proliferation was evaluated using the CCK-8 assay. DHL6 and LY3 cells (1 × 10⁵ cells/100 µL) were seeded in 96-well plates and exposed to varying concentrations of Selinexor (0, 10, 50, 100, 200, 300, 400, 500, 600, 800, and 1000 nM). DMSO-treated cells served as the control. After 24 and 48 h of treatment, 10 µL of CCK-8 solution (Boster, Wuhan, China) was added to each well and incubated at 37 °C for 2 h. Optical density (OD) at 450 nm was measured using a plate reader. Results represent the average of three replicates. The concentration range and IC50 value determination were based on the protocol and data reported by Deng et al. [ 7 ]. Cell viability assays were used to define treatment conditions. IC50 analyses demonstrated differential sensitivity to Selinexor, with GCB-type cell lines (DHL6 and LY7) exhibiting earlier growth inhibition than the non-GCB LY3 cells. To achieve comparable biological response states while avoiding excessive cytotoxicity, exposure durations were adjusted accordingly: GCB-type cells were treated for 24 h, whereas the non-GCB model was exposed for 48 h in subsequent transcriptomic and proteomic experiments. Exposure time was kept consistent within each molecular subtype to ensure robust subtype-specific analyses. Thus, downstream comparisons were performed under matched drug-response conditions rather than fixed exposure durations across cell lines.

Cell cycle assay

A 1 mL single-cell suspension was collected from each group after Selinexor treatment. Cells were rinsed with ice-cold PBS and fixed in 70% ethanol at 4 °C for 24 h. After two additional PBS washes, they were resuspended in 100 µL PBS and stained with 1 mL of propidium iodide (BD Biosciences, San Jose, CA, USA) at 4 °C for 30 min. Cell cycle analysis was conducted using an FC-500 flow cytometer (Beckman Coulter, Brea, CA, USA), and the data were processed using Multicycle AV software (version 275, Phoenix Flow Systems, San Diego, CA, USA). The proliferation index (PI) was calculated as follows:

Apoptosis assay

Cells were washed with cold PBS and resuspended in 100 µL of 1X binding buffer (BD Biosciences, San Jose, CA, USA). Annexin V-PE (5 µL; BD Biosciences) was added, followed by incubation on ice in the dark for 15 min. Afterward, 400 µL of 1X binding buffer and 5 µL of 7-AAD (BD Biosciences) were added, and the mixture was incubated on ice for another 15 min in the dark. Following a final PBS wash, the cells were resuspended in 1 mL PBS and analyzed for early and late apoptosis using an FC-500 flow cytometer. The apoptotic rate was calculated using EXPO 32 ADC v1.2 software (Beckman Coulter DxFLEX, Brea, CA, USA).

RNA sequencing (RNA-seq)

Selinexor was applied to DHL6 (300 nM, 24 h) and LY3 (100 nM, 48 h) cells, with untreated cells serving as controls. Each group included three replicates, containing at least 1 × 10⁶ cells per sample. RNA-seq was conducted to investigate the transcriptomic effects of Selinexor. Total RNA was extracted using Trizol Reagent (Invitrogen, Carlsbad, CA, USA) from triplicate samples of both cell lines. RNA concentrations ranged from 23.8 to 41 ng/µL, with RIN values of ~ 9.4 assessed by an Agilent Bioanalyzer 2100 (Agilent Technologies, Palo Alto, CA, USA).

Paired-end libraries were prepared using the VAHTS Stranded mRNA-seq Library Prep Kit for Illumina ® (Vazyme Biotech, Nanjing, China) according to a modified strand-specific protocol. Library construction and sequencing were performed by Shanghai Biotechnology Corporation (China). Differentially expressed genes (DEGs) were identified with a False Discovery Rate (FDR) < 0.05 and a fold-change ≥ 2. Enrichment analysis of DEGs was conducted to identify gene ontology (GO) terms and pathways from the Kyoto Encyclopedia of Genes and Genomes (KEGG) database [ 10 , 11 ].

4D label-free proteomic analysis

For proteomic profiling, DHL6 and LY3 cells were exposed to Selinexor (300 nM for 24 h and 100 nM for 48 h, respectively), alongside untreated controls. Three independent biological replicates were prepared for each condition, with no fewer than 1 × 10⁶ cells per replicate. Following centrifugation, cell pellets were washed three times with ice-cold DPBS and stored at − 80 °C. Proteins were extracted using SDS-based lysis buffer, reduced, alkylated, and digested with trypsin. Peptide mixtures were desalted and analyzed by 4D label-free LC–MS/MS using a timsTOF Pro system (Bruker Daltonics, Bremen, Germany) coupled with a nanoElute UHPLC. MS data were processed in MaxQuant (version 2.0) for peptide identification and label-free quantification. Differentially expressed proteins (fold-change ≥ 1.5, adjusted P < 0.05) were subjected to GO and KEGG enrichment analyses. Detailed procedures are provided in the Supplementary Materials.

Gene-set enrichment and pathway activity analyses (GSEA/ssGSEA and downstream statistics)

Bulk RNA-seq from DLBCL cohorts (TCGA n = 48; CTSP n = 41; NCICCR n = 481) was normalized to log2 (TPM + 1). Gene sets were from MSigDB. Analyses were performed on Sangerbox 3.0 (vip.sangerbox.com; cited per platform guidance).

GSEA: Genes were ranked by log2 fold-change between phenotypes; 1,000 permutations; gene-set size 15–500. Significance was defined by NES and FDR q -value (default threshold q < 0.25 unless stated).

ssGSEA: Single-sample enrichment scores were computed per pathway and per sample ( z -scaled when indicated) and used for downstream statistics.

Correlation scatter: Pearson’s r between SLC1A5 expression and pathway scores (Spearman’s ρ as sensitivity), plotted with OLS regression line and 95% CI; P values FDR-adjusted (Benjamini–Hochberg).

Violin plots: Samples dichotomized at the median SLC1A5 (High vs. Low); group differences in ssGSEA scores tested by the Wilcoxon rank-sum test (Welch’s t-test if normality/variance assumptions held).

Correlation matrix: Pairwise Pearson correlations among selected pathways, visualized as bubble heatmaps (bubble area ∝ |r|; color encodes sign/magnitude); FDR-adjusted.

All tests were two-sided; exact n, statistics and P values are reported in figure legends. The enrichment plots in the Results were generated directly from Sangerbox [ 12 , 13 ].

For cell line RNA-seq and proteomics data, pathway activity inference and interaction analyses were performed in R (v4.3.1) using internal scripts. Single-sample gene set enrichment analysis (ssGSEA) was conducted to compute enrichment scores for predefined hallmark and curated pathways (MSigDB c2.cp and c6.all) across each biological replicate ( n = 3 per group). To assess the interaction between p53 signaling and metabolic reprogramming, paired control and post-treatment comparisons were performed using the Wilcoxon signed-rank test. Spearman’s rank correlation coefficients ( ρ ) were calculated between pathway scores to evaluate co-regulation trends. In parallel, principal component analysis (PCA) was applied to ssGSEA score matrices to visualize treatment-induced shifts and pathway co-variation. All P values were two-sided and adjusted using the Benjamini–Hochberg method.

All transcriptomic and proteomic experiments were conducted using three independent biological replicates per condition ( n = 3), which is a commonly accepted standard in exploratory omics-based mechanistic studies. Statistical significance was determined using appropriate multiple-testing correction methods (Benjamini–Hochberg FDR control), ensuring robustness of pathway-level inferences.

Real time fluorescence quantitative reverse transcription-polymerase chain reaction (qRT-PCR)

Total RNA from the aforementioned cells was extracted using RNA-easy Isolation Reagent-R701 (Vazyme, Nanjing, China) following the manufacturer’s protocol. RNA concentration and purity were determined by spectrophotometry at 260 nm. A total of 1 µg of RNA was reverse-transcribed into cDNA using the SureScript™ First-Strand cDNA Synthesis Kit (GeneCopoeia, Rockville, MD, USA). The primer sequences used are listed in Supplementary Table S2 . Real-time fluorescence qRT-PCR was conducted on a Real-Time PCR system (ThermoFisher Scientific, Waltham, MA, USA) using reaction mixtures (20 µl) containing cDNA, specific primer pairs, and NovoStart ® SYBR qPCR SuperMix Plus (E096, Novoprotein, Suzhou, China).

Digital polymerase chain reaction (dPCR)

Total RNA extraction and cDNA synthesis followed the same protocol as described for qRT-PCR. The specific PCR primers for the genes are detailed in Supplementary Table S2 . Digital PCR was performed using the Sniper DQ24 digital PCR system (manufactured by Sniper, Suzhou, Jiangsu, China). Gene expression levels were quantified and reported as copies per µL.

Western blot (WB)

Approximately 1 × 10 6 cells were collected from T25 culture dishes (Biosharp, Hefei, China), washed with PBS (Gibco, Grand Island, NY, USA), and lysed in RIPA buffer (Biosharp, Hefei, China). The lysates were sonicated for 30 s using an ultrasonic disruptor (Wonbio, Shanghai, China) and centrifuged at 12,000 rpm (Eppendorf, Hamburg, Germany) to collect the supernatant. Protein concentrations were quantified using a BCA protein assay kit (Boster, Wuhan, China) based on the standard curve. Proteins were mixed with loading buffer, separated by SDS-PAGE, and transferred onto a PVDF membrane (Beyotime, Shanghai, China). After blocking with fast blocking solution (Biosharp, Hefei, China) for 1 h, the membrane was incubated overnight with a primary antibody, followed by washing and secondary antibody incubation. Detection was performed using an enhanced chemiluminescence system (Affinity, Changzhou, China). Band intensities were quantified using ImageJ software for grayscale analysis.

High content screening (HCS) analysis

Cells in 24-well plates were fixed with 4% paraformaldehyde (Solarbio, Beijing, China) for 15 min at room temperature, then permeabilized with 0.2% Triton X-100 (Sigma-Aldrich, St. Louis, MO, USA) in PBS for 10 min. Blocking was carried out using 10% normal goat serum (Beyotime, Shanghai, China) at 37 °C for 1 h. Cells were incubated overnight at 4 °C with primary antibodies, followed by a 1-hour incubation at 37 °C with Alexa Fluor-conjugated secondary antibodies (Dylight 488, Goat Anti-Rabbit IgG, 1:500; Abcam, Cambridge, UK). Nuclei were stained with DAPI (Beyotime, Shanghai, China), and cells were mounted with PVP-containing anti-fade medium (Beyotime, Shanghai, China) to prevent photobleaching. Fluorescence images were captured using an ImageXpress Micro High Content Screening System (Molecular Devices, San Jose, CA, USA) equipped with a 10× objective lens.

Immunohistochemistry (IHC)

Tissue samples from DLBCL patients (GCB and non-GCB subtypes) and reactive lymphadenitis (control) were formalin-fixed, paraffin-embedded, and stained for PCNA, GTSE1, CKS1B, RRM2, and TRIM22. Due to challenges in obtaining post-treatment samples, reactive lymphadenitis was used as a control.

Slides were examined under a 20× objective, and protein expression was semi-quantitatively scored using the Immunoreactive Score method, incorporating staining intensity and positive cell proportions.

Details regarding antibody dilution ratios used in IHC, WB, and HCS experiments are provided in the Supplementary Table S3 .

Statistical analysis

Data are expressed as mean ± SEM from three independent experiments. Statistical analyses were performed using GraphPad Prism (v8.0.2) for two-tailed Student’s t-tests and one-way ANOVA, while SPSS software (v20) was used for additional calculations. Statistical significance was defined as P < 0.05.

Results

Clinical outcomes

We evaluated 36 lymphoma patients treated with Selinexor, of whom 25 (69.4%) had DLBCL. The cohort was 52.8% male, and approximately 50% of patients were 60 years or older. Among DLBCL, 4 patients (16%) had double-hit DLBCL (all GCB), and 2 patients had concurrent malignancies (one lung carcinoma and one duodenal stromal tumor). High-risk features were frequent: Ann Arbor stage III–IV in 69.4%, IPI > 2 in 66.7%, elevated β2-microglobulin in 50.0%, elevated LDH in 55.6%, bulky disease (≥ 5 cm) in 77.8%, and extranodal involvement in 52.8%. In the DLBCL subset, 60% of cases were of the non-GCB subtype. Patients were heavily pretreated, with a median of 3 prior lines of therapy (Table 1 ).

By December 2023, 26 of 36 patients (72.2%) had progressed and 11 (30.6%) had died, including deaths from non-lymphoma causes such as stroke and COVID-19. Median OS and PFS for the entire cohort were both 1,314 days (~ 3.6 years). Despite adverse baseline features, Selinexor achieved an overall response rate (ORR) of 47.2% (CR 25.0% [ n = 9], PR 22.2% [ n = 8]); among DLBCL, ORR was 52% (CR 32%, PR 20%). Responses were durable in select cases, with the longest ongoing CR reaching 1,532 days (~ 4.2 years). All 25 DLBCL patients received Selinexor in combination with other agents. Common co-administered therapies included chemotherapeutic backbones (e.g., R-CHOP, R-EPOCH) and novel targeted agents such as PD-1 inhibitors, lenalidomide, or BTK inhibitors. Treatment was administered across a broad range of lines, from first-line to sixth-line settings. In addition to systemic therapy, 3 patients received radiotherapy, 2 underwent autologous stem cell transplantation (ASCT), and 1 received CAR T-cell therapy during the treatment course. Notably, no patient received Selinexor monotherapy (Fig. 1 E). Treatment-related adverse events-primarily nausea, vomiting, appetite loss, and insomnia-were generally mild and manageable with supportive care or dose adjustment.

Univariate Kaplan–Meier analyses identified adverse factors for OS in the 36-patient cohort: IPI ≥ 2, no radiotherapy, GCB subtype, elevated β2-microglobulin, and greater number of prior lines. Shorter PFS was associated with double-hit status, IPI ≥ 2, TP53 mutation (NGS), no radiotherapy, DLBCL histology (vs. other lymphoma types), worse ECOG performance status, elevated β2-microglobulin, elevated LDH, and more prior lines. In the DLBCL subgroup ( n = 25), cell-of-origin (GCB vs. non-GCB) and the number of prior therapy lines remained significant prognostic factors for both OS and PFS (Fig. 1 A-D).

On multivariate Cox regression, no independent predictors for OS were identified in the 36-patient cohort. However, GCB subtype and receiving Selinexor in the fourth-line or later setting were independent predictors of shorter PFS. In multivariate analysis restricted to the DLBCL patients, no factor remained statistically significant for OS or PFS. Overall, classical high-risk markers (higher IPI, advanced stage, elevated LDH/β2-microglobulin, extensive prior therapy) correlated with worse outcomes, whereas non-GCB patients appeared to derive greater benefit from Selinexor than GCB patients (Table 1 , Supplementary Table S1 , Supplementary Fig. S2 , Fig. S3 ).

Selinexor induces cytostatic effects with G 1 arrest and apoptosis in DLBCL cell lines

Using CCK-8 assays, Selinexor showed potent anti-proliferative activity across DLBCL cell lines, with IC50 values of 300 nM in DHL6 and LY7 and 100 nM in LY3 (Fig. 2 A, Supplementary Fig. S4 A), indicating greater sensitivity in LY3 and aligning with prior reports (Fig. 2 A) [ 7 ]. Flow cytometry confirmed robust pro-apoptotic effects in DHL6, LY3 (Fig. 2 B), and LY7 (Fig. Supplementary Fig. S4 B). Concordantly, cell-cycle profiling demonstrated a pronounced G 1 -phase accumulation with reduction of S-phase fractions versus untreated controls, consistent with G 1 /S checkpoint blockade (Fig. 2 C, Supplementary Fig. S4 C). Together, these data verify that Selinexor suppresses lymphoma cell proliferation while inducing G 1 arrest and apoptosis, providing functional evidence to support subsequent transcriptomic and proteomic interrogation of mechanism.

Global omics remodeling and nomination of two mechanistic axes

To determine whether Selinexor induces subtype-specific molecular responses in DLBCL, integrated transcriptomic and proteomic analyses were performed in GCB-type (DHL6) and non-GCB-type (LY3) models. Unsupervised heatmaps revealed coordinated expression alterations at both mRNA (Fig. 3 A, E) and protein (Fig. 3 B, F) levels, with a core gene/protein subset exhibiting concordant regulatory patterns. Notably, key components of the p53 signaling axis (GTSE1, PCNA, PCLAF, CKS1B) and metabolic regulators (SLC1A5, RRM2, TRIM22) are prominently perturbed.

Pathway enrichment analyses consistently identified two major programs involving p53 signaling associated with cell-cycle regulation, DNA repair, and apoptosis, alongside metabolic pathways related to amino-acid transport, mTORC1–glycolysis, and nucleotide biosynthesis (Fig. 3 C-D, G-H). To further investigate the interplay between the two mechanistically implicated axes—p53 signaling and metabolic reprogramming—we conducted pathway interaction analyses based on our in-house RNA-seq and proteomic datasets.

In GCB-type DHL6 cells, ssGSEA analysis revealed an inverse correlation between p53 signaling and metabolic pathway activities at the transcriptomic level (Spearman ρ = − 0.86, P < 0.05; Fig. 3 I), accompanied by opposite enrichment trends following Selinexor treatment (Fig. 3 J). Although protein-level correlations did not reach statistical significance (Supplementary Fig. S5 A), a consistent tendency toward reciprocal pathway regulation was observed.

In contrast, non-GCB LY3 cells exhibited a coordinated response pattern, with positive associations between p53 and metabolic pathway activities at both transcriptomic and proteomic levels (Fig. 3 K-L). Supplementary pathway-level analyses demonstrated concordant directional trends consistent with the primary observations despite the absence of statistical significance (Supplementary Fig. S5 B-C). While statistical significance was limited by the use of three biological replicates, consistent concurrent increases across pathways were observed, indicating a distinct subtype-associated response to XPO1 inhibition.

Collectively, these findings demonstrate subtype-dependent differences in the interaction between p53 signaling and metabolic reprogramming following Selinexor treatment, with GCB-like cells showing a tendency toward reciprocal pathway regulation and non-GCB cells displaying coordinated pathway activation.

Selinexor activates the p53 pathway to induce cell cycle arrest, DNA repair, and apoptosis in DLBCL

GSEA analysis revealed that the leading-edge genes GTSE1, CKS1B, and PCLAF were all enriched in p53 signaling-related modules, governing DNA replication, homologous recombination, mismatch repair, cell cycle regulation, and apoptosis pathways (Fig. 4 A). Notably, PCLAF (also known as PCNA clamp-associated factor) is a PCNA-binding protein that participates directly in DNA damage repair and G 1 /S progression. PCLAF acts as a cofactor with PCNA (the proliferating cell nuclear antigen) to regulate DNA polymerase binding to DNA, and silencing of PCLAF has been shown to impair DNA repair and cell cycle progression [ 14 , 15 ]. Although PCNA did not yield an optimal individual GSEA curve, its clamp partner PCLAF was a leading-edge, supporting suppression of the PCNA-dependent replication machinery (Fig. 4 B).

Consistent with the p53-axis predictions, functional assays previously detailed confirmed apoptosis and G 1 -phase accumulation with S-phase reduction in both GCB and non-GCB cells.

At the molecular level, Selinexor downregulated the expression of key cell-cycle and DNA-repair genes. Dose-dependent qRT-PCR and WB in DHL6 cells showed reduced mRNA and protein levels of GTSE1, CKS1B, PCNA, and PCLAF after Selinexor treatment (Fig. 4 C, D). A similar trend was seen in the LY7 line (Supplementary Fig. S6 A, B). The PCNA suppression observed by qPCR/WB was cross-supported by PCLAF in GSEA and by a significant PCLAF-PCNA transcriptional correlation, reinforcing replication-complex inhibition.

In the non-GCB cells, Selinexor markedly suppressed CKS1B protein levels by HCS analysis (Supplementary Fig. S6 C), indicating that the drug’s effect on cell-cycle regulators extends to non-GCB subtype as well. Finally, IHC of clinical DLBCL specimens showed high expression of GTSE1, CKS1B, and PCNA in both GCB- and non-GCB-subtype tumors, mirroring our cell-based findings (Fig. 4 E). Collectively, Selinexor activates p53 signaling in DLBCL, represses cell-cycle/DNA-repair programs, and enforces G 1 arrest and apoptosis across both GCB and non-GCB subtypes, highlighting XPO1 inhibition as a p53-dependent vulnerability.

Selinexor induces metabolic reprogramming with restricted amino acid transport and suppressed nucleotide biosynthesis

GSEA analysis identified significant downregulation of amino acid transport and nucleotide biosynthesis gene sets. In particular, components of glutamine transport and nucleotide synthesis were among the most depleted pathways. Correlation heatmaps further showed that key metabolic genes-including the glutamine transporter SLC1A5, the ribonucleotide reductase subunit RRM2, and the interferon-stimulated factor TRIM22-co-clustered with these pathways (Fig. 5 A).

Among amino acid transporters, SLC1A5 (ASCT2) emerged as a top candidate. Consistent with this, GSEA showed significant enrichment of SLC1A5-related transport pathways in the Selinexor transcriptome. Correlation analysis confirmed that SLC1A5 expression tracked with the suppressed amino acid transport network (Fig. 5 B-D). Selinexor markedly downregulated SLC1A5 mRNA in LY3 cells by dPCR validation (Fig. 5 E).

Genes involved in deoxynucleotide synthesis were also suppressed. Notably, RRM2, encoding the small subunit of ribonucleotide reductase (RNR), was significantly downregulated. In DHL6, qRT-PCR showed a marked, dose-dependent reduction in RRM2 transcripts with a concordant loss of protein by WB (Fig. 5 F, G); the GCB replicate LY7 exhibited the same directionality (Supplementary Fig. S6 A, B). Similarly, dPCR in LY3 confirmed strong downregulation of RRM2 transcripts (Fig. 5 H).

Consistent with enrichment of interferon/translation-control programs, TRIM22 was induced in LY3 after Selinexor (Fig. 5 I-K), supporting activation of a stress-adaptive module that accompanies metabolic down-shifts.

Immunohistochemistry in primary DLBCL specimens showed expression patterns of SLC1A5, RRM2, and TRIM22 consistent with the cellular data, underscoring clinical relevance of Selinexor-induced metabolic reprogramming across cell-of-origin subtypes (Fig. 5 L).

These complementary lines of evidence from transcriptomics to protein assays, which consistently support a model of XPO1 inhibition driving metabolic reprogramming in DLBCL, especially in the non-GCB subtype.

Secondary gene modulation

Beyond the two core regulatory axes, we identified consistent yet modest alterations in the expression profiles of CENPM, GMNN, CD48, and S100A11. In LY3 cells, dPCR analysis revealed significant downregulation of CENPM, GMNN, and CD48, accompanied by upregulation of S100A11 (Supplementary Fig. S7 A); These directional changes were validated in DHL6 cells via qRT-PCR, albeit with attenuated effect magnitudes (Supplementary Fig. S7 B). GSEA provided functional corroboration (Supplementary Fig. S7 C): CENPM and GMNN demonstrated enrichment in suppressed E2F/M-phase cell cycle programs, CD48 in immune activation pathways, and S100A11 in stress response/translation-interferon signatures. These observations serve as complementary evidence rather than primary mechanistic determinants.

Discussion

In a real-world cohort of 36 lymphoma patients (25 DLBCL patients), Selinexor achieved an overall response rate of ~ 52%, including ~ 32% complete responses, demonstrating activity in R/R DLBCL. High IPI, multiple prior treatment lines, TP53 mutations, and elevated β2-microglobulin all correlated with shorter survival, consistent with known prognostic factors. Patients with GCB subtype had worse survival than non-GCB patients, likely due to an overrepresentation of high-grade disease in the GCB subset (double-hit lymphoma in ~ 16% of GCB cases versus ~ 5% typically [ 16 ], along with frequent TP53 mutations). On multivariate analysis, only the number of prior treatment lines independently predicted OS/PFS; patients treated in second–third line had the best outcomes, whereas efficacy declined in later lines. Selinexor was generally well tolerated with manageable toxicity. These findings support Selinexor as an effective option in R/R DLBCL and warrant further prospective studies to refine its optimal use and long-term benefit.

Although the number of patients included in the clinical cohort was relatively limited, this reflects the current real-world accessibility of Selinexor. As a recently approved selective inhibitor of nuclear export, Selinexor is primarily used for relapsed or refractory DLBCL, and eligible patients who can provide complete clinical data and follow-up information remain relatively few. Within this context, our single-center cohort represents a relatively well-characterized and comprehensive patient set for this treatment setting. Nevertheless, ongoing efforts are being made to expand patient recruitment and establish multi-center validation to strengthen the clinical evidence base.

By inhibiting XPO1, Selinexor causes nuclear accumulation of tumor suppressors like p53, triggering a p53-dependent G 1 /S checkpoint arrest and apoptosis [ 17 ]. Consistent with this mechanism, our data revealed a p53-centric transcriptional program in DLBCL cells. Selinexor downregulated several S-phase promoters (GTSE1, PCNA, PCLAF, CKS1B), especially in GCB cells. GTSE1 normally binds and inhibits p53, so its suppression relieves p53 inhibition and promotes apoptosis [ 18 ]. PCNA and its cofactor PCLAF are essential for DNA replication and repair, their loss likely impairs DNA synthesis and imposes replication stress [ 19 ]. CKS1B, a cyclin-dependent kinase regulator that drives S-phase entry, was likewise suppressed, reinforcing the G 1 arrest. Collectively, these changes reflect enhanced p53 pathway activity leading to cell-cycle arrest and apoptosis. Interestingly, emerging evidence suggests that effective tumor suppression may also occur through cell-cycle regulatory mechanisms independent of canonical p53 signaling. Rajabi et al. demonstrated that activation of the p21 pathway can induce selective cytotoxicity in lymphoma cell models even in the absence of functional p53 activity, indicating that therapeutic responses to nuclear export inhibition may partially involve alternative checkpoint mechanisms beyond strict p53 dependence [ 20 ]. This aligns with reports that Selinexor induces apoptosis mainly in TP53-wildtype lymphoma cells. Importantly, XPO1 inhibition has shown efficacy in both GCB and non-GCB DLBCL, including high-risk “double-hit” cases without TP53 mutations [ 7 , 17 ].

In parallel, Selinexor triggered a metabolic stress program, with the strongest effects observed in the non-GCB subtype (LY3). It downregulated SLC1A5 and RRM2, two genes critical for cellular metabolism and DNA synthesis. SLC1A5 is the principal glutamine importer in lymphoma, fueling the TCA cycle and nucleotide biosynthesis, and it also activates mTORC1 signaling to drive anabolic growth [ 21 , 22 ]. Thus, loss of SLC1A5 likely starves cells of glutamine and other amino acids, effectively mimicking nutrient deprivation and dampening proliferative signals. Meanwhile, RRM2 is the rate-limiting subunit of ribonucleotide reductase required for de novo dNTP production [ 23 ]. Its suppression would deplete nucleotide pools and stall DNA replication, exacerbating replication stress. Together, these effects suggest that Selinexor not only reactivates tumor-suppressor pathways but also cuts off the cell’s metabolic fuel supply – a two-pronged attack on lymphoma viability. An interferon-inducible stress sensor, TRIM22, was also upregulated, indicating an integrated nutrient-stress response [ 24 , 25 ]. Collectively, these readouts define a Selinexor-responsive “metabolic axis” that is enriched in non-GCB relative to GCB subtype, consistent across RNA-seq, proteomics, and targeted validations. Selinexor’s dual effects illustrate how p53 reactivation and metabolic stress can synergize to suppress DLBCL cell growth. Nuclear export inhibition leads to p53 accumulation and activation, which in turn triggers a metabolic checkpoint response [ 26 , 27 ]. Activated p53 engages the energy sensor AMPK and suppresses mTORC1 signaling, effectively mimicking a nutrient-deprived state [ 28 ]. This p53–AMPK–mTOR cross-talk forms a feedback loop: p53-induced mTORC1 inhibition curtails amino acid and nucleotide biosynthesis while also reinforcing p53 stability (by preventing its dephosphorylation and MDM2-mediated degradation) [ 29 ]. At the same time, nutrient and nucleotide scarcity imposes replication stress that activates a DNA damage response, further stabilizing p53 [ 30 ].

In GCB-type DLBCL cells, integrative transcriptomic and proteomic analyses during Selinexor treatment revealed that p53 pathway activation was strongly associated with metabolic pathway suppression (ssGSEA correlation ρ = − 0.86, P < 0.05). This suggests that p53 acts as a metabolic checkpoint, redirecting resources from biosynthesis toward apoptosis—an antagonism evident at both mRNA and protein levels. In contrast, non-GCB cells exhibited parallel upregulation of p53 and metabolic pathways after Selinexor, indicating a stress-adaptive response wherein p53 activation coexists with compensatory metabolic remodeling. Although individual pathway shifts were modest, these subtype differences imply that robust p53 activation in GCB cells suppresses metabolic reprogramming, whereas in non-GCB cells it fails to repress, and may even enhance metabolic activity. This divergence likely reflects inherent differences between GCB and non-GCB biology, underscoring the importance of cell-of-origin context in p53–metabolism crosstalk.

Context-dependent molecular behavior is increasingly recognized in hematologic malignancies. For example, Galectin-9 has been shown to exert dual and even opposing functions depending on cellular and microenvironmental context [ 31 ]. This regulatory plasticity supports the biological plausibility of the subtype-dependent pathway coordination observed in our study and highlights its potential therapeutic relevance. Growing evidence further indicates that metabolic remodeling critically influences therapeutic responsiveness and immune regulation, prompting integrated strategies targeting both signaling and metabolic networks [ 32 ]. In this framework, the coordinated activation pattern observed in non-GCB models may represent a biologically permissive state for sustained response to XPO1 inhibition. While Selinexor activated p53-associated programs across DLBCL subtypes, the distinct coordination between p53 signaling and metabolic pathways suggests that treatment responses may differ according to cell-of-origin. However, prospective clinical validation will be required to determine whether molecular subtype can serve as a predictive biomarker.

In recent years, several studies have reinforced the link between p53 pathway alterations and metabolic reprogramming in DLBCL. For instance, Du et al. found in a cohort of over 2600 patients that TP53-mutated cases had worse prognosis and transcriptomic signatures of metabolic dysregulation [ 33 ]. Similarly, Zhang et al. identified TP53 as a top driver gene in a high-risk DLBCL subset, which exhibited selective upregulation of amino acid and lipid metabolism pathways [ 34 ]. Moreover, Cai et al. reported enhanced glutamine metabolism in DLBCL and demonstrated that α-KG treatment activated p53-mediated ferroptosis in cell lines, illustrating a metabolite-to-p53 feedback loop [ 35 ]. Together, these studies show that p53 pathway aberrations and metabolic rewiring are functionally intertwined in DLBCL. In this context, our data indicate that Selinexor triggers p53 pathway activation while suppressing key metabolic genes (e.g., SLC1A5, RRM2) in both GCB and non-GCB models. This dual effect highlights the rationale for a therapeutic strategy co-targeting p53 reactivation and metabolic vulnerabilities in DLBCL. By reactivating p53 while crippling metabolism, Selinexor may drive DLBCL cells into irreversible cell-cycle arrest and metabolic starvation. This outcome exemplifies a therapeutic strategy: restore p53’s tumor-suppressor function while exploiting metabolic vulnerabilities. Notably, inducing nucleotide scarcity (e.g., via RRM2 inhibition) can stabilize p53 and amplify its tumor-suppressive effects, further supporting this co-targeting approach. Coordinated targeting of the p53 pathway and cancer metabolism could thus enhance DLBCL treatment efficacy.

Beyond these two major pathways, additional molecular changes are likely to fine-tune the Selinexor response rather than fundamentally alter it. Downregulation of CENPM (a centromere/kinetochore protein) and GMNN (geminin, a replication licensing inhibitor) would further reinforce cell-cycle checkpoint control, aligning with reports that oncogenic CENPM overexpression and geminin dysregulation cause replication stress and genomic instability [ 36 – 38 ].

Meanwhile, changes in CD48 and S100A11 indicate that Selinexor also affects immune and stress-response pathways. CD48, a ligand for the 2B4 receptor on NK/T cells, modulates cytotoxic synapses; altered CD48 expression could influence immunosurveillance as metabolic stress builds. S100A11 is a calcium-binding protein involved in inflammatory and translational regulation, and its induction is consistent with interferon-related translational changes under glutamine and nucleotide deprivation. Together, these secondary signals suggest that Selinexor fine-tunes the cell-cycle machinery (via CENPM/GMNN) and the immune/stress environment (via CD48/S100A11).

[ 39 – 41 ]. Despite these findings, several limitations should be acknowledged. First, the clinical cohort included in this study was relatively small, which may restrict the generalizability of the observed therapeutic outcomes. Although the clinical observations provided supportive translational context, the limited sample size precludes definitive conclusions regarding subtype-specific treatment benefit. In addition, the mechanistic insights were primarily derived from in vitro models, and further validation in larger prospective cohorts and functional in vivo systems will be necessary to confirm the clinical relevance of these subtype-dependent responses.

In summary, this study demonstrates that XPO1 inhibition induces subtype-dependent molecular adaptations in DLBCL, characterized by distinct interactions between p53 signaling and metabolic reprogramming in GCB and non-GCB contexts. The coordinated activation pattern observed in non-GCB models suggests a potentially greater biological vulnerability to Selinexor, providing a mechanistic rationale for subtype-oriented therapeutic stratification. By integrating multi-omics profiling with functional pathway analysis, our findings highlight the importance of considering molecular heterogeneity when interpreting treatment responses. Future studies incorporating larger clinical cohorts and in vivo validation will be essential to determine whether these subtype-specific mechanisms can be translated into predictive biomarkers and optimized precision treatment strategies for patients with DLBCL.

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

Supplementary Material 1

Acknowledgements

We sincerely thank Antengene Corporation for providing the Selinexor stock solution and financial support for this study. We also extend our gratitude to the Department of Lymphoma, Tianjin Medical University Cancer Institute and Hospital, particularly Professors Hui-lai Zhang and Xian-huo Wang’s team, for generously providing the LY7 cell line. For proteomic analysis, we thank Shanghai Applied Protein Technology Co., Ltd. for technological assistance.

Author contributions

WZ was responsible for conducting the cellular experiments, writing the manuscript, and contributing to the experimental design. SC conducted the cellular experiments and contributed to data interpretation. LS and YL were responsible for data analysis, statistical interpretation, and updating reference lists. CH prepared all the figures and tables. YPL and XXW performed the immunohistochemistry experiments and contributed to interpreting the results. LL, YG, LD, LW, XW, ST, WL, and BS managed clinical patients, contributed to data collection and analysis, and provided feedback on the report. CH was responsible for designing and supervising the overall experiments and providing critical feedback on the manuscript. All authors reviewed the manuscript.

Funding

This work was supported by the Hebei Provincial Health Commission Medical Science Research Project (Grant No. 20231222).

Data availability

The RNA sequencing data generated and analyzed in this study have been deposited in the Gene Expression Omnibus (GEO) database and are accessible via GSE283388 . Proteomics data have been uploaded to the iProX database: SU-DHL6 cell line data are available under accession number IPX0010431001, and OCI-LY3 cell line data under IPX0010431002. These datasets are publicly accessible for validation and further analysis.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of the Fourth Hospital of Hebei Medical University (Approval No. 2025KY169). Written informed consent was obtained from all participants prior to their inclusion in the study. For participants under 18 years of age, consent was obtained from their legal guardians. For the cell line experiments, all procedures were conducted in accordance with institutional guidelines for laboratory research.

Consent for publication

All authors have reviewed the final manuscript and consented to its publication. Written informed consent for publication of any potentially identifiable data was obtained from the study participants.

Competing interests

The authors declare no conflict of interest.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

1. Armitage JO, Weisenburger DD. New approach to classifying non-Hodgkin’s lymphomas: clinical features of the major histologic subtypes. Non-Hodgkin’s lymphoma classification project. J Clin Oncol. 1998;16:2780–95.

2. Aghaei M, JalaliFar MA, Yousefi-Avarvand A, Bahreiny SS, Karamali N, Mansouri Z, et al. Update Prognostic Potency of Vascular Endothelial Growth Factor-A in the Acute Lymphoblastic Leukemia Landscape: A Meta-Analysis. Cancer Rep (Hoboken). 2025;8:e70351.

3. Roschewski M, Staudt LM, Wilson WH. Diffuse large B-cell lymphoma-treatment approaches in the molecular era. Nat Rev Clin Oncol. 2014;11:12–23.

4. Parikh K, Cang S, Sekhri A, Liu D. Selective inhibitors of nuclear export (SINE)--a novel class of anti-cancer agents. J Hematol Oncol. 2014;7:78.

5. Turner JG, Dawson J, Sullivan DM. Nuclear export of proteins and drug resistance in cancer. Biochem Pharmacol. 2012;83:1021–32.

6. Ben-Barouch S, Kuruvilla J. Selinexor (KPT-330) - a selective inhibitor of nuclear export (SINE): anti-tumor activity in diffuse large B-cell lymphoma (DLBCL). Expert Opin Investig Drugs. 2020;29:15–21.

7. Deng M, Zhang M, Xu-Monette ZY, Pham LV, Tzankov A, Visco C, Fang X, et al. XPO1 expression worsens the prognosis of unfavorable DLBCL that can be effectively targeted by Selinexor in the absence of mutant p53. J Hematol Oncol. 2020;13:148.

8. Kalakonda N, Maerevoet M, Cavallo F, Follows G, Goy A, Vermaat J, Casasnovas O, et al. Selinexor in patients with relapsed or refractory diffuse large B-cell lymphoma (SADAL): a single-arm, multinational, multicentre, open-label, phase 2 trial. Lancet Haematol. 2020;7:e511–22.

9. Karimi F, Aghaei M, Saki N. Impact of Genetic Polymorphisms on Treatment Outcomes of Proteasome Inhibitors and Immunomodulatory Drugs in Multiple Myeloma. Curr Treat Options Oncol. 2025;26:197–212.

10. Trapnell C, Pachter L, Salzberg SL. TopHat: discovering splice junctions with RNA-Seq. Bioinformatics. 2009:1105–11.

11. Sun L, Luo H, Bu D, Zhao G, Yu K, Zhang C, Liu Y, et al. Utilizing sequence intrinsic composition to classify protein-coding and long non-coding transcripts. Nucleic Acids Res. 2013:e166.

12. Shen W, Zhang X, Zhang Z, et al. Sangerbox: a comprehensive, interaction-friendly clinical bioinformatics analysis platform. iMeta. 2022;3:e36.

13. Chen D, Xu L, Xing H, Shen W, et al. Sangerbox 2: Enhanced functionalities and update for a comprehensive clinical bioinformatics data analysis platform. iMeta. 2024;3:e238.

14. Zhang Q, Yuan J, Liu Y, Liu X, Lv T, Zhou K, Song Y. KIAA0101 knockdown inhibits cell proliferation and induces cell cycle arrest and cell apoptosis in chronic lymphocytic leukemia cells. Ann Transl Med. 2021;9:487.

15. Jayaraman S, Pazhani J, Priyaveeraraghavan V, Raj AT, Somasundaram DB, Patil S. PCNA and Ki-67: prognostic proliferation markers for oral cancer. Oral Oncol. 2022;130:105943.

16. Riedell PA, Smith SM. Double hit and double expressors in lymphoma: definition and treatment. Cancer-Am Cancer Soc. 2018;124:4622–32.

17. Ehsan H, Robinson M, Voorhees PM, Cassetta K, Borden S, Atrash S, Bhutani M, Varga C, Pineda-Roman M, Friend R, Paul BA. Efficacy of Selinexor in relapsed/refractory multiple myeloma (RRMM) patients with del17p and other high-risk abnormalities (a retrospective single-center study). Life (Basel). 2024;14.

18. Monte M, Benetti R, Buscemi G, Sandy P, Del SG, Schneider C. The cell cycle-regulated protein human GTSE-1 controls DNA damage-induced apoptosis by affecting p53 function. J Biol Chem. 2003;278:30356–64.

19. Kelman Z. PCNA: structure, functions and interactions. Oncogene. 1997;14:629–40.

20. Rajabi L, Ebrahimdoost M, Mohammadi SA, et al. Aqueous and ethanolic extracts of Moringa oleifera leaves induce selective cytotoxicity in Raji and Jurkat cell lines by activating the P21 pathway independent of P53. Mol Biol Rep. 2025;52:102.

21. Yoo HC, Park SJ, Nam M, Kang J, Kim K, Yeo JH, Kim JK, Heo Y, Lee HS, Lee MY, Lee CW, Kang JS, Kim YH, Lee J, Choi J, Hwang GS, Bang S, Han JM. A variant of SLC1A5 is a mitochondrial glutamine transporter for metabolic reprogramming in cancer cells. Cell Metab. 2020;31:267–83.

22. Chen P, Jiang Y, Liang J, Cai J, Zhuo Y, Fan H, Yuan R, Cheng S, Zhang Y. SLC1A5 is a novel biomarker associated with ferroptosis and the tumor microenvironment: a pancancer analysis. Aging (Albany NY). 2023;15:7451–75.

23. Zhou Z, Song Q, Yang Y, Wang L, Wu Z. Comprehensive landscape of RRM2 with immune infiltration in pan-cancer. Cancers (Basel). 2022;14.

24. Vicenzi E, Poli G. The interferon-stimulated gene TRIM22: a double-edged sword in HIV-1 infection. Cytokine Growth Factor Rev. 2018;40:40–7.

25. Pagani I, Poli G, Vicenzi E. TRIM22. A multitasking antiviral factor. Cells-Basel. 2021;10.

26. Jones RG, Plas DR, Kubek S, Buzzai M, Mu J, Xu Y, Birnbaum MJ, Thompson CB. AMP-activated protein kinase induces a p53-dependent metabolic checkpoint. Mol Cell. 2005;18:283–93.

27. Zhang XD, Qin ZH, Wang J. The role of p53 in cell metabolism. Acta Pharmacol Sin. 2010;31:1208–12.

28. Budanov AV, Karin M. P53 target genes sestrin1 and sestrin2 connect genotoxic stress and mTOR signaling. Cell. 2008;134:451–60.

29. Cui D, Qu R, Liu D, Xiong X, Liang T, Zhao Y. The cross talk between p53 and mTOR pathways in response to physiological and genotoxic stresses. Front Cell Dev Biol. 2021;9:775507.

30. Cao J, Chen X, Chen L, Lu Y, Wu Y, Deng A, Pan F, Huang H, Liu Y, Li Y, Tong X, Du J. DHODH-mediated mitochondrial redox homeostasis: a novel ferroptosis regulator and promising therapeutic target. Redox Biol. 2025;85:103788.

31. Aval OS, Ahmadi A, Al-Athari H. Galectin-9: A double-edged sword in Acute Myeloid Leukemia. Ann Hematol. 2025;104:3077–90.

32. Sherafat NS, et al. Rationale of using immune checkpoint inhibitors (ICIs) and anti-angiogenic agents in cancer treatment from a molecular perspective. Clin Experimental Med. 2025;1:238.

33. Du KX, Wu YF, Hua W, et al. Identify truly high-risk TP53-mutated diffuse large B cell lymphoma patients and explore the underlying biological mechanisms. Cell Commun Signal. 2024;22:401.

34. Zhang H, Zhang Z, Li L, et al. Genomic mutation profile and metabolic alterations in diffuse large B-cell lymphoma with abdominal bulky mass. Ann Hematol. 2025;104:3359–75.

35. Cai Y, Lv L, Lu T, et al. α-KG inhibits tumor growth of diffuse large B-cell lymphoma by inducing ROS and TP53-mediated ferroptosis. Cell Death Discov. 2023;9:182.

36. Shi YX, Dai PH, Jiang YF, Wang YQ, Liu W. A pan-cancer landscape of centromere proteins in tumorigenesis and anticancer drug sensitivity. Transl Oncol. 2023;31:101658.

37. Zhang H. Regulation of DNA replication licensing and re-replication by cdt1. Int J Mol Sci. 2021;22.

38. Qi N, Niu Y, Li Z, Xiao L, Tang D, Gao W. The prognostic value and mechanisms of centromere protein m in patients with lung adenocarcinoma. Transl Cancer Res. 2022;11:3471–90.

39. Sun L, Gang X, Li Z, Zhao X, Zhou T, Zhang S, Wang G. Advances in understanding the roles of CD244 (SLAMF4) in immune regulation and associated diseases. Front Immunol. 2021;12:648182.

40. Buller CW, Mathew PA, Mathew SO. Roles of NK cell receptors 2B4 (CD244), CS1 (CD319), and LLT1 (CLEC2D) in cancer. Cancers (Basel). 2020;12.

41. Mohammed TO, Lin YR, Akter L, Weissenbruch K, Ngo KX, Zhang Y, Kodera N, Bastmeyer M, Miyanari Y, Taoka A, Franz CM. S100A11 promotes focal adhesion disassembly via myosin ii-driven contractility and piezo1-mediated Ca2 + entry. J Cell Sci. 2024;137.

Associated Data

Supplementary Materials

Supplementary Material 1

Data Availability Statement

The RNA sequencing data generated and analyzed in this study have been deposited in the Gene Expression Omnibus (GEO) database and are accessible via GSE283388 . Proteomics data have been uploaded to the iProX database: SU-DHL6 cell line data are available under accession number IPX0010431001, and OCI-LY3 cell line data under IPX0010431002. These datasets are publicly accessible for validation and further analysis.