KPT-330-mediated XPO1 inhibition impairs homologous recombination and enhances radiosensitivity in extranodal NK/T-cell lymphoma Authors: Huijie Zhou, Qiuluo Liu, Kexing Ren, Qian Luo, Chunli Yang, Tian Fang, Xi Chen, Liqun Zou Published: 2026 ABSTRACT AbstractBackgroundExtranodal NK/T-cell lymphoma (ENKTL) is a rare, aggressive lymphoma in which radioresistance remains a major cause of treatment failure in the relapsed/refractory (R/R) setting.MethodsWe analysed XPO1 expression in ENKTL and assessed its role in radiosensitization using monoallelic XPO1-knockout models and KPT-330 in vitro and in xenografts. Mechanistic studies focused on the c-Myc–RAD51/CHEK1 axis, and clinical efficacy was evaluated in two R/R patients.ResultsImmunohistochemistry showed XPO1 overexpression in primary treatment-naïve ENKTL specimens relative to nasal polyp controls, and high XPO1 expression was associated with inferior overall survival. Monoallelic XPO1 knockout impaired homologous recombination (HR) repair, establishing a DNA repair defect exploitable as a radiosensitizing vulnerability. Pharmacologic inhibition of XPO1 with KPT-330 recapitulated these HR defects and synergised with radiotherapy. Mechanistically, KPT-330 disrupts the XPO1–c-Myc–RAD51/CHEK1 axis by blocking c-Myc nuclear export, reducing c-Myc abundance and promoter occupancy at the RAD51 and CHEK1 loci, thereby impairing HR. In two heavily pretreated R/R ENKTL patients, radiotherapy rechallenge plus low-dose KPT-330 achieved one partial response and one complete response with manageable toxicity.ConclusionsXPO1 inhibition impairs HR and enhances radiosensitivity by disrupting the c-Myc–RAD51/CHEK1 axis. These findings support prospective evaluation of KPT-330–based radiosensitization in R/R ENKTL. Subject terms: Cancer therapy, Haematological cancer, Preclinical research Introduction Extranodal NK/T-cell lymphoma (ENKTL) is a rare Epstein–Barr virus (EBV)-associated lymphoma with a poor prognosis in the relapsed/refractory (R/R) setting [1]. Asparaginase-based chemotherapy combined with involved-field radiotherapy has improved outcomes in patients with localised disease [1, 2]; however, intrinsic and acquired radioresistance remain important causes of treatment failure [2]. Although radiotherapy is a cornerstone of both curative and palliative management in ENKTL, the mechanism-based radiosensitizing strategies are lacking. Radiotherapy exerts its antitumour effect largely by inducing DNA double-strand breaks (DSBs) that exceed the repair capacity of tumour cells. Among the canonical DSB repair pathways, homologous recombination (HR) is essential for high-fidelity repair and for survival after irradiation [3]. Dysregulation of HR and broader DNA damage repair (DDR) networks have been implicated in radioresistance in solid tumours and haematological malignancies, and pharmacological DDR inhibitors have been developed as potential radiosensitizers [3–5]. However, the clinical use of these inhibitors is often limited by narrow therapeutic windows and haematological or gastrointestinal toxicity [5]. XPO1 (Exportin-1, also known as CRM1) is a nuclear export receptor for numerous cargo proteins, including tumour suppressors, cell-cycle regulators and oncogenic transcription factors such as c-Myc [6]. XPO1 is overexpressed in multiple cancers and is associated with poor prognosis and therapeutic resistance [7]. Small-molecule selective inhibitors of nuclear export (SINEs), such as KPT-330 (selinexor), bind XPO1 in a slowly reversible manner, inhibit XPO1-mediated nuclear export and have demonstrated therapeutic efficacy in R/R multiple myeloma and diffuse large B-cell lymphoma (DLBCL) [8, 9]. We found that XPO1 was overexpressed in ENKTL tumour tissues, and that its inhibition suppressed HR. In this context, we aimed to investigate the role of XPO1 in HR regulation and radioresistance in ENKTL, and to evaluate XPO1 inhibition as a potential radiosensitizing strategy. Here, we identify XPO1 as a central node that links nuclear export to HR and radiosensitivity in ENKTL. We show that pharmacologic XPO1 inhibition with KPT-330 disrupts the XPO1–c-Myc–RAD51/CHEK1 axis, impairs the repair of radiation-induced DSBs, and enhances the efficacy of radiotherapy in preclinical ENKTL models and in heavily pretreated R/R ENKTL patients. These findings establish XPO1 as a critical therapeutic target and provide a strong biological rationale for evaluating KPT-330–mediated XPO1 inhibition as a radiosensitizing strategy in biomarker-defined R/R ENKTL. Methods Cell lines and cell culture SNK-6 and SNT-8 human ENKTL cell lines (kindly provided by Prof. Mingzhi Zhang, First Affiliated Hospital of Zhengzhou University) were maintained under standard conditions and routinely screened for mycoplasma contamination using a Mycoplasma PCR Detection Kit (#C0301S, Beyotime Biotechnology, Shanghai, China; last negative test September 2025). SNK-6 cultures were supplemented with recombinant human interleukin-2 (700 U/mL; Shandong Quangang Pharmaceutical Co., Ltd., Jinan, Shandong, China). CRISPR–Cas9–mediated monoallelic XPO1 knockout and validation Single-guide RNAs (sgRNAs) targeting human XPO1 were designed, cloned, and subsequently inserted into the LentiCRISPRv2 (U6-gRNA-EFS-mCherry) lentiviral vector. The sgRNA sequences were: sgRNA1-F: CACCGGTGATATTGTTGGAGCAAGT sgRNA1-R: AAACACTTGCTCCAACAATATCACC sgRNA2-F: CACCGCAGATGACGTCTTGATAATG sgRNA2-R: AAACCATTATCAAGACGTCATCTGC sgRNA3-F: CACCGGGATTATGTGAACAGAAAAG sgRNA3-R: AAACCTTTTCTGTTCACATAATCCC Monoallelic knockout clones were selected by limiting dilution and validated by western blotting, quantitative Reverse Transcription PCR (qRT‒PCR), TA cloning and Sanger sequencing. In vitro and in vivo irradiation procedures Irradiation was delivered using a Rad Source RS2000 Pro irradiator (Rad Source Technologies; 320 kV, 12.5 mA). Radiation doses were selected according to the specific objective of each experiment. In vitro, cells were exposed to 5 Gy for clonogenic survival and apoptosis assays, whereas 10 Gy was used for DNA damage and repair analyses (comet assays and γ-H2AX immunofluorescence) to generate a sufficiently robust signal for evaluating repair kinetics. In vivo, local tumour radiotherapy was delivered as a single 3 Gy fraction on day 7. Tumour-bearing mice were anaesthetised with isoflurane and positioned on a custom platform, with the body shielded by lead in such a way that only the tumour-bearing area was exposed to irradiation. The 3 Gy in vivo dose was selected on the basis of pilot experiments as a well-tolerated, nonablative dose that preserved an adequate therapeutic window for assessing radiosensitization. Cell viability assay and drug–radiation interaction analysis Cell viability was assessed using a Cell Counting Kit-8 (CCK-8; #C0038; Beyotime Biotechnology, Shanghai, China). Cells (1 × 105) were seeded into 96-well plates for the indicated treatments. At the designated time points, CCK-8 solution (10%) was added and the cells were incubated for 2 h. Absorbance was measured at 450 nm using a Multiskan FC microplate reader (Thermo Fisher Scientific, Waltham, MA, USA). For interaction analyses, SNK-6 and SNT-8 cells were exposed to KPT-330 (0–7.5 μM) and irradiation (0–12 Gy). Cell viability was assessed 72 h after treatment using a CCK-8 assay. The fraction affected (Fa) was calculated as Fa = 1 − (treated group viability/control group viability). Combination index (CI) values were calculated using the Chou–Talalay method and are displayed as isobolograms (CI < 1 synergy; CI = 1 additivity; CI > 1 antagonism). Western blotting The membranes were blocked and incubated with primary antibodies against XPO1 (#sc-74454; Santa Cruz Biotechnology; Dallas, TX, USA), c-Myc (#ab32072; Abcam, Cambridge, UK), RAD51 (#sc-8349; Santa Cruz Biotechnology; Dallas, TX, USA) and CHK1 (#sc-8408; Santa Cruz Biotechnology; Dallas, TX, USA). qRT–PCR Total RNA was extracted using TRIzol reagent (#15596026CN; Thermo Fisher Scientific; Waltham, MA, USA) and reverse transcribed with PrimeScript RT Master Mix (#RR036A; Takara, Kusatsu, Shiga, Japan). qRT‒PCR was performed on a LightCycler® 96 (Roche Diagnostics, Basel, Switzerland) using TaqMan Gene Expression Master Mix (#4369016; Thermo Fisher Scientific, Waltham, MA, USA). Primer sequences for XPO1 were: qP-xpo-F1: ATCTGACCCAACTTGTGTAGAGA qP-xpo-R1: TGGTCCTACTTGCTCCAACAAT qP-xpo-F2: CTGAAAACTCAACCGAGATGGAA qP-xpo1-R2: CAAAGACTGGAACAGGCACC qP-xpo-F3:TGGGCTGAAAACTCAACCGAG qP-xpo1-R3: TTGCTGATGCTGTAGCTCCC Chromatin immunoprecipitation (ChIP) assay ChIP was performed using a commercial ChIP kit (#P2080S; Beyotime Biotechnology, Shanghai, China) according to the manufacturer’s instructions. Briefly, the chromatin was sheared using a Scientz 08-II sonicator (Scientz, Ningbo, China) with 30-s pulses and 30-s intervals for a total of 15–20 min, which yields ~60–90% of the chromatin fragments <1 kb. Immunoprecipitation was performed using an anti-c-Myc antibody or an IgG control. Purified DNA was analysed by qPCR using primers flanking the c-Myc binding sites in the RAD51 and CHEK1 promoters. Primer sequences for RAD51 were: qP-rad51-F1: AGATGGGGTTTTGCCATCTT qP-rad51-R1: CAAAACATGTAAGGAGAGGCCA qP-rad51-F2: GATGGGGTTTTGCCATCTTGG qP-rad51-R2: ACATGTAAGGAGAGGCCAGG qP-rad51-F3: TCCTGAACTAAATGTAACCTTCCAG qP-rad51-R3: TCTTGGGCGAAGTGGAAGAC qP-rad51-F4: CCTGAACTAAATGTAACCTTCCAG qP-rad51-R4: GGATTCTTGGGCGAAGTGGA Primer sequences for CHEK1 were: qP-chek1-F1: TGGGGTCTCCCTATGTTCCC qP-chek1-R1: AGTGGCTCCTGCCTGTAATTC qP-chek1-F2: CCTCCCAAGTAGCTAGGGCT qP-chek1-R2: GAAGGCGAGGCAGGAGAATG qP-chek1-F3: GTGACTTGGGATGCTGAGGT qP-chek1-R3: GAGACAGGGTGTGTTGCTCT qP-chek1-F4: TGGGATGCTGAGGTAGGAGG qP-chek1-R4: GTGTGTTGCTCTGTCACCCA Immunofluorescence staining and γ-H2AX foci quantification A 0.5-mL cell suspension (2 × 106 cells/mL) was seeded onto coverslips in 24-well plates and centrifuged at 1200 rpm for 10 min to promote cell attachment. The supernatant was aspirated, and coverslips were air-dried at room temperature for 15–20 min to stabilise the adherent cells. The cells were fixed using 4% paraformaldehyde (#BL539A; Biosharp, Beijing, China) for 10 min, and then the cells were washed twice with phosphate-buffered saline (PBS). Permeabilization was achieved with 0.5% Triton X-100 in PBS (#X100RS; Sigma‒Aldrich, St. Louis, MO, USA) for 3 min, after which the cells were washed twice more with PBS. Nonspecific binding was blocked with 5% bovine serum albumin (BSA; #4240GR; BioFroxx, Einhausen, Germany) in PBS for 30 min. Coverslips were then incubated overnight at 4 °C with anti–γ-H2AX primary antibody (1:100; #sc-517348; Santa Cruz Biotechnology, Dallas, TX, USA). Following equilibration at 37 °C for 45 min and two PBS washes, the cells were incubated with fluorescein isothiocyanate (FITC)-conjugated AffiniPure goat anti-rabbit IgG (H + L) (1:1000; #BA1105; Boster, Wuhan, China) for 1 h at room temperature in the dark. Nuclei were counterstained with DAPI (#C1005; Beyotime Biotechnology, Shanghai, China), and fluorescence signals were captured using a laser confocal microscope (Nikon Eclipse Ti2; Tokyo, Japan). Comet assay The comet assay was performed using a Trevigen kit (#4250-050-K; Trevigen, Gaithersburg, MD, USA) according to the manufacturer’s instructions. Briefly, 2 × 105 cells were layered into low-melting agarose (at 37 °C) and immobilised onto a comet assay glass slide. After lysis and neutral electrophoresis (25 V for 25 min), the slides were dried and fixed in 70% ethanol (30 min at room temperature). After being stained with propidium iodide reagent (10 min at room temperature), the slides were imaged under a fluorescence microscope. Comet tail moments were measured using ImageJ (v1.54a). Co-immunoprecipitation and nuclear–cytoplasmic fractionation Cells (2 × 107) were lysed in NP-40 buffer (#P0013F, Beyotime Biotechnology, Shanghai, China) and subjected to immunoprecipitation with an anti-XPO1 antibody or control IgG. The complexes were then incubated with Protein A/G Magnetic Beads (#88803, Thermo Fisher Scientific, Waltham, MA, USA), and the eluted proteins were subsequently analysed by western blotting. Cytoplasmic and nuclear proteins were extracted through the Nuclear and Cytoplasmic Protein Extraction Kit (#P0027, Beyotime Biotechnology, Shanghai, China). Histone H3 (#9715, CST, Danvers, MA, USA) and GAPDH (#sc-47724, Santa Cruz Biotechnology, Dallas, TX, USA) served as nuclear/cytoplasmic markers. Colony formation assay The colony formation assay was carried out using Methylcellulose cell cloning kit (#GMS10003, GENMED SCIENTIFICS INC, USA), according to the manufacturer’s instructions. Cells (1 × 103) were seeded in 24-well plates with methylcellulose semi-solid medium at low density, and incubated for 10–14 days. Flow cytometry (apoptosis, EdU incorporation and cell-cycle analysis) Apoptosis was measured using Annexin V-allophycocyanin (APC)/DAPI Apoptosis Kit (#E-CK-A258, Elabscience, Wuhan, China) as described in the manufacturer’s manual. EdU incorporation: The cell proliferation assay was performed using the EdU-647 Cell Proliferation Assay Kit (#C0081, Beyotime Biotechnology, Shanghai, China). Briefly, cells (1.2 × 106 cells/well) were seeded in 6-well plates and subjected to the corresponding treatments. Subsequently, the cells were incubated with EdU for 2 h, fixed with 4% paraformaldehyde for 15 min, and then permeabilized with 0.3% Triton X-100 for another 10 min. The cells were then incubated with the Click Additive Solution for 30 min at room temperature in the dark. Detection was performed via the APC channel. Cell cycle distribution was monitored by flow cytometry using 7-aminoactinomycin D (7-AAD) Cell Cycle Analysis Kit (#FXP0311, 4 A Biotech, Beijing, China). Cells were ethanol-fixed, incubated with ribonuclease A (37 °C for 30 min), and stained with 7-AAD (30 min at room temperature in the dark). In vivo xenograft mouse model and therapeutic treatment Six-week-old female BALB/c nude mice (Gempharmatech, Nanjing, Jiangsu, China) were housed under specific pathogen-free conditions. The experimental unit was an individual mouse. SNK-6 or SNT-8 cells (1 × 107 in 100 μL 50% Matrigel) were injected subcutaneously into the right axilla. Once tumours reached 100–150 mm³ (day 5), mice were randomly assigned by simple randomisation to receive vehicle, KPT-330 (5 mg/kg by oral gavage on days 5, 7 and 9), single-dose local tumour irradiation (3 Gy on day 7; Rad Source RS2000 Pro, 320 kV, 12.5 mA), or the combination. Cages were balanced across groups according to baseline body weight and tumour volume, and rack positions were rotated weekly. The primary outcome measure was tumour volume over time; secondary outcome measures included body weight, serum biochemistry, haematologic parameters and histopathological assessment of major organs. Tumour volume was calculated as V = π/6 × length × width2. Tumours and body weight were measured every 2 days. Mice were euthanized on day 25. No formal a priori sample size calculation was performed. Sample sizes were determined empirically based on prior experience with the model and are comparable to those used in similar published studies. Investigators were not blinded to group allocation. No a priori exclusion criteria were applied other than predefined humane endpoints; no animals met exclusion criteria and no samples were excluded from analysis. For in vivo treatment, at least 3 mice per group were used. All procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of Sichuan University (protocol 20240131005). Immunohistochemistry (IHC) and histopathology Formalin-fixed paraffin-embedded sections (4 μm) were dewaxed, rehydrated and subjected to antigen retrieval. Sections were incubated with primary antibodies against XPO1, RAD51, or CHK1 overnight, followed by secondary antibody and 3,3′-diaminobenzidine (DAB) detection. Staining intensity was quantified as integrated optical density (IOD) using ImageJ. For H&E staining of mouse organs, sections were stained with haematoxylin and eosin. Epstein–Barr virus-encoded RNA (EBER) in situ hybridisation and interphase fluorescence in situ hybridization (FISH) EBV positivity in tumour biopsies from both patients was confirmed by EBER in situ hybridization, and circulating plasma EBV DNA was quantified by qPCR. MYC status was evaluated by interphase FISH using a dual-colour 5′/3′ MYC probe reagent (Wuhan HealthCare Biotechnology Co., Ltd., Wuhan, China) with DAPI counterstaining. Statistical analysis Data are expressed as mean ± s.d. unless otherwise specified. All reported P values were two-sided unless otherwise specified. Between-group comparisons were performed using unpaired two-tailed Student’s t-tests, and multiple-group comparisons using one- or two-way ANOVA with Tukey’s post hoc test. Before parametric testing, normality was assessed using the Shapiro–Wilk test and homogeneity of variances using Levene’s test. When these assumptions were not met, the corresponding non-parametric tests were used. Variance was similar between groups being statistically compared unless otherwise indicated. Statistical analyses were conducted with GraphPad Prism 6 (GraphPad Software, San Diego, CA, USA), and combination indices (CI) for drug–radiation interactions were calculated with CompuSyn (ComboSyn Inc., Paramus, NJ, USA). P < 0.05 was considered statistically significant. Experiments were performed in technical duplicate or triplicate and independently repeated at least three times. All experiments were performed under an IACUC-approved protocol at Sichuan University in accordance with institutional and national guidelines. Results XPO1 is overexpressed and is required for ENKTL cell proliferation and tumour growth We first evaluated XPO1 expression in primary ENKTL specimens. XPO1 expression was markedly higher in ENKTL tumour tissues than in nasal polyps (Fig. 1a, Supplementary Fig. 1A). Elevated XPO1 expression showed a trend toward inferior overall survival in patients with ENKTL (Supplementary Fig. 1B, P = 0.17). To further assess the prognostic relevance of XPO1, we analysed the GSE58445 cohort. High XPO1 expression was significantly associated with inferior overall survival (log-rank P = 0.0246; HR, 1.58; 95% CI, 1.06–2.36; n = 81 per group) (Fig. 1b). CRISPR–Cas9–mediated disruption of XPO1 in SNK-6 cells yielded monoallelic knockout clones, as confirmed by western blotting, qRT–PCR, TA cloning and Sanger sequencing (Fig. 1c, Supplementary Figs. 1C–E), with residual XPO1 expression from the remaining intact allele. Three independent sgRNAs targeting distinct XPO1 exons significantly impaired proliferation in SNK-6 cells (Supplementary Figs. 2A, B), and xenografts derived from monoallelic XPO1-KO cells exhibited attenuated tumour growth compared with those derived from control cells (Fig. 1d, Supplementary Fig. 2C). These data support XPO1 as an oncogenic driver and potential therapeutic target in ENKTL. XPO1 is overexpressed in ENKTL and genetic or pharmacologic XPO1 inhibition impairs tumour growth and HR signalling. a Representative immunohistochemical staining of XPO1 in primary treatment-naïve ENKTL biopsies (n = 17) and nasal polyps (n = 8). Scale bar, 100 μm. b Kaplan–Meier analysis of overall survival in the GSE58445 cohort stratified by median XPO1 expression. Patients with high XPO1 expression (n = 81) had inferior overall survival compared with those with low XPO1 expression (n = 81) (P = 0.0246, HR 1.58, 95% CI 1.06–2.36). c Western blot analysis of XPO1 in SNK-6 cells transduced with empty vector or three independent monoallelic XPO1-knockout (XPO1-KO) clones generated by CRISPR–Cas9; GAPDH served as a loading control. d Growth of subcutaneous SNK-6 xenografts derived from vector control or monoallelic XPO1-KO cells. e KEGG pathway enrichment analysis of differentially expressed genes in monoallelic XPO1-KO versus vector control SNK-6 cells. f Western blot analysis of XPO1, CHK1 and RAD51 in SNK-6 cells transduced with empty vector or monoallelic XPO1-KO clones; GAPDH served as a loading control. g Haematoxylin and eosin (H&E) staining and immunohistochemistry for XPO1, RAD51 and CHK1 in xenograft tumours derived from control or monoallelic XPO1-KO SNK-6 cells. Scale bar, 100 μm. h Western blot analysis of XPO1, CHK1 and RAD51 in SNK-6 and SNT-8 cells treated with KPT-330 at the indicated concentrations for 24 h; GAPDH served as a loading control. XPO1 loss is associated with HR deficiency To delineate the downstream consequences of XPO1 loss, we performed RNA sequencing on monoallelic XPO1-knockout (XPO1-KO) SNK-6 cells. Differential expression analysis revealed 1623 deregulated genes, with KEGG pathway enrichment for HR, PI3K–Akt and Ras signalling (Fig. 1e). In keeping with the HR defect, western blotting in vitro and immunohistochemistry in xenografts revealed marked downregulation of the expression of the HR effectors RAD51 and CHK1 (CHEK1 gene product; UniProt ID O14757) in XPO1-KO cells (Fig. 1f, Supplementary Fig. 2D) and tumours (Fig. 1g). These findings indicate that monoallelic XPO1-KO impairs HR and results in the establishment of a DNA repair defect that may be exploitable as a radiosensitizing vulnerability. KPT-330 phenocopies monoallelic XPO1 knockout and induces HR defects We next examined whether pharmacologic XPO1 inhibition could phenocopy the HR-deficient state induced by monoallelic XPO1 disruption. We selected KPT-330 (chemical structure shown in Supplementary Fig. 3A), a second-generation SINE compound that binds XPO1 in a slowly reversible manner, thereby modelling partial, sustained XPO1 inhibition and phenocopying the effects of monoallelic XPO1 knockout. Moreover, KPT-330 has regulatory approval in R/R haematologic malignancies and has shown improved tolerability. Accordingly, KPT-330 dose-dependently decreased cell viability in SNK-6 and SNT-8 cells (Supplementary Fig. 3B) and triggered G1-phase cell-cycle arrest at 2.5 μM, as determined by flow cytometric analysis (Supplementary Fig. 3C). KPT-330 reduced the RAD51 and CHEK1 mRNA levels (Supplementary Fig. 3D) and led to the concentration-dependent depletion of the XPO1, RAD51 and CHK1 proteins (Fig. 1h, Supplementary Fig. 3E). Collectively, these results validate KPT-330 as a pharmacologic surrogate for monoallelic XPO1-KO that recapitulates HR deficiency. KPT-330 synergises with radiation in vitro On the basis of these findings, we hypothesised that pharmacologic XPO1 inhibition with KPT-330 would enhance ENKTL radiosensitivity by suppressing HR-mediated DNA repair. In the clonogenic assays, either KPT-330 or radiation alone reduced colony formation, whereas the combination almost completely abolished the clonogenic survival of the SNK-6 and SNT-8 cells (Fig. 2a). Annexin V–based flow cytometry revealed a greater proportion of apoptotic cells with the combination therapy than with either monotherapy (Fig. 2b). Combination index analysis revealed synergistic interactions between KPT-330 (0–7.5 μM) and radiation (0–12 Gy) in SNK-6 and SNT-8 cells, with the majority of the data points falling below the additivity line (Fig. 2c). Collectively, these results indicate that XPO1 inhibition with KPT-330 potentiates the cytotoxic effects of radiation in ENKTL cells. KPT-330 synergises with radiation to suppress ENKTL cell growth in vitro. a Clonogenic survival of SNK-6 and SNT-8 cells assessed by methylcellulose colony-formation assay after treatment with vehicle (control), radiation (5 Gy), KPT-330 (2.5 μM), or the combination. Rod-shaped structures in the images represent methylcellulose in the semi-solid medium. Left, representative images of colonies grown in methylcellulose; right, quantification of surviving fraction expressed as percentage of control. Bars represent mean ± s.d. Scale bar, 200 μm. b Apoptosis in SNK-6 cells treated with vehicle, KPT-330 (2.5 μM), radiation (5 Gy), or the combination for 48 h, assessed by flow cytometry (Annexin V–APC/DAPI staining). Left, representative contour plots; right, total apoptotic cells (Annexin V–positive) expressed as percentage of all events. Bars represent mean ± s.d. c Isobolograms constructed from fraction inhibition (Fa) and Chou–Talalay combination index (CI) values. SNK-6 and SNT-8 cells were exposed to escalating doses of KPT-330 (0–7.5 μM) and radiation (0–12 Gy), and cell viability was assessed 72 h after treatment via CCK-8 assay. Fa = 1 − (treated group viability/control group viability), where viability was determined by CCK-8 assay. CI values were derived from CCK-8 assay data using the Chou–Talalay equation. Data points on the isobologram line (additive effect, CI = 1) indicate additivity; points below the line (synergistic effect, CI < 1) indicate synergy; points above the line (antagonistic effect, CI > 1) indicate antagonism. In all panels, P < 0.05 was considered statistically significant (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). KPT-330 potentiates the antitumour activity of radiation in vivo To evaluate this combination in vivo, BALB/c nude mice bearing subcutaneous xenografts were randomised to receive vehicle control, KPT-330 alone, radiation alone, or the combination of KPT-330 and radiation (Fig. 3a). In this model, compared with either monotherapy, combination therapy with KPT-330 and radiation resulted in a pronounced increase in antitumour activity (Fig. 3b). The regimen was well tolerated, with stable body weight, serum biochemistry and haematologic parameters remaining within normal limits, and no treatment-related histopathological abnormalities were detected in major organs (Supplementary Figs. 4–6). Collectively, these data indicate that KPT-330 potentiates the antitumour efficacy of radiotherapy in ENKTL xenografts with no evident systemic toxicity. KPT-330 enhances the antitumour effect of radiation in ENKTL xenografts. a Schematic of the treatment schedule for BALB/c nude mice bearing subcutaneous ENKTL xenografts. Six-week-old female mice were injected subcutaneously with SNK-6 or SNT-8 cells on day 0, and treatment was initiated when tumours reached 100–150 mm³. Mice were randomised to receive vehicle control, KPT-330 (5 mg/kg by oral gavage every other day for three doses), radiation alone (single 3 Gy fraction on day 7), or the combination of KPT-330 and radiation, and were euthanized on day 25. b Tumour growth curves and representative tumours from SNK-6 (n = 7 mice per group) and SNT-8 xenografts (n = 6 per group). Tumour volume was measured every 2 days and is presented as mean ± s.d.; statistical comparisons between groups were performed using two-way ANOVA with appropriate multiple-comparisons testing (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). KPT-330 impairs the repair of radiation-induced DSBs To further elucidate how KPT-330 enhances radiosensitivity, we examined markers of DNA DSB induction and repair. γ-H2AX immunofluorescence demonstrated comparable numbers of γ-H2AX foci across treatment groups 1 h after 10 Gy irradiation, indicating similar initial levels of DSB formation; however, compared with radiation monotherapy, KPT-330–pretreated ENKTL cells retained substantially more γ-H2AX foci at 24 h (Fig. 4a). Consistently, comet assays performed 24 h after irradiation revealed an increased comet tail moment in cells treated with KPT-330 plus radiation compared with those treated with radiation alone (Fig. 4b). Collectively, these data indicate that KPT-330 impairs the repair of radiation-induced DNA DSBs. KPT-330 impairs the repair of radiation-induced DNA damage in ENKTL cells. a γ-H2AX foci (green) in SNK-6 and SNT-8 cells treated with KPT-330 (2.5 μM) and/or irradiation (10 Gy). Cells were fixed at 1 and 24 h after treatment. Nuclei were counterstained with DAPI (blue). Representative confocal images are shown; right, quantification of the percentage of γ-H2AX–positive cells from three independent experiments (mean ± s.d.). Scale bar, 5 μm. b Comet assay of SNK-6 cells treated with vehicle, KPT-330 (2.5 μM, 24 h), irradiation (10 Gy), or the combination. The extent of DNA damage is reflected by comet tail moment. Left, representative comet images after staining with propidium iodide; right, quantification of tail moment (mean ± s.d.). Scale bar, 20 μm. In all panels, P < 0.05 was considered statistically significant (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). XPO1 inhibition disrupts the XPO1–c-Myc–RAD51/CHEK1 axis We next sought to elucidate how XPO1 inhibition leads to HR defects. MYC is overexpressed in ENKTL and is associated with poor prognosis [10, 11], in keeping with its established role in driving oncogenic programmes, including glutamine and nucleotide metabolic dependencies, that support its candidacy as a therapeutic target [12, 13]. As a transcriptional regulator, c-Myc promotes radioresistance by transcriptionally activating key HR genes, such as RAD51 and CHEK1, through promoter binding [14]. Notably, c-Myc is an export cargo of XPO1 [15, 16]. On this basis, we hypothesised that pharmacologic XPO1 inhibition would disrupt c-Myc–dependent HR and thereby enhance radiosensitivity. Consistent with this model, c-Myc is listed as an XPO1 cargo in the Nuclear Export Signal Database (NESdb), and co-immunoprecipitation demonstrated an association between c-Myc and XPO1 in ENKTL cells (Fig. 5a). After 24 h of exposure to KPT-330, ENKTL cells showed a marked reduction in c-Myc protein expression accompanied by increased γ-H2AX expression, indicating aggravated DNA damage (Fig. 5b, Supplementary Fig. 7). Nuclear–cytoplasmic fractionation revealed a transient accumulation of c-Myc in the nucleus followed by depletion, together with sustained loss from the cytoplasm, which was consistent with the blockade of nuclear export and subsequent protein destabilization (Fig. 5c). KPT-330 disrupts the XPO1–c-Myc axis and reduces c-Myc chromatin occupancy at the RAD51 and CHEK1 loci. a Co-immunoprecipitation (co-IP) of XPO1 and c-Myc in SNK-6 and SNT-8 cells. Cell lysates were immunoprecipitated with anti-XPO1 antibody or control IgG and analysed by Western blotting for XPO1 and c-Myc. b Western blot analysis of c-Myc and γ-H2AX in SNK-6 and SNT-8 cells treated with KPT-330 at the indicated concentrations for 24 h; GAPDH served as a loading control. c Nuclear–cytoplasmic fractionation showing the time-dependent redistribution of c-Myc after KPT-330 treatment in SNK-6 and SNT-8 cells. Nuclear and cytoplasmic extracts collected at the indicated time points were analysed by western blotting for c-Myc. Histone H3 and GAPDH served as nuclear and cytoplasmic markers, respectively, and as loading controls. SNK-6 cells were treated with 2.5 μM KPT-330 for 24 h before ChIP–qPCR analysis of c-Myc occupancy at the RAD51 (d) and CHEK1 (e) loci. Enrichment is shown as % input. P < 0.05 was considered statistically significant (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). To directly link XPO1 inhibition to HR suppression, we performed ChIP assays to quantify c-Myc occupancy at critical HR repair gene promoters. ChIP–qPCR analysis revealed that KPT-330 treatment of ENKTL cells resulted in a marked decrease in c-Myc occupancy at the RAD51 and CHEK1 loci (Fig. 5d, e). Collectively, these data indicate that KPT-330 disrupts XPO1-mediated c-Myc export and the transcriptional activation of HR genes, resulting in impaired HR repair and providing a mechanistic basis for radiosensitization in ENKTL. KPT-330 enhances the radiosensitivity of heavily pretreated R/R ENKTL patients with radioresistance or post-radiotherapy recurrence Finally, we explored the translational relevance of these findings in two heavily pretreated patients with R/R ENKTL harbouring XPO1 overexpression and MYC amplification (Fig. 6a). Both patients had received multiple prior lines of therapy, including radiotherapy and presented with radioresistant or post-radiotherapy relapsed disease. Clinical responses to radiotherapy rechallenge plus KPT-330 in two heavily pretreated R/R ENKTL patients with radioresistant or post-radiotherapy relapsed disease. a Immunopathologic characterisation. Left, XPO1 immunohistochemistry in tumour tissue versus nasal polyps from Case 1 and Case 2. Scale bar, 100 μm. Right, interphase FISH demonstrating MYC amplification in both cases (5′/3′ MYC probes; DAPI counterstain). Scale bar, 5 μm. b Case 1: 55-year-old man with cardiac-involved R/R ENKTL and primary radioresistance. Top, serial contrast-enhanced CT scans acquired after initial radiotherapy and after radiotherapy rechallenge plus KPT-330; green arrows indicate the tumour lesions. Bottom, treatment timeline depicting systemic regimens (P-GEMOX; anti–PD-1 plus pegaspargase and chidamide; PP-GEMOX plus chidamide), radiotherapy courses, and KPT-330. c Case 2: 59-year-old woman with R/R ENKTL and post-radiotherapy relapse. Top, serial ¹⁸F-FDG PET/CT images before and after radiotherapy rechallenge plus KPT-330, showing resolution of metabolically active disease (green circles). Bottom, treatment timeline summarising prior P-GEMOX, chidamide plus radiotherapy, anti–PD-1 plus chidamide, and KPT-330 with radiotherapy, with best response per RECIST 1.1 indicated. Month 0 indicates initiation of first-line therapy. P-GEMOX, pegaspargase-containing gemcitabine/oxaliplatin regimen; PP-GEMOX plus chidamide, P-GEMOX combined with anti–PD-1 antibody and chidamide. PD progressive disease, PR partial response, SD stable disease, CR complete response. The first case involved a 55-year-old man who experienced cardiac-involved R/R ENKTL and had progressed through three prior lines of therapy, including radiotherapy in combination with pegaspargase, chidamide, and a programmed cell death protein 1 (PD-1) inhibitor. The patient presented with rapidly progressive disease confined to the mediastinum, involving the anterior mediastinum, pericardium, pleura, and both atria and was in critical condition (New York Heart Association class IV) with atrial fibrillation and a rapid ventricular response. Emergent palliative radiotherapy rechallenge (50 Gy/25 fractions, gross mediastinal lesion volume) was administered for rapid symptom relief, together with compassionate-use low-dose KPT-330 (40 mg orally once weekly), resulting in rapid clinical improvement. Interval CT at 4 months revealed a 60% reduction in tumour burden, indicating a partial response (PR) according to RECIST 1.1 (Fig. 6b). The second case involved a 59-year-old woman who experienced disease progression after three lines of therapy. She exhibited a limited response to first-line asparaginase-based chemotherapy (P-GEMOX; best response: stable disease). Following relapse (right arm mass) after second-line therapy, the patient was completely refractory to salvage anti–PD-1 therapy combined with chidamide. Given the lack of approved treatments after progression with three lines of treatment, we rechallenged radiotherapy (50 Gy/25 fractions, the involved right arm lesion) with the compassionate addition of KPT-330 (40 mg orally once weekly). Follow-up PET/CT demonstrated a complete metabolic response, indicating a complete response (CR). The patient achieved a progression-free survival (PFS) of 6 months (Fig. 6c). Both patients showed a reduction in circulating plasma EBV DNA levels after combination treatment with KPT-330 and radiotherapy (Supplementary Fig. 8). KPT-330 was most commonly administered at dosages of 60–80 mg twice weekly, and the adverse events—predominantly nausea, vomiting, fatigue, thrombocytopenia and anaemia—were dose-dependent (Supplementary Table 1). In this study, two patients received low-dose KPT-330 (40 mg once weekly), and treatment was generally well tolerated. Both patients experienced grade 1–2 nausea and fatigue. The first patient developed grade 3 neutropenia (on a background of prior grade 4 myelosuppression), which resolved with recombinant human granulocyte colony-stimulating factor (rhG-CSF) support; no thrombocytopenia was observed. The second patient experienced grade 1 neutropenia and grade 2 leucopenia in the context of prior grade 3 myelosuppression (Supplementary Fig. 9). Liver and renal function tests remained within or near the reference ranges in both patients throughout the treatment (Supplementary Fig. 10). Discussion ENKTL remains a therapeutically challenging lymphoma subtype, particularly in the R/R setting [17]. Radiotherapy is a cornerstone of both curative and palliative management in ENKTL; however, intrinsic and acquired radioresistance frequently compromise treatment efficacy, and mechanism-based radiosensitizing strategies are lacking [18, 19]. Here, we show that XPO1 is overexpressed in ENKTL and associated with inferior overall survival. Our mechanistic and preclinical findings support a model in which XPO1 modulates HR–associated DNA repair capacity, thereby supporting XPO1 inhibition as a biologically informed radiosensitizing strategy in ENKTL. Preliminary clinical observations in molecularly selected cases (characterised by XPO1 overexpression and MYC amplification) further support the translational relevance of this approach. Over the past decade, genomic and functional studies have revealed a complex oncogenic landscape in ENKTL, with recurrent alterations affecting JAK–STAT signalling, NF-κB activation, TP53 and multiple epigenetic regulators, and these insights have provided a rationale for biologically informed approaches, including immune checkpoint blockade and epigenetic therapy [17, 18, 20, 21]. These efforts have focused primarily on proliferative signalling, immune escape and epigenetic reprogramming. Emerging evidence supports biological tractability of DDR pathways in ENKTL, including links between checkpoint dysregulation, genomic instability and treatment resistance [22, 23]. Our transcriptomic analysis of monoallelic XPO1-knockout cells revealed disruption of the HR pathway, suggesting that XPO1 contributes to the maintenance of HR competence in ENKTL. In this context, XPO1 inhibition appears to induce an HR-deficient state that can be therapeutically exploited to enhance radiosensitivity. Therefore, our study defines a distinct and clinically relevant therapeutic axis linking nuclear export dependency to DNA repair competence and radiation response. This study indicates that XPO1 is not merely overexpressed as a transport receptor in ENKTL but also functions as a regulator of HR competence and radiation tolerance. Prior studies have shown that XPO1 inhibition exerts antitumour activity in other T-cell lymphoma contexts, mainly through growth arrest and apoptosis-related mechanisms [24]. More broadly, across solid and haematologic malignancies, the antitumour effects of XPO1 inhibition have most often been attributed to restoration of nuclear tumour-suppressive programmes and suppression of pro-survival signalling, culminating in cell-cycle arrest and apoptosis [7]. Our findings extend this by suggesting that, in ENKTL, XPO1 inhibition appears to create a therapeutically exploitable DNA repair vulnerability through the disruption of a c-Myc–dependent transcriptional programme that sustains RAD51 and CHEK1 expression. This interpretation is consistent with broader evidence that MYC can transcriptionally regulate DNA damage checkpoint and repair programmes and thereby promote radiotolerance [25, 26]. In this context, the radiosensitizing effect of KPT-330 can be interpreted as a consequence of impaired recovery from radiation-induced DSBs, thereby shifting ENKTL cells from damage tolerance towards persistent genotoxic stress and cell death. This model provides a mechanistic framework that links nuclear export, oncogenic transcriptional control and DDR in ENKTL. These observations also have translational implications. Our results suggest that XPO1 inhibition may represent an orthogonal therapeutic strategy by targeting a non-metabolic therapeutic axis that is distinct from established asparaginase-based approaches and directly relevant to radioresistance. Although based on only two illustrative cases, the co-occurrence of XPO1 overexpression and MYC amplification in both responders raises the hypothesis that tumours dependent on this axis may be particularly susceptible to XPO1-directed radiosensitization. Taken together, our study supports the clinical evaluation of KPT-330 plus radiotherapy in patients with biomarker-selected ENKTL and provides a rationale for future work to determine whether XPO1/MYC co-alteration can serve as a predictive marker of response. Compared with direct inhibitors of DDR kinases, which may enhance the effect of radiation at the cost of narrowing the therapeutic window between tumour and normal tissue [27], XPO1 inhibition may offer a more clinically tractable strategy by targeting upstream regulatory dependency rather than directly blocking core repair enzymes. This potential advantage may be particularly relevant for KPT-330, which binds XPO1 in a slowly reversible manner [6], thereby preserving partial nuclear export and potentially limiting off-target toxicity [6, 7]. Preclinical evidence further suggests that compared with conventional DDR-targeting radiosensitizers, KPT-330 does not impair protein synthesis in nonmalignant cells [6], raising the possibility of a more favourable therapeutic window. Such a profile could be especially relevant in clinically fragile patients requiring repeated or palliative radiotherapy, although this hypothesis will require formal prospective validation. Our study has several limitations. The clinical observations are restricted to two patients and should be interpreted as hypothesis-generating rather than definitive evidence of efficacy. Larger, prospective studies are needed to define the safety, optimal dosing and antitumour activity of KPT-330–radiotherapy combinations in broader ENKTL populations and to determine whether XPO1/MYC co-alteration or HR-related expression signatures can serve as predictive biomarkers of response. In summary, our findings support XPO1 as a biologically relevant therapeutic target linked to HR competence and radiosensitivity in ENKTL. Further prospective studies are needed to establish the clinical utility of XPO1 inhibition as a radiosensitizing strategy in R/R ENKTL and to optimise patient selection. Supplementary information Supplementary Figures and Tables Acknowledgements We thank Prof. Mingzhi Zhang of Zhengzhou University First Affiliated Hospital for kindly gifting the ENKTL cell lines. We also thank Antengene Corporation Co., Ltd. for providing KPT-330. Author contributions HJZ, QLL, KXR designed and conducted experiments and analysed the data. LQZ designed the project and experiments. HJZ, QLL, KXR, QL, CLY, XC, TF performed the experiments. HJZ, QLL, KXR analysed the data and prepared the figures. HJZ wrote the paper, with input from all the authors. Data availability Additional data related to this paper may be requested from authors. Competing interests The authors declare no competing interests. Ethics approval All experiments were performed in accordance with an IACUC-approved protocol at Sichuan University and other relevant guidelines and regulations (2024-01-31, 20240131005). All procedures involving human participants, human data and human tissue were conducted in accordance with the Declaration of Helsinki and were approved by the Ethics Committee of West China Hospital, Sichuan University (2025-04-24, 2025/875). Patient consent The privacy rights of human subjects have been observed and all patients provided written informed consent. Declaration of generative AI and AI-assisted technologies in the writing process During the preparation of this work the authors used ChatGPT in order to improve language. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. These authors contributed equally: Huijie Zhou, Qiuluo Liu, Kexing Ren. Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41416-026-03492-0. References Associated Data Data Availability Statement Additional data related to this paper may be requested from authors.