Sex as a biological variable. Human samples were deliberately sourced from male and female individuals. For mouse experiments, to avoid interference from androgen in antitumor effects, only female animals were used in this study.

Mice. Female C57BL/6J and BALB/c mice (female, 7 weeks) were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. Female NSG mice (age, 7 weeks) were purchased from SPF Biotechnology Co., Ltd. All mice were fed under specific pathogen–free conditions in the animal facilities of Shandong Cancer Hospital and Institute, affiliated with Shandong First Medical University and Shandong Academy of Medical Sciences (temperature: 22°C; humidity: 40%–60%; with free water and fed a 12-hour light/12-hour dark cycle).

Cell lines and culture condition. The murine colorectal cancer cell line MC38 and the melanoma cell line B16F10 were provided by Liufu Deng (Shanghai Jiao Tong University, School of Pharmacy, Shanghai, China). The murine fibrosarcoma cell line MCA205 was purchased from the BeNa Culture Collection (BNCC339717). The murine breast cancer cell line 4T1, murine colorectal cancer cell line CT26, and murine melanoma cell line B16 were obtained from the Cell Bank of the Chinese Academy of Sciences (SCSP-5056, SCSP-523, and SCSP-5096). The murine lung carcinoma cell line CMT167 was purchased from Otwo Biotech Inc. (HTX3130). Luciferase-expressing CMT167 (CMT167-luci) cells were established using lentiviruses purchased from GenePharma (2022-38569). The MC38 cells, CMT167 cells, 4T1 cells, and CT26 cells were cultured in DMEM (Thermo Fisher Scientific, C11995500BT) supplemented with 10% FBS (ExCellBio, FSP500) and 100 U/mL penicillin/streptomycin (Solarbio, P1400) at 37°C with 5% CO2. B16, B16F10, and MCA205 cells were cultured in RPMI-1640 (Thermo Fisher Scientific, C11875500BT) supplemented with the same ingredients.

CD8+ T cell isolation and culture. To determine the effects of IL-21 on the function of mouse and human CD8+ T cells in vitro, mouse CD8+ T cells were obtained from the spleens of C57BL/6J mice (7 weeks old) through negative selection using the EasySep Mouse CD8+ T Cell Isolation Kit (STEMCELL Technologies, 19853), and human CD8+ T cells were obtained from the peripheral blood of healthy donors using the EasySep Human CD8+ T Cell Isolation Kit (STEMCELL Technologies, 17953) according to the manufacturer’s standard protocols. Subsequently, anti-mouse/human CD3 (2 μg/mL, BioGems, Inc., 05111-20 and 05112-25) and anti–mouse/human CD28 (5 μg/mL, BioGems, Inc., 10311-20 and 10312-25) were used to stimulate CD8+ T cells in the presence or absence of IL-21 (Shanghai Junshi Biosciences Co., Ltd., JS014, 5 μg/mL), followed by flow cytometry analysis after 48 hours.

Mouse models and treatment. Unilateral subcutaneous tumor models were established using MC38, MCA205, B16, B16F10, CMT167, 4T1, and CT26 cells (1 × 106 cells) as previously reported. Briefly, the indicated cells resuspended in 100 μL PBS were subcutaneously injected into the right flank of C57BL/6J or BALB/c mice. For the bilateral subcutaneous tumor models, 1 × 106 MC38 cells or B16 cells were subcutaneously inoculated in the right flank as the first tumor followed by radiation, and 5 × 105 MC38 cells or B16 cells were subcutaneously inoculated in the left flank on the same day as the second tumor. When the unilateral tumor or first tumor volume reached approximately 100 mm3, the mice were randomly grouped and intraperitoneally administered either IL-21 (Shanghai Junshi Biosciences Co., Ltd., JS014) or PBS (Solarbio, P1020) of equal volume twice a week, and the unilateral tumors or first tumors were irradiated with 1 fraction of 15 Gy using the RS2000-225 Biological Irradiator (Rad Source Technologies, Inc.) the next day. Tumor volume was monitored at least twice a week with calipers and calculated using V = (length × width2)/2. The mice were euthanized at the indicated time or when the tumor volume reached 1,500 mm3.

An orthotopic lung cancer model was established using CMT167-luci cells from C57BL/6J mice. First, 5 × 104 CMT167-luci tumor cells resuspended in PBS were mixed with Matrigel (Corning, 356234) at a ratio of 1:1 (5 μL/mouse) and injected directly into the left mouse lung on day 0. The tumor-bearing mice were then randomly grouped and intraperitoneally administered IL-21 (twice per week) from day 6 (total of 9 doses), followed by irradiation on lung tumors with 1 fraction of 8 Gy using the Small Animal Radiation Research Platform (Xstrahl, SN2019077) on day 7. Tumor size and luciferase signal were monitored using IRIS PET/CT (Inviscan SAS) and IVIS Spectrum CT (PerkinElmer Inc.), respectively.

For in vivo depletion of CD8+ T cells, tumor-bearing mice were intraperitoneally administered an anti–mouse CD8α antibody (200 μg/mouse, BioXcell, BE0061) 2 days before and twice weekly thereafter. For triple therapy, tumor-bearing mice were intraperitoneally administered anti–PD-1 antibody (200 μg/mouse, BioXcell, BE0146) or solvent of equal volume with or without simultaneous administration of IL-21 twice every week, followed by irradiation 1 day after the first administration of IL-21 or anti–PD-1. Tumor volume was monitored and calculated as above. The mice were euthanized at the indicated time or when the tumor volume reached 1,500 mm3.

Humanized mice were established as previously reported (62, 63). Briefly, A549 cells (1 × 107) were subcutaneously inoculated into the right flank of the NSG mice on day 0. When the mean volume of the tumors reached approximately 100 mm3 on day 25 after tumor challenge, 1 × 107 PBMCs were obtained using human lymphocyte separation fluid (Lymphoprep Density Gradient Medium, STEMCELL Technologies Inc., 07851) from healthy donors according to the manufacturer’s standard protocols via density gradient centrifugation and introduced into mice via tail vein injection. Subsequently, the mice were randomly grouped and administered IL-21 (1.0 mg/kg, twice a week, a total of 3 doses administered on days 25, 29, and 32) with or without irradiation of 1 fraction of 15 Gy after PBMC transplantation on day 25. Tumor volume was monitored and calculated as above; mice were euthanized 37 days after tumor inoculation and subsequently subjected to flow cytometry assay of the TME. To evaluate the antitumor effects of the combination of IL-21 and radiation in immunodeficient NSG mice without PBMC engraftment, A549 cells (1 × 107) were subcutaneously inoculated into the right flank of the NSG mice. When the tumor volume reached 100 mm3, the mice were randomly grouped and intraperitoneally administered either IL-21 (Shanghai Junshi Biosciences Co., Ltd., JS014) or PBS (Solarbio, P1020) of equal volume twice a week with or without radiation treatment of 1 fraction of 15 Gy as described above. Tumor volume was monitored and calculated as above. The mice were euthanized at the indicated time or when the tumor volume reached 1,500 mm3.

Flow cytometry. Subcutaneous tumors were collected 8–10 days after tumor irradiation, minced with ophthalmic forceps, and subsequently digested with 1 mg/mL collagenase type 4 (Worthington Biochemical Corp., LS004186) and 0.2 mg/mL DNase I (Sigma-Aldrich, DN25) at 37°C for 30 minutes. Digestion was terminated using RPMI-1640 with 2% FBS, and the solution was filtered using a 70 μm cell strainer (Biosharp Life Sciences, BS-70-CS) to obtain single-cell suspensions, followed by cell staining and flow cytometry. For surface staining, the single-cell suspensions from mice were incubated with anti–mouse CD16/CD32 antibody (1:50, BD Biosciences, 553141) for 15 minutes at 4°C to block nonspecific antibody binding, washed with FACS buffer (2% FBS in PBS), and subsequently stained with an antibody cocktail against Fixable Viability Stain 780 (1:1,000, BD Biosciences, 565388), CD45.2 (1:200, BioLegend, 103149; 1:200, BD Biosciences, 552848), CD3 (1:200, BioLegend, 1100204/100233; 1:200, BD Biosciences, 552774), CD8 (1:200, BioLegend, 100730), CD4 (1:200, BioLegend, 100546/100451), CD11b (1:200, BioLegend, 117310), CD11c (1:200, BioLegend, 117349), CD19 (1:200, BioLegend, 115506), MHC-II (1:200, BioLegend, 107626), Ly6C (1:200, BioLegend, 128033), Ly6G (1:200, BioLegend, 127639), F4/80 (1:200, BioLegend, 123132), CD69 (1:200, BioLegend, 104505), CD279 (1:200, BioLegend, 135231), and CD366 (1:200, BioLegend, 134012). For intracellular staining to evaluate the function of T cells in the TME, single-cell suspensions were treated with Cell Activation Cocktail (with Brefeldin A) (1:200, BioLegend, 423304) for 4 hours before cell staining and stained with an antibody cocktail against CD45.2 (1:200, BioLegend, 103149) and CD8 (1:200, BioLegend, 100730). The samples were fixed in fixation/permeabilization buffer (Thermo Fisher Scientific, 00-5523-00) at 4°C for 1 hour, washed with 1× permeabilization/wash buffer, and incubated with the corresponding antibody cocktails against GZMB (1:200, BioLegend, 372204/396410) and IFN-γ (1:200, BioLegend, 505831) in permeabilization buffer in the dark at 4°C for 40–60 minutes.

To determine the effects of IL-21 on human CD8+ T cells in the TME after radiotherapy, A549 tumors from humanized mice were collected 37 days after the tumor challenge. Single-cell suspensions were prepared as described above and then subjected to surface staining against Fixable Viability Stain 780, CD45 (1:200, BioLegend, 304047), CD3 (1:200, BioLegend, 317323), CD4 (1:200, BioLegend, 317435), CD8 (1:200, BioLegend, 344747), PD-1 (1:200, BioLegend, 367428), and TIM-3 (1:200, BioLegend, 345016), followed by intracellular staining against GZMB (1:200, BioLegend, 372221), TNF-α (1:200, BioLegend, 502913), and IFN-γ (1:200, BioLegend, 502527).

To analyze the effects of IL-21 on mouse CD8+ T cells in vitro, the stimulated cells were subjected to surface staining against Fixable Viability Stain 780, CD45.2 (1:200, BioLegend, 103149), CD8 (1:200, BioLegend, 100730), and CD69 (1:200, BioLegend, 104505), followed by intracellular staining against GZMB (1:200, BioLegend, 372204), IFN-γ (1:200, BioLegend, 505831), TNF-α (1:200, BioLegend, 506339), PD-1 (1:200, BioLegend, 135231), and Tim-3 (1:200, BioLegend, 134012). Similarly, the stimulated human CD8+ T cells were stained with Fixable Viability Stain 780, CD8 (1:200, BioLegend, 344747), GZMB (1:200, BioLegend, 372221), TNF-α (1:200, BioLegend, 502913), IFN-γ (1:200, BioLegend, 502527), PD-1 (1:200, BioLegend, 367428), and TIM-3 (1:200, BioLegend, 345016).

Flow cytometry was performed on an LSR Fortessa system (BD Biosciences), and the output was analyzed using FlowJo software (version 10.4).

scRNA-Seq. Single-cell suspensions (prepared as mentioned above) were obtained 9 days after MC38 tumor irradiation. Tumor-infiltrating lymphocytes were isolated via density gradient centrifugation using Ficoll (GE Healthcare, 17-5442-02). The obtained tumor-infiltrating lymphocytes were stained with an antibody cocktail against Fixable Viability Stain 510 (1:1,000, BD Biosciences, 564406) and CD45.2 (1:200, BioLegend, 103116). Live CD45+ immune cells were sorted using a FACSAria III (BD Biosciences), followed by single-cell transcriptome sequencing using the droplet-based 10× Genomics platform (CapitalBio). Stringent data quality-control was conducted during downstream analysis. The initial dataset contained 64,010 cells, which were filtered using the following parameters to exclude outliers: maximum percentage mito (Percent_mito) = 20%, maximum number of unique molecular identifiers = 60,000, minimum number of genes = 300, and maximum number of genes = 7,500 (64). After discarding poor-quality cells, 27,818 cells were retained for downstream analyses. T-distributed stochastic neighbor embedding (t-SNE) plots and uniform manifold approximation and projection (UMAP) plots were drawn based on the single-cell analysis platform provided by CapitalBio (follow-up visual analysis was also based on the platform). The cluster/subcluster was preliminarily defined through singleR (ImmGenData), and each cell type was determined based on the multiple cell type–specific marker genes identified in the previous literature (64–66) as follows: Cd3e, Cd3d, and Cd3g for T cells; Cd8a for CD8+ T cells; Cd4 for CD4+ T cells; Cd4 and Foxp3 for Treg cells; Cd19, Ms4a1, and Cd79a for B cells; Ly6c2, Ccr2, and S100a4 for monocytes; Adgre1, Cd68, and Arg1 for macrophages; S100a8, S100a9, and Retnlg for neutrophils; Ccl22, Cd209a, and Batf3 for DCs; and Gzma, Klrb1c, and Ncr1 for NK cells. Subsequently, Nightingale rose charts and heatmaps were used to illustrate the proportions of different immune cell types. For deep analysis of CD8+ T cells, 4 subclusters were identified after removal of low-quality data: naive CD8+ T cells (Ccr7, Lef1, and Tcf7), effector and memory CD8+ T cells (Gzma, Ccl5, Cxcr3, and Gzmk), exhausted CD8+ T cells (Pdcd1, Tnfrsf9, Ifng, and Havcr2), and proliferating CD8+ T cells (Mki67 and Top2a). Scatter plots were used to display the expression levels and distribution of marker genes in each subset of CD8+ T cells. To infer the differentiation trajectories of CD8+ T cells, a pseudo-time analysis of individual cells was conducted using Monocle (https://bioconductor.org/packages/release/bioc/html/monocle.html) with the DDRTree algorithm based on characteristic genes. To analyze the function of CD45+ immune cells and CD8+ T cells from different treatment groups, the analysis was conducted using the GSEA function in clusterProfiler (https://bioconductor.org/packages/release/bioc/html/clusterProfiler.html, 4.4.4) with default parameters (P-value cutoff = 0.05) with data from the GO and the KEGG databases. Next, for subgroups of CD8+ T cells, the pairwise.wilcox.test() function was used to analyze cytotoxic marker genes and exhaustion marker genes between the IR and IL-21+IR groups.

Multiplex immunofluorescence assay and analysis. To further verify the relationship between the expression of IL-21/IL-21R and the TME, multicolor fluorescent immunohistochemistry was performed on lung adenocarcinoma tumor microarrays purchased from ShanDong PhenoScience Biotechnology Ltd., according to standard protocols (Akoya Biosciences, NEL871001KT); the relevant clinical information is listed in Supplemental Table 1. The following markers were used: panCK (1:200, ZSGB-BIO, ZM-0069) labeled with Akoya Opal fluorophores 480, IL-21R (1:50, Proteintech, 10533-1-AP) labeled with Akoya Opal fluorophores 570, CD3 (1:200, Abcam, ab16669) labeled with Akoya Opal fluorophores 690, CD45 (1:300, Proteintech, 20103-1-AP) labeled with Akoya Opal fluorophores 620, IL-21 (1:50, ABclonal Biotechnology Co., Ltd., A7235) labeled with Akoya Opal fluorophores 520, and CD8 (1:300, Abcam, ab199016) labeled with Akoya Opal fluorophores 780, and the nuclei were labeled with DAPI (1:100, Akoya Biosciences). CD3+ cells, CD8+ cells, CD45+ cells, IL-21+ cells, IL-21R+ cells, and tumor areas were photographed and analyzed using the Vectra Polaris system (Akoya Biosciences).

To further verify the relationship between IL-21 expression and the TME after radiotherapy, multicolor fluorescent immunohistochemistry was conducted on specimens from patients with esophageal squamous cell carcinoma who had undergone neoadjuvant radiotherapy provided by Shandong Cancer Hospital. The relevant clinical information is listed in Supplemental Table 2. The following markers were used: panCK (1:200, ZSGB-BIO, ZM-0069) labeled with Akoya Opal fluorophores 480, PD-1 (1:200, ZSGB-BIO, ZM-0381) labeled with Akoya Opal fluorophores 570, GZMB (1:200, Abcam, ab255598) labeled with Akoya Opal fluorophores 690, TIM-3 (1:1000, Abcam, Ab241332) labeled with Akoya Opal fluorophores 520, IL-21 (1:50, ABclonal Biotechnology Co., Ltd., A7235) labeled with Akoya Opal fluorophores 620, CD8 (1:300, Abcam, ab199016) labeled with Akoya Opal fluorophores 780, and the nuclei were labeled with DAPI (1:100, Akoya Biosciences). CD8+ cells, PD-1+ cells, TIM-3+ cells, IL-21+ cells, GZMB+ cells, and tumor areas were photographed and analyzed using the Vectra Polaris system (Akoya Biosciences).

To determine the function of IL-21 in the TME after radiotherapy, multicolor fluorescent immunohistochemistry was performed on mouse tumor tissues obtained 9 days after radiation, fixed in formalin, and embedded in paraffin. The following markers were used: GZMB (1:200, Abcam, ab255598) labeled with Akoya Opal fluorophores 480, PD-1 (1:200, Abcam, ab214421) labeled with Akoya Opal fluorophores 570, IL-21 (1:50, ABclonal Biotechnology Co., Ltd., A7235) labeled with Akoya Opal fluorophores 520, IFN-γ (1:1000, ABclonal Biotechnology Co., Ltd., A12450) labeled with Akoya Opal fluorophores 690, TIM-3 (1:1,000, Abcam, ab241332) labeled with Akoya Opal fluorophores 620, CD8 (1:800, Abcam, ab217344) labeled with Akoya Opal fluorophores 780, and the nuclei were labeled with DAPI (1:100, Akoya Biosciences). CD8+ cells, PD-1+ cells, TIM-3+ cells, IL-21+ cells, GZMB+ cells, and IFN-γ+ cells were photographed and analyzed using the Vectra Polaris system (Akoya Biosciences).

H&E staining. To explore the safety of systematic administration of IL-21, the vital organs (brain, heart, lung, thymus, liver, kidney, spleen, and lymph nodes) of mice systematically administered IL-21 and subjected to H&E staining. Briefly, the sections were deparaffinized in xylene, followed by hydration by immersion in 100%, 95%, 80%, and 70% ethanol. Subsequently, staining was conducted using hematoxylin staining solution (Beyotime, C0107) and eosin staining solution (Beyotime, C0109). Finally, the sections were subjected to gradient dehydration and mounting and observed under a microscope.

Statistics. The data are presented as mean ± SEM and were compared using a 2-tailed Student’s t test for 2 groups and 1-way ANOVA followed by multiple-comparison test for more than 2 groups. Tumor curve data were compared using 2-way ANOVA. Survival analysis was performed using the Kaplan-Meier method with a log-rank test. Statistical significance was set at P less than 0.05.

Study approval. Studies using human samples were approved by the ethics committee of Shandong Cancer Hospital and Institute (no. 202511043) and conform to the guidelines of the 2000 Helsinki Declaration. All animal studies were performed according to the protocol approved by the IACUC of Shandong Cancer Hospital and Institute (no. SDTHEC2024006125).

Data availability. All data necessary for the evaluation of the conclusions in the article are presented in the article and/or in the supplemental materials and methods. The survival data of the IL-2 family analyzed in this study can be accessed on the Kaplan-Meier plotter (https://kmplot.com/analysis/) and TCGA database (https://www.cancer.gov/ccg/research/genome-sequencing/tcga). The scRNA-Seq data of MC38 tumors in study are available in the National Genomics Data Center under accession PRJCA052467 (https://ngdc.cncb.ac.cn/bioproject/). Values for all data points in graphs are reported in the Supporting Data Values file.