Abstract Objective This study investigates the mechanisms underlying the enhancement of radiosensitivity in esophageal carcinoma (ESCC) cells through the coupling of pH-responsive nanobubbles with Nivolumab, a CD8^+ T cell activator. Methods Single-cell transcriptomics analysis was used to identify radiation-sensitive cancer cells in ESCC tissues. An in situ mouse model of ESCC was established to study the effects of radiation therapy on CD8^+ T cells using high-throughput sequencing. Machine learning algorithms were employed to identify key genes associated with ESCC. CRISPR/Cas9 technology was used to knock out key genes, while lentivirus was used to overexpress them. In vitro assays were conducted to evaluate the impact of these key genes on CD8^+ T cell activity, proliferation, migration, invasion, apoptosis, and sensitivity to radiation therapy. Nanobubbles conjugated with antibodies were prepared and their uptake by CD8^+ T cells was observed. A humanized mouse model of ESCC was used to assess the effectiveness of the nanobubbles in enhancing CD8^+ T cell activity and cytotoxicity. Results The analysis revealed a close relationship between tumor cell radiosensitivity and CD8^+ T cells. The key gene PD-1 was found to play a resistant role in the response to radiation therapy. PD-1 inhibited the activity and cytotoxicity of CD8^+ T cells in ESCC tissues. The development of pH-responsive nanobubbles conjugated with Nivolumab (αPD1-O[2]-nivolumab (NB), enhanced CD8^+ T cell cytotoxicity and increased the sensitivity of ESCC cells to radiation therapy. The release of oxygen by the nanobubbles further improved the efficacy of Nivolumab. In vivo experiments confirmed that αPD1-O[2]-NB increased CD8^+ T cell activity and cytotoxicity, thereby improving the sensitivity of ESCC cells to radiation therapy. Conclusion PD-1 promotes resistance to radiation therapy in ESCC cells by suppressing the activity and cytotoxicity of CD8^+ T cells. pH-responsive αPD1-O[2]-NB enhance CD8^+ T cell activity and improve the sensitivity of ESCC cells to radiation therapy. Graphical Abstract [36]graphic file with name 12951_2025_3552_Figa_HTML.jpg Supplementary Information The online version contains supplementary material available at 10.1186/s12951-025-03552-2. Keywords: PD-1, ESCC, CD8^+ T cells, ^125I radioactive particles, Radiotherapy, αPD1-O[2]-NB Introduction Esophageal carcinoma (ESCC), a common malignant tumor of the digestive system, has shown an increasing incidence worldwide [[37]1]. Despite advancements in the treatment of ESCC, its high invasiveness and propensity for lymphatic metastasis pose significant challenges, resulting in poor prognosis for patients [[38]2–[39]4]. Radiation particle therapy, an important approach for managing ESCC, has achieved some success. However, resistance to radiation therapy in some patients limits its effectiveness [[40]5, [41]6]. Therefore, it is important to thoroughly understand the molecular mechanisms of radiation therapy resistance and identify strategies to enhance therapeutic sensitivity [[42]7, [43]8]. To uncover the molecular mechanisms of radiation therapy sensitivity in ESCC cells, researchers have extensively applied single-cell transcriptomic techniques, which offer high-resolution sensitivity and comprehensive gene expression information, enabling the study of cellular heterogeneity. Through integrated analysis of single-cell transcriptomics, researchers can gain a better understanding of the roles of cell types and genes in radiation therapy sensitivity. One of our research goals is to employ this technique to elucidate the molecular mechanisms of radiation therapy sensitivity in ESCC cells and identify key cell types and genes associated with it. In our study, we identified a crucial gene, PD-1, that plays a pivotal role in radiation particle therapy for ESCC cells, using high-throughput sequencing analysis. Our functional experiments revealed that PD-1 promotes resistance to radiation therapy in ESCC cells by inhibiting the activity and cytotoxicity of CD8^+ T cells. This underscores the significance of PD-1 in the mechanism of radiation therapy resistance and provides clues for further investigation into the regulatory mechanisms of this pathway [[44]9–[45]11]. To enhance radiation therapy sensitivity, we developed a pH-responsive oxygen-carrying nanobubble called αPD1-O[2]-NB, which was combined with Nivolumab. Our experimental results demonstrated that αPD-1-O[2]-NB specifically targets CD8^+ T cells, enhances their cytotoxic activity, and the released oxygen enhances the effects of Nivolumab [[46]12–[47]14]. Further in vivo experiments validated the effectiveness of αPD-1-O[2]-NB, as it not only enhanced the activity and cytotoxicity of CD8^+ T cells but also increased the radiation therapy sensitivity of ESCC cells [[48]15, [49]16]. This study aims to delve deeper into the molecular mechanisms of radiation therapy sensitivity in ESCC cells and identify strategies to enhance therapeutic effectiveness. It not only provides new strategies for the treatment of ESCC but also contributes to improving patient survival rates and quality of life. Further research is warranted to refine and apply these methods to make greater contributions to the treatment of ESCC. Materials and methods Obtaining single-cell sequencing data related to ESCC The single-cell RNA sequencing (scRNA-seq) dataset [50]GSE188900, related to ESCC, was obtained from the Gene Expression Omnibus (GEO). This dataset includes seven ESCC tumor tissue samples and one normal tissue sample from five different patients. The analysis of the [51]GSE188900 data was performed using the “Seurat” package in R software [[52]17]. Data quality control was conducted by setting the nFeature_RNA between 200 and 5000 and percent.mt less than 25. Subsequently, 2000 genes with highly variable expression were selected from the dataset [[53]18]. UMAP clustering analysis and cell annotation The dimensionality of the scRNA-Seq dataset was reduced through principal component analysis based on the top 2000 highly variable genes. The first 30 principal components were selected for downstream analysis using the Elbowplot function in the Seurat package. The FindClusters function in Seurat was then utilized to identify major cell subgroups, with the resolution set to 1. Subsequently, the UMAP algorithm was applied to the scRNA-seq data for non-linear dimensionality reduction. Cell annotation was carried out by integrating known lineage-specific marker genes and utilizing the CellMarker online tool available at [54]http://xteam.xbio.top/CellMarker/. For the list of marker genes used for cell type annotation, please refer to Table [55]S1 [[56]19]. Malignant cell extraction After extracting epithelial cells, the subset function was used in conjunction with the inferCNV package for CNV analysis. Next, the K-means algorithm was applied to eliminate non-malignant cells from the epithelial cell population, followed by further UMAP clustering analysis of the malignant cells. Finally, the AddModuleScore function was employed to score the radiosensitivity of the malignant cells [[57]20, [58]21]. Construction of the in situ ESCC mouse model SPF-level female C57BL/6J mice were purchased from Hunan Slac Jingda Laboratory Animal Co., Ltd., aged 4–6 weeks, weighing 14–18 g. The mice were individually housed in SPF-level animal facilities under conditions of 60-65% humidity and a temperature of 25 ± 2℃. They were maintained on a 12-hour light-dark cycle and provided with ample food and water. After one week of acclimatization, the mice were observed for their health status prior to the experiment. This experimental procedure and animal usage plan were approved by the Institutional Animal Ethics Committee [[59]22]. To establish the in situ esophageal cancer mouse model, 4-NQO (Nitroquinoline 1-oxide; obtained from MedChemExpress, USA) was added to the drinking water. Initially, 4-NQO was dissolved in 5 mL of propylene glycol (HY-Y0921, MedChemExpress) to prepare a stock solution of 5 mg/mL. Then, 3 mL of the stock solution was mixed with 147 mL of water sterilized under high pressure, resulting in a final concentration of 100 µg/mL in the drinking water. The drinking water was changed once or twice a week. The control group was provided with 147 mL of drinking water containing 3 mL of propylene glycol. On days 0, 113, and 162, the mice were euthanized, and tumor development was evaluated by hematoxylin and eosin staining of esophageal tissue Sects. [[60]23, [61]24]. Cell culture The human ESCC cell line TE-1 (TCHu 89, deposited by the Cell Bank of the Chinese Academy of Sciences, China) and human CD8^+ T cells (1506, LDEBIO, China) were cultured in RPMI-1640 medium (A1049101, Gibco, USA) supplemented with 10% fetal bovine serum (12484028, Gibco, USA) and 1% penicillin/streptomycin (15140148, Gibco, USA). The human ESCC cell line KYSE30 (CL-0577, Procell, China) was cultured in RPMI-1640 medium supplemented with Ham’s F-12 (11765054, Gibco, USA), 2 mM L-Glutamine (25030149, Gibco, USA), 10% fetal bovine serum, and 1% penicillin/streptomycin. Isolated CD8^+ T cells were stimulated with CD3/CD28 activator (11161D, Gibco, USA) for 24 h prior to the experiments. The 293T cell line was obtained from ATCC (CRL-3216) and cultured in DMEM medium (11965092, Gibco, USA) supplemented with 10% fetal bovine serum, 10 µg/mL streptomycin, and 100 U/mL penicillin. All cells were maintained in a humidified incubator at 37 °C with 5% CO[2]. Passaging was performed when cell confluence reached 80–90% [[62]25]. Construction and transfection of overexpression and knockdown lentiviral vectors The potential target sequences for short hairpin RNA (shRNA) against PD-1 were analyzed based on the mouse cDNA sequences obtained from GeneBank. Three target sequences specific to PD-1 were designed, with a negative control sequence (sh-NC) lacking any interfering sequence. The primer sequences can be found in Table [63]S2, and the oligonucleotides were synthesized by GenePharma^® (Shanghai, China). The lentiviral packaging system was constructed using pLKO.1 (a lentiviral gene silencing vector). The packaging virus and the target vector were co-transfected into 293T cells using Lipofectamine 2000 (11668500, Invitrogen, USA) when cell confluence reached 80–90%. The supernatant was collected 48 h after cell culture, and the virus particles were present in the filtered and centrifuged supernatant. The virus particles were collected from cells at the exponential growth phase, and the viral titer was determined. The overexpression lentivirus for PD-1 was constructed and packaged by Genechem (Shanghai, China), with the lentiviral overexpression vector named LV-PDGFRA. The packaging virus and the target vector were co-transfected into 293T cells using Lipofectamine 2000 (11668500, Invitrogen, USA) when cell confluence reached 80–90%. The supernatant was collected 48 h after cell culture, and the virus particles were present in the filtered and centrifuged supernatant. The virus particles were collected from cells at the exponential growth phase, and the viral titer was determined. Additionally, in order to facilitate subsequent bioluminescence imaging experiments, the silencing or overexpression vectors were carrying the GFP/mCherry genes. When CD8^+ T cells reached the logarithmic growth phase, they were digested and dispersed using trypsin to obtain a cell suspension of 5 × 10^4 cells/mL, which was then seeded in a 6-well plate with 2 mL per well. Prior to establishing the in vitro cell model, the overexpression lentivirus targeting PD-1 (MOI = 10, viral titer of 1 × 10^8 TU/mL) was separately added to the CD8^+ T cell culture medium. In order to enhance the transfection efficiency, 6 µg/mL of Polybrene (40804ES76, YEASEN, China) was added to the medium. Stable cell lines were selected using 2 µg/mL of puromycin (UC0E03, Sigma-Aldrich, USA) [[64]26–[65]29]. Construction of PD-1 knockout cells using CRISPR/Cas9 PD-1 knockout cells were generated using the CRISPR/Cas9 technique. The sgRNA sequences used are detailed in Table [66]S3. The sgRNAs were inserted into the Lenti-CRISPR v2 vector (HanBio, Shanghai, China) containing the Streptococcus pyogenes Cas9 nuclease gene. CD8^+ T cells were transduced with the lentiviral Lenti-CRISPR v2 vector and subsequently edited using the CRISPR/Cas9 system to generate PD-1 knockout cells. Cells transfected with sgRNA plasmids and donor sequences were selected using 4 µg/mL puromycin (A1113803, Gibco, USA). Surviving cells were then subjected to limiting dilution cloning, and PD-1 knockout cells were screened using RT-qPCR and Western blot analysis. Confirmation of PD-1 knockout was achieved through DNA sequencing [[67]30, [68]31]. Co-culture of cells CD8^+ T cells were co-cultured with ESCC cells TE-1/KYSE30 at a ratio of 5:1 in vitro. The co-culture was continued for 48 h, after which the supernatant and ESCC cells were collected for further experiments. ESCC cells were separated using EpCAM antibody through flow cytometry cell sorting [[69]32–[70]35]. The cell co-culture groups were as follows: (1) Control group: co-culture of wild-type CD8^+ T cells and KYSE30 cells; (2) Treat/Radiotherapy (RT) group: co-culture of wild-type CD8^+ T cells, KYSE30 cells, and ^125I radiation particles; (3) PD-1-WT + RT group: co-culture of CD8^+ T cells with PD-1 knockout using CRISPR/Cas9 empty vector transduction, KYSE30 cells, and ^125I radiation particles; (4) PD-1-KO + RT group: co-culture of PD-1 knockout CD8^+ T cells, KYSE30 cells, and ^125I radiation particles; (5) oe-NC + RT group: co-culture of CD8^+ T cells transduced with empty vector, KYSE30 cells, and ^125I radiation particles; (6) oe-PD-1 + RT group: co-culture of CD8^+ T cells transduced with PD-1 overexpression vector, KYSE30 cells, and ^125I radiation particles; (7) Blank group: co-culture of PBS solution, wild-type CD8^+ T cells, TE-1/KYSE30 cells; (8) αPD1 group: co-culture of 1 mg/mL Nivolumab solution, wild-type CD8^+ T cells, and TE-1/KYSE30 cells; (9) N[2]-NB group: co-culture of nitrogen-carrying nanobubbles, wild-type CD8^+ T cells, and TE-1/KYSE30 cells; (10) O[2]-NB group: co-culture of oxygen-carrying nanobubbles, wild-type CD8^+ T cells, and TE-1/KYSE30 cells; (11) αPD1-N[2]-NB group: co-culture of Nivolumab-conjugated nitrogen-carrying nanobubbles, wild-type CD8^+ T cells, and TE-1/KYSE30 cells; (12) αPD1-O[2]-NB group: co-culture of Nivolumab-conjugated oxygen-carrying nanobubbles, wild-type CD8^+ T cells, and TE-1/KYSE30 cells; (13) ^125I group: co-culture of PBS solution, wild-type CD8^+ T cells, TE-1/KYSE30 cells, and ^125I radiation particles; (14) αPD1 + ^125I group: co-culture of 1 mg/mL Nivolumab solution, wild-type CD8^+ T cells, TE-1/KYSE30 cells, and ^125I radiation particles; (15) N[2]-NB + ^125I group: co-culture of nitrogen-carrying nanobubbles, wild-type CD8^+ T cells, TE-1/KYSE30 cells, and ^125I radiation particles; (16) O[2]-NB + ^125I group: co-culture of oxygen-carrying nanobubbles, wild-type CD8^+ T cells, TE-1/KYSE30 cells, and ^125I radiation particles; (17) αPD1-N[2]-NB + ^125I group: co-culture of Nivolumab-conjugated nitrogen-carrying nanobubbles, wild-type CD8^+ T cells, TE-1/KYSE30 cells, and ^125I radiation particles; (18) αPD1-O[2]-NB + ^125I group: co-culture of Nivolumab-conjugated oxygen-carrying nanobubbles, wild-type CD8^+ T cells, TE-1/KYSE30 cells, and ^125I radiation particles. High-throughput sequencing and analysis of CD8^+ T cells in ESCC tumor tissues in situ mouse model After constructing the ESCC mouse model, the tumor surface was cleansed with iodine disinfectant solution. Once anesthetized, the tumor was punctured vertically using an 18G needle from the Japanese company Kakko, with the needle tip positioned at the center of the tumor. Subsequently, an implantation apparatus (model 6711) from Ningbo Junan Pharmaceutical Technology Co. Ltd was used to inject ^125I radioactive particles purchased from the same company into the tumor through the needle. Following the procedure, sterile gauze was applied for hemostasis if necessary. For both the control group (n = 3) and the ^125I-treated group (n = 3) of ESCC mice, random samples of ESCC tissues were collected. These samples were washed in cold PBS to remove residual tissue other than tumor tissue. Once tissue disruption occurred, 1 mg/mL of collagenase (C2674) from Sigma-Aldrich (USA) was added, and the tissue was digested at 37 °C for 10 min. Further incubation at 37 °C for 5 min with trypsin/EDTA (25200072) from Gibco (USA) was carried out to prepare a single-cell suspension. The cells were then labeled with fluorescently conjugated anti-CD3 (58-0032-82) and anti-CD8α (11-0081-82) antibodies, both purchased from Invitrogen (USA). CD8^+ T cells marked with CD3 and CD8α were sorted using a flow cytometer. Total RNA from the CD8^+ T cells was extracted from the six tissue samples using the Total RNA Isolation Kit (12183555) from Invitrogen (USA). The OD value of the total RNA was quantified using a UV-Vis spectrophotometer (BioSpectrometer basic) from Eppendorf (USA). The integrity of the total RNA was assessed using agarose gel electrophoresis. High-quality total RNA was reverse transcribed into cDNA, and RNA libraries were constructed for sequencing using the Illumina NextSeq 500 platform. The raw sequencing image data was converted into raw reads through base calling. To ensure the quality of the raw reads, the cutadapt tool was used to remove sequencing adapter sequences and low-quality sequences, resulting in the generation of “clean reads”. The Hisat2 software was employed to align these sequences to the mouse reference genome, and gene expression was quantified using R software packages, generating a gene expression matrix [[71]36]. The high-throughput sequencing data was analyzed using the “limma” package in R to identify differentially expressed genes. The criteria for filtering were defined as |log2FC| > 2 and P.value < 0.001. Volcano plots were generated using the ggplot2 package, and heatmaps were generated using the pheatmap package in R. The “clusterProfiler”, “org.Hs.eg.db”, “enrichplot”, and “ggplot2” packages in R were used for pathway enrichment analysis and visualization of the DEGs using the DO pathway. GO and KEGG pathway enrichment analysis of the DEGs was performed using the SangerBox database ([72]http://sangerbox.com/home.html). Protein-protein interaction analysis of the candidate target genes was conducted using the STRING database ([73]https://cn.string-db.org/), and the protein interaction network was visualized using Cytoscape v3.10.0 software. The GEPIA database ([74]http://gepia.cancer-pku.cn/detail.php) was utilized to analyze gene expression in TCGA-ESCC. The Xiantaozi Academic Database ([75]https://www.xiantaozi.com/) was employed to assess the impact of single-gene expression levels on immune cell infiltration in TCGA-ESCC (MID: 35774278, PMID: 36046241, and PMID: 36267311). The grouping information of the mouse models is as follows (6 mice per group): Control group (ESCC mouse models), RT group (ESCC mouse models treated with ^125I radioactive particles), sh-NC + RT group (ESCC mouse models treated with ^125I radioactive particles and injected with empty vector slow virus), sh-PD-1 + RT group (ESCC mouse models treated with ^125I radioactive particles and injected with PD-1 knockdown vector slow virus), oe-NC + RT group (ESCC mouse models treated with ^125I radioactive particles and injected with empty vector slow virus for overexpression), and oe-PD-1 + RT group (ESCC mouse models treated with ^125I radioactive particles and injected with PD-1 overexpression vector slow virus). Except for the Control and RT groups, mice in the other groups were injected with 1 × 10^8 TU/mL slow virus into the tumor tissue through surgery after successful modeling, and bioluminescence imaging was used to record tumor volume for each group every 7 days. The formula used for the growth curve was (a*b^2)/2, where “a” represents the longest diameter of the tumor and “b” represents the shortest diameter. After 2 weeks, the CRi Maestro in vivo imaging system (Cambridge Research & Instrumentation, Massachusetts, USA) was used to analyze the bioluminescent signals of firefly luciferase in the mice. Mice were anesthetized with 2% isoflurane and injected with D-luciferin (150 mg/kg; 122799, PerkinElmer, USA) into the abdomen. Two sets of photos were taken after a 15-minute interval, with an exposure time of 10 min [[76]37]. After 28 days, the mice were euthanized with CO[2], and tumor tissue was harvested from their bodies. The tumor weight was measured, and the tumor tissue was collected for subsequent tests [[77]22]. Construction of the ^125I irradiation model In accordance with previous methods, a ^125I irradiation model was constructed, as shown in Figure [78]S1. The model consisted of a 3 mm thick polystyrene panel, with a layer of radioactive particle tray at the bottom and a layer of cell culture tray on top. In the seed tray, eight seeds with equal activity were evenly distributed around a 35 mm diameter (D) circumference. In the cell culture tray, a similar gap surrounded a 35 mm diameter circumference, aligned with the vertical line of the seed tray to accommodate a 35 mm diameter culture dish during the experiment. The height between the seed tray and the bottom of the culture dish was 12 mm, with a D/H ratio of 2.9. The purpose of this design was to achieve a relatively uniform dose distribution at the bottom of the culture dish. The polystyrene components were enclosed by a 3 mm thick lead shield, equipped with a ventilation hole to allow the placement of the entire model in a culture incubator during the study. The culture incubator maintained a constant cell culture environment, ensuring protective conditions. The model employed ^125I radioactive particles provided by China Ningbo Junan Pharmaceutical Technology Co., Ltd., and subsequent experiments were carried out after irradiation at a dose of 4 Gy for 92 h [[79]38, [80]39]. Synthesis of acetalated dextran In a 50 mL round-bottom flask, 2 g of dextran were added and flushed with nitrogen gas. Then, 20 mL of anhydrous DMSO was added to the flask and the mixture was stirred for approximately 20 min until it dissolved completely. Subsequently, 30.2 mg of pyridinium p-toluenesulfonate (82815, Sigma-Aldrich, USA) and 6.8 mL of 2-methoxypropene (174645, Sigma-Aldrich, USA) were added to initiate the reaction. The reaction was stopped after 3 h by adding 2 mL of triethylamine (TEA; 8.08352, Sigma-Aldrich, USA), and the crude product was precipitated in 200 mL of deionized water. The white precipitate was collected by centrifugation at 4600 rpm for 10 min and washed twice with 100 mL of deionized water at approximately pH 8.0. The remaining moisture was removed by freeze-drying, resulting in the production of Acetalated Dextran (AC-DEX) as a fine white powder [[81]40]. Preparation of Nivolumab-conjugated oxygen-carrying nanobubbles In this experiment, nanobubbles were prepared using the method of emulsifier solvent evaporation and internal phase separation. To summarize briefly, the method was as follows: 40 mg of Acetalated Dextran was dissolved in 1 mL of dichloromethane, and then 400 µL of octane was added as the organic phase. Next, 15 mg of 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC; P0763, Sigma-Aldrich, USA), 15 mg of 1,2-dipalmitoyl-sn-glycero-3-phospho-(1’-rac-glycerol) (sodium salt) (DPPG; P9789, Sigma-Aldrich, USA), and 10 mg of 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-2000] (DSPE-PEG2000; 880135, Avanti Polar Lipids Inc., USA) were dissolved in 4 mL of chloroform, and then PEG 4000 was added to the mixture (mass ratio of 99:1). Subsequently, the lipid mixture was freeze-dried according to a predetermined program to form a lipid film. The lipid film was immersed in an 8 mL solution of 10% (w/v) glycerol at 70 °C, hydrated for 20 min, to form a lipid solution. Using an ultrasonic cell disruptor (SFX150, Branson, USA), with 100% amplitude and a switch mode of 1 s on and 1 s off, the organic phase was mixed with 5 mL of the lipid solution to prepare a nanoemulsion. The nanoemulsion solution was diluted into 15 mL of lipid solution and stirred at a speed of 300 rpm for 2 h to remove dichloromethane. The obtained nanocapsule solution was transferred to a dialysis bag and dialyzed against deionized water (pH ~ 8.0) for 24 h to remove excess lipids and glycerol. 1 gram of PEG was added to the purified nanocapsule solution, which was then frozen in liquid nitrogen for 5 min. The frozen sample was freeze-dried for 72 h to remove moisture and octane as the core. The freeze-dried matrix was then transferred to a 5 mL glass bottle with a rubber septum, and the air in the bottle was replaced with oxygen, resulting in the freeze-dried material of oxygen-carrying nanobubbles (O[2]-NB). Nitrogen-filled nanobubbles were used as the control group (N[2]-NB). Next, 0.025 mg/mL of avidin (434401, Invitrogen, USA) [[82]41] was added to the mixture at room temperature and incubated for 30 min, followed by rinsing and centrifugation under the same conditions. Then, the biotinylated nanobubbles bound to avidin were mixed with biotinylated FITC-labeled Nivolumab (1 mg/mL; T9907, TargetMol, USA) (1 mL: 1 mL), incubated at room temperature, and then rinsed and centrifuged following the previous steps, finally obtaining Nivolumab-conjugated nanobubbles (αPD1-O[2]-NB / αPD1-N[2]-NB). The prepared nanobubbles were stored overnight at 4 °C to reach gas equilibrium. Prior to use, 5 mL of PBS solution was injected into the bottle, and gentle agitation was applied to reassemble the nanobubble solution [[83]40, [84]42, [85]43]. Characterization of the physical properties of nano-bubbles The polydispersity index (PDI), size, and zeta potential of oxygen-carrying nano-bubbles in a pH 7.4 PBS solution were measured at a scattering angle of 90° using a Zeta Plus instrument (Brookhaven Instruments, USA) in an Eppendorf UVette centrifuge tube. Three measurements were conducted. The structure and morphology of the nano-bubbles were examined using a transmission electron microscope (TEM) (JEM-2100 F, JEOL, Japan) and a scanning electron microscope (SEM) (JEOL 6700, JEOL, Japan). For TEM imaging preparation, approximately 3 µL of the nano-bubble solution was dropped onto a plasma-treated copper grid and left for 2 min to settle. The excess solution was removed with filter paper. The grid was air-dried overnight before TEM imaging. For SEM examination, approximately 5 µL of the nano-bubble suspension was dropped onto a clean silicon wafer and prepared after overnight drying. The diameter and concentration changes of the nano-bubbles (NBs) were observed after 0, 1, 2, 4, 6, 8, 12, and 24 h of preparation. All prepared NBs were kept at 4 °C. The size and concentration at different time points were measured using the same method. Each sample was measured five times. A homemade agarose mold was used to compare the imaging capability of NBs at different concentrations (5 × 10^5, 1 × 10^6, 5 × 10^6, 1 × 10^7, and 5 × 10^7/mL). The images were captured and analyzed using the clinical IU22 ultrasound scanner from Philips Healthcare. The mechanical index (MI) was set at 0.06 with a contrast gain of 90%. The ultrasound characteristics of nano-bubbles were analyzed using a diagnostic ultrasound system (Acuson S2000, Siemens, Germany) to confirm their hollow structure and evaluate their contrast-enhancing performance. Ultrasound images of the nano-bubbles were obtained in pulse sequence (CPS) mode with a frequency of 8 MHz and MI of 0.2. The reconstituted nano-bubble solution at the desired concentration was added to the sample well of a biomimetic agarose tissue-mimicking phantom, which was immersed in a water tank. The average grayscale value of the ultrasound image was obtained using Image J for subsequent quantitative analysis. The contrast-enhancing performance was expressed by the contrast tissue ratio (CTR), defined as the ratio of the average intensity between the sample with contrast agent and the sample without contrast agent [[86]44]. Dynamic evaluation of oxygen release Low oxygen solutions were prepared by bubbling nitrogen gas (oxygen partial pressure pO[2] = 0.4 mg/L) into a solution. Then, 2 mL of a concentrated oxygen-carrying nanobubble solution with a concentration of 2.5 mg/mL was separately injected into 30 mL of PBS solution with pH values of 7.4 and 6.5. The samples were incubated at 37 °C under a nitrogen atmosphere. The oxygen partial pressure (pO[2]) of the mixed solution was continuously recorded at specific time points using a blood oxygen meter (JPB-607 A, INESA Scientific Instrument, China) for 30 min [[87]40]. Cell adhesion detection of NBs to CD8^+ T cells The specific binding of NBs to T lymphocytes was demonstrated through optical microscopy and confocal laser scanning microscopy (CLSM). Approximately 5 × 10^6 cells were incubated with 1 × 10^8 NBs for 30 min. The binding of NBs to T lymphocytes was evaluated using a 40x magnification optical microscope. For CLSM imaging, cells were fixed with 4% paraformaldehyde for 10 min, followed by the same incubation process as mentioned above with DiI-labeled NBs (C1036, Beyotime Biotechnology Co., Ltd, Shanghai, China). To confirm the binding rate of NBs to T lymphocytes, 5 × 10^6 T lymphocytes were incubated separately with 1 × 10^8 DiI-labeled NBs for 30 min. The cells were washed three times with PBS and then analyzed using flow cytometry [[88]43]. Enzyme-linked immunosorbent assay (ELISA) In accordance with the manufacturer’s instructions, the levels of IFN-γ and Granzyme B in human tumor cells and mouse tumor tissues were determined using the ELISA kit purchased from Invitrogen (USA), as described in Table [89]S4 [[90]45, [91]46]. Relative expression levels of target genes were detected by RT-qPCR Total RNA was extracted from tissues or cells using Trizol reagent (15596026, Invitrogen, USA), and the concentration and purity of the total RNA were measured at 260/280 nm using NanoDrop LITE (ND-LITE-PR, Thermo Scientific™, Germany). The extracted total RNA was then synthesized into cDNA using the PrimeScript RT Reagent Kit with gDNA Eraser (RR047Q, TaKaRa, Japan). Subsequently, the SYBR Green PCR Master Mix reagents (4364344, Applied Biosystems, USA) and ABI PRISM 7500 Sequence Detection System (Applied Biosystems) were utilized to perform RT-qPCR for each gene. The primers for each gene were synthesized by TaKaRa (Table [92]S5), with GAPDH serving as the reference gene. The relative expression levels of each gene were analyzed using the 2^-ΔΔCt method, where ΔΔCt = (average Ct value of the target gene in the experimental group - average Ct value of the reference gene in the experimental group) - (average Ct value of the target gene in the control group - average Ct value of the reference gene in the control group). All RT-qPCR experiments were repeated three times [[93]47–[94]49]. Western blot First, cells or tissues were collected and lysed using an enhanced RIPA lysis buffer containing a protease inhibitor (P0013B, Beyotime Biotechnology Co., Ltd., Shanghai, China). The concentration of protein was determined using the BCA protein assay kit (P0012, Beyotime Biotechnology Co., Ltd., Shanghai, China). The proteins were separated by 10% SDS-PAGE and then transferred to a PVDF membrane. Blocking was performed with 5% BSA at room temperature for 2 h to minimize non-specific binding. Subsequently, the diluted primary antibody (rabbit anti-human, see details in Table [95]S6) was added and incubated at room temperature for 1 h. After washing the membrane, it was incubated with an HRP-conjugated secondary antibody, either goat anti-rabbit IgG (ab6721, 1:2000, abcam, UK) or goat anti-mouse IgG (ab6785, 1:1000, abcam, UK), at room temperature for 1 h. Equal amounts of Solution A and Solution B (Pierce™ ECL Western Blotting Substrate, 32209, Thermo Scientific, Germany) were mixed in the dark, added to the membrane, and then exposed in a gel imaging system. The images were captured using the Bio-Rad Image Analysis System (BIO-RAD, USA) and analyzed for grayscale quantification of protein bands using Image J software, with GAPDH as the loading control [[96]48]. Each experiment was performed in triplicate. MTT and CCK-8 assays were used to measure cell viability MTT method: CD8^+ T cells were seeded onto a 96-well cell culture plate at a density of 3–5 × 10^4 cells/mL and incubated for 48 h. MTT solution (10 mg/mL, ST316, Beyotime Biotechnology Co., Ltd, Shanghai, China) was added to the cell suspension and incubated for 4 hours, followed by the addition of dimethyl sulfoxide (DMSO) and shaking for 10 min. Absorbance (OD 570 nm) was measured using a spectrophotometer (Laspec, China) [[97]50]. CCK-8 method: ESCC cells in each group were digested, resuspended, and adjusted to a cell density of 1 × 10^5 cells/mL. A volume of 100 µL was seeded into a 96-well plate for routine cultivation. After the cells adhered, drugs were added and incubated overnight. At 0, 12, 24, and 48 h after cultivation, cell viability was assessed according to the instructions of the CCK-8 assay kit (C0041, Beyotime, Shanghai). For each measurement, 10 µL of CCK-8 detection reagent was added to each well and incubated in a cell incubator for 4 h. Absorbance at 450 nm was then measured using an enzyme-linked immunosorbent assay (ELISA) reader, and cell viability was calculated. Cell viability = (ΔA_sample - ΔA_blank) / ΔA_control, where ΔA_sample represents the absorbance difference of the sample, ΔA_blank represents the absorbance difference of the blank, and ΔA_control represents the absorbance difference of the control group [[98]51, [99]52]. EdU experiment for detecting cell proliferation ESCC cells were cultured in a 24-well plate and incubated with EdU ([100]C10337, Invitrogen, USA) at a concentration of 10 µmol/L for 2 h in a cell culture incubator. The culture medium was then removed, and the cells were fixed with a PBS solution containing 4% paraformaldehyde for 15 min. After washing twice with PBS containing 3% BSA, the cells were incubated with a PBS solution containing 0.5% Triton-100 at room temperature for 20 min and washed twice again with PBS containing 3% BSA. Next, 100 µL of a staining solution was added to each well and incubated at room temperature in the absence of light for 30 min. This was followed by staining with DAPI solution for 5 min. After mounting the slides, 6–10 random fields were observed under a fluorescence microscope (BX63, Olympus, Japan), and the number of positive cells was recorded in each field. The EdU labeling rate was calculated as the percentage of positive cells out of the total number of cells (positive + negative) multiplied by 100% [[101]53]. Each experiment was performed in triplicate. Transwell assay for measuring cell migration and invasive ability ECM gel (E1270, Sigma-Aldrich, Germany) was added to the upper chamber of a 24-well Transwell plate (8 μm) and the plate was incubated at 37℃ for 30 min to allow the gel to solidify. ESCC cells transfected for 48 h were collected, resuspended in a serum-free medium at a density of 10^5 cells, and seeded into the upper chamber. Each well was seeded with 200 µL of cell suspension (2 × 10^4 cells/well) in the upper chamber and 800 µL of medium containing 20% FBS was added to the lower chamber. After 24 h of incubation at 37℃, the Transwell plate was removed, washed twice with PBS, fixed with formalin for 10 min, and then washed three times with water. The cells were stained with 0.1% crystal violet, placed at room temperature for 30 min, washed twice with PBS, and the cells on the upper surface were wiped off with a cotton ball. The stained invasive cells were photographed using an inverted light microscope (CKX53, Olympus, Japan), and Image J software was used for cell counting and analysis of the invasive capability. For the migration assay, Transwell plates without ECM gel coating were used, and the remaining steps were the same as the invasion assay [[102]54, [103]55]. TUNEL staining The tumor tissue or ESCC cells were fixed with 4% paraformaldehyde at room temperature for 15 min and permeabilized with 0.25% Triton X-100 at room temperature for 20 min. Subsequently, the samples were blocked with 5% bovine serum albumin (BSA, 36101ES25, Yeasen Biotechnology (Shanghai) Co., Ltd., China) and stained using the TUNEL assay (C1086, Beyotime Biotechnology Co., Ltd., Shanghai, China). The slides were then counterstained in the dark using a DAPI staining solution (C1002, Beyotime Biotechnology Co., Ltd., Shanghai, China). The images of apoptotic cells were captured under a confocal microscope (LSM 880, Carl Zeiss AG, Germany). TUNEL-positive cells (green fluorescence) indicated cells undergoing apoptosis, while DAPI labeled the cell nuclei, appearing as blue fluorescence. Five different fields of view were selected for each group to calculate the apoptotic cell rate, determined as the percentage of apoptotic cells relative to the total cell count [[104]56]. Radiation clonogenic survival assay Cells were seeded into six-well plates with each well containing three replicates. After 24 h, cells were irradiated with different doses (0, 2, 4, 6, and 8 Gy) using a ^125I irradiation model. Cells were cultured at 37℃ for 9–12 days until visible colonies appeared. The colonies were stained with crystal violet staining solution and counted. In the cell cycle enrichment cloning experiment, cells were transfected with ES-FUCCI plasmid (provided by biovectorNTCC plasmid carrier cell gene preservation center). FUCCI-labeled cells were collected using the BD FACSAria Fusion Sorter (BD Biosciences, USA), with the G1 phase being detected using a red filter and the S, G2, and M phases being detected using a green filter. The sorted cells were immediately plated onto 60 mm tissue culture dishes and irradiated with 0 Gy and 8 Gy, respectively. After 10–14 days of incubation, the colonies were fixed with methanol/acetic acid and stained with 0.5% crystal violet. Colonies or colony-forming units (CFUs) containing at least 50 cells were counted using a dissecting microscope, and the survival fraction was calculated. The survival fraction was determined by the formula: (number of irradiated colonies/number of plated cells)/(number of unirradiated colonies/number of plated cells) [[105]57]. The experiment was repeated three times. Flow cytometry analysis The cells to be tested were stained with the antibodies from Table [106]S7, thoroughly mixed, and then incubated at 4 °C in the dark for 30 min. Subsequently, 2 mL of PBS solution (Sigma-Aldrich, USA, P4417) was added, and the mixture was centrifuged at 1500 g for 10 min at 4 °C. The supernatant was discarded, and 2% paraformaldehyde (Sigma-Aldrich, USA, 30525-89-4)/PBS solution was added to fix the cells. The cells were then stored at 4 °C in the dark and were analyzed within 24 h using the FACS Aria II flow cytometer (BD Bioscience, USA). The data of the stained cells were collected using Flowjo CE software, and further analyzed using FlowJo software [[107]58]. ESCC patient-derived tumor xenograft model The animal experiment was conducted using 6 to 8-week-old non-obese diabetic-severe combined immunodeficiency (NOD-SCID) female mice (Hunan Slaik Jingda Experimental Animal Co., Ltd., Hunan, China). A total of 1 × 10^7 pre-treated and co-cultured TE-1/KYSE30 cells were subcutaneously injected into the mice’s axillary region. The tumor growth was monitored timely and recorded through photography. Tumor volume was checked every other day. After 1 week of cell inoculation, 1 × 10^5 human peripheral blood mononuclear cells (hPBMCs) were intravenously injected through the tail vein. On the 13th and 23rd day, the concentration of human CD45 in the mice’s blood was measured [[108]59–[109]61]. When the tumor size reached an average volume of approximately 160 mm3, 100 µL of a nano-bubble solution labeled with DIR (40757ES25, YEASEN, China) at a concentration of 2.5 mg/mL was injected into the mice’s bodies through the tail vein. The subsequent steps of ^125I particle therapy were consistent with the aforementioned procedure. At 2, 4, 6, 8, and 24 h after injection, the whole-body fluorescence imaging of the mice was performed using an infrared dual-band small animal in vivo imaging system (Photon). At 24 h after injection, the liver, spleen, kidney, heart, lung, and tumor were dissected, and the fluorescence intensity of each organ or tissue was measured. For intratumoral imaging of the tumor tissue, the collected tumor tissue was subjected to ex vivo fluorescence imaging, fixed in 4% paraformaldehyde for 24 h, and then placed in a 15% sucrose PBS solution for 24 h until sedimentation occurred. Afterward, the tumor tissue was sectioned into slices with a thickness of 20 μm, followed by staining with 1 mg/mL DAPI at room temperature for 10 min. After being washed twice with PBS (pH 7.4), the slices were immediately examined under a confocal laser scanning microscope (CLSM) [[110]62, [111]63]. The mice were euthanized with CO[2] after 28 days, and the transplanted tumors were removed and weighed. The tumor tissue was collected for Western blot and immunohistochemical analyses [[112]22]. The mice were randomly divided into 12 groups, with 6 mice in each group: (1) Blank group: injected with PBS solution and not treated with ^125I particle therapy; (2) αPD1 + DIR group: injected with 1 mg/mL Nivolumab solution and DIR solution, without ^125I particle therapy; (3) ^125I group: injected with PBS solution and treated with ^125I particle therapy; (4) O[2]-NB group: injected with oxygen-carrying nano-bubbles, without ^125I particle therapy; (5) αPD1-N[2]-NB group: injected with Nivolumab-conjugated nitrogen-carrying nano-bubbles, without ^125I particle therapy; (6) αPD1-O[2]-NB group: injected with Nivolumab-conjugated oxygen-carrying nano-bubbles, without ^125I particle therapy; (7) αPD1 + DIR + ^125I group: injected with 1 mg/mL Nivolumab solution, DIR solution, and treated with ^125I particle therapy; (8) O[2]-NB + ^125I group: injected with oxygen-carrying nano-bubbles and treated with ^125I particle therapy; (9) αPD1-N[2]-NB + ^125I group: injected with Nivolumab-conjugated nitrogen-carrying nano-bubbles and treated with ^125I particle therapy; (10) αPD1-O[2]-NB + ^125I group: injected with Nivolumab-conjugated oxygen-carrying nano-bubbles and treated with ^125I particle therapy. Histopathological staining The method for hematoxylin and eosin (H&E) staining is described as follows: Firstly, tissue samples from the study group are obtained and fixed. Then, wax blocks are sectioned, and the sections are placed in a wax remover (xylene). Subsequently, they are rehydrated in 100% ethanol, followed by 95% ethanol and 70% ethanol, and finally embedded or rinsed in water. Next, the prepared sections are placed in the hematoxylin staining solution (H8070, Solarbio, Beijing, China) for staining at room temperature for 5–10 min. The slides are then rinsed with distilled water, dehydrated in 95% ethanol, and immersed in an eosin staining solution (G1100, Solarbio, Beijing) for 5–10 min. Finally, routine dehydration, clearing, and slide mounting are carried out [[113]64]. IHC staining A list of antibodies can be found in Table [114]S8. Mouse tumor tissue was first fixed overnight with 4% paraformaldehyde and then cut into 4 μm thick paraffin sections. The sections were deparaffinized with xylene, followed by rehydration with three 3-minute washes in each of the following: anhydrous ethanol, 95% ethanol, and 75% ethanol. Antigen retrieval was performed by boiling the sections in a 0.01 M citrate buffer for 15–20 min, followed by incubation at room temperature for 30 min with 3% H[2]O[2] to inactivate endogenous peroxidase. Afterward, the sections were blocked with goat serum at room temperature for 20 min and excess serum was removed. Primary antibodies were then added and incubated at room temperature for 1 h, followed by PBS washes. Subsequently, rabbit anti-goat IgG secondary antibodies were added and incubated at 37℃ for 20 min, followed by more PBS washes. Staining was achieved by adding SP avidin peroxidase at 37℃ for 30 min, and after another round of PBS washes, DAB chromogenic substrate (P0202, Beyotime Biotechnology Co., Ltd) was added. After 5–10 min of color development, the reaction was terminated by washing with water for 10 min. Counterstaining was done using hematoxylin (C0107, Beyotime Biotechnology Co., Ltd), followed by differentiation with hydrochloric acid alcohol and another 10-minute wash with water. Finally, the sections were dehydrated with a gradient series of alcohols (in xylene) and mounted with 2–3 drops of neutral resin. Observation and quantification were performed using an upright microscope: 5 random high-power fields were selected per section, with 100 cells observed in each field, in order to calculate the Ki67-positive cell rate [[115]65]. Statistical analysis Our study utilized R version 4.2.1 programming language for data analysis. RStudio integrated development environment, version 2022.12.0-353, was used for R language compilation. We employed Perl version 5.30.0 for file processing. Additionally, GraphPad Prism software version 8.0 was utilized. Mean ± standard deviation (SD) was used to represent quantitative data. Independent sample t-test was employed for comparing two groups of data [[116]66]. One-way analysis of variance (ANOVA) was utilized for comparing data among different groups, while two-way ANOVA was applied for comparing data among different time points. Post-hoc tests were conducted using the Bonferroni method. The significance threshold was set at P < 0.05 [[117]67]. Results Subtyping and marker gene annotation of ESCC cells ESCC is one of the most lethal malignant tumors worldwide, posing significant challenges in treatment and prognosis. It imposes immense physical and psychological burdens on patients and their families. Despite advancements in medical technology and the continual optimization of comprehensive treatment strategies for ESCC, the lack of obvious early symptoms often leads to late-stage diagnosis for the majority of patients [[118]68–[119]70]. Accordingly, radiotherapy, one of the primary methods for ESCC treatment, plays a vital role in prolonging patient survival and alleviating symptoms. However, the effectiveness of radiotherapy is significantly influenced by the hypoxic and acidic conditions in the tumor microenvironment, as well as the biological heterogeneity of the tumor itself, posing a formidable challenge in ESCC treatment [[120]71–[121]73]. To gain deeper insights into the development of ESCC and the changes in the tumor microenvironment, we retrieved ESCC-related single-cell transcriptome sequencing data from the GEO database, specifically the [122]GSE188900 dataset. This dataset encompasses seven ESCC tumor tissue samples (T1-7) from five different patients, along with one normal tissue sample (N1) (Fig. [123]1A). The Seurat package was employed for data integration, initially examining the gene count (nFeature_RNA), mRNA molecule count (nCount_RNA), and percentage of mitochondrial genes (percent.mt) in all cells of the scRNA-seq data (Figure [124]S2A). Using the criteria of 200 < nFeature_RNA < 5000 and percent.mt < 25, a quality control step was conducted to remove low-quality cells and duplicated genes, resulting in an expression matrix comprising 20,197 genes and 31,695 cells. The calculation of sequencing depth correlations revealed a correlation coefficient of r = 0.19 between filtered data nCount_RNA and percent.mt, r = 0.86 between nCount_RNA and nFeature_RNA, and r = -0.03 between nCount_RNA and percent.HB (Figure [125]S2B), indicative of good-quality filtered cell data. Fig. 1. [126]Fig. 1 [127]Open in a new tab Single-cell RNA sequencing Data Cell Clustering and Annotation. Note: (A) Diagram illustrating the process of single-cell data sample acquisition and analysis; (B) Variance analysis for selecting highly variable genes, with the top 2000 highly variable genes in red, lowly variable genes in black, and the names of the top 10 highly variable genes labeled (N = 8); (C) PCA analysis showing the distribution of cells in PC_1 and PC_2, with each point representing a cell (N = 8); (D) Distribution of standard deviations for principal components (PCs), highlighting important PCs with larger standard deviations (N = 8); (E) Visualization of UMAP clustering results, showing the aggregation and distribution of cells from ESCC samples (N = 7) and adjacent normal tissues (N = 1) at a resolution of 1. Each color represents a cluster; (F) Expression of known lineage-specific marker genes in ESCC samples (N = 7) and adjacent normal tissue (N = 1) across different cells, where deeper purple indicates higher average expression levels and larger circles indicate a higher number of cells expressing that gene; (G) Visualization of cell annotations in ESCC samples (N = 7) and adjacent normal tissue (N = 1) based on UMAP clustering, with each color representing a cell subpopulation; (H) UMAP clustering visualization showing the aggregation and distribution of cells in ESCC samples (N = 7) and adjacent normal tissue (N = 1). The “Normal” group contains one adjacent normal tissue sample, while the “Tumor” group includes seven ESCC tumor tissue samples from five different patients. CD79A is a marker gene for B cells, SPRR1A for Basal cells, GZMB for CD8 + T cells, SSR4 for Dendritic cells, PLVAP for Endothelial cells, KRT5 for Epithelial cells, GZMK for Helper T cells, C1QB for Macrophages, TRSAB1 for Mast cells, G0S2 for Neutrophils, S100A8 for Progenitor cells, TNFRSF4 for Regulatory T cells, and PIGR for Secretory cells To further analyze the filtered cells, we identified highly variable genes based on gene expression variance and selected the top 2000 genes with the highest variability for downstream analysis (Fig. [128]1B). We computed the cell cycle scores of the samples using the CellCycleScoring function (Figure [129]S2C) and performed preliminary data normalization. Subsequently, we employed principal component analysis to linearly reduce the dimensionality of the data and visualized the results in a scatter plot (Fig. [130]1C). The heatmap in Figure [131]S2D shows the expression patterns of the top correlated genes with principal components PC_1 to PC_6. Batch correction of the sample data was performed using the harmony package (Figure [132]S2E-F), and the standard deviations of the principal components were ranked using the ElbowPlot method (Fig. [133]1D). The results indicate that PC_1 to PC_30 effectively capture the information contained in the highly variable genes with significant analytical implications. Next, we employed the UMAP algorithm to perform nonlinear dimensionality reduction of the top 30 principal components and conducted cluster analysis using a resolution of 1 (Figure [134]S3). Through this clustering analysis, we identified 31 clusters and determined marker gene expression profiles for each cluster (Fig. [135]1E). We annotated the cells by identifying known cell lineage-specific marker genes from the literature and incorporating information from the online database CellMarker (Fig. [136]1F, Figure [137]S2G). We identified a total of 13 cell types: B Cell, Basal Cell, CD8^+ T Cell, Dendritic Cell, Endothelial Cell, Epithelial Cell, Helper T Cell, Macrophage, Mast Cell, Neutrophil, Progenitor Cell, Regulatory T Cell, and Secretory Cell (Fig. [138]1G-H). Taken together, these results demonstrate that the ESCC samples, along with their adjacent normal samples, can be divided into 31 clusters, representing 13 distinct cell subtypes. Variability and radiation sensitivity of malignant epithelial cells in esophageal squamous cell carcinoma To investigate the variations of the aforementioned cell types in ESCC samples, we first validated the cell annotation results, which showed accurate annotation (Figure [139]S4). Subsequently, we extracted malignant cells from the Epithelial Cell using the “inferCNV” package and scored their radiosensitivity based on the expression of radioresistant genes (LAMA5, LAMB2, LAMB3, and ITGA6) in esophageal cancer cells [[140]74] to distinguish between radiotherapy-sensitive cancer cells and radiotherapy-resistant cancer cells (Fig. [141]2A). Fig. 2. [142]Fig. 2 [143]Open in a new tab Extraction of Malignant Cells and Cell Communication Analysis in the [144]GSE188900 Dataset. Note: (A) Schematic of the analysis for selecting radiation-sensitive cancer cells from scRNA-seq data in the [145]GSE188900 dataset; (B) CNV analysis from scRNA-seq data, with macrophages as reference, CD8^+ T cells in yellow, Epithelial Cells in cyan, blue indicating DNA copy number loss, and red indicating DNA copy number gain; (C) K-means clustering of cells from “observation” in diagram A; (D) CNV scores after K-means clustering; (E) UMAP clustering of scRNA-seq data after malignant cell extraction in the [146]GSE188900 dataset; (F) UMAP clustering of malignant cells; (G) Gene ranking after re-sorting to exclude genes with a q-value greater than 0.05 in malignant cells; (H) Visualization of cell annotations in ESCC samples (N = 7) based on UMAP clustering, with each color representing a cell subpopulation; (I) UMAP clustering of scRNA-seq data in the [147]GSE188900 dataset after selection of radiation-sensitive cancer cells; (J-K) Cell communication difference circle plots from cellchat analysis in ESCC samples (N = 7). The thickness of the lines in panel J represents the number of pathways, while in panel K, it represents interaction strength; (L) Differential cell communication between RS Cells and RR Cells in ESCC samples (N = 7), with line thickness indicating interaction strength. The Normal group contains one adjacent normal tissue sample, and the Tumor group consists of seven ESCC tumor tissue samples from five different patients As a control, we used CD8^+ T cells to extract malignant cells through the detection of large-scale copy number variations (Copy Number Variation, CNV). The results showed that most of the epithelial cells underwent CNV (Fig. [148]2B). Subsequently, we clustered the normal and malignant cells into eight categories using the “Observation” method. The results showed that the second category almost exclusively contained normal cells with the lowest CNV score. After excluding the second category, we obtained 4417 malignant cells (Fig. [149]2C-D). Furthermore, we clustered the obtained malignant cells into eight categories (Fig. [150]2D, Figure [151]S5). We also clustered the malignant epithelial cells into eight categories (Fig. [152]2E-F). Then, using the AddModuleScore function and the radioresistant gene set, we scored the radiosensitivity of these eight categories of malignant cells, as shown in Fig. [153]2G. Clusters 0, 2, 3, 5, 6, 8, and 11 had significantly higher radiosensitivity scores compared to clusters 1, 4, 7, 9, and 10. Therefore, we identified clusters 0, 2, 3, 5, 6, 8, and 11 as radiotherapy-sensitive cells (RS cells) and clusters 1, 4, 7, 9, and 10 as radiation-resistant cells (RR cells) (Fig. [154]2H-I). To further explore the functional differences between different cell types in the Normal and Tumor groups, we used the R package “CellChat” to investigate pathway activity between different cells. Figure [155]2J-K showed the number of cell-cell communications and interaction strengths between the eight cell subtypes, where RS cells and CD8^+ T cells exhibited higher numbers and strengths of interactions. Additionally, we compared the differences in cell communication between the two groups of samples, and the results showed that RS cells had significantly stronger contact and interaction with CD8^+ T cells compared to RR cells (Fig. [156]2L). This result suggests a close relationship between the radiosensitivity of tumor cells and CD8^+ T cells. Several studies have found that tumor cells inhibit CD8^+ T cell infiltration and develop radioresistance after radiation therapy [[157]75–[158]77]. In summary, the Epithelial Cells derived from ESCC samples can be divided into malignant epithelial cells and non-malignant epithelial cells. Furthermore, malignant cells can be further classified into radiotherapy-resistant cancer cells and radiotherapy-sensitive cancer cells. The results of cell communication analysis suggest the radiosensitivity of tumor cells may be closely related to CD8^+ T cells. Resistance of PD-1 in ^125I radioactive particle therapy for ESCC In recent years, radioactive plant therapy using ^125I radioactive particles has emerged as a widely accepted palliative treatment method for advanced esophageal cancer [[159]78, [160]79]. Compared to external radiation therapy, ^125I radioactive particle therapy is a low-energy gamma radiation treatment method with high tissue penetration depth and a relatively long half-life. By implanting radioactive plants into the tumor site, this therapy can deliver higher doses of radiation to the target area while minimizing collateral damage and complications [[161]80, [162]81]. Numerous studies have indicated that ^125I radioactive particles can induce DNA damage and cell apoptosis directly and indirectly through ionization and the generation of reactive oxygen species (ROS) [[163]82, [164]83]. Therefore, we constructed a murine xenograft model to observe the esophageal tissue. The esophagus of normal mice exhibited an intact structure with clear delineation between layers, while the esophagus of the modeled mice showed disrupted tissue structure, increased cell density, and the presence of tumor cells, confirming the successful establishment of the ESCC mouse model (Figure [165]S6A). We selected ^125I radioactive particles as the treatment modality for ESCC and performed high-throughput sequencing on CD8^+ T cells in the ESCC tissue of mice before and after treatment to explore genes or pathways closely related to the sensitivity of CD8^+ T cells to ESCC radiation therapy (Fig. [166]3A). The sequencing data of the ESCC tissue from the control group (pre-treatment) and the treatment group (post-treatment) were processed and differentially analyzed, revealing 135 significantly differentially expressed mRNAs in CD8^+ T cells of the Treat group compared to the Control group (Fig. [167]3B). Gene enrichment analysis of these 135 genes showed enrichment in biological processes such as mRNA processing, positive regulation of cellular catabolic process, and regulation of translation. KEGG enrichment analysis showed that differentially expressed genes were mainly enriched in signaling pathways such as the T cell receptor signaling pathway, PD-L1 expression and PD-1 checkpoint pathway in cancer, and apoptosis (Figure [168]S6B-C). GO and KEGG enrichment analysis results indicate a close association between differentially expressed genes in CD8^+ T cells of ESCC tissue and pathways such as PD-1. Fig. 3. [169]Fig. 3 [170]Open in a new tab Machine Learning Algorithm Selection of ESCC Radiation Sensitivity-related mRNA. Note: (A) Schematic of mouse modeling and high-throughput sequencing analysis process; (B) Volcano plot of differentially expressed mRNAs between CD8^+ T cells from 3 control and 3 treated mice in ESCC tissues; (C) Candidate target gene interaction network, with each circle representing a gene, lines indicating interactions between genes, darker colors signifying more interaction partners and higher degree values indicating a more central role in the network; (D) Core gene adjacency node count statistics, with the x-axis representing adjacency node counts and the y-axis gene names; (E) Heatmap revealing differential expression of 7 candidate target genes in sequencing data between control (N = 3) and treated (N = 3) samples; (F) Lasso coefficient screening plot, where the curve shows how the model’s deviation changes under different λ (regularization parameter) values. The numbers below the curve indicate the number of non-zero coefficients in the model at the corresponding λ values. The most stable part of the curve represents the optimal λ value, balancing the trade-off between minimized deviation and model complexity (number of non-zero coefficients); (G) Random forest algorithm results, where each point represents a feature mRNA, and its position indicates its importance in the random forest model. Mean Decrease Gini measures the quality of a feature’s split, with higher values indicating greater importance in the model; (H) SVM-RFE analysis results, where the curve shows how the model’s prediction error (expressed as Root Mean Square Error, RMSE) changes as the number of features decreases (represented on the horizontal axis). A decrease in RMSE indicates improved model accuracy, and the point where RMSE starts to increase significantly represents the optimal number of features; (I) Venn diagram showing the intersection of related mRNAs selected by lasso regression, random forest algorithm, and SVM-RFE; (J) Expression of PDCD-1 in sequencing data from control (N = 3) and treated (N = 3) groups, ***P < 0.001. In the volcano plot, blue points represent significantly downregulated mRNAs in the Treat group, red points represent significantly upregulated mRNAs, and gray points represent mRNAs with no significant difference. The Control group represents RNA-seq results from CD8^+ T cells in mouse ESCC tissues before 125I radiotherapy, while the Treat group represents RNA-seq results from CD8^+ T cells in mouse ESCC tissues after 125I radiotherapy Further analysis was conducted on the 135 candidate target genes, and a gene interaction network (Fig. [171]3C) was constructed. The number of adjacent nodes for each gene in the network was calculated, revealing that seven genes had an adjacency node count greater than or equal to 10 (RETNLA, GM49394, EDNRB, PDCD1, CSF2, ALDH1A2, and IGF2) (Fig. [172]3D). GM49394, CSF2, and PDCD1 showed downregulated expression in the sequencing data, while RETNLA, EDNRB, ALDH1A2, and IGF2 demonstrated upregulated expression (Fig. [173]3E). Subsequently, these seven genes underwent LASSO regression for multivariate Cox analysis (Fig. [174]3F), and the random forest algorithm was utilized to assess gene importance (Fig. [175]3G). Additionally, the related genes were extracted using the SVM-RFE analysis method (Fig. [176]3H). Finally, the key mRNA obtained was PD-1 (Fig. [177]3I), and the expression pattern of PD-1 in the sequencing data is illustrated in Fig. [178]3J. PDCD1, also known as PD-1, is a regulatory protein in T lymphocytes that inhibits immune cytotoxicity and typically binds to the ligand protein PD-L1 on tumor cells, enabling PD-L1-expressing tumor cells to evade clearance by T lymphocytes [[179]84–[180]86]. These results indicate that the high-throughput sequencing analysis identified one essential gene for ESCC, namely PD-1, which may play a role in conferring resistance during the process of ^125I radioisotope therapy for ESCC cells. Regulation of PD-1 on the activity and cytotoxicity of CD8^+ T cells in esophageal cancer under ^125I radioisotope therapy To verify the aforementioned bioinformatics analysis results, we first co-cultured human CD8^+ T cells with ESCC cells, establishing an in vitro co-culture model. Additionally, we constructed an external cell model for ^125I radiotherapy by utilizing a ^125I irradiation model. KYSE30 cell line was chosen for the mechanistic validation since it is commonly used for radiosensitivity-related experiments. First, we examined the expression level of PD-1 in CD8^+ T cells before (Control) and after (Treat) ^125I radiotherapy. The results showed a consistent increase in PD-1 expression after the radiotherapy (Fig. [181]4A). To investigate the role of PD-1 in CD8^+ T cells during ^125I radiotherapy for esophageal cancer, we employed CRISPR/Cas9 gene editing technology to construct PD-1 knockout (PD-1-KO) CD8^+ T cells, using PD-1 wild type (PD-1-WT) cells as the control (Figure [182]S7A). The expression level of PD-1 was detected in the PD-1-KO CD8^+ T cells by RT-qPCR and Western blot, followed by the selection of monoclonal cells with zero PD-1 expression for subsequent expansion and cultivation as experimental subjects (Fig. [183]4B). Concurrently, we constructed CD8^+ T cells overexpressing PD-1 (oe-PD-1) using lentivirus, with oe-negative control (oe-NC) as the control. RT-qPCR and Western blot were employed to validate the transfection efficiency of PD-1, and the results demonstrated a significant increase in PD-1 expression after transfection (Fig. [184]4C, Figure [185]S7B). Fig. 4. [186]Fig. 4 [187]Open in a new tab The Effect of PD-1 on CD8^+ T Cell Activity and Cytotoxicity in Esophageal Cancer Tissues After ^125I Radiation Therapy. Note: (A) RT-qPCR analysis of PD-1 expression levels in CD8^+ T cells from an in vitro co-culture model before and after 125I radiation therapy; (B) RT-qPCR and Western blot analysis of PD-1 mRNA and protein expression levels in PD-1-KO cells constructed using CRISPR/Cas9 gene-editing technology and in PD-1-overexpressing cells transfected with lentivirus; (C) RT-qPCR and Western blot detection of PD-1 mRNA and protein expression levels in CD8 + T cells from mouse tumor tissues after PD-1 knockdown or overexpression by lentivirus; (D) MTT assay measuring the activity of CD8 + T cells after PD-1 silencing or overexpression; (E) Mechanism diagram of CD8 + T cells killing tumor cells (CD8 + T cells secrete GZMB and IFN-γ, which recognize and kill tumor cells); (F) ELISA detection of GZMB and IFN-γ levels in CD8 + T cell culture media after PD-1 silencing or overexpression; (G-H) Flow cytometry analysis of the number of IFN-γ and GZMB-positive CD8^+ T cells after PD-1 silencing or overexpression; (I) Immunohistochemical staining to detect the protein expression levels of CD8α in mouse tumor tissues (scale bar: 50 μm); (J) ELISA detection of GZMB and IFN-γ levels in tumor tissues. *P < 0.05, **P < 0.01 compared to the sh-PD-1 group, PD-1-WT + RT group or sh-PD-1 + RT group, #P < 0.05 compared to the oe-NC, oe-NC + RT group, $P < 0.05, $$P < 0.01 compared to Control group. Cell experiments were repeated three times, with 6 mice per group We then utilized an MTT assay kit to assess CD8 + T cell viability. The results revealed a significant increase in CD8 + T cell viability in the 125I RT group compared to the untreated control group. Furthermore, CD8 + T cell viability in the PD-1 knockout treatment group (PD-1-KO + RT) was notably higher than in the PD-1-WT + RT group. Conversely, the viability of CD8 + T cells in the PD-1 overexpression treatment group (oe-PD-1 + RT) was significantly reduced compared to the oe-NC + RT group (Fig. [188]4D). Literature has shown that granzyme B (GZMB) and interferon-γ (IFN-γ) secreted by CD8 + T cells enhance their ability to identify and kill tumor cells [[189]87] (Fig. [190]4E). ELISA assays measuring the secretion of GZMB and IFN-γ in cell supernatants from different groups indicated a substantial increase in these cytokines in the RT and PD-1-KO + RT groups compared to the Control and PD-1-WT + RT groups, while a significant decrease was observed in the oe-PD-1 + RT group compared to the oe-NC + RT group (Fig. [191]4F). Flow cytometry analysis showed that the number of GZMB and IFN-γ positive cells was significantly increased in the PD-1 knockout group and reduced in the PD-1 overexpression group (Figs. [192]4G-H). Subsequently, we injected 1 × 10^8 TU/mL of lentivirus into the tumor tissues of mice to knock down or overexpress PD-1. The transfection efficiency was validated by RT-qPCR and Western blot, confirming successful overexpression in mice. Among the knockdown constructs, sh-PD-1#2 showed the best efficiency, so sh-PD-1#2 was selected for subsequent experiments (Fig. [193]4B). Additionally, immunohistochemistry and ELISA results revealed that compared with the Control group, CD8 + T cells and the secretion of GZMB and IFN-γ were significantly increased in the RT group. Compared with the sh-NC + RT group, CD8 + T cells and the secretion of GZMB and IFN-γ were significantly increased in the sh-PD-1 + RT group. However, compared with the oe-NC + RT group, CD8 + T cells and the secretion of GZMB and IFN-γ were significantly reduced in the oe-PD-1 + RT group (Figs. [194]4I-J). These results suggest that ^125I radiotherapy can promote the infiltration of CD8^+ T cells in tumor tissue, while PD-1 can suppress the activity and cytotoxic effects of CD8^+ T cells in esophageal cancer tissue after ^125I radiotherapy. Impact of PD-1 Inhibition on the radiotherapy resistance of ESCC cells mediated by CD8^+ T cells To further investigate the impact of PD-1 expression levels in CD8^+ T cells on ESCC cells and tissue, we assessed the vitality, proliferation, migration, invasion, and apoptosis of KYSE30 cells using CCK-8, EdU labeling, Transwell assay, and TUNEL staining (Fig. [195]5A). Additionally, we conducted a radiation clonogenic assay to examine the influence on the radiosensitivity of KYSE30 cells. The results showed that compared to the Control group, the RT group exhibited significantly reduced cell vitality, proliferation, migration, and invasion, accompanied by increased apoptosis. When compared to the PD-1-WT + RT group, the PD-1-KO + RT group displayed significantly decreased cell vitality, proliferation, migration, and invasion, as well as increased apoptosis, resulting in a lower survival rate following ionizing radiation. Moreover, in comparison to the oe-NC + RT group, the oe-PD-1 + RT group showed significantly increased cell vitality, proliferation, migration, and invasion, along with reduced apoptosis and a higher survival rate following ionizing radiation (Fig. [196]5B-G). These findings indicate that knocking out PD-1 increases the sensitivity of KYSE30 cells to radiation therapy, while overexpressing PD-1 leads to increased resistance. Fig. 5. [197]Fig. 5 [198]Open in a new tab The Effect of PD-1 on the Biological Functions of KYSE30 Cells. Note: (A) Schematic of experiments assessing the impact of PD-1 on the biological functions of esophageal cancer cells, including MTT, EdU, Transwell, and TUNEL staining; (B) MTT assay to evaluate the effects of RT (radiotherapy) or co-incubation with PD-1 knockout or overexpressed CD8 + T cells on the viability of KYSE30 cells in each group; (C) EdU assay to assess the proliferation capacity of KYSE30 cells in each group after RT or co-incubation with PD-1 knockout or overexpressed CD8 + T cells (scale bar: 25 μm); (D-E) Transwell assay to detect the migration and invasion abilities of KYSE30 cells in each group after RT or co-incubation with PD-1 knockout or overexpressed CD8 + T cells (scale bar: 50 μm); (F) TUNEL assay to measure the apoptosis rate of KYSE30 cells in each group after RT or co-incubation with PD-1 knockout or overexpressed CD8 + T cells (scale bar: 25 μm); (G) Radiation clonogenic survival assay to assess the survival rate of KYSE30 cells after RT or co-incubation with PD-1 knockout or overexpressed CD8 + T cells; (H) Morphological characteristics of tumor tissues in mice from each group after RT or treatment with PD-1-silenced or overexpressed lentivirus; (I) Effects of RT or treatment with PD-1-silenced or overexpressed lentivirus on tumor growth in mice from each group; (J) Effects of RT or treatment with PD-1-silenced or overexpressed lentivirus on tumor weight in mice from each group; (K) Immunohistochemical staining to detect the effects of RT or treatment with PD-1-silenced or overexpressed lentivirus on the protein expression levels of Ki67 in tumor tissues of mice from each group (scale bar: 50 μm); (L) TUNEL assay to measure the apoptosis rate in tumor tissues of mice from each group after RT or treatment with PD-1-silenced or overexpressed lentivirus (scale bar: 50 μm). *P < 0.05 compared to PD-1-WT + RT group or sh-PD-1 + RT group, #P < 0.05 compared to oe-NC + RT group, $P < 0.05 compared to Control group. Cell experiments were repeated three times. Each mouse group consisted of six mice Furthermore, we conducted in vivo animal experiments to validate the aforementioned results. The results demonstrated that both ^125I radioactive particle treatment and PD-1 knockout from CD8^+ T cells significantly inhibited tumor growth. However, compared to the oe-NC + RT group, the tumor tissue in the oe-PD-1 + RT group exhibited a significant increase in size (Fig. [199]5H-J). TUNEL staining and Ki67 staining were employed to assess tumor cell apoptosis and proliferation rates, respectively. The results revealed an increase in radiation-induced tumor cell apoptosis in the RT group and sh-PD-1 + RT group compared to the Control group and sh-NC + RT group (Fig. [200]5L). Furthermore, the proportion of Ki67-positive cells in tumor tissue was significantly lower after radiation particle treatment compared to the untreated group, and overexpression of PD-1 resulted in a significant increase in the proportion of Ki67-positive cells, while knocking down PD-1 led to a significant decrease in Ki67-positive cells in tumor tissue (Fig. [201]5K). These findings demonstrate that PD-1 promotes resistance of ESCC cells to ^125I radiation particle treatment by inhibiting the activity and cytotoxicity of CD8^+ T cells. Enhancement of CD8^+ T cell activity and cytotoxicity by Nivolumab-conjugated oxygen nanobubbles after ^125I radiotherapy The resistance of tumors to radiotherapy and the poor therapeutic effect of ^125I radioactive particles are related to tumor hypoxia [[202]88–[203]90]. We have previously demonstrated that the upregulation of PD-1 in CD8^+ T cells leads to the resistance of ESCC to ^125I radioactive particle therapy by impairing their activity and cytotoxicity. In order to address this issue, we propose to develop a nanomaterial that can selectively inhibit the PD-1 protein in CD8^+ T cells within tumor tissue, thereby enhancing the sensitivity of ESCC to ^125I radioactive particle therapy. Additionally, we aim to utilize an oxygen-enhanced environment to further reverse the radiotherapy resistance in ESCC cell growth. Nivolumab is an immune checkpoint inhibitor that belongs to a class of monoclonal antibodies that enhance the immune system’s attack on cancer cells by inhibiting the immune checkpoint molecule, PD-1 (programmed cell death 1) [[204]91–[205]93]. Ethylenediamine-modified dextran oxygen-carrying nanobubbles, which exhibit excellent stability and biocompatibility, can spontaneously generate oxygen in response to a slight decrease in pH in the tumor microenvironment. Encapsulated in an acetylated dextran polymer shell, these nanobubbles serve as a powerful barrier to prevent gas dissolution in the circulating blood, thereby retaining the majority of the oxygen payload. Their pH-responsive characteristics allow for the sudden release of oxygen in mildly acidic tumor microenvironments [[206]40]. In this study, they are utilized as oxygen carriers. ICG, a photosensitizer in this study, has high absorption and fluorescence emission peaks in the near-infrared (NIR) region, which helps to avoid the interference from visible light in fluorescence imaging and achieve real-time high-resolution imaging of tumors. Furthermore, it can produce ROS under laser irradiation, exerting toxicity on cells [[207]94]. The synthetic schematic is illustrated in Fig. [208]6A. Fig. 6. [209]Fig. 6 [210]Open in a new tab Characterization of NBs and Their Impact on CD8^+ T Cells. Note: (A) Schematic of the synthesis of pH-responsive oxygen-carrying αPD1-O2-NB; (B) Scanning electron microscopy (top) and transmission electron microscopy (bottom) images of pH-responsive oxygen-carrying nanobubbles; (C) Measurement of the average diameter of N2-NB, O2-NB, αPD1-N2-NB and αPD1-O2-NB group; (D) DLS measurement of the zeta potential of NB solutions in each group; (E) PDI results for each NB group; (F) Fluorescence microscopy observation of antibody conjugation results for each NB group, with fluorescence indicating successful coupling with biotinylated antibody Nivolumab; (G) Changes in dissolved oxygen concentration over time in NB solutions at pH = 6.5 or pH = 7.4; (H) Schematic of AC-DEX shell degradation of nanobubbles under acidic conditions; (I) Attachment of NBs to CD8^+ T cells in each group; (J-K) Flow cytometry detection of binding rates of NBs to CD8 T cells; (L) MTT assay for the effect of each NB group on the vitality of CD8^+ T cells co-cultured with KYSE30 cells; (M-N) ELISA measurement of the effect of each NB group on GZMB and IFN-γ secretion by CD8^+ T cells co-cultured with KYSE30 cells; (O-Q) Flow cytometry analysis of the number of IFN-γ and GZMB positive CD8^+ T cells co-cultured with KYSE30 cells in each group. *P < 0.05, **P < 0.01, ***P < 0.001, cell experiments were repeated three times The prepared NBs did not show any significant detachment phenomenon after standing at room temperature. Optical microscopy and transmission electron microscopy images revealed that the NBs were spherical, uniformly sized, and evenly distributed (Fig. [211]6B). There were no significant differences in the average diameter, ζ-potential, and PDI among the NBs groups (Fig. [212]6C-E). The diameter distribution of each group of NBs is shown in Figure [213]S8A-D. After incubation with FITC-labeled secondary antibodies, αPD1-N[2]-NB and αPD1-O[2]-NB exhibited green fluorescence, indicating successful binding with biotinylated antibodies (Fig. [214]6F). Flow cytometric analysis quantitatively determined the antibody binding efficiency of each group of NBs, showing that αPD1-N[2]-NB and αPD1-O[2]-NB groups achieved over 90% successful binding with anti-CD3 antibodies or control IgG (Figure [215]S8E). The average size and concentration of the N[2]-NB and O[2]-NB groups and the αPD1-N[2]-NB and αPD1-O[2]-NB groups were similar at different time points (Figure [216]S8F-G). The NBs in all groups demonstrated good stability in terms of diameter (Figure [217]S8F) and concentration (Figure [218]S8G) within 6 h after preparation. The pH-responsive oxygen-carrying NBs were investigated by monitoring the kinetics of oxygen release in low-oxygen solutions with different pH values. Various groups of NBs were added to acidic (pH = 6.5) or neutral (pH = 7.4) low-oxygen solutions. The dissolved oxygen concentration in the mixed solutions was recorded over time using a portable oxygen meter under a nitrogen atmosphere (Fig. [219]6G). It was observed that the N[2]-NB and αPD1-N[2]-NB groups showed no significant change in dissolved oxygen concentration and remained close to zero. On the other hand, the O[2]-NB and αPD1-O[2]-NB groups exhibited a sharp increase in dissolved oxygen concentration within one minute, primarily due to the contribution of oxygen from the oxygen-carrying nanobubbles. The concentration then slightly decreased and reached a relatively stable level as the dissolved oxygen diffused into the nitrogen atmosphere. It is worth noting that the rate of decrease in dissolved oxygen concentration was slower in acidic solution compared to normal solution, which might be attributed to faster degradation of AC-DEX shell of nanobubbles under acidic conditions, resulting in faster oxygen release (Fig. [220]6H). Further analysis of the attachment of Nivolumab or prepared NBs to CD8^+ T cells was carried out. Figure [221]6I revealed that a significant number of nanobubbles from the αPD1, αPD1-N[2]-NB, and αPD1-O[2]-NB groups were bound to CD8^+ T cells, while only a few nanobubbles from the N[2]-NB and O[2]-NB groups were attached to the cell surface. Flow cytometry further confirmed the binding rate with CD8^+ T cells. When compared to the N[2]-NB and O[2]-NB groups, CD8^+ T cells incubated with diI-labeled αPD1, αPD1-N[2]-NB, and αPD1-O[2]-NB groups showed a higher percentage of positive fluorescence (Fig. [222]6J-K). Subsequently, the effect of Nivolumab or prepared NBs in inhibiting PD-1 to enhance the activity and killing effect of CD8^+ T cells in esophageal cancer tissue after ^125I radiotherapy was examined to further improve the sensitivity of esophageal cancer cells to ^125I radiation therapy. Initially, CD8^+ T cells co-cultured with various groups of NBs were incubated with esophageal cancer cells. The viability of CD8^+ T cells was measured using the MTT assay, and the secretion of GZMB and IFN-γ cytokines in the supernatant of different cell groups was analyzed using ELISA. The results showed that compared to the untreated group (Blank), the CD8^+ T cells’ viability and the secretion of GZMB and IFN-γ significantly increased in the ^125I radiotherapy group (^125I). The N[2]-NB + ^125I and O[2]-NB + ^125I groups showed no significant changes in the viability of CD8^+ T cells and the secretion of GZMB and IFN-γ when compared to the ^125I group, indicating the low toxicity and therapeutic effects of NBs carriers on CD8^+ T cells. However, compared to the ^125I group, the αPD1 + ^125I group, αPD1-N[2]-NB + ^125I group, and αPD1-O[2]-NB + ^125I group exhibited significantly increased CD8^+ T cell viability and increased secretion of GZMB and IFN-γ. Moreover, the therapeutic effect of the αPD1-O[2]-NB + ^125I group was more pronounced than that of the αPD1 + ^125I group and αPD1-N[2]-NB + ^125I group (Fig. [223]6L-N, Figure [224]S9A-C). The trends revealed by flow cytometry were consistent with the results obtained from ELISA (Fig. [225]6O-Q, Figure [226]S9D-F). Coupling Nivolumab with pH-responsive oxygen-carrying nanobubbles can enhance the sensitivity of esophageal cancer cells to ^125I radiation therapy by inhibiting the PD-1 pathway of CD8^+ T cells, and the oxygen carried by the nanobubbles can enhance the efficacy of Nivolumab. In summary, the pH-responsive oxygen-carrying nanobubbles coupled with Nivolumab can inhibit PD-1, thereby enhancing the activity and killing effect of CD8^+ T cells in esophageal cancer tissue after ^125I radiotherapy, and further improving the sensitivity of esophageal cancer cells to ^125I radiation therapy. Enhancement of sensitivity of ESCC to radiation therapy by αPD1-O[2]-NB To further explore the effects of αPD1-O2-NB on enhancing the activity and cytotoxicity of CD8^+ T cells against ESCC cells and tissues, we used the CCK-8 assay to detect the viability of KYSE30 and TE-1 cells, the EdU assay to assess the proliferation capacity of KYSE30 and TE-1 cells, the Transwell assay to measure the migration and invasion abilities of KYSE30 and TE-1 cells, the TUNEL staining assay to evaluate apoptosis in KYSE30 and TE-1 cells, and the radiation clonogenic assay to examine the effects on radiosensitivity of KYSE30 and TE-1 cells. Western Blot was used to detect the expression of DNA damage markers pATM and γ-H2AX in each group of cells. The results showed that, compared with the Blank group, the 125I group exhibited significantly reduced cell viability, proliferation, migration, and invasion abilities, with an increase in apoptosis and the expression of DNA damage markers pATM and γ-H2AX. Compared with the non-conjugated Nivolumab groups (N[2]-NB + ^125I / O2-NB + ^125I), the Nivolumab-conjugated groups (αPD1 / αPD1-N[2]-NB + ^125I / αPD1-O[2]-NB + ^125I) showed significantly reduced cell viability, proliferation, migration, and invasion abilities, with a marked increase in apoptosis, lower survival rates in response to IR, and increased expression of pATM and γ-H2AX. Compared with nitrogen-carrying nanobubbles, the oxygen-carrying nanobubble treatment group exhibited significantly suppressed cell viability, proliferation, migration, and invasion abilities, with a notable increase in apoptosis, lower survival rates in response to IR, and increased expression of pATM and γ-H2AX, indicating that oxygen can enhance the sensitivity of ESCC cells to ^125I radiotherapy. Compared with the αPD1 + ^125I group, the αPD1-N[2]-NB + ^125I group showed a significant decrease in cell viability, proliferation, migration, and invasion abilities, with a marked increase in apoptosis, lower survival rates in response to IR, and increased expression of pATM and γ-H2AX, which may be due to the fact that delivering Nivolumab via nanobubbles can reduce its degradation. Compared with the αPD1-N[2]-NB + ^125I group, the αPD1-O[2]-NB + ^125I group showed a significant decrease in cell viability, proliferation, migration, and invasion abilities, with a marked increase in apoptosis, lower survival rates in response to IR, and increased expression of pATM and γ-H2AX (Fig. [227]7A-G, Figure [228]S10A-G). Fig. 7. [229]Fig. 7 [230]Open in a new tab The Impact of αPD1-O2-NB on the Biological Functions of KYSE30 Cells. Note: (A) MTT assay to evaluate the effects of ^125I or various groups of NBs on the viability of KYSE30 cells; (B) EdU assay to assess the effects of ^125I or various groups of NBs on the proliferation capacity of KYSE30 cells (scale bar: 25 μm); (C-D) Transwell assay to detect the effects of ^125I or various groups of NBs on the migration and invasion abilities of KYSE30 cells (scale bar: 50 μm); (E) TUNEL assay to measure the apoptosis rate of KYSE30 cells after treatment with ^125I or various groups of NBs (Scale bar = 25 μm); (F) Radiation clonogenic survival assay to assess the survival rate of KYSE30 cells after treatment with ^125I or various groups of NBs; (G) Western Blot to detect the expression of DNA damage markers pATM and γ-H2AX in cells of each group. *P < 0.05 compared to N [2] -NB +  ^125I group, #P < 0.05 compared to αPD1-N [2] -NB +  ^125I group, $P < 0.05 compared to Blank group, & P < 0.05 compared to ^125I group, cell experiments were repeated three times Mechanism of αPD1-O[2]-NB-enhanced ^125I radiotherapy Finally, the mechanisms underlying the enhanced therapeutic effect of αPD1-O[2]-NB with ^125I radioisotope were validated in vivo. First, a humanized subcutaneous xenograft model (PDX) in mice was established by subcutaneous injection of KYSE30/TE0-1 cells. Subsequently, hPBMCs were injected to construct a humanized PDX model, and the levels of human CD45 in mouse blood were measured 13 and 23 days after hPBMC injection (Fig. [231]8A). It was observed that the levels of human CD45 protein increased by more than 30% after 23 days of hPBMC injection, indicating the successful construction of a humanized model in mice (Figure [232]S11A). Fig. 8. [233]Fig. 8 [234]Open in a new tab In Vivo Validation of αPD1-O2-NB Regulation of KYSE30 Radiation Therapy Sensitivity. Note: (A) Schematic of constructing humanized subcutaneous xenograft (PDX) mouse models; (B) Observation of NIR fluorescence imaging on the impact of ^125I or various groups’ NBs on the in vivo fluorescence distribution in tumor-bearing nude mice; (C) NIR fluorescence imaging to observe the fluorescence distribution in major organs and tumor tissues of each nude mouse group; (D) Immunohistochemical staining for CD8α protein expression levels in tumor tissues from each mouse group (scale bar: 50 μm); (E) ELISA measurement of GZMB and IFN-γ levels in tumor tissues; (F) Morphology of tumor tissues from each mouse group; (G) Growth of tumors in each mouse group; (H) Weight of tumor tissues from each mouse group; (I) Immunohistochemical staining for Ki67 protein expression levels in tumor tissues from each mouse group (scale bar: 50 μm); (J) Western Blot analysis of the expression of DNA damage markers pATM and γ-H2AX in tumor tissues. *P < 0.05, **P < 0.01, ***P < 0.001, each group consisted of 6 mice Next, DIR-labeled Nivolumab or NBs were injected into mice via the tail vein, and the biodistribution of each group of NBs was evaluated by NIR imaging. In vivo imaging was performed within 24 h after tail vein injection of NBs, and organs and tumors were dissected for ex vivo imaging 24 h later. The results showed that within 24 h, the fluorescence signals of the αPD1 + DIR group, O[2]-NB group, αPD1-N[2]-NB group, and αPD1-O[2]-NB group were mainly located in the kidneys and tumor tissues. The αPD1 + DIR group and O[2]-NB group were rapidly metabolized from the kidneys due to the non-targeted nature and smaller particle size of DIR, resulting in a gradual decrease in accumulation at the tumor site. However, the fluorescence signals in tumor tissues of the αPD1-N[2]-NB group and αPD1-O[2]-NB group persisted for a longer duration, with the αPD1-N[2]-NB group and αPD1-O[2]-NB group showing the strongest fluorescence intensity. This was attributed to the targeting ability of Nivolumab, which prolonged its accumulation time at the tumor site (Fig. [235]8B-C, Figure [236]S11B-C). Immunohistochemical and ELISA experimental results revealed that compared to the Blank group, the proliferation ability of CD8^+ T cells in tumor tissues as well as the secretion of GZMB and IFN-γ significantly increased in the ^125I/O[2]-NB/αPD1 + DIR group. Comparing with the αPD1 + DIR/αPD1 + DIR + ^125I group and O[2]-NB/O[2]-NB + ^125I group, the proliferation ability of CD8^+ T cells in tumor tissues as well as the secretion of GZMB and IFN-γ significantly increased in the αPD1-N[2]-NB/αPD1-N[2]-NB + ^125I group and αPD1-O[2]-NB/αPD1-O[2]-NB + ^125I group of mice. Compared to the αPD1-N[2]-NB/αPD1-N[2]-NB + ^125I group, the proliferation ability of CD8^+ T cells in tumor tissues as well as the secretion of GZMB and IFN-γ significantly increased in the αPD1-O[2]-NB/αPD1-O[2]-NB + ^125I group of mice (Fig. [237]8D-E, Figure [238]S11D-E). Next, the weight and volume of tumor tissues in each mouse group were measured. Treatment with ^125I radioactive particles, αPD1 + DIR, αPD1-N[2]-NB, or αPD1-O[2]-NB resulted in significant inhibition of tumor growth. Compared to the αPD1 + DIR group, the αPD1-N[2]-NB group exhibited a noticeable reduction in tumor size. Furthermore, the tumor tissues in the αPD1-O[2]-NB/αPD1-O[2]-NB + ^125I group were significantly smaller compared to the αPD1-N[2]-NB/αPD1-N[2]-NB + ^125I group. Interestingly, the addition of O[2] further enhanced the therapeutic effect of ^125I radiation, consistent with previous literature reports [[239]88–[240]90] (Fig. [241]8F-H, Figure [242]S11F-H). Ki67 staining was conducted to assess tumor cell apoptosis and proliferation rates. After radiation treatment, the number of Ki67-positive cells in tumor tissues was significantly lower compared to the untreated group. Following treatment with αPD1 + DIR, αPD1-N[2]-NB, or αPD1-O[2]-NB, the proportion of Ki67-positive cells gradually decreased. Compared to the αPD1 + DIR group, the αPD1-N[2]-NB group exhibited a significant reduction in the number of Ki67-positive cells in mouse tumor tissues. Additionally, the αPD1-O[2]-NB/αPD1-O[2]-NB + ^125I group had significantly fewer Ki67-positive cells in tumor tissues compared to the αPD1-N[2]-NB/αPD1-N[2]-NB + ^125I group. Importantly, the addition of O[2] further lowered the percentage of Ki67-positive cells during ^125I radiation treatment (Fig. [243]8I, Figure [244]S11I). Western Blot detection of DNA damage markers pATM and γ-H2AX in tumor tissues: After radiotherapy with 125I seeds, the expression of pATM and γ-H2AX proteins in tumor tissues was significantly higher than in the untreated group. After treatment with αPD1 + DIR, αPD1-N2-NB, or αPD1-O2-NB, the levels of pATM and γ-H2AX gradually increased. Compared with the αPD1 + DIR group, the expression of pATM and γ-H2AX in the tumor tissues of the αPD1-N2-NB group was significantly higher. Moreover, the expression of pATM and γ-H2AX in the αPD1-O2-NB / αPD1-O2-NB + 125I group was significantly higher than in the αPD1-N2-NB / αPD1-N2-NB + 125I group. Additionally, oxygen further increased the expression of pATM and γ-H2AX proteins during 125I radiotherapy (Fig. [245]8J, Figure [246]S11J). These results indicate that pH-responsive oxygen-carrying nanobubbles conjugated with Nivolumab (αPD1-O2-NB) enhance DNA damage in ESCC cells after 125I radiotherapy by strengthening the function of CD8^+ T cells. Subsequently, major organs and tumor tissues were removed after completion of the treatment for H&E staining. No significant pathological changes were observed in the major organs, indicating that αPD1-O[2]-NB has good biocompatibility (Figure [247]S12). These results collectively demonstrate that the pH-responsive oxygen-carrying nanobubbles coupled with Nivolumabα (PD1-O[2]-NB) enhance the sensitivity of ESCC cells to ^125I radiation therapy by enhancing the function of CD8^+ T cells. Discussion This study elucidates the mechanism by which esophageal cancer cells enhance sensitivity to ^125I radiation therapy, accomplished by employing single-cell transcriptomics technology. This finding provides new theoretical and experimental groundwork for improving the efficacy of radiotherapy for esophageal cancer [[248]73, [249]95, [250]96]. Our research results demonstrate that after radiation therapy, the quantity and cytotoxic function of CD8^+ T cells in esophageal cancer cells increase, which is consistent with other research findings. Additionally, we have discovered that PD-1 plays a critical role in inhibiting the vitality and cytotoxic effects of CD8^+ T cells. Compared to previous studies, this research is innovative in its elucidation of the mechanism behind the sensitivity of esophageal cancer to radiotherapy. We clearly establish the relationship between the radiosensitivity of tumor cells and CD8^+ T cells through single-cell transcriptomic analysis and uncover the inhibitory role of PD-1. While consistent with existing research findings, our study further reveals the role of PD-1 in regulating the function of CD8^+ T cells. This discovery provides new targets for improving the response of tumor cells to radiotherapy [[251]97–[252]99]. The combined analysis of single-cell transcriptomics employed in this study enables the identification of radiation-sensitive cancer cells from the overall cell population. This provides a new approach to understanding the mechanism of tumor radiosensitivity. Compared to traditional whole-cell analysis, single-cell transcriptomics can provide more accurate information on cell types and states, thereby better deciphering the complexity and heterogeneity of tumors [[253]100, [254]101]. Therefore, the methodology of this study holds significant importance in uncovering the mechanisms of radiosensitivity. This study provides a comprehensive investigation into the mechanisms underlying the enhancement of radiotherapy sensitivity in esophageal cancer, with significant scientific and clinical value. Our findings reveal that the inhibitory effect of PD-1 leads to resistance of esophageal cancer cells to radiotherapy, thus providing a novel target for overcoming radiotherapy resistance. Additionally, we have developed a pH-responsive oxygen-carrying nanobubble coupled with Nivolumab, demonstrating its potential for clinical application. This approach enhances the effectiveness of immunotherapy while reducing unnecessary toxic side effects. Looking ahead to future research directions, further exploration of other potential mechanisms and targets regulating radiotherapy sensitivity is warranted. For example, the role of other immune checkpoint molecules in CD8^+ T cell function regulation can be investigated, as well as pathways co-regulated with PD-1 [[255]102, [256]103]. Moreover, it is essential to refine the technology and preparation methods of the pH-responsive oxygen-carrying nanobubble coupled with Nivolumab for deeper clinical investigations, ensuring its safety and efficacy. Based on the above results, we can tentatively draw the following conclusions: PD-1 promotes resistance to ^125I radiotherapy in ESCC cells by inhibiting the activity and cytotoxicity of CD8^+ T cells. The pH-responsive oxygen-carrying nanobubbles coupled with Nivolumab enhance the activity and cytotoxicity of CD8^+ T cells, thereby increasing the sensitivity of ESCC cells to ^125I radiotherapy. This study provides a novel therapeutic strategy to overcome resistance to conventional ^125I radiotherapy in ESCC, which is difficult to treat. Our research offers a theoretical basis for understanding the occurrence of ESCC and developing new therapeutic targets. Despite the important findings and potential applications of this study, there are still some limitations. Firstly, this study was conducted in a mouse model, and further clinical experiments are necessary to verify its feasibility and safety. Secondly, the study only involves a subset of key genes and cellular pathways, and there are other factors that may influence radiotherapy sensitivity that need to be further explored. The scope of this study is limited to mouse models, and to gain a more in-depth understanding of the mechanisms underlying radiotherapy sensitivity, future research can expand the sample size and conduct more clinical experiments. Additionally, attention should be given to other factors that may affect radiotherapy sensitivity, and further exploration of treatment strategies and targeted drugs is warranted. In summary, this study provides an important theoretical and experimental foundation for uncovering the mechanisms of radiotherapy sensitivity in esophageal cancer and developing novel radiotherapy sensitization strategies. Electronic supplementary material Below is the link to the electronic supplementary material. [257]12951_2025_3552_MOESM1_ESM.jpg^ (422KB, jpg) Supplementary Material 1: Figure S1. Schematic of the 125I irradiation model. [258]12951_2025_3552_MOESM2_ESM.jpg^ (452.4KB, jpg) Supplementary Material 2: Figure S2. Quality Control and PCA Dimension Reduction of scRNA-seq Data. Note: (A) Violin plots showing the distribution of gene counts (nFeature_RNA), mRNA molecule counts (nCount_RNA), and mitochondrial gene percentage (percent.mt) for each cell (N=8); (B) Scatter plots showing the correlation between nCount_RNA and percent.mt, nCount_RNA and nFeature_RNA, and nCount_RNA and percent.HB after data filtering (N=8); (C) Cell cycle status for each cell in the scRNA-seq data, with S.Score indicating S phase and G2M.Score indicating G2M phase (N=8); (D) Heatmap of the top 20 genes correlated with PC_1 – PC_6 in PCA, where yellow indicates upregulated expression and purple indicates downregulated expression (N=8); (E) Harmony batch correction process graph, with the x-axis representing interaction counts; (F) Cell distribution in PC_1 and PC_2 after Harmony batch correction, with each point representing a cell; (G) Expression of known lineage-specific marker genes in different clusters of ESCC samples (N=8). The Normal group contains one adjacent normal tissue sample, while the Tumor group consists of seven ESCC tumor tissue samples from five different patients. The marker gene for Basal Cells is SPRR1A, for CD8+ T Cells is GZMB, for Dendritic Cells is SSR4, for Endothelial Cells is PLVAP, for Epithelial Cells is KRT5, for Helper T Cells is GZMK, for Macrophages is C1QB, for Mast Cells is TRSAB1, for Neutrophils is G0S2, for Progenitor Cells is S100A8, for Regulatory T Cells is TNFRSF4, and for Secretory Cells is PIGR. [259]12951_2025_3552_MOESM3_ESM.jpg^ (903.8KB, jpg) Supplementary Material 3: Figure S3. UMAP Clustering Dendrogram of scRNA-seq Data. Note: Different colors represent different resolutions (RNA_snn_res.), the size of the circles represents the number of cells (size), the transparency of the arrows represents the proportion (in_prop), and the color of the arrows represents the count number (count). [260]12951_2025_3552_MOESM4_ESM.jpg^ (1.2MB, jpg) Supplementary Material 4: Figure S4. Distribution Scatter Plot of Marker Genes in scRNA-seq Cells. Note: The expression of marker genes for 13 cell types across different cell subtypes in adjacent normal tissue (Normal, N=1) and ESCC samples (Tumor, N=7) (B Cell: CD79A; Basal Cell: SPRR1A; CD8+ T Cell: NKG7, GZMB; Dendritic Cell: SSR4; Endothelial Cell: PLVAP; Epithelial Cell: KRT5, KRT6A; Helper T Cell: GZMK; Macrophage: C1QB; Mast Cell: TRSAB1; Neutrophil: G0S2; Progenitor Cell: S100A8; Regulatory T Cell: TNFRSF4; Secretory Cell: PIGR). The deeper the red color, the higher the average expression level. [261]12951_2025_3552_MOESM5_ESM.jpg^ (1.3MB, jpg) Supplementary Material 5: Figure S5. UMAP Clustering Dendrogram of Malignant Cell scRNA-seq Data. Note: Different colors represent different resolutions (RNA_snn_res.), the size of the circles represents the number of cells (size), the transparency of the arrows represents the proportion (in_prop), and the color of the arrows represents the count number (count). [262]12951_2025_3552_MOESM6_ESM.jpg^ (1.7MB, jpg) Supplementary Material 6: Figure S6. In Vivo Modeling Validation and Pathway Enrichment Analysis of Differentially Expressed Genes in High-throughput Sequencing Data. Note: (A) Esophageal appearance (left) and H&E-stained tissue sections (right) of normal (Normal group) mice and ESCC model (Tumor group) mice, scale bar: 200 µm, each group consisted of 6 mice; (B) Bubble chart of GO enrichment analysis for differentially expressed genes before and after 125I radiation therapy, BP represents Biological Process, CC represents Cell Component, MF represents Molecular Function; (C) KEGG clustering dendrogram for differentially expressed genes before and after ^125I radiation therapy. In both the GO enrichment bubble plot and the KEGG clustering tree, the color of the dots represents the P-value, and the size of the dots represents the counts. [263]12951_2025_3552_MOESM7_ESM.jpg^ (1.9MB, jpg) Supplementary Material 7: Figure S7. Validation of PD-1 Knockout and Overexpression Efficiency. Note: (A) Schematic diagram of PD-1 knockout CD8+ T cells constructed using CRISPR/Cas9 gene-editing technology; (B) Schematic diagram of PD-1-overexpressing CD8+ T cells constructed using lentiviral transfection. [264]12951_2025_3552_MOESM8_ESM.jpg^ (1.9MB, jpg) Supplementary Material 8: Figure S8. Physical Characterization of NBs. Note: (A-D) Average diameter distribution of N2-NB, O2-NB, αPD1-N2-NB and αPD1-O2-NB group; (E) Flow cytometry detection of coupling results between each NB group and Nivolumab; (F-G) Changes in average size and concentration of each NB group at different time points. Cell experiments were repeated three times. [265]12951_2025_3552_MOESM9_ESM.jpg^ (3MB, jpg) Supplementary Material 9: Figure S9. The Impact of NBs on CD8+ T Cells. Note: (A) MTT assay for the effect of N2-NB, O2-NB, αPD1-N2-NB and αPD1-O2-NB group on the vitality of CD8+ T cells co-cultured with TE-1 cells; (B-C) ELISA measurement of the effect of each NB group on GZMB and IFN-γ secretion by CD8+ T cells co-cultured with TE-1 cells; (D-F) Flow cytometry analysis of the number of IFN-γ and GZMB positive CD8+ T cells co-cultured with TE-1 cells in each group. *P < 0.05, **P < 0.01, ***P < 0.001, cell experiments were repeated three times. [266]12951_2025_3552_MOESM10_ESM.jpg^ (3.5MB, jpg) Supplementary Material 10: Figure S10. The Impact of αPD1-O2-NB on the Biological Functions of TE-1 Cells. Note: (A) MTT assay to evaluate the effects of 125I or various groups of NBs on the viability of TE-1 cells; (B) EdU assay to assess the effects of 125I or various groups of NBs on the proliferation capacity of TE-1 cells (scale bar: 25 μm); (C-D) Transwell assay to detect the effects of 125I or various groups of NBs on the migration and invasion abilities of TE-1 cells (scale bar: 50 μm); (E) TUNEL assay to measure the apoptosis rate of TE-1 cells after treatment with 125I or various groups of NBs (Scale bar=25 μm); (F) Radiation clonogenic survival assay to assess the survival rate of TE-1 cells after treatment with 125I or various groups of NBs; (G) Western Blot analysis to detect the expression of DNA damage markers pATM and γ-H2AX in each group of cells. *P < 0.05 compared to N2-NB + 125I group, #P < 0.05 compared to αPD1-N2-NB + 125I group, $P < 0.05 compared to Blank group, &P < 0.05 compared to 125I group, cell experiments were repeated three times. [267]12951_2025_3552_MOESM11_ESM.jpg^ (980.6KB, jpg) Supplementary Material 11: Figure S11. In Vivo Validation of αPD1-O2-NB Modulating TE-1 Cell Sensitivity to Radiation Therapy. Note: (A) Flow cytometry to detect the successful establishment of humanized PDX model mice; (B) NIR fluorescence imaging to observe the effects of 125I or various groups of NBs on the in vivo fluorescence distribution in tumor-bearing nude mice; (C) NIR fluorescence imaging to observe the effects of 125I or various groups of NBs on the fluorescence distribution in major organs and tumor tissues of nude mice; (D) Immunohistochemical staining to assess the effects of 125I or various groups of NBs on the protein expression levels of CD8α in mouse tumor tissues (scale bar: 50 μm); (E) ELISA to detect the effects of 125I or various groups of NBs on the levels of GZMB and IFN-γ in tumor tissues; (F) Effects of 125I or various groups of NBs on the morphology of mouse tumor tissues; (G) Effects of 125I or various groups of NBs on tumor growth in mice; (H) Effects of 125I or various groups of NBs on tumor tissue weight in mice; (I) Immunohistochemical staining to assess the effects of 125I or various groups of NBs on the protein expression levels of Ki67 in mouse tumor tissues (scale bar: 50 μm); (J) Western Blot analysis to detect the expression of DNA damage markers pATM and γ-H2AX in tumor tissues. *P < 0.05, **P < 0.01, ***P < 0.001, with 6 mice per group. [268]12951_2025_3552_MOESM12_ESM.jpg^ (11.4MB, jpg) Supplementary Material 12: Figure S12. H&E Staining of Major Organs and Tumor Tissues in Mice. Note: The images show H&E stained sections of the lungs, heart, liver, kidneys, spleen, and tumor tissues of mice treated with 125I or various groups' NBs. [269]Supplementary Material 13^ (23.1KB, docx) Acknowledgements