Accelerating tumor evolution and enhancing immunotherapy efficacy in lung adenocarcinoma based on EXO1 inhibition

Introduction

Immune checkpoint inhibitors (ICIs) have revolutionized lung cancer therapy (1,2). Tumor mutational burden (TMB) quantified by whole-exome sequencing (WES) has emerged as a biomarker for predicting programmed death receptor 1/programmed death-ligand 1 (PD-1/PD-L1) blockade efficacy (3,4). Elevated TMB correlates with increased generation of tumor-specific neoantigens, which are presented by tumor cells and enhance immune recognition and cytotoxicity, ultimately improving immunotherapy responses (5). Dysregulation of cellular DNA damage repair (DDR) mechanisms is intrinsically linked to accumulation of somatic mutations. Mismatch repair (MMR) deficiency induces hypermutator phenotypes characterized by microsatellite instability (MSI) and elevated TMB in colorectal cancers, rendering tumors exquisitely sensitive to ICIs (6). Analogous MMR defects are infrequent in lung cancers but are still associated with durable ICI benefits (7). Homologous recombination (HR) deficiency, exemplified by BRCA1/2 loss in breast cancer, promotes insertions/deletions (indels), copy-number variations, and large-scale genomic rearrangements (8). Emerging evidence suggests that HR deficiency-mediated genomic instability in lung cancers may correlate with responsiveness to neoadjuvant immunotherapy (9). Although DDR defects broadly associate with increased TMB and immunotherapy outcomes in lung malignancies, the mechanistic underpinnings remain elusive.

Our preliminary investigations identified exonuclease 1 (EXO1) as a DDR-related molecule associated with genomic instability in lung adenocarcinoma (LUAD), where its overexpression portends poor prognosis. EXO1, a 5’-3’ exonuclease, serves as a mediator of HR and MMR (10). Its oncogenic upregulation has been documented across diverse malignancies, including Lynch syndrome (11) and breast (12), oral (13), lung (14), and hepatocellular cancers (15). In prostate cancer models, EXO1 overexpression accelerates tumor progression via P53 suppression, alleviating transcriptional repression of sterol regulatory element-binding protein 1 (SREBP1) to drive lipid accumulation (16). Nevertheless, the contextual role of EXO1 in LUAD remains enigmatic. Despite growing recognition of genomic instability as an immunogenic catalyst, the immunological consequences of EXO1-mediated DNA repair fidelity—given the dual HR/MMR functionality of EXO1—remain unexplored. We hypothesize that EXO1 ablation induces a therapeutically exploitable “controlled genomic chaos” state, accelerating error-prone repair to amplify tumor immunogenicity while maintaining proliferative viability, based on oncological findings from mouse model. With multidimensional interrogation of DNA repair kinetics, clonal evolution, and immune microenvironment remodeling, this study mechanistically links DDR pathway bias to neoantigen-driven T-cell activation, establishing EXO1 suppression as a potential strategy to potentiate checkpoint immunotherapy. We present this article in accordance with the ARRIVE reporting checklist (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-1-1357/rc).

Methods

Bioinformatics analysis of public data

TCGA-LUAD data download

We downloaded WES, RNA-seq, and DNA methylation data from The Cancer Genome Atlas (TCGA) LUAD datasets. Clinical parameters encompassing demographic characteristics (age, gender), tumor staging, and survival outcomes were retrieved to support subsequent integrative analyses.

Multi-omics integration via Multi-Omics VI Clustering and Subtyping (MOVICS)

Integrative clustering of WES, DNA methylation, and RNA-seq datasets was performed using the MOVICS algorithm. A unified feature matrix integrating mutation status, mean methylation values, and maximal expression values served as input for MOVICS to derive multi-omics-based LUAD subtypes.

Machine learning modeling

A random forest (RF) classifier was constructed via the randomForest package (ntree =2,000) using RNA expression, methylation, and mutations, with performance evaluated by 10-fold cross-validation. The top 30 feature genes were identified by mean decreased accuracy ranking. Support vector machine-recursive feature elimination implemented in caret selected optimal feature subsets (step size =1.5, 10-fold cross-validation), yielding 30 core discriminative genes.

Survival analysis, least absolute shrinkage and selection operator (LASSO) regression, and stepwise Cox modeling

Survival data of TCGA-LUAD patients were merged with DDR-related gene expression profiles to generate an integrated dataset for prognostic modeling. Univariate Cox proportional hazards regression was performed using the coxph function to assess associations between individual gene expression levels and overall survival, with statistical significance defined at P<0.05.

Cell culture

Lewis lung carcinoma cell (LLC), LLC-Ovalbumin (OVA), B16-F10, and HEK-293T cells were bought from the American Type Culture Collection (ATCC) and cultured in DMEM (Gibco) supplemented with 10% FBS (Gibco), 100 U/mL penicillin (Gibco), and 100 µg/mL streptomycin (Gibco). Cells were maintained in a humidified atmosphere at 37 ℃ with 5% CO₂.

3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay (MTT) proliferation assay

Cellular proliferation was quantified by MTT assay. LLC, LLC-OVA, or B16-F10 cells were seeded in 96-well plates (2,000 cells/well, 100 µL medium/well), with blank controls containing medium and MTT reagent only. Triplicate wells were analyzed at 24, 48, and 72 h. Following incubation (37 ℃, 5% CO₂), 20 µL MTT solution (5 mg/mL) was added per well, and cells were incubated for 4 h. Formazan crystals were solubilized with 50 µL lysis buffer, and absorbance was measured at 562 nm using a Multiskan GO microplate reader (Thermo Scientific). Background-adjusted optical density was used to calculate proliferation rates relative to baseline measurements.

Subcutaneous tumor xenografts

All C57BL/6 mice were housed and maintained in a specific pathogen-free (SPF) animal facility. Six-week-old C57BL/6 mice were subcutaneously inoculated in the flank with 1×10⁶ LLC, B16-F10, or LLC-OVA cells suspended in 100 µL PBS. Tumor dimensions were measured using digital calipers every 3 days starting on post-inoculation day 7, with tumor volume calculated as 0.523×length×width². Mice bearing LLC or B16-F10 tumors were euthanized approximately 3 weeks post-implantation. LLC-OVA tumor-bearing mice were monitored for 4 weeks because of slower growth kinetics. For anti-PD-1 therapeutic evaluation in LLC models, tumors were harvested 2 weeks post-inoculation. Tumors and tumor-draining lymph nodes were surgically excised immediately after euthanasia.

Euthanasia: At the experimental endpoint, all C57BL/6 mice were euthanized. To ensure humane euthanasia, mice were first placed in an induction chamber and anesthetized with 3–4% isoflurane (in oxygen) until loss of consciousness was confirmed by the absence of pedal reflex (toe pinch). Immediately following confirmation of deep anesthesia, euthanasia was performed by cervical dislocation (also referred to as cervical dislocation or neck snap). This procedure was carried out swiftly and proficiently by trained personnel in accordance with the American Veterinary Medical Association (AVMA) Guidelines for the Euthanasia of Animals, which recognize cervical dislocation as an acceptable method for mice when performed correctly by skilled individuals.

Tumor processing for single-cell suspensions

Excised tumors were mechanically dissociated into 1–2 mm³ fragments and digested in RPMI-1640 medium containing 0.1 mg/mL collagenase IV (Sigma) and 0.02 mg/mL DNase I (Sigma) supplemented with 10% fetal bovine serum (FBS) at 37 ℃ for 40 min. Digested tissues were filtered through 70-µm cell strainers, centrifuged (700 ×g, 5 min), and treated with red blood cell lysis buffer (Sigma). After PBS washes, single-cell suspensions were prepared in phosphate buffered saline (PBS).

Flow cytometry analysis

Single-cell suspensions were pre-treated with anti-CD16/32 Fc block (BD Biosciences) at 4 ℃ for 15 min prior to surface marker staining using fluorochrome-conjugated antibodies (30 min, 4 ℃). For intracellular staining, cells were fixed/permeabilized using Invitrogen buffers (2 h, 4 ℃), followed by overnight incubation with antibodies against forkhead box P3 (Foxp3), granzyme B (GzmB), or interferon-γ (IFN-γ) in permeabilization buffer. Compensation controls included single-stained positive and unstained negative samples. Data acquisition was performed on a Beckman Coulter CytoFLEX instrument, with subsequent analysis using FlowJo v10.8. All antibodies were procured from BD Biosciences unless otherwise specified.

Intratumoral T-cell functional profiling

Tumor-derived single-cell suspensions (5×10⁶ cells/mL) were cultured in RPMI-1640/10% FBS medium supplemented with 1:1,000 monensin, 1:1,000 ionomycin, and 1:2,500 phorbol 12-myristate 13-acetate for 5 h (37 ℃, 5% CO₂) to induce cytokine production. Cells were then trypsinized, neutralized with complete medium, and subjected to Fc blocking followed by surface marker staining. Intracellular cytokine detection was performed postfixation/permeabilization using anti-GzmB and anti-IFN-γ antibodies, with flow cytometric quantification of cytokine-positive T cells.

In vitro tumor/T-cell coculture

LLC cells (2×10⁵/well) were plated in 12-well plates overnight. Splenic CD8⁺ T cells isolated from C57BL/6 mice using a MojoSort™ CD8⁺ T-Cell Isolation Kit (Mojoy Biotech) were activated on CD3/CD28 antibody-coated 48-well plates for 4 h prior to coculture with LLC cells at a 2:1 effector-to-target ratio. For antigen-specific interactions, OT-I–derived CD8⁺ T cells were cocultured with LLC-OVA cells under identical conditions. Cellular morphology was monitored daily, with triplicate wells maintained for experimental rigor. Post-coculture cells were trypsinized, washed, and analyzed by flow cytometry for effector molecules (IFN-γ, GzmB).

Transcriptomic profiling of murine tumor samples

Total RNA was isolated from murine tumor tissues using the MJzol Animal RNA Isolation Kit (Majorivd) following manufacturer protocols. Purification was performed with the RNAClean XP Kit (Beckman Coulter) and RNase-Free DNase Set (Qiagen). RNA integrity was assessed using an Agilent 2100 Bioanalyzer/4200 TapeStation (Agilent Technologies). Quantification and purity were determined using a Qubit 2.0 Fluorometer (Thermo Fisher Scientific) and a NanoDrop ND-2000 spectrophotometer (Thermo Fisher Scientific). Sequencing libraries were constructed and quantified using Qubit 2.0, with fragment size distribution analyzed on the Agilent 4200 TapeStation. Libraries were sequenced on the Illumina NovaSeq6000 platform (PE150 mode) to generate paired-end 150-bp reads. Raw reads underwent quality filtering using Seqtk, followed by spliced alignment to the GRCm39 genome via Hisat2 (v2.0.4) to generate BAM files. Gene expression was normalized to transcripts per million (TPM) for cross-sample comparability. Differential expression analysis was conducted using DESeq2 (Bioconductor), applying thresholds of |log2fold change| >1 and adjusted P<0.05. Enriched Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) pathways were identified with clusterProfiler using hypergeometric testing. Protein-protein interaction (PPI) networks were constructed from differentially expressed genes (DEGs) using the STRING database and visualized in Cytoscape.

Exome sequencing analysis of murine tumors

Genomic DNA was fragmented to 200–300 bp, end-repaired, and ligated with adaptors for library preparation. Hybridization-based capture using exome probes was followed by linear amplification and sequencing on the DNBSEQ platform to generate paired-end reads. Raw data were filtered via SOAPnuke to obtain clean reads, which were aligned to the murine reference genome (GRCm39) using BWA. Variant calling was performed via GATK4 HaplotypeCaller on single samples, or via Mutect2 on normal lung matched samples, with stringent hard-filtering applied. Variants were annotated using Annodb for functional impact prediction. Tumor clonality and evolutionary dynamics were analyzed through: (I) variant allele frequency (VAF) calculation via GATK’s GetPileupSummaries (VAF = mutant reads/total reads); (II) clonal architecture inference via SciClone (v0.99.0) with maximum likelihood estimation; (III) Bayesian clustering of mutations via PyClone-VI and phylogenetic optimization via CITUP (v1.0). Clonal frequency dynamics were visualized using Timescape (v1.0).

RNA isolation and quantitative reverse-transcription PCR (RT-qPCR)

RNA extraction was performed by lysing PBS-washed cell pellets in 1 mL TRIzol with 15 min incubation on ice. Following phase separation via chloroform addition (200 µL) and centrifugation (12,000 rpm, 15 min, 4 ℃), RNA was precipitated from the aqueous phase using isopropanol (30 min, ice), washed with 75% ethanol, and resuspended in DEPC-treated water. RNA purity (A260/A280>1.8) and concentration were quantified using a NanoDrop ND-2000. Reverse transcription used 1 µg total RNA with the Vazyme 5× HiScript^® III RT SuperMix kit under conditions of 50 ℃ for 15 min, followed by enzyme inactivation at 85 ℃ for 5 s. RT-qPCR reactions were conducted in 20 µL volumes containing SYBR^® Green Pro Taq HS master mix (Vazyme), gene-specific primers (0.4 µL each, 10 µM), and 2 µL cDNA template. Amplification on a QuantStudio™ 7 system included UDG pretreatment (50 ℃, 2 min), initial denaturation (95 ℃, 30 s), 40 cycles of denaturation/annealing (95 ℃, 10 s/60 ℃, 30 s), and melt curve analysis. Relative gene expression was calculated by the ΔΔCt method, with primer specificity confirmed by single-peak melt curves.

Immunofluorescence staining

LLC cells (2×10⁴/well) grown on coverslips were fixed with 4% paraformaldehyde (15 min, room temperature), permeabilized with 0.1% Triton X-100 (10 min), and blocked with 5% BSA (1 h, room temperature). Primary antibodies were applied overnight (4 ℃), followed by Alexa Fluor 594-conjugated secondary antibodies (2 h, room temperature) and DAPI counterstaining. Coverslips were mounted with antifade medium and imaged on a Zeiss LSM 880 confocal microscope. Image processing and foci quantification were performed using ZEN software.

Statistical analysis

Statistical analysis and plotting were performed using R software (v4.3.3) or GraphPad Prism 9. Unpaired t-test was used to compare two groups that conformed to the normal distribution, and ANOVA combined with Tukey multiple comparisons was used to compare multiple measures that conformed to the chi-square distribution. Significance markers were ****P<0.0001, ***P<0.001, **P<0.01, *P<0.05.

Results

High EXO1 expression associated with genomic instability and poor prognosis in LUAD

The top 10 pathways associated with TMB in the TCGA-LUAD data were concentrated in DDR-related processes, including DNA replication, HR, MMR, base excision repair, and nucleotide excision repair (Figure 1A), suggesting that DDR signaling is enhanced during periods of genomic instability. Multi-omics subtyping via unsupervised consensus clustering based on the mutation status, methylation status, and mRNA expression levels of 276 reported DDR-associated genes (17) classified the TCGA-LUAD patients into two subtypes, CS1 and CS2. The CS2 subtype exhibited elevated mutation burden and expression of DDR-related genes (Figure 1B), higher proportions of progressive disease and short survival time (log-rank P=0.026; Figure 1C,1D), and higher genomic instability compared with the CS1 subtype across all metrics evaluated, including aneuploidy score (P=2.6×10⁻¹²), fraction of genome (P=2.8×10⁻⁶), MSI sensor score (P=0.0023), and TMB (P=2.7×10⁻⁹; Figure 1E).

Figure 1 High EXO1 expression is strongly associated with genomic instability and poor prognosis in lung adenocarcinoma. (A) Analysis of lung adenocarcinoma data from TCGA revealed that DNA damage repair pathways had the highest correlation with tumor mutation burden. (B) Clustering and subtyping based on transcriptomic, methylation, and DNA mutation data allowed classification of TCGA lung adenocarcinoma data into two subtypes, CS1 and CS2. (C) Patients with the CS1 subtype were staged earlier, whereas the CS2 subtype had a higher proportion of patients with advanced disease. (D) Survival analysis showed a poorer prognosis for patients with the CS2 subtype. (E) Patients with the CS2 subtype have higher genomic instability, as reflected in aneuploidy score, fraction of genome altered, MSIsensor score, and tumor mutation load. (F) A random forest model to distinguish between CS1 and CS2 subtypes had high accuracy in both the training and the test sets (100% and 87.4%, respectively). (G) The random forest model had the lowest cross-validation error when constructed using the top 10 genes in terms of importance. (H) A support vector machine model to distinguish between CS1 and CS2 subtypes had high accuracy in both the training and the test sets (97.1% and 93.3%, respectively). (I) Construction with genes ranked in the top 100 in terms of importance led to the highest accuracy in the support vector machine model (cross-validation). (J) Bar graph depicts the top 30 genes of importance in the random forest model. (K) Bar graph depicts the top 30 genes of importance in the support vector machine model. (L) Top 30 genes with highest correlation to tumor mutation burden. (M) Venn diagram shows 15 genes that are important in both the random forest and the support vector machine models and are strongly associated with tumor mutation burden. (N) The protein-protein interaction network analysis and weight scores suggest that EXO1 is a hub gene. (O) Construction of a survival model based on DNA damage repair-related genes. (P) The relationship between λ in the LASSO model and the partial likelihood deviance. (Q) The final survival model includes only a single gene for EXO1, and survival analyses of TCGA lung adenocarcinoma data show that patients with high EXO1 expression had poor prognosis. ***, P<0.001. LASSO, least absolute shrinkage and selection operator; RF, random forest; SVM, support vector machine; TCGA, The Cancer Genome Atlas; TMB, tumor mutational burden.

Given the unique clinical and genomic manifestations exhibited by the CS2 subtype, two machine learning algorithms were employed to extract important features for classification. The training set for machine learning comprised 70% of the TCGA-LUAD samples, with the remaining 30% serving as the test set. Prediction was conducted using both sets. The RF model demonstrated 100% and 87.4% accuracy in the training and test sets, respectively, exhibiting the lowest cross-validation error when the number of included variables was 10 (Figure 1F,1G). The support vector machine model demonstrated 97.1% and 93.3% accuracy in the training and test sets, respectively, exhibiting the highest cross-validation accuracy when 100 variables were included (Figure 1H,1I). The top 30 variables for each model’s importance score were extracted separately, and the top 30 variables with the highest correlation with TMB were obtained (Figure 1J-1L). These were intersected to obtain 15 variables (all mRNA expressions of corresponding genes; Figure 1M) to construct PPI networks, which identified EXO1 as a hub gene (Figure 1N).

To investigate the prognostic significance of DDR-related genes in TCGA-LUAD patients, univariate Cox proportional hazards regression analysis initially identified 80 DDR genes associated with overall survival (Figure 1O). We implemented LASSO regression for feature dimension reduction, followed by 10-fold cross-validation with optimal penalty coefficient selection (λ=0.11), which retained 21 genes with non-zero coefficients (Figure 1O,1P). In a subsequent stepwise multivariate Cox regression analysis, EXO1 expression maintained independent prognostic value (hazard ratio =1.047, 95% confidence interval: 1.03–1.064, P=3.72×10⁻⁸). Kaplan-Meier survival analysis demonstrated reduced median survival time in patients with high EXO1 expression (> median value) compared with the low-expression group (19.9 vs. 22.7 months, log-rank P=2.82×10⁻⁵, Figure 1Q).

These findings demonstrate that EXO1 expression is associated with tumor genomic instability and plays a role in DDR-related molecular subtyping of LUAD. EXO1 was the sole independent prognostic factor correlated with survival outcomes in multivariate analysis, highlighting its potential as a critical biomarker that warrants further mechanistic investigation.

Exo1 deficiency inhibits tumor growth in immunocompetent mice by increasing terminally differentiated CD8⁺ T cells

Using the pre-designed short hairpin RNA (shRNA) library (Sigma-Aldrich), we identified two high-efficiency targeting sequences (shRNA 1-2) specific to murine Exo1 mRNA. Stable Exo1-knockdown LLC lines were established through lentiviral transduction. RT-qPCR validation confirmed 60–90% reduction of Exo1 mRNA expression in knockdown groups compared with scramble controls. In vitro proliferation assays revealed no significant divergence in growth kinetics between Exo1-knockdown and control cells over 72 h (Figure 2A). Cellular doubling rates were comparable across experimental groups (control =37.1 h; knockdown =36.5–37.1 h).

Figure 2 Reduced Exo1 expression inhibits LLC and B16-F10 cell growth in immunocompetent mice. (A) MTT assay suggests that reduced Exo1 expression does not affect the growth rate of LLC cells in vitro. (B) Tumors were induced in C57 mice by subcutaneous injection of Exo1-knockdown and control LLC cell lines. (C) Mass of Exo1-knockdown and control LLC subcutaneous tumors. (D) Proportion of CD4⁺ and CD8⁺ T-cell infiltration in subcutaneous Exo1-knockdown and control LLC tumors. (E) PD-1 and TCF-1 expression in CD8⁺ T cells in Exo1-knockdown and control LLC subcutaneous tumors. (F) Proportion of exhausted CD8⁺ T-cell infiltration in Exo1-knockdown and control LLC subcutaneous tumors. (G) Proportion of granzyme B-positive exhausted CD8⁺ T cells in Exo1-knockdown and control LLC subcutaneous tumors after in vitro culture. (H) Proportion of PD-1⁺CD39⁺ CD8⁺ T cells in Exo1-knockdown and control LLC subcutaneous tumors after in vitro culture. (I) Proportion of GzmB⁺CD39⁺ CD8⁺ T cells in Exo1-knockdown and control LLC subcutaneous tumors after in vitro culture. (J) MTT assay suggests that reduced Exo1 expression does not affect the growth of B16-F10 cells in vitro. (K) Tumors were induced in C57 mice by subcutaneous injection of Exo1-knockdown and control B16-F10 cells. (L) Mass of Exo1-knockdown and control B16-F10 subcutaneous tumors. (M) PD-1 and TCF-1 expression in CD8⁺ T cells in Exo1-knockdown and control B16-F10 subcutaneous tumors. (N) Proportion of exhausted CD8⁺ T-cell infiltration in Exo1-knockdown and control B16-F10 subcutaneous tumors. ns, no significance; *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001. LLC, Lewis lung carcinoma cell; MTT, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay; PD-1, programmed death receptor-1; TCF-1, T cell factor 1.

Exo1-knockdown (LLC-Exo1-sh1/sh2) and control (LLC-PLKO.1) LLC cells were subcutaneously implanted into the flanks of C57BL/6 mice (n=7/group, Figure 2B). At 3 weeks post-implantation, tumor mass was reduced in Exo1-deficient tumors compared with control tumors (Control vs. sh1/sh2: 0.344±0.123 vs. 0.167±0.067/0.145±0.081 g, P<0.01 for each knockdown vs. control; Figure 2C).

Infiltration of CD4⁺/CD8⁺ T cells in tumor leukocyte populations was comparable among all groups (Figure 2D). Given the critical role of CD8⁺ T-cell dynamics in antitumor immunity and ICI responsiveness, we used exhaustion-associated biomarkers to stratify the CD8⁺ T cells into four subsets, including T cell factor 1 (TCF-1)⁺ programmed death receptor-1 (PD-1)⁻ (naïve) and TCF-1⁻PD-1⁺ (terminally differentiated). Exo1-deficient tumors exhibited expansion of PD-1⁺TCF-1⁻ populations compared with control tumors (P<0.01, Figure 2E). Consistent with the TCF-1 axis findings, PD-1⁺ T-cell immunoglobulin and mucin-domain containing-3 (TIM-3)⁺ dual-positive cells (canonical exhausted CD8⁺ T cells) were expanded in Exo1-deficient tumors compared with control tumors (P<0.01; Figure 2F), corroborating enhanced T-cell exhaustion.

Although exhausted CD8⁺ T cells are considered to have diminished cytotoxic capacity (18), their intratumoral proportion was inversely correlated with tumor mass. To resolve this paradox, we subjected tumor-derived T cells to monensin-blocked intracellular staining, which revealed that >95% of PD-1⁺TIM-3⁺ cells secreted granzyme B (GzmB), indicating sustained cytotoxicity in terminally exhausted populations (Figure 2G).

Emerging evidence suggests that CD39⁺CD8⁺ T cells are tumor-specific cytotoxic lymphocytes with characteristics of exhaustion, tumor reactivity, and clonal expansion (19,20). Exo1-deficient tumors demonstrated enrichment of PD-1⁺CD39⁺ coexpressing cells compared with controls (P<0.01; Figure 2H). Flow cytometry revealed near-complete CD39 positivity within the tumor-derived PD-1⁺TIM-3⁺ populations, establishing this subset as tumor-specific exhausted effectors. Parallel detection of surface CD39 and intracellular GzmB in tumor-derived, monensin-treated T cells showed increased GzmB⁺CD39⁺ double-positive cells in the Exo1-deficient groups (P<0.0001; Figure 2I). These findings demonstrate that Exo1 suppression remodels the tumor immune microenvironment by expanding tumor-specific cytotoxic CD8⁺ T cells with dual exhaustion/effector phenotypes, ultimately driving tumor regression. Comprehensive immune profiling by flow cytometry showed comparable Treg (Foxp3⁺CD25⁺), neutrophil (CD11b⁺Ly6G⁺CD45⁺), monocyte (CD11b⁺Ly6C⁺CD45⁺), and macrophage (CD11b⁺F4/80⁺CD45⁺) infiltration between Exo1-deficient tumors and controls (Figure S1).

To determine whether the tumor-suppressive phenotype of Exo1 knockdown exhibits neoplastic lineage dependency, we constructed Exo1-knockdown B16-F10 cell lines with the same shRNA sequences. The suppression of Exo1 expression did not impair the proliferation of B16-F10 cells in vitro (Figure 2J). Stable Exo1-knockdown (B16- Exo1-sh1/sh2) and control (B16-PLKO.1) B16-F10 cells were subcutaneously implanted into the flanks of C57BL/6 mice (n=7/group; Figure 2K). Exo1 deficiency reduced the B16 tumor mass compared with control tumors (Control vs. sh1: 0.496±0.209 vs. 0.286±0.144 g, P=0.045; Control vs. sh2: 0.496±0.209 vs. 0.047±0.060 g, P<0.0001; Figure 2L). PD-1⁺TCF-1⁻ and PD-1⁺TIM-3⁺ populations of CD8⁺ T cells exhibited marked enrichment in the Exo1-deficient tumors compared with control tumors across both cancer models (P<0.05, Figure 2M,2N), highlighting the potential pan-cancer significance of Exo1 inhibition.

Exo1 deficiency enhances antitumor effects

Whole-transcriptome sequencing of LLC tumors from C57BL/6 mice identified 389 DEGs (P<0.05, |log₂FC|>1). The most significantly upregulated gene was basic leucine zipper transcription factor ATF-like 2 (Batf2) (P<0.01), an AP-1 family member known to enhance antitumor immunity through dendritic-cell differentiation and CD36-BATF2/MYB pathway regulation, with tumor-suppressive roles across gastric, lung, and hepatocellular carcinomas. Conversely, claudin 4 (Cldn4) (P<0.01) and Janus kinase and microtubule interacting protein 3 (Jakmip3) (P<0.01) showed notable downregulation. Cldn4 overexpression correlates with epithelial tumor progression (lung/gastric/hepatocellular/bladder cancers) and malignant pleural effusion recurrence, whereas Jakmip3 promotes oncogenesis via JAK-STAT pathway activation (Figure 3A). Comparative Gene Ontology pathway analysis revealed that the upregulated genes in Exo1-deficient tumors were enriched in tumoricidal pathways (Figure 3B), whereas the downregulated genes clustered within pro-tumorigenic processes (Figure 3C). These findings demonstrate that Exo1 suppression orchestrates an antitumor transcriptional program coupled with oncogenic pathway inhibition, consistent with the in vivo tumor-suppressive phenotype.

Figure 3 Reduced Exo1 expression promotes antitumor immunity. (A) Volcano plot depicts differentially expressed genes in Exo1-knockdown versus control LLC subcutaneous tumors. (B) KEGG pathway enrichment analysis of upregulated genes in the Exo1-knockdown group compared with the control group. (C) KEGG pathway enrichment analysis of downregulated genes in the Exo1-knockdown group compared with the control group. (D) Heatmap shows the expression of major antitumor immune genes in the Exo1-knockdown group versus the control group. (E) Protein-protein interaction network analysis and weight scores suggest that Ifng is the hub gene among the upregulated genes in the Exo1-knockdown group. (F) Coculture of LLC cells with CD8⁺ T cells activated by CD3 and CD28 in vitro suggests that reduced expression of Exo1 slightly promotes granzyme B secretion by CD8⁺ T cells. (G) CD8⁺ T cells activated by CD3 and CD28 fail to kill LLC cells in vitro. (H) Coculture of LLC-OVA cells with total CD8⁺ T cells collected from spleens of OT-I mice in vitro suggests that reduced expression of Exo1 slightly promotes granzyme B secretion by CD8⁺ T cells. (I) Reduced Exo1 expression did not affect in vitro killing of LLC-OVA cells by OT-I CD8⁺ T cells, suggesting that reduced Exo1 expression does not affect antigen presentation and response to CD8⁺ T cells in LLC cells. ns, no significance; *, P<0.05; ***, P<0.001; ****, P<0.0001. KEGG, Kyoto Encyclopedia of Genes and Genomes; LLC, Lewis lung carcinoma cell; OVA, ovalbumin.

Profiling of immune-related molecules across experimental cohorts revealed pan-upregulation of immunological mediators in Exo1-deficient tumors, encompassing antigen presentation machinery [murine major histocompatibility complex (MHC) components H2-Aa, H2-Eb1, H2-D1, H2-K1], costimulatory molecules (Cd80, Cd86), T cell-recruiting chemokines (Cxcl9, Cxcl10), T-cell lineage markers (Cd3e, Cd3d, Cd3g, Cd8a, Cd4), and cytotoxic effectors (Gzma, Gzmb, Prf1, Ifnb1, Ifng; Figure 3D). PPI network analysis identified IFN-γ, encoded by Ifng, as the topological hub (maximal clique centrality score =18), with its downstream targets Cxcl9, Cxcl10, and H2-Aa showing coordinated overexpression (Figure 3E). This positions IFN-γ as a master regulator of Exo1-mediated immune reprogramming, characterized by enhanced effector T-cell function.

To delineate whether Exo1 inhibition directly modulates CD8⁺ T-cell cytotoxicity, we established an in vitro coculture system employing magnetic bead-sorted splenic CD8⁺ T cells (purity >95%) from immunocompetent C57BL/6 mice. Following 12-h CD3/CD28 pre-activation, T cells were cocultured with LLC-PLKO.1 (control) or LLC- Exo1-sh1/sh2 cells. Modest enhancement of GzmB⁺CD8⁺ T cells was observed in the Exo1-deficient groups at 24 h (+9% vs. control, P<0.01; Figure 3F). However, annexin V/live-dead dual staining revealed comparable tumor apoptosis rates across groups at 48 h (P>0.05, Figure 3G). To address MHC-antigen recognition limitations in polyclonal T cells, we employed OVA-specific OT-I CD8⁺ T cells with LLC-OVA target cells. Antigen-restricted cocultures recapitulated the GzmB⁺ elevation pattern (+4%, P<0.01; Figure 3H). Marked apoptosis of LLC cells was observed at 48 h, but there was no intergroup variation in apoptosis (Figure 3I). These complementary experiments demonstrate that Exo1 deficiency induces marginal cytotoxic potentiation without altering antigen-specific tumor clearance, suggesting that its primary immunomodulatory effects occur through microenvironmental remodeling rather than direct cytolytic pathway activation.

Exo1 deficiency accelerates tumor evolution

WES of LLC tumors (n=2/group) revealed elevated somatic mutations in Exo1-deficient cohorts, with increased total single nucleotide polymorphisms (SNPs, including synonymous and missense), indels (frameshift predominant), and overall mutational load versus controls (Figure 4A-4D). VAF-based subclonal deconvolution demonstrated expanded numbers of cellular clusters in Exo1-deficient tumors (Figure 4E), with elevated numbers of clonally dominant populations (VAF >70%; Figure 4F). More clonal SNP sites were also detected in Exo1-deficient tumors (Figure 4G,4H). PyClone/citup phylogenetic reconstruction confirmed accelerated clonal evolution in Exo1-deficient tumors (Figure 4I).

Figure 4 Reduced Exo1 expression promotes mutation and accelerates clonal evolution of LLC tumors. (A) Whole-exome sequencing suggests that LLC tumors with reduced Exo1 expression harbor more total SNPs than control LLC tumors. (B) More coding SNPs (including synonymous and missense mutations) are detected in LLC tumors with reduced Exo1 expression compared with controls. (C) Whole-exome sequencing suggests that LLC tumors with reduced Exo1 expression harbor more total indels than control LLC tumors. (D) More coding indels (including frameshift mutations) are detected in LLC tumors with reduced Exo1 expression compared with controls. (E) VAF-based clonal inference of tumors using SciClone showed that reduced Exo1 expression promotes an increased number of clones. The x-axis represents the VAF, defined as the proportion of sequencing reads supporting the mutant allele out of the total reads covering that locus; the height at any point reflects the relative concentration of mutations with VAFs in that vicinity. Each peak in the plot corresponds to a cluster of mutations sharing similar VAF values, suggesting they may originate from the same tumor subclone. (F) The clonal clusters (VAF ³70%) and subclonal clusters (VAF <70%) in Exo1-knockdown (n=2) and control (n=2) tumors. (G) The clonal mutation spots (VAF ³70%) and subclonal mutation spots (VAF <70%) in Exo1-knockdown (n=2) and control (n=2) tumors. (H) The clonal mutation spots as a percentage of total spots in Exo1-knockdown (n=2) and control (n=2) tumors. (I) A tumor evolutionary tree constructed using PyClone in combination with Citup suggests that reduced Exo1 expression promotes clonal evolution of tumor cell lines. (J) A tumor evolutionary tree constructed using normal lung tissue as a matched germline control suggests that Exo1 deficiency promotes evolution within LLC cell lines and facilitates immune-mediated elimination of clonal populations in vivo. LLC, Lewis lung carcinoma cell; SNPs, single nucleotide polymorphisms; VAF, variant allele frequency.

Initial TCGA analysis revealed that elevated EXO1 expression correlated with high TMB in LUAD, while our in vitro experiments demonstrated that Exo1 knockdown in murine lung cancer cell lines led to an increased mutation load. This seemingly paradoxical observation prompted us to hypothesize that the genomic instability associated with EXO1 in clinical samples may differ qualitatively from that induced by experimental Exo1 suppression. To explore this, we performed a multivariable Cox regression analysis incorporating EXO1 expression, TMB, their interaction, and key clinical covariates (age, gender, stage, smoking status) in the TCGA-LUAD cohort. The results showed that EXO1 expression and tumor stage were independent prognostic factors, whereas TMB was not significantly associated with overall survival. Moreover, no significant interaction between EXO1 and TMB was detected, indicating that EXO1 does not influence patient prognosis by modulating TMB (Figure S2A). Next, we applied COSMIC mutational signature decomposition to WES data from TCGA-LUAD tumors stratified by EXO1 expression. Notably, the EXO1-high group exhibited a significant enrichment of Signature 4 (S4), a signature tightly linked to tobacco smoking. This finding suggests that environmental carcinogen exposure, rather than EXO1 itself, is the primary driver of the mutational landscape in these tumors (Figure S2B). Gene set enrichment analysis (GSEA) further revealed that EXO1-high tumors were enriched in hallmark gene sets related to Myc and E2F transcription factor targets, as well as the G2M checkpoint, pointing to enhanced proliferative capacity (Figure S2C). Collectively, these results indicate that in TCGA-LUAD patients, high EXO1 expression is more likely a consequence of environmental mutagenesis (e.g., smoking) that sustains tumor proliferation, fundamentally differing from the experimentally induced Exo1-knockdown scenario where loss of Exo1 directly promotes mutagenesis. This distinction reconciles the observed differences between clinical and experimental findings, highlighting the context-dependent roles of EXO1 in genomic instability and tumor progression.

To evaluate the impact of tumor cell line evolution on host anti-tumor immunity, we then performed WES on both Exo1-deficient and control in vivo tumors, along with their corresponding cell lines at the time of inoculation. Murine normal lung tissue served as the germline control. Following standard processing steps including quality control, alignment, and variant annotation, clonal evolutionary trees were reconstructed using the integrated pipeline of PyClone-VI, CITUP, and Timescape. The results demonstrated that the Exo1-deficient cell lines evolved a greater number of distinct clonal clusters and exhibited more complex branching architectures compared to their control counterparts. Notably, the Exo1-deficient in vivo tumors displayed a significant reduction in the number of detectable clonal clusters. In contrast, the clonal cluster composition and evolutionary branching patterns observed in the control in vivo tumors were conserved relative to their corresponding inoculated cell lines (Figure 4J). These findings indicate that Exo1 deficiency not only promotes evolution within LLC tumors but also potentiates the immune system’s capacity to recognize and eliminate emergent clonal populations, mechanistically linking Exo1-mediated genomic instability to antitumor immunity through mutation-driven immunoediting (21,22).

Subsequently, neoantigen prediction was performed using pVACtools on all in vitro and in vivo samples. Indeed, the total number of neoantigens and the number of unique neoantigens (after collapsing identical mutations) were largely comparable between the in vitro control cell lines (Figure S3A,S3B). However, the SH group exhibited a marked reduction in unique neoantigens in vivo compared to their in vitro counterparts (Figure S3C). Venn diagram analysis revealed that, relative to their respective in vitro tumors, the EV group lost 115 unique neoantigens in vivo, whereas the SH group lost 131 (Figure S3D). This observation suggests that, while Exo1 knockdown did not significantly increase the overall count of unique neoantigens, it promoted their immune-mediated elimination in vivo. Differential analysis between Exo1-SH and EV in vitro tumors identified 36 Exo1-SH specific neoantigens. Intersecting these with Exo1-SH in vivo tumors revealed that the majority [33] were lost in vivo, indicating that Exo1-SH specific neoantigens are predisposed to immune clearance (Figure S3E-S3G). Compared to retained neoantigens, these Exo1-SH specific lost neoantigens exhibited higher NetMHCpanEL MT Scores, suggesting they possess enhanced immunogenicity (Figure S3H). Gene structure analysis of the 33 neoantigens was performed using the MEME Suite, revealing three significantly enriched motif sequences. Notably, the majority of Exo1-SH specific lost neoantigens harbored Motif 2 (consensus sequence: AGAAAGGAAAAAAGAGAGAAAAAAAGAAAGGAAGGGAGAAAGGAAGGAAG, E-value =4.1e−58) within their promoter regions (Figure S3I), suggesting that Exo1 deficiency-induced neoantigens share common structural features. Collectively, these results indicate that Exo1 knockdown does not directly increase neoantigen quantity. Instead, the neoantigens induced by Exo1 knockdown may promote greater tumor evolutionary branching and possess heightened immunogenicity, ultimately rendering tumors more susceptible to immune clearance in vivo.

Exo1 deficiency accelerates inaccurate DSB repair by delaying HR repair and enhancing NHEJ

The functional impact of EXO1 as a mediator of HR for double-strand break (DSB) repair and MMR was mechanistically probed using zeocin-induced genotoxic stress. Acute DSB challenge (100 µg/mL×4 h) caused comparable proliferation inhibition across Exo1-deficient and control LLC cells (Figure 5A). However, when LLC cells were chronically exposed to 100 µg/mL zeocin, Exo1-deficient cells had superior proliferation compared with controls on the second and fourth days of exposure. Furthermore, the survival advantages of Exo1 deficiency became more significant with increasing doses of zeocin, revealing dose-dependent resistance to DSBs in Exo1-deficient cells (Figure 5B).

Figure 5 Reduced Exo1 expression promotes inaccurate double-strand break repair by delaying homologous recombination and accelerating non-homologous end joining. (A) The growth rates of Exo1-knockdown and control LLC cells were similar in vitro when double-strand breaks were induced by transient treatment with zeocin. (B) The growth rate of Exo1-knockdown LLC cells was higher than that of controls in vitro under continuous treatment with different concentrations of zeocin. (C) The γ-H2AX content of Exo1-knockdown LLC cells was less than that of controls when treated with different concentrations of zeocin for 2 h. (D) Western blots showed that the γ-H2AX content of Exo1-knockdown LLC cells was less than that of control cells 0, 4, and 16 h after treatment with 100 mg/mL zeocin for 2 h. (E) Immunofluorescence assay showed that the γ-H2AX content of Exo1-knockdown LLC cells was less than that of control cells 0, 4, and 16 h after treatment with 100 mg/mL zeocin for 2 h (scale bars: 4 µm). (F) Western blots showed that the γ-H2AX content of Exo1-knockdown LLC cells was less than that of control cells without treatment or early (£20 min) during treatment with 100 mg/mL zeocin. With longer treatment time, the γ-H2AX content of Exo1-knockdown LLC cells increased to the same level as that of the control cells. (G) Line graph shows changes in γ-H2AX content within Exo1-knockdown and control LLC cells. (H) Rad51 antibody was used to show the intensity of homologous recombination in LLC cells, followed by confocal fluorescence microscopy to view and count the number of foci within individual cells. Up to 80 min after the start of zeocin treatment, the number of Rad51 foci within Exo1-knockdown cells was significantly less than that within control cells. At 120 min and during the repair stage of zeocin treatment, the number of Rad51 foci within Exo1-knockdown cells persisted, whereas the foci within the control cells largely disappeared (scale bars: 4 µm). (I) Antibody to 53bp1 was used to show the strength of non-homologous end joining in LLC cells, followed by confocal fluorescence microscopy to view and count the number of foci within individual cells. The number of 53bp1 foci within Exo1-knockdown cells was significantly greater than that within control cells up to 40 min after the start of zeocin treatment; at 80, 120 min, and during the repair phase of zeocin treatment, the number of 53bp1 foci in Exo1-knockdown cells gradually decreased to essentially disappear, whereas the number of 53bp1 foci persisted in control cells (scale bars: 4 µm). *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001. LLC, Lewis lung carcinoma cell.

Levels of gamma-H2A histone family member X (γ-H2AX), a canonical marker of unrepaired DSBs, declined in Exo1-deficient LLC cells following acute zeocin exposure (100–200 µg/mL ×2 h; Figure 5C). To observe the repair kinetics in more detail, we exposed LLC cells to 100 µg/mL zeocin for 2 h and measured γ-H2AX content immediately and after 4 or 26 h to allow for DSB repair. We found that γ-H2AX levels in Exo1-deficient cells were less than those in control cells at each time point (Figure 5D). Immunofluorescence confirmed accelerated foci resolution throughout the repair windows (Figure 5E). The γ-H2AX content was also assayed at earlier time points (20, 40, and 80 min) to assess whether the Exo1-deficient cells were defective in damage detection. Although the Exo1-deficient cells had less γ-H2AX content than the control cells initially (0 and 20 min), by 80 min the intracellular γ-H2AX content of the two groups was comparable (Figure 5F). This temporal resolution of intracellular γ-H2AX levels suggests that Exo1 deficiency promotes tolerance of zeocin by accelerating DSB repair rather than by impairing damage sensing (Figure 5G).

Given the attenuated growth suppression and accelerated γ-H2AX resolution in Exo1-deficient LLC cells under chronic zeocin exposure, we hypothesized that HR and non-homologous end joining (NHEJ) dynamics were temporally modulated. High-resolution immunofluorescence tracking of Rad51 recombinase (Rad51) and Tumor suppressor p53-binding protein 1 (53BP1) marker foci across sequential repair phases [baseline, zeocin exposure (20–120 min), and post-damage recovery (4 h)] revealed distinct pathway kinetics. Control cells exhibited rapid Rad51 accumulation during acute damage (20–80 min), peaking at 80 min, whereas Exo1-deficient cells showed delayed HR activation, with maximal Rad51 recruitment at 120 min and persistence of residual foci during recovery (Figure 5H). This aligned with the canonical role of Exo1 in HR-mediated repair. Conversely, 53BP1 dynamics demonstrated baseline elevation in Exo1-deficient cells, enhanced early recruitment (20–40 min), and near disappearance during recovery, indicating accelerated NHEJ completion (Figure 5I).

To functionally validate the impact of Exo1 knockdown on DSB repair pathway usage, we then established stable LLC cell lines expressing either an RFP-tagged Exo1-shRNA plasmid or a control vector. Knockdown efficiency was confirmed by flow cytometry sorting followed by qPCR validation. Subsequently, these cells were transfected with the pDR-GFP (HR reporter) or EJ5-GFP (NHEJ reporter) plasmids, selected with puromycin, and then transiently transfected with an I-SceI-expression vector to induce site-specific DSBs. Flow cytometric analysis of GFP-positive cells revealed that Exo1 knockdown resulted in a modest reduction in HR efficiency compared to controls. In contrast, NHEJ-mediated repair was substantially enhanced in Exo1-deficient cells (Figure S4A,S4B). These quantitative findings are consistent with our earlier immunofluorescence observations. To assess whether the accelerated resolution of γ-H2AX foci upon Exo1 knockdown reflected checkpoint dysfunction rather than enhanced repair, we examined the phosphorylation of checkpoint kinase 2 (p-Chk2), a key transducer of the DNA damage checkpoint, in response to Zeocin treatment. LLC cells expressing control or Exo1-shRNA were treated with increasing concentrations of Zeocin for 24 hours, and p-Chk2 levels were analyzed by western blotting. Zeocin treatment induced p-Chk2 in a dose-dependent manner in both control and Exo1-knockdown cells, with no significant differences observed between the two groups at any concentration (Figure S4C). These results indicate that Exo1 deficiency does not impair the activation of the DNA damage checkpoint, supporting the notion that the accelerated disappearance of γ-H2AX reflects genuine repair progression rather than failure to sustain the damage signal. In conclusion, the kinetic imbalance between delayed HR and expedited NHEJ drove preferential utilization of error-prone repair, mechanistically explaining the increased indels and clonal evolution observed in Exo1-deficient tumors. This repair-pathway switching created a permissive mutational landscape that enhanced tumor immunogenicity, linking Exo1-mediated genomic instability to immune activation.

Beyond its canonical role in HR repair, EXO1 also participates in MMR. To investigate whether EXO1-driven tumor regression involves MMR pathway modulation, we genetically perturbed two upstream MMR regulators, AT-rich interactive domain-containing protein 1A (ARID1A) and mutS homolog 2 (MSH2), in LLC cells by lentiviral shRNA knockdown. Suppression of both regulators induced in vitro growth inhibition, contrasting with the proliferation-neutral phenotype of Exo1-deficient cells (Figure S5A). Subcutaneous implantation in C57BL/6 mice led to attenuated tumor growth in Arid1a/Msh2-deficient groups versus controls at 3 weeks post-inoculation (P<0.0001; Figure S5B,S5C). Flow cytometry demonstrated comparable CD4⁺/CD8⁺ T-cell infiltration across groups (Figure S5D), except for marginal CD8⁺ elevation in LLC-Arid1a-sh2 tumors (P<0.05). Crucially, all MMR-deficient tumors exhibited reduced PD-1⁺TCF-1^- and PD-1⁺TIM-3⁺ exhausted CD8⁺ T-cell subsets compared with controls—an immunophenotype diametrically opposed to that in Exo1-deficient tumors (Figure S5E,S5F). To directly assess whether Exo1 knockdown compromises MMR function, we performed MSI analysis using PCR coupled with capillary electrophoresis. A panel of murine MSI markers was examined. Msh2 knockdown induced a marked mobility shift at the D3Mit22 locus, indicative of MSI and confirming MMR dysfunction. In contrast, Exo1-deficient cells exhibited mononucleotide repeat profiles indistinguishable from those of control cells across all markers tested (Figure S6A-S6D). This mechanistic dissociation conclusively demonstrates that Exo1-mediated tumor suppression occurs through an imbalance between HR and NHEJ rather than through canonical MMR pathway inhibition.

Inhibition of Exo1 expression reverses resistance of LLC tumors to anti-PD-1 therapy

The Exo1-deficient phenotype theoretically predisposes tumors to enhanced anti-PD-1 responsiveness. To validate this hypothesis, LLC-PLKO.1 (control) and LLC- Exo1-sh cells were subcutaneously implanted in C57BL/6 mice (n=10/group), with anti-PD-1 (100 µg/dose) or PBS administered intravenously on days 10 and 13 post-inoculation following baseline tumor stratification (day 7). Longitudinal monitoring revealed attenuated tumor growth in Exo1-deficient cohorts versus controls by day 16 (Control vs. sh: 0.058±0.024 vs. 0.018±0.011, P=0.009; Figure 6A), with anti-PD-1 treatment eliciting pronounced mass reduction specifically in knockdown groups (EV vs. EV+a-PD1: 0.058±0.024 vs. 0.045±0.040, P=0.574; sh vs. sh + a-PD1: 0.018±0.011 vs. 0.004 ±0.003, P=0.026) compared with marginal effects in controls (Figure 6B,6C). Exo1-deficient tumors exhibited 77.3% mass reduction with anti-PD-1 versus 21.2% in controls, indicating synergistic effects of Exo1inhibition and anti-PD-1 therapy.

Figure 6 Reduced Exo1 expression reverses resistance of LLC cells to anti-PD-1 therapy. (A) Growth curves of Exo1-knockdown and control subcutaneous tumors treated with anti-PD-1 therapy. Anti-PD-1 treatment was applied every 3 days starting on day 10 after tumor cell injection for a total of two doses. (B) Tumors in C57 mice induced by subcutaneous injection of Exo1-knockdown and control LLC cells were treated with anti-PD-1 therapy. (C) The tumor mass of Exo1-knockdown and control subcutaneous tumors treated with anti-PD-1 therapy. (D) Proportion of CD4⁺ and CD8⁺ T-cell infiltration in Exo1-knockdown and control subcutaneous LLC tumors treated with anti-PD-1 therapy. (E) PD-1 and TCF-1 expression in CD8⁺ T cells in Exo1-knockdown and control subcutaneous LLC tumors treated with anti-PD-1 therapy. (F) Granzyme B-positive CD8⁺ T cells in Exo1-knockdown and control subcutaneous LLC tumors treated with anti-PD-1 therapy. (G) Interferon-γ-positive CD8⁺ T cells in Exo1-knockdown and control subcutaneous LLC tumors treated with anti-PD-1 therapy. ns, no significance; *, P<0.05; **, P<0.01; ***, P<0.001. LLC, Lewis lung carcinoma cell; PD-1, programmed death receptor-1; TCF-1, T cell factor 1.

Flow cytometry of Exo1-deficient versus control tumors at the 2-week therapeutic endpoint revealed distinct immune microenvironmental dynamics following anti-PD-1 checkpoint blockade. Baseline analysis demonstrated elevated CD8⁺ T-cell infiltration in Exo1-deficient tumors (P<0.01; Figure 6D), which was maintained but not amplified after anti-PD-1 treatment. Both cohorts exhibited comparable reductions in PD-1⁺TCF-1⁻ CD8⁺ T-cell frequencies after treatment (Figure 6E), although GzmB⁺ cytotoxicity remained unaltered across conditions (Figure 6F). Notably, anti-PD-1 selectively enhanced IFN-γ⁺CD8⁺ T-cell populations in Exo1-deficient tumors (P<0.05), contrasting with marginal effects in control tumors (Figure 6G). This temporal pattern of immune activation mechanistically aligns with the observed growth inhibition, suggesting that EXO1 suppression synergizes with checkpoint blockade by amplifying early-phase IFN-γ-mediated antitumor immunity.

Discussion

DDR defects constitute a pivotal driver of tumor genomic instability. MMR deficiency-induced MSI and ultra-hypermutation mediated by loss of POLE/POLD1 function have been clinically validated as biomarkers for enhanced ICI responsiveness (23,24). Approximately 50% of non-small cell lung cancer patients (132/266) harboring pathogenic DDR mutations exhibited elevated TMB and superior objective response to PD-(L)1 blockade (25), and WES of ICI-treated cohorts revealed that deleterious mutations in DDR regulators (ATM, MSH2, BRCA2, POLE) correlated with prolonged overall survival (26). However, the mechanisms linking DDR aberrations to immune microenvironment remodeling remain incompletely characterized.

We identified EXO1 as a DDR effector associated with genomic instability and poor prognosis in LUAD. EXO1 expression correlates with LUAD subtypes based on immune-infiltration patterns (27), with elevated EXO1 levels linked to lymph node metastasis, pleural invasion, and poorly differentiated histology (28,29). Our survival analysis confirmed EXO1 as an independent prognostic factor, aligning with clinical observations of its pro-tumorigenic role. Paradoxically, EXO1 suppression in human LUAD cells attenuated proliferative and migratory capacities (27,28), but in murine models it preserved in vitro growth kinetics and resulted in robust tumor suppression mediated by expanded tumor-specific cytotoxic CD8⁺ T-cell infiltrates (PD-1⁺CD39⁺GzmB⁺). This dichotomy underscores the microenvironment-dependent functionality of EXO1.

Integrated RNA-seq and coculture assays excluded direct modulation of CD8⁺ T-cell antigen recognition and cytotoxicity, but WES revealed EXO1 deficiency-driven clonal mutation expansion and accelerated tumor evolution, which are associated with CD8⁺ T-cell recruitment and sensitivity to ICIs (21,22). Temporal resolution of DDR kinetics demonstrated that EXO1 ablation confers tolerance to persistent DSBs via γ-H2AX reduction, a phenotype that is conserved in human cell models28. Immunofluorescence tracking delineated a dual repair-kinetic shift: delayed HR activation coupled with expedited NHEJ resolution, fostering error-prone repair and immunogenic mutation accrual. Crucially, MMR core-protein inhibition failed to recapitulate EXO1-deficient immune profiles, excluding canonical MMR involvement.

The coexistence of clonal mutation accumulation and enrichment of PD-1⁺TIM-3⁺ exhausted CD8⁺ T cells in Exo1-deficient tumors unveil a novel therapeutic vulnerability. In vivo validation confirmed that Exo1 ablation overcame intrinsic resistance to anti-PD-1 therapy, achieving 100% objective response versus 40% in controls, with 77.3% tumor mass reduction versus 21.2%. These findings establish EXO1 as a DDR regulator that governs HR/NHEJ equilibrium, thereby calibrating immunogenic mutation landscapes to potentiate checkpoint immunotherapy. These findings decipher the immunogenomic consequences of repair-kinetics modulation and nominate EXO1 inhibition as a precision strategy to reverse ICI resistance in DDR-proficient tumors.

Limitations

There are several limitations in this study. Although we have demonstrated the impact of EXO1 deficiency on accelerating tumor evolution from multiple molecular biology perspectives, it failed to elucidate the specific molecular signals that lead to the acceleration of NHEJ following delayed HR. Furthermore, how accelerated tumor evolution promotes the increased infiltration of antigen-specific PD1⁺ CD8 T cells and the secretion of effector molecules requires further in-depth investigation. The subcutaneous tumor model used in this study exhibits good homogeneity within groups, but it still differs from the microenvironment of primary lung tumors. Employing conditional EXO1 knockout orthotopic tumor models in future studies will enhance the reliability of the conclusions.

Conclusions

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the ARRIVE reporting checklist. Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-1-1357/rc

Data Sharing Statement: Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-1-1357/dss

Peer Review File: Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-1-1357/prf

Funding: The study was supported by grants from the National Natural Science Foundation of China (No. 92478203) and the Natural Science Foundation of Shanghai Municipality (No. 22ZR1439200).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-1-1357/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Experiments were performed under a project license (No. SINH-2024-LV-1) granted by the Animal Ethics Committee of Shanghai Institute of Nutrition and Health, in compliance with institutional ethical guidelines for the care and use of animals.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Mountzios G, Remon J, Hendriks LEL, et al. Immune-checkpoint inhibition for resectable non-small-cell lung cancer - opportunities and challenges. Nat Rev Clin Oncol 2023;20:664-77. [Crossref] [PubMed]
2. Zhou F, Qiao M, Zhou C. The cutting-edge progress of immune-checkpoint blockade in lung cancer. Cell Mol Immunol 2021;18:279-93. [Crossref] [PubMed]
3. Hellmann MD, Ciuleanu TE, Pluzanski A, et al. Nivolumab plus Ipilimumab in Lung Cancer with a High Tumor Mutational Burden. N Engl J Med 2018;378:2093-104. [Crossref] [PubMed]
4. Rizvi NA, Hellmann MD, Snyder A, et al. Cancer immunology. Mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer. Science 2015;348:124-8.
5. McGranahan N, Swanton C. Neoantigen quality, not quantity. Sci Transl Med 2019;11:eaax7918. [Crossref] [PubMed]
6. Williams CJM, Peddle AM, Kasi PM, et al. Neoadjuvant immunotherapy for dMMR and pMMR colorectal cancers: therapeutic strategies and putative biomarkers of response. Nat Rev Clin Oncol 2024;21:839-51. [Crossref] [PubMed]
7. Yang SR, Gedvilaite E, Ptashkin R, et al. Microsatellite Instability and Mismatch Repair Deficiency Define a Distinct Subset of Lung Cancers Characterized by Smoking Exposure, High Tumor Mutational Burden, and Recurrent Somatic MLH1 Inactivation. J Thorac Oncol 2024;19:409-24. [Crossref] [PubMed]
8. Tarsounas M, Sung P. The antitumorigenic roles of BRCA1-BARD1 in DNA repair and replication. Nat Rev Mol Cell Biol 2020;21:284-99. [Crossref] [PubMed]
9. Zhou Z, Ding Z, Yuan J, et al. Homologous recombination deficiency (HRD) can predict the therapeutic outcomes of immuno-neoadjuvant therapy in NSCLC patients. J Hematol Oncol 2022;15:62. [Crossref] [PubMed]
10. Keijzers G, Bakula D, Petr MA, et al. Human Exonuclease 1 (EXO1) Regulatory Functions in DNA Replication with Putative Roles in Cancer. Int J Mol Sci 2018;20:74. [Crossref] [PubMed]
11. Wang S, Lee K, Gray S, et al. Role of EXO1 nuclease activity in genome maintenance, the immune response and tumor suppression in Exo1D173A mice. Nucleic Acids Res 2022;50:8093-106. [Crossref] [PubMed]
12. Sokolenko AP, Preobrazhenskaya EV, Aleksakhina SN, et al. Candidate gene analysis of BRCA1/2 mutation-negative high-risk Russian breast cancer patients. Cancer Lett 2015;359:259-61. [Crossref] [PubMed]
13. Tsai MH, Tseng HC, Liu CS, et al. Interaction of Exo1 genotypes and smoking habit in oral cancer in Taiwan. Oral Oncol 2009;45:e90-4. [Crossref] [PubMed]
14. Chen Q, Chen S, Wang J, et al. Construction and validation of a hypoxia-related risk signature identified EXO1 as a prognostic biomarker based on 12 genes in lung adenocarcinoma. Aging (Albany NY) 2023;15:2293-307. [Crossref] [PubMed]
15. Bayram S, Akkız H, Bekar A, et al. The significance of Exonuclease 1 K589E polymorphism on hepatocellular carcinoma susceptibility in the Turkish population: a case-control study. Mol Biol Rep 2012;39:5943-51. [Crossref] [PubMed]
16. Wang Z, Chao Z, Wang Q, et al. EXO1/P53/SREBP1 axis-regulated lipid metabolism promotes prostate cancer progression. J Transl Med 2024;22:104. [Crossref] [PubMed]
17. Knijnenburg TA, Wang L, Zimmermann MT, et al. Genomic and Molecular Landscape of DNA Damage Repair Deficiency across The Cancer Genome Atlas. Cell Rep 2018;23:239-254.e6. [Crossref] [PubMed]
18. Giles JR, Globig AM, Kaech SM, et al. CD8(+) T cells in the cancer-immunity cycle. Immunity 2023;56:2231-53. [Crossref] [PubMed]
19. Duhen T, Duhen R, Montler R, et al. Co-expression of CD39 and CD103 identifies tumor-reactive CD8 T cells in human solid tumors. Nat Commun 2018;9:2724. [Crossref] [PubMed]
20. Palomero J, Panisello C, Lozano-Rabella M, et al. Biomarkers of tumor-reactive CD4(+) and CD8(+) TILs associate with improved prognosis in endometrial cancer. J Immunother Cancer 2022;10:e005443. [Crossref] [PubMed]
21. Litchfield K, Reading JL, Puttick C, et al. Meta-analysis of tumor- and T cell-intrinsic mechanisms of sensitization to checkpoint inhibition. Cell 2021;184:596-614.e14. [Crossref] [PubMed]
22. McGranahan N, Furness AJ, Rosenthal R, et al. Clonal neoantigens elicit T cell immunoreactivity and sensitivity to immune checkpoint blockade. Science 2016;351:1463-9. [Crossref] [PubMed]
23. Lu C, Guan J, Lu S, et al. DNA Sensing in Mismatch Repair-Deficient Tumor Cells Is Essential for Anti-tumor Immunity. Cancer Cell 2021;39:96-108.e6. [Crossref] [PubMed]
24. Ma X, Dong L, Liu X, et al. POLE/POLD1 mutation and tumor immunotherapy. J Exp Clin Cancer Res 2022;41:216. [Crossref] [PubMed]
25. Ricciuti B, Recondo G, Spurr LF, et al. Impact of DNA Damage Response and Repair (DDR) Gene Mutations on Efficacy of PD-(L)1 Immune Checkpoint Inhibition in Non-Small Cell Lung Cancer. Clin Cancer Res 2020;26:4135-42. [Crossref] [PubMed]
26. Pan D, Hu AY, Antonia SJ, et al. A Gene Mutation Signature Predicting Immunotherapy Benefits in Patients With NSCLC. J Thorac Oncol 2021;16:419-27. [Crossref] [PubMed]
27. Zhang L, Jiang B, Lan Z, et al. Immune infiltration landscape on prognosis and therapeutic response and relevant epigenetic and transcriptomic mechanisms in lung adenocarcinoma. Front Immunol 2022;13:983570. [Crossref] [PubMed]
28. Sun B, Zhang J, Wang N, et al. The bioinformatics analysis and experimental validation of the carcinogenic role of EXO1 in lung adenocarcinoma. Front Oncol 2024;14:1492725. [Crossref] [PubMed]
29. Chang CC, Hsieh MS, Lin MW, et al. Novel Genetic Prognostic Signature for Lung Adenocarcinoma Identified by Differences in Gene Expression Profiles of Low- and High-Grade Histological Subtypes. Biomolecules 2022;12:160. [Crossref] [PubMed]