IntroductionCirculating T cells are critical for maintaining systemic immune homeostasis, orchestrating inter-organ immune coordination, and modulating local microenvironments.1 The discovery of T cell subtypes/states (Ts/s) and identities have expanded rapidly with the advent of single-cell RNA sequencing (scRNA-seq) and the development of cell identity marker gene panels (ciMGPs). Spatial transcriptomics (Stereo-seq) demonstrates that the distribution and dynamics of Ts/s within tumor microenvironments (TME) are correlated with cancer malignancy, tumor invasion, formation of anti-tumor immune niches, and therapeutic responsiveness.2,3,4,5 Circulating T cells play a critical role in stabilizing systemic immune function, mimicking spatialized microenvironmental defense, and reflecting inter-organ communication.2,3,6 scRNA-seq can refine the abundance, diversity, and characteristics of activated or exacerbated T cells and receptors, as biomarkers for the severity and progression of immune-related adverse events in melanoma patients treated with immune checkpoint blockade agents.7,8 The delineation of Ts/s defined by scRNA-seq provides novel insights into the T cell atlas, disease-specific T cell patterns, and interactions between Ts/s, which are expected to be translated into clinical applications as part of the clinical biochemistry of hematology.9,10,11 However, the clinical translation of scRNA-seq-defined Ts/s remains challenging due to cost burden, technical complexity, regulations, equipment certifications, standardization, repeatability, data interpretation, bioinformatic security, variability in physician expertise, and legal frameworks. Among these, a major hurdle lies in the precise selection of cell Ts/s, which is complicated by the multifaceted nature of T cell development, localization, regulation, functional plasticity, and metabolic heterogeneity.12The accuracy and specificity of Ts/s are highly dependent on the reproducibility and overlap of the selected marker gene number and function, the dynamic transitions from naïve to active Ts/s, and disease-specific phenomes. The evidence-based evaluation systems should represent the disease specificity and clinical utility of ciMGPs for diagnostic purposes. In order to facilitate clinical translation, a robust framework for cell identity assessment is urgently required to quantify and classify the biological and clinical specificities of ciMGPs. Analysis of confidence intervals-expression quantitative trait loci demonstrated that the definition of cell types and identities in peripheral blood mononuclear cells (PBMCs) is essential for elucidating the immune functions, cellular heterogeneity, and disease pathogenesis.13 Nevertheless, it remains unclear precisely how circulating Ts/s profiles correspond to clinical data in patients with lung cancer. Taking into consideration that Ts/s in peripheral blood are accessible in the clinical setting, there is a pressing need for precise and efficient strategies to delineate characteristics specific to disease nature, severity, stage and response of Ts/s, and to define the gene expression overlaps and changes of ciMGPs for each Ts/s. Expression profiles of Ts/s-specific ciMGPs reflect systemic immune cell development, compositional patterns, transcriptional programs, and genetic variants associated with immune dysregulation or cancer progression.13,14,15 The prior evidence suggests that the specificity and diversity of Ts/s reflect the spatialization of tumor microenvironmental inflammation, balance between T cell activation and exhaustion, and patient prognosis.2,3,16,17 With the rapid advancement of single-cell measurements, numerous immune cell subtypes were identified for deeper insights into the molecular mechanisms and disease pathogenesis.18The clinical specificity and utility of T cell ciMGPs require multidimensional validation to delineate the physiological range and pathological alterations of circulating Ts/s. The primary aim of the present study was to establish a new evaluation system for the validation of novel subtypes of T regulatory cells (Treg) in the circulation of patients with non-small cell lung cancer (NSCLC). This was achieved by systematically assessing the diversity and specificity of Ts/s in healthy individuals, the correspondence between circulating and tissue-resident T cell profiles in lung cancer, dynamic changes prior to and following surgical intervention, spatial distribution across various organs, and disease-specific signatures across various lung diseases and extrapulmonary malignancies. A regional overlap-expression rate (rOER) metric, a screening and evaluation system, was developed to quantify Ts/s ciMGPs expression specificity. Ts/s ciMGPs were categorized through bibliometric and nomenclature-based screening from our own and public databases. NSCLC-specific Ts/s ciMGPs were selected and validated based on the expression dynamics, disease specificity, altered patterns, and prognostic values by profiling single-cell transcriptomes of peripheral T cells from paired pre- and post-operative blood samples, and matched NSCLC and para-NSCLC tissues. The specificity of these ciMGPs was confirmed by comparing the expression patterns with Ts/s profiles of pulmonary diseases to validate the NSCLC specificity, with Ts/s profiles of other extrapulmonary diseases to confirm the disease specificity, and with other organs/tissues to define the tissue specificity.One of the Treg subtypes, Tregfci with CD4, CD45, FOXP3, CTLA4, and IL2RA (5g-Tregfci) and Tregfci with FOXP3, CTLA4, and IL2RA (3g-Tregfci) were compared, and the latter was used as the representative panel description, as clarified in Section 10 of the Supplementary Materials. This panel was used in the current study to enhance classification accuracy and robustness, particularly to mitigate gene dropout effects in scRNA-seq and Stereo-seq data. The disease-, organ-, and time-specific characteristics of Tregfci were validated across various spatializations, temporalizations, diseases, and prognoses. Moreover, the present study characterized the biological functions of Tregfci under healthy and disease conditions, with or without stimulations, and in co-culture with NSCLC cells or human normal airway epithelial cells, following individual or combined downregulation of Tregfci ciMGPs. During the evaluation, ETS1 (ETS Proto-Oncogene 1, Transcription Factor) emerged as one of the key Tregfci-specific regulators and exhibited significant alterations in transcription-factor networks. Functional perturbations of ETS1 introduced through knockin, knockdown, or knockout approaches in cells and animal models demonstrated its critical role in regulating Tregfci ciMGPs and immune functions during lung cancer. The roles of ETS1 in reprogramming of transcriptomic and metabolic profiles were confirmed in Tregfci from human PBMCs treated with ETS1 inhibitor or from mouseETS1-wt or miceETS1-cKO blood and lung tissues.ResultsSelection of rOER-based Ts/s ciMGPsTo initiate the selection of T cell ciMGPs, Ts/s ciMGPs were collected from our own data, 11 self-established scRNA-seq databases, and publicly available scRNA-seq repositories (Fig. 1a). About 300 candidates of ciMGPs were shortlisted from an initial screening and selection for T cell annotations in the subsequent scRNA-seq analyses. A refined list of 170–262 ciMGPs was retained for downstream analysis, after the removal of duplicate entries or synonymous gene sets across species and tissues. The corresponding T-cell subsets were consistently renamed based on standardized CD markers or characteristic functional genes to ensure uniform nomenclature, as detailed in Supplementary Table 1. T cell ciMGPs in human blood samples and tissues were subjected to specificity evaluation by assessing the cutoff values, number of overlaps, rOER filtering, and ciMGP sorting (Fig. 1a).Fig. 1The alternative text for this image may have been generated using AI.Full size imageWorkflow and classification of T ciMGPs across immune compartments and clinical conditions. a A total of 300 candidate marker gene panels associated with distinct T cell subsets and functional states across species (human, mouse, and zebrafish) from various sources and selections. b Strategy for evaluating the specificity of ciMGPs using regional overlap expression rate (rOER). Columns 1–9, which represent the mean expression level of each ciMGP across the scRNA-seq datasets under evaluation that were used for the repeatability assessment. The red boxplot summarizes the distribution (Q1–median–Q3) of rOER values for the target ciMGPs within the focal T-cell subtype. The corresponding black boxplots, which represent non-target ciMGPs used as internal controls for specificity. c Subset-specific ciMGPs (ss-ciMGPs) as ciMGPs in fewer than the threshold number of subsets with rOER >30%. d Subset-associated ciMGPs (sa-ciMGPs) as ciMGPs with intermediate rOER values (30–60%) in a greater-than-threshold number of subsets. e Subset-reference ciMGPs (sr-ciMGPs) were characterized as ciMGPs expressed in more than the threshold number of subsets with rOER >60%, representing broadly expressed ciMGPs with limited discriminatory power. f Schematic overview of the study design for validation of circulating T ciMGPs. g UMAP visualization of T cell subset distribution based on rOER-defined subset specificity. h Upset plot illustrating intersections of identified T ciMGPs across NSCLC patient blood and lung tissue samples. i Heatmap displaying the presence (orange) or absence (green) of ss-ciMGPs in pre- or post-operative blood, NSCLC, or para-NSCLC lung tissues. j Heatmap showing appearance and variation of rOER-defined T cell ciMGPs in pre- or post-operative blood. k The impact of the low (black line) and high expression (red line) of 5g- or 3g-Tregfci ciMGP signatures on the long-term overall survival rate of 672 patients with lung cancer over 200 months by Kaplan–Meier survival analysisTo define the specificity of each ciMGP in the annotation of scRNA-seq data, the overlapping regions of the target ciMGP mRNA expression were quantified and visualized using boxplots. The target panel box between the corresponding Q3 and Q1 was defined as the repeatable range of non-target ciMGPs, as explained in Fig. 1b. The cutoff value was set at 10% of the total number of ciMGPs identified per sample by scRNA-seq and adjusted according to the number of detected subtypes, as described in Section 6 of the Supplementary Materials. The threshold of the selection was defined by re-calculating and establishing more than 70% sensitivity, less than 25% false positive rates, and more than 60% reproducibility and consistency in the regional line of Number30%/Number60% and the ternary classification-ss ciMGP (Supplementary Table 13). Based on the proportion of ciMGPs overlapping within the repeatable range, ss-ciMGPs were defined as those with Number>30% 30% >cutoff value and Number>60% < cutoff value (Fig. 1d), and sr-ciMGPs as those with Number >60% >cutoff value (Fig. 1e). The trade-off values between sensitivity and specificity were detailed in Supplementary Table 17. The intermediate thresholds between 10–20% provided the optimal balance between sensitivity, specificity, and reproducibility, as detailed in Section 7 of the Supplementary Materials and listed in Supplementary Table 17.To expound the presence of selected ciMGPs in normal PBMCs, 89,982 PBMCs were clustered from a cohort of healthy individuals (n = 31), of which 123 Ts/s were detectable. Clinical reference ranges for those Ts/s mainly included the mean, median, standard deviation, and standard error values (Table 1 and Supplementary Table 2.1), to mimic the report format of clinical biochemistry or hematology, although the number of clinical samples was limited and the accuracy requires further validation. Among the identified ciMGPs, 29 (39%) were defined as ss-ciMGPs, 8 (11%) as sa-ciMGPs, and 37 (50%) as sr-ciMGPs in healthy individuals (Supplementary Table 2.2). To explore the distribution and repeatability, the number (Supplementary Table 3) and rOER values (Supplementary Table 4) of ss-ciMGPs (Supplementary Tables 3.1 and 4.1), sa-ciMGPs (Supplementary Tables 3.2 and 4.2), and sr-ciMGPs (Supplementary Tables 3.3 and 4.3) were quantified across 10% rOER intervals. The ss-ciMGPs, sa-ciMGPs, and sr-ciMGPs exhibited distinct rOER distributions. The largest overlapping region of the ss-ciMGPs was 15, and >20, as detailed in Section 7 of the Supplementary Materials. An FC > 3 was selected as a biologically meaningful and statistically balanced threshold for downstream functional annotation and overlap analysis, as seen in Supplementary Table 16. The potential impacts of different threshold choices on the specificity and performance were evaluated by comparing the influences of various thresholds in the sensitivity, specificity, and reproducibility, as described in Section 7 of the Supplementary Materials.Validation of Ts/s-associated survival ratesSurvival curves of T cell subtypes were estimated by the Kaplan–Meier method, with statistical significance calculated by the log-rank test. To evaluate patient outcomes, overall survival (OS), post-progression survival (PPS), and progression-free survival (PFS) were generated using the online KMplot tool (http://kmplot.com/analysis/). The average expression level of the constituent genes of each ciMGP was calculated for each patient sample to represent the subtype abundance. For each T cell subtype, patients were stratified into high- and low- expression groups based on median ciMGPs levels, and Kaplan–Meier curves were plotted accordingly. Hazard ratios with 95% confidence intervals were calculated and the corresponding log-rank p values were reported to assess the prognostic significance of each subtype.Validation of disease- and organ/tissue-specificityThe specificity of lung cancer was evaluated by comparing with COPD as the representative of chronic local inflammation, idiopathic pulmonary fibrosis (IPF) as the representative of chronic tissue remodeling, and interstitial lung disease in systemic sclerosis (SSC) as the representative of chronic lung disease secondary to systemic immune disorder. The organ/tissue specificities of selected Ts/s were evaluated by comparing with 17 organs/tissues, including the nasal mucosa, nose, breast, gastric, tooth, lung, pancreatic, bone marrow, eye, renal, skin, small intestine, cortex, cervical, liver, mouth and choroid. The disease specificities of selected Ts/s were determined by comparing with 31 extrapulmonary diseases. Detailed information on validation of specificities was described in Section 8 of Supplementary Materials.Validation of Tregfci spatializationIn order to define the biological differences between 5g-Tregfci and 3g-Tregfci, the spatial consistency was evaluated in two sets of data: our in-house spatial transcriptomics dataset (n = 4, described above); and public datasets of spatial transcriptomes, including breast cancer (BRCA, n = 6), colorectal cancer (CRC, n = 4), hepatocellular carcinoma (HCC, n = 3), ovarian cancer (OV, n = 8), pancreatic ductal adenocarcinoma (PDAC, n = 3), and squamous cell carcinoma (SCC, n = 4), from the SpatialTME database website.12 The spatial distribution patterns of 5g-Tregfci or 3g-Tregfci differed as depicted in Supplementary Fig. 8a–d. To determine the significance of Tregfci localization, spatial gene expression data were integrated with TME annotations, as detailed in Section 9 of the Supplementary Materials.All slides were normalized prior to downstream analysis to ensure data comparability across different spatial spots and tissue samples. To investigate the spatial distribution and relationship, FOXP3, CTLA4, IL2RA, and ETS1 were selected and examined, and spatial gene expression patterns were visualized using the “SpatialFeaturePlot” function. To quantify expression differences across spatially annotated regions in the SpatialTME database (e.g., tumor core, invasive margin, stromal compartments), the “DotPlot” and “VlnPlot” functions from the Seurat package (version 3.0.2) were utilized. The Tregfci spatialization was assessed by calculating a Tregfci signature score using the “AddModuleScore” function in the Seurat package. The spatial distribution of the composite score was visualized using the same spatial plotting approach to maintain consistency. Statistical comparisons between spatial regions were performed using the Wilcoxon rank-sum test (two groups) or Kruskal–Wallis test (multiple groups) (version 0.6.0).48Validation of Tregfci functionsConstruction of stable multi-gene knockdownTo evaluate the influence of each Tregfci ciMGP or combination of genes on cell functions, mono-target and multi-target short hairpin RNA (shRNA) lentiviral delivery systems targeting FOXP3, CTLA4, and IL2RA were constructed. The shRNA sequences (selected from 3–5 candidates per gene) were validated and synthesized (GeneChem Co., Ltd., Shanghai, China). After the sequencing and function of each mono-gene for knockdown were screened and selected, the FOXP3, CTLA4, and IL2RA genes in T cells were simultaneously knocked-down by constructing a recombinant lentiviral vector that encoded multiple shRNA sequences, each of which specifically targeted a distinct gene of interest. The shRNA sequences used in the study were listed in Section 10 of Supplementary Materials. For combinatorial gene silencing, a recombinant lentiviral construct encoding three tandem U6 promoter-driven shRNA cassettes was designed to simultaneously target FOXP3, CTLA4, and IL2RA, according to a previous protocol.49 The equal mixtures of three mono lentiviral vectors were produced for co-infection and validated for specificity and knockdown efficiency. The lentiviral particles were produced by transient co-transfection of HEK-293T cells with the shRNA plasmid(s) and the packaging plasmids pMD2.G and psPAX2, using Lipofectamine 2000 (Thermo Fisher Scientific). Viral supernatants were collected at 48 and 72 h following the transfection, filtered through a 0.45 μm filter (Thermo Fisher Scientific), and concentrated via ultracentrifugation at 25,000 rpm for 2 h at 4 °C. The viral titer was determined using a p24 antigen ELISA kit (QuickTiter™ Lentivirus Titer Kit, Cell Biolabs), and adjusted to 1–5 × 10⁸ TU/mL prior to use. T cells were seeded at a density of 5 × 10⁵ cells per 60 mm dish and transduced with the multi-target lentivirus or a mixture of the mono-target lentiviruses at a multiplicity of infection of 10–20, in the presence of 8 μg/mL Polybrene (GeneChem, Shanghai, China). After 24 h, the medium was replaced, and the cells were cultured for 48 h before being selected with puromycin (3 μg/mL, Beyotime, ST551) for 5–7 days. Transduction efficiency was initially assessed via fluorescent microscopy or flow cytometry, based on the co-expression of a green fluorescent protein (GFP) reporter gene encoded in the lentiviral backbone. Cell population with >80% GFP positivity were considered to be successfully transduced. The knockdown efficiency of each target gene was confirmed via qRT-PCR, wherein T cells transduced with a non-targeting shRNA lentivirus (shGFP) served as negative controls. The resulting stable cell lines with triple or mono gene knockdown were designated as Tregfci-KD, Tregfoxp3-KD, Tregil2ra-KD, Tregctla4-KD, and Tregets1-KD or ETS1 up-regulated cell-lines as Tregets1-OE for subsequent functional analyses.Validation of Tregfci ciMGPs and ETS1 protein expressionThe protein expression and distribution of FOXP3, CTLA4, IL2RA, and ETS1 in lung tissues of NSCLC patients were examined through immunohistochemistry. Lung cancer tissue samples were collected from young and old patients with NSCLC (n = 21; 11 females, 10 males; 8 young, 13 old), followed by formalin fixation and paraffin embedding in blocks. Sections of 4 µm thickness were cut from the blocks and mounted on glass slides for immunohistochemical staining. To enhance the exposure of epitopes masked during the fixation process, heat-induced antigen retrieval was performed. Sections were deparaffinized, rehydrated through a graded ethanol series, and subjected to heat-induced antigen retrieval in a pH 7.8 buffer for 10 min at 100 °C using a pressure cooker. To block nonspecific antibody binding and reduce background staining contamination, sections were incubated in a blocking solution containing 5% normal goat serum in phosphate-buffered saline (PBS) for 30 min at room temperature. Primary antibodies specific to FOXP3 (Cat#AB3199), IL2RA (Cat#AB2709), CTLA4 (Cat#AB3659) and ETS1 (Cat#AB3669) were purchased from Xiamen Zhiwei Electronic and Pharmaceutical Co., Ltd. The sections were incubated with the primary antibodies overnight at 4 °C in a humidified chamber. After washing the sections with PBS thrice, a secondary antibody conjugated to horseradish peroxidase was applied. The sections were incubated with the secondary antibody for 60 min at room temperature. The bound antibodies were visualized using the EnVision Kit (Agilent DAKO, Santa Clara, USA) according to the manufacturer’s instructions, to produce a brown chromogenic precipitate at the site of antibody binding. The nuclei were counterstained with hematoxylin. The stained sections were examined under a light microscope (Olympus BX51). The staining intensity and distribution of marker proteins were assessed in tumor cells and tumor-infiltrating lymphocytes (TILs).The protein expression and distribution of ETS1 in T cells were examined via cellular immunofluorescence staining. T cells were collected, washed twice with PBS, loaded onto poly-L-lysine-coated glass slides, and cytospun at 800 × g for 5 min. The cells were immediately fixed in 4% paraformaldehyde (Sigma-Aldrich, USA) at room temperature for 15 min, followed by three washes with PBS. The cell membranes were permeabilized with 0.1% Triton X-100 (Solarbio, China) in PBS for 10 min, followed by blocking with 5% bovine serum albumin (Sigma-Aldrich) in PBS for 1 h at room temperature. The cells were incubated with anti-ETS1 antibody (Abclonal, China) and Cy3-conjugated antibody (Beyotime, China) in blocking buffer overnight at 4 °C in a humidified chamber. After three washes with PBS, the nuclei were counterstained with 1 μg/mL 4′,6-diamidino-2-phenylindole (DAPI) (Thermo Fisher Scientific) for 5 min in the dark. Slides were prepared with the cell and nuclei samples, followed by mounting with an anti-fade fluorescence mounting medium (Vectashield, Vector Laboratories) and sealing with coverslips. Fluorescence images were acquired using the Perkin-Elmer high-content automatic imaging system. Cy3 and DAPI signals were captured using appropriate excitation/emission filters. The quantitative fluorescence intensity analysis of ETS1 localization was performed using the ImageJ software (NIH, USA). For each randomly selected field, the mean fluorescence intensity of ETS1 within the cell cytoplasm and nucleus was measured. Each experiment was repeated at least 6 times to ensure reproducibility.Immunoblotting assayCell samples were homogenized in RIPA lysis buffer (P0013B, Beyotime, Shanghai, China) and centrifuged at 14,000 × g for 30 min at 4 °C for Western blotting, as previously described.50 Equal quantities of protein (30 μg) were loaded onto 7.5% − 12.5% SDS-PAGE gel, followed by electrophoresis and transfer to nitrocellulose membranes (0.45 µm for proteins >50 kDa, 0.2 µm for proteins ≤50 kDa). The membranes were blocked with 5% non-fat blocking grade milk (Bio-Rad, Hercules, CA, USA) and incubated overnight at 4 °C with the following primary antibodies: anti-caspase3 (1:1000, A0214, Abclonal, China), anti-cleaved caspase 3 (1:1000, 25128-1-AP, Proteintech, Wuhan, China), anti-Bax (1:1000, A19684, Abclonal, China), anti-Bcl2 (1:1000, A19693, Abclonal, China), anti-GAPDH (1:5000, A19056, Abclonal, China), anti-IL4 (1:1000, A23277, Abclonal, China), anti-IL10 (1:1000, A12255, Abclonal, China), anti-CD73 (1:1000, A25914, Abclonal, China), anti-CD29 (1:1000, A23497, Abclonal, China), anti-CD69 (1:1000, A26620, Abclonal, China), and anti-CD223 (1:1000, A2996, Abclonal, China). On the following day, the membranes were incubated with the appropriate secondary antibody (1:2000, AS097, Abclonal, Wuhan, China) at room temperature for 1 h. The immunoblots were visualized using an ECL Plus chemiluminescence reagent kit (P0018S, Beyotime, Shanghai, China). For quantitative analysis, the integrated density of each protein band was calculated using ImageJ software (ImageJ 1.5, NIH, USA). The resulting values were normalized to GAPDH as an internal control.Construction of Treg-specific Ets1-deficient miceTo investigate the functional role of Ets1 in Tregs, a Treg-specific Ets1 conditional knockout mouse model was generated using the Cre-LoxP system. Guide RNAs (gRNAs) targeting the intronic regions flanking exons 7 and 8 were designed using CRISPR design tools to minimize off-target effects and synthesized. Cas9 mRNA (100 ng/µL) was co-injected with gRNAs and a donor plasmid containing 1–2 kb homologous arms on each side of the targeted region to facilitate homology-directed repair. Off-target sites with high similarity to the gRNA sequences were evaluated in silico, and gRNAs with minimal predicted off-target activity were selected. The gRNAs, Cas9 mRNA, and donor vector were microinjected into fertilized C57BL/6 mouse zygotes. The mouse embryos were transferred into the oviducts of pseudopregnant females to obtain founder (F0) mice. F0 offspring were screened via PCR and validated via Sanger sequencing to identify mice with the correct loxP-flanked alleles. Positive founders were bred with wild-type C57BL/6JCya mice to obtain heterozygous Ets1-floxed mice (Ets1fl/+), which were interbred to obtain homozygous Ets1fl/fl mice. To achieve Treg-specific deletion of Ets1, Ets1fl/fl mice were crossed with Foxp3YFP-Cre transgenic mice, wherein Cre recombinase was driven by the endogenous FOXP3 promoter. Breeding was performed by mating male Foxp3YFP-Cre+/−; Ets1+/fl mice with female Ets1fl/fl mice, generating experimental mice (Foxp3YFP-Cre+/−; Ets1fl/fl, hereafter referred to as Ets1Treg-KO) and control littermates (Foxp3YFP-Cre−/−; Ets1fl/fl, referred to as Ets1fl/fl). All animals were maintained on a pure C57BL/6 genetic background to minimize strain-related immunological variability. Tail biopsies were collected for genotyping at 2–3 weeks of age using PCR with primers specific to the loxP-flanked region and the Foxp3YFP-Cre transgene. The PCR products were resolved on agarose gels and visualized under UV illumination.Ets1flox/flox control and Ets1flox/flox;FOXP3-Cre conditional knockout mice (6–8 weeks old, female) were sourced from Cyagen Biosciences. To establish the lung cancer mouse model, 10⁶ Lewis lung carcinoma cells (LLC cells, ATCC, CRL-1642) were suspended in 100 µL of sterile PBS and injected subcutaneously into the left axillary region of each mouse. Tumor development was monitored every 2–3 days using calipers to measure the tumor size. Mice were randomly assigned to different experimental groups (n = 6–8 animals/group). Tumor growth was monitored for approximately 30 days, or until tumors reached a maximum volume of 1500 mm³, or if mice exhibited signs of distress or impaired mobility, at which point animals were euthanized according to institutional ethical guidelines. All animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of Zhongshan Hospital, Fudan University, and conducted in accordance with institutional and national ethical guidelines. Mice were housed under specific pathogen-free conditions with a 12-h light/dark cycle and provided with autoclaved food and water ad libitum.Ets1 roles in metabolomicsTo deeply define the potential influence of Ets1 on metabolism, systemic (circulation) and local metabolism (lung) were measured in Ets1flox/flox control and Ets1flox/floxFOXP3-Cre conditional knockout mice (n = 6/group) using untargeted metabolomics. Plasma samples were obtained from peripheral blood of mice, and lung tissues were dissected, with half of the lobes reserved for metabolomics analysis. For untargeted metabolomics, plasma and lung tissue samples were kept on ice and vortexed thoroughly before extraction. Briefly, 300 μL of 80% methanol containing 2-chloro-phenylalanine (4 ppm) was added to each sample, vortexed for 1 min, and centrifuged at 12,000 rpm for 10 min at 4 °C. The supernatant was collected, filtered through a 0.22 μm membrane, and transferred into LC-MS vials for analysis. Chromatographic separation was performed on an ACQUITY UPLC® HSS T3 column (2.1 × 100 mm, 1.8 μm; Waters) using a Vanquish UHPLC system (Thermo Fisher Scientific, USA) with a flow rate of 0.3 mL/min and an injection volume of 5 μL. The mobile phase consisted of 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B) for positive mode, and 5 mM ammonium format in water (A) and acetonitrile (B) for negative mode. Mass spectrometry was carried out on an Orbitrap Exploris 120 mass spectrometer (Thermo Fisher Scientific, USA) equipped with an electrospray ionization (ESI) source operating in positive and negative ion modes. The MS parameters were set as follows: spray voltage 3.50 kV (positive) and –2.50 kV (negative), capillary temperature 325 °C, sheath gas 40 arb, auxiliary gas 10 arb, full MS resolution 60,000, scan range m/z 100–1000, and data-dependent MS/MS resolution 15,000 with normalized collision energy of 30%. Quality control samples were prepared by pooling equal aliquots of all test samples and were injected at regular intervals to monitor instrument stability.Cell culture and reagentsHuman T cell acute lymphoblastic leukemia cell lines (Tfci), human lung adenocarcinoma epithelial cell lines (A549), and human bronchial epithelial cell lines (HBE), human small cell lung carcinoma epithelial cell line (H446), human lung large cell lung cancer cell line (H460), human lung adenocarcinoma (SPC-A1), human small cell lung carcinoma cell line (H1688), human lung large cell carcinoma (H661), and human lung adenocarcinoma cell line (LTEP), were obtained from ATCC. All cell lines were cultured in RPMI 1640 medium (Keygen Bio, KGM31800-500, Shanghai, China) supplemented with 10% fetal bovine serum (Hyclone, SH30084.03, Logan, UT, USA), 100 U/mL penicillin, and 100 μg/mL streptomycin at 37 °C in a humidified chamber containing 5% CO2. For co-culture experiments, Tfci stably transfected with either control vector (V- Tfci-NC), PHA-stimulated control vector (PHA-Tfci-NC), Tregfci ciMGPs knockdown vector (V-Tfci-KD), or PHA-stimulated fci knockdown vector (PHA-Tfci-KD) were used (n = 6/group), as detailed in Section 11 of Supplementary Materials. The co-culture was performed in 24-well plates equipped with Transwell inserts (0.4 μm pore size, Corning). Adherent epithelial or lung cancer cells were seeded in the lower chamber, while the different Tregfci subtypes were seeded in the upper insert. The co-culture system was maintained for 24 h under standard culture conditions.Quantitative reverse transcription PCR (qRT-PCR)Universal RNA Purification Kits (EZB-RN4, eBiosciences, Shanghai, China) were used to extract total RNA according to the manufacturer’s protocol. RNA concentration and purity were determined by measuring A260/A280 ratios using a Nanodrop spectrophotometer (Thermo Fisher Scientific, USA), and RNA integrity was assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies, USA). A total of 500 ng of total RNA was reverse-transcribed into cDNA using the PrimeScript RT Master Mix (RR036A, Takara, Japan). The real-time PCR was performed using TB Green Premix Ex Taq (RR420A, Takara, Japan) with the primers listed in Supplementary Table 2 on the ABI 7000 PCR instrument (Applied Biosystems, USA) by two-stage program parameters. The expression level of the target gene was normalized using the GAPDH housekeeping gene and the expression level was calculated by the comparative method (2−ΔΔCt). The primer design and DNA oligonucleotide sequences of GAPDH, FOXP3, CTLA4, IL2RA, IL-10, TGFβ, EBI3, TLR4, IL-27, IL-4, IL-6, IL-1β ETS1, ETS2, ETV1, ETV4, ETV6, TNFα, and IL-8 were detailed in the Section 12 of Supplementary Materials.Measurements of cytokine proteinsLevels of TNF-α, IL-1β, IL-6, and IL-8 proteins in culture supernatants across different experiments were measured using ELISA. We repeated the measurement of supernatant levels of TNF-α, IL-1β, IL-6, and IL-8 proteins 24 h after Tfci-NC or Tfci-KD culture pre-treated with vehicle or PHA (V-Tfci-NC, PHA-Tfci-NC, V-Tfci-KD, and PHA-Tfci-KD), or co-culture of A549 or HBE with Tfci-NC or Tfci-KD pre-treated with vehicle or PHA (n = 6/group). Each was performed with technical duplicates, and analyzed using standard ELISA procedures. The detailed methodology was described in Section 13 of the Supplementary Materials.Co-culture assayT cells were co-cultured with either A549 or HBE at a 10:1 ratio for 24 h. For the indirect co-culture model, A549 or HBE were seeded on the lower chamber of a 24-well plate at a density of 1.5 × 105 per well and allowed to adhere overnight. T cells were added to the upper chamber at a density of 1.5 × 106 cells per insert and cultured until the cells were confluent. Co-cultures were maintained in RPMI 1640 medium supplemented with 10% FBS for 24 h.Crystal violet stainingViable cell numbers were determined via crystal violet staining. A549, HBE, H446, H460, SPC-A1, H1688, H661 and LTEP cells were plated in 6-well plates/24-well plates overnight. After treatment, the RPMI 1640 culture medium was removed, and the cells were washed with PBS, fixed with paraformaldehyde for 10 min at room temperature, and washed twice. Cells were stained with 0.5% crystal violet (Beyotime Institute of Biotechnology, Cat number: KGA317, Jiangsu, China) for 15 min at room temperature. After staining, cells were washed twice with distilled water to remove excess dye, air-dried, and then imaged under a microscope. Images of stained cells were captured using an inverted microscope (Olympus IX71) at ×20 magnification. The number of viable cells was quantified by counting stained cells in randomly selected fields using ImageJ software (NIH, USA). All experiments were repeated twice (n = 6/group).Cell Counting Kit-8 (CCK-8) assayCell viability was assessed using CCK-8 (Keygen, Catalog KGA317, Jiangsu, China) according to the manufacturer’s instructions. Briefly, 2000 cells/well were seeded into 96-well plates and subjected to transient transfection with plasmid vectors. At pre-determined time points, 10 µL of CCK-8 reagent was added to each well, and cells were incubated for 2 h at 37 °C. Absorbance at 450 nm was measured using a microplate reader (BioTek, USA). Each experiment was repeated thrice (n = 6/group).Apoptosis assayFor analyzing apoptosis, single-cell suspensions were subjected to flow cytometric staining. The T cells were harvested and processed using the Apoptosis Detection Kit (Beyotime, Shanghai, China) and analyzed using a Beckman FACS flow cytometer.10. The percentage of apoptotic cells was determined via flow cytometry using the Alexa Fluor 488/annexin V/PI Cell Apoptosis Kit (Beyotime, Shanghai, China) according to the manufacturer’s protocol.Reactive oxygen species (ROS) assayLevels of intracellular ROS were measured via flow cytometry using an ROS detection kit (S0033, Beyotime, Shanghai, China). Briefly, the cells were washed with PBS after treatment, incubated with 15 μM 2',7'-dichlorofluorescin diacetate (DCFH2-DA) for 30 min at 37 °C in the dark, and analyzed using a FACSCalibur flow cytometer (BD Biosciences). Intracellular ROS levels were expressed as the mean DCFDA fluorescence intensity of the cells.Intracellular calcium assayIntracellular levels of calcium signaling were measured using a Fluo-4 Calcium Assay Kit according to the manufacturer’s protocol (S1061S, Beyotime, Shanghai, China). After treatment, medium from the culture plate was removed, and 100 µl of Fluo-4 AM dye loading solution was added to each well quickly. The cells were incubated at 37 °C for 30 min, followed by an additional 30 min at room temperature in the dark. Intracellular Ca²⁺ signals were measured using flow cytometry with excitation at 494 nm and emission at 516 nm. Potential fluorescence by organic transporters outside the cells was inhibited by adding probenecid, and the baseline signal was reduced.Mitochondrial stain and high-content screening assayCellular mitochondria were stained with MitoTracker Deep Red (MTDR) (40743ES50, Yeasen, Shanghai, China) and the fluorescence was determined. To assess the colocalization of nuclei and mitochondria, cells were stained with 25 nM Mito Tracker Deep Red FM (MTDR; Yeasen, cat# 40743ES50) in pre-warmed culture medium for 30 min at 37 °C under 5% CO₂ to label functional mitochondria. After staining, cells were washed with PBS and fixed with 4% paraformaldehyde for 15 min at room temperature. Following fixation, nuclei were incubated with 0.5 μg/mL Hoechst 33342 (Beyotime, Shanghai, China) at room temperature for 10 min. The emission/excitation wavelengths of the mitochondria and Hoechst stain were 665/644 nm and 460/346 nm, respectively. The cells were screened using the Perkin-Elmer high-content automatic imaging system in confocal mode, and the results were analyzed in the Perkin-Elmer Harmony image analysis software (PerkinElmer Life Sciences).Oxygen consumption rate (OCR) measurementThe OCR (in pmol/min) of cultured T cells was determined using a Seahorse XF96 Extracellular Flux Analyzer (Seahorse Bioscience) as per the manufacturer’s instructions. T cells were plated at a density of 200,000 or 50,000 cells/well in 24 or 96-well Seahorse assay plates and incubated for 1 h in a 37 °C incubator without CO2. The culture medium was replaced with XF assay medium containing unbuffered DMEM, 5.5 mM glucose, and 0.5 mM carnitine. Values of OCR were measured using the XF96 analyzer in XF base medium (pH 7.4) containing 1 µM mitochondrial respiration inhibitors oligomycin (OLI), 2 µM carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP), and 1 µM Antimycin A (AA). The data were analyzed using the Seahorse XF Mito Stress Test Reporter Generator package.Cell trajectory assaysT cells at 2 × 105 cells/mL/well were plated on 24-well plates for imaging using the Perkin-Elmer high-content automatic imaging system (PerkinElmer Life Sciences). The Perkin-Elmer Harmony image analysis software (PerkinElmer Life Sciences) was used to calculate the cell numbers, mean square displacement (MSD), and velocity of the T cells post-migration. The values calculated for the areas were inversely proportional to the amount of cell migration. Following a 48-h migration period, cell migration data were acquired through a high-content screening approach in the Operetta high-content screening system (Perkin Elmer). Data were presented as the mean value per group, along with the standard error of the mean (SEM).Transmission electron microscopyT cells were harvested, washed twice with PBS, and immediately fixed in 2.5% glutaraldehyde (Biossci, Wuhan, China) in 0.1 M phosphate buffer (pH 7.4) at 4 °C overnight. After three washes with phosphate buffer (10 min each), the samples were post-fixed with 1% osmium tetroxide for 2 h at room temperature. Cells were then dehydrated through a graded ethanol series with each step lasting 15 min (30%, 50%, 70%, 90%, and 100%), followed by two 15 min washes in 100% ethanol. The dehydrated samples were embedded in epoxy resin (SPI-PON 812), polymerized at 60 °C for 48 h, and sectioned into ultrathin slices (70–90 nm) using an ultramicrotome (Leica EM UC7, Leica Microsystems, Germany). The sections were mounted onto copper grids and stained with 2% uranyl acetate for 15 min, followed by 2.6% lead citrate for 10 min at room temperature. Transmission electron microscopy was performed using a Hitachi HT7800 electron microscope (Hitachi, Japan) at an accelerating voltage of 80 kV. Imaging and data acquisition were conducted by Biohao Biotechnology Co., Ltd. (Wuhan, China).Human PBMC isolation assayPBMCs were harvested from the healthy subjects and patients with NSCLC and isolated via the density gradient centrifugation using Ficoll (#1114546, Axis-Shield/Alere Technologies AS, Germany). Blood was carefully layered over Ficoll and centrifuged at 400 × g for 25 min at room temperature without brake. The PBMC layer was collected, washed twice with PBS, and counted using a hemocytometer. The collection and analysis of the patient samples were performed in accordance with the Declaration of Helsinki and the protocols were approved by the Institutional Review Board of Zhongshan Hospital. Written informed consent for genomic analysis was obtained from the participant.Flow cytometry assaySurface staining for flow cytometry was performed as per standard flow cytometry protocols. Briefly, PBMCs (~1 × 106 cells 100 µL per group) were collected from patients with NSCLC or healthy subjects, and first stained with FVS510 Live/Dead dye (Cat#564406, BD Biosciences) for 20 min at room temperature in the dark. Surface markers were labeled using the following antibodies for the classical Treg identification (CD3⁺CD4⁺CD25⁺CD127⁻CD45⁺): FITC-conjugated anti-CD3 (Cat#555339, clone UCHT1), APC-conjugated anti-CD4 (Cat#555349, clone RPA-T4), PE-conjugated anti-CD25 (Cat#555432, clone M-A251), BV421-conjugated anti-CD127 (Cat#562436, clone HIL-7R-M21), and APC-Cy7-conjugated anti-CD45 (Cat#557833, clone HI30), and for Tregfci identification (CD3⁺CD4⁺FOXP3⁺IL2RA⁺CTLA4⁺): FITC-conjugated anti-CD3 (Cat#555339, clone UCHT1); APC-conjugated anti-CD4 (Cat#555349, clone RPA-T4); PE-conjugated anti-IL2RA/CD25 (Cat#555432, clone M-A251); PE-Cy7-conjugated anti-CTLA4/CD152 (Cat#567341, clone BNI3) and AF647-conjugated anti-FOXP3 (Cat#560045, clone PCH101, BioLegend). For intracellular staining, AF647-conjugated anti-FOXP3 antibody was applied for 30 min at 4 °C in the dark, after the cells were fixed and permeabilized using BD Cytofix/Cytoperm kit (BD Biosciences) or FOXP3/Transcription Factor Staining Buffer Set (Cat#00-5523-00, eBioscience) according to the manufacturer’s instructions. Data were collected on an LSR III flow cytometer (BD Biosciences, California, USA) and analyzed using the FlowJo software v10.7.1 (FlowJo LLC). Gating strategies were verified using single-stain. All experiments were performed in three independent replicates.Bulk RNA-seq assayTotal RNA was extracted using TRIzol reagent (Beyotime, R0016, Suzhou, China), and RNA quality and integrity were assessed using the Agilent Bioanalyzer 2100 system (Agilent Technologies, CA, USA). The samples with an RIN > 7.0 were used for library construction. The poly(A)+ mRNA was enriched using oligo(dT)-conjugated magnetic beads (New England Biolabs), followed by fragmentation at 94 °C and synthesis of double-stranded cDNA using the NEBNext Ultra Directional RNA Library Prep Kit. After undergoing end repair and adapter ligation, PCR amplification was performed (typically 15 cycles). The resulting libraries were purified using AMPure XP beads (Beckman Coulter, USA). Library quality was evaluated via qRT-PCR, and libraries with an effective concentration >2 nM were pooled according to the target sequencing depth. Sequencing was performed using NovaSeq 6000 platform (paired-end) by Shanghai Biotechnology Corporation (Shanghai, China).Treg effects on autologous NSCLC organoidAutologous PBMC and cancer tissues were collected from four patients with NSCLC into tumor storage solution. Organoids were generated, passaged for 3–4 generations, and cryopreserved, as detailed in Section 14 of the Supplementary Materials. The recovered organoids were between the 4th and 6th generations, and the culture medium was changed every 3 days. PBMCs were cultured overnight at 37 °C and used for stimulating the autologous NSCLC organoids. After collection, cells were washed twice with staining buffer (DPBS supplemented with 1% FBS) and stained with Live/dead, Annexin V-FITC/PI, CD3, CD4, CD25, CD127, PD1, IFNγ. The cells were collected on BD Lyric flow cytometer and the analysis was performed using FlowJo software. Prior to co-culture with autologous T cells, tumor organoids were pre-stimulated with IFNγ to enhance antigen presentation. Plate-bound anti-CD28 and IL-2 were added to provide co-stimulation and support T cell proliferation, respectively. PBMC were stimulated weekly with autologous tumor organoids. Tumor recognition by CD8+ T cells was evaluated at baseline and after 2 weeks of co-culture, by staining for IFNγ and PD1.ETS1 roles in human and mouse TregfciTo evaluate direct roles of ETS1 in Tregfci, human Tregfci were isolated from PBMCs of healthy volunteers (n = 22) and treated with vehicle (1% DMSO) or ETS1 inhibitor (TK216, HY-122903, ShangHai Caerulum Pharma Discovery Co., Ltd., Shanghai, China) at 0.25 μM/ml for 24 h (n = 11/group). TK216, an orally active ETS inhibitor, inhibits ETS family protein function through the protein-protein interactions of EWS-FLI1.70 Afterward, cells were collected and prepared for scRNA-seq as described above. Peripheral WBCs and lung tissues were obtained from twelve mice (6 Ets1fl/fl control and 6 Ets1fl/fl; Foxp3YFP-Cre+/−). WBCs were isolated following red blood cell lysis, while lung lobes were immediately dissociated into single-cell suspensions within 30–60 min of harvest for scRNA-seq. Tissues were transported on ice, minced on ice, and digested with 0.25% Trypsin (Thermo Fisher, Cat. no. 25200-072) and 10 μg/mL DNase I (Sigma, Cat. no. 11284932001) in PBS supplemented with 5% FBS (Thermo Fisher, Cat. no. SV30087.02) at 37 °C with gentle shaking (50 rpm) for 40 min. Cell suspensions were collected every 20 min to maximize yield and viability, filtered through a 40 μm nylon strainer (Corning, Cat. no. 352340), and subjected to red blood cell lysis (Thermo Fisher, Cat. no. 00-4333-57). The suspensions with >90% viability were used for library preparation and were loaded at 2 × 10⁵ cells/mL onto the ABcellar® Single Cell Chip Set (ABclonal, RK30150). Gel Beads-in-emulsions (GEMs) were generated using the ABcellar® Single Cell Encapsulation Instrument (ABclonal, AI50001), containing the single cell and barcode-labeled gel bead for mRNA capture. The cDNAs were amplified and used for library construction according to the ABcellar® Single Cell 3' RNA-seq Lib Prep Kit protocol (ABclonal, RK20390). Library quality was evaluated with an Agilent High Sensitivity DNA Chip on a Bioanalyzer 2100 (Agilent Technologies) and quantified with the Qubit High Sensitivity DNA Assay Kit (Thermo Fisher Scientific). Sequencing was performed on an Illumina NovaSeq 6000 platform with 2 × 150 bp paired-end reads. Primary data processing was conducted with the ABcellar v1.2.5 pipeline using default parameters. FASTQ files generated from Illumina sequencing were aligned to the mouse reference genome (GRCm38) using the STAR algorithm. Unique molecular identifiers were counted to generate gene-barcode matrices. The resulting expression matrix was imported into Seurat (v3.0.2) in R for quality control, normalization, dimensionality reduction, clustering, and downstream analyses.Statistical analysisData are presented as the means ± SEM. Group comparisons were performed using the Mann–Whitney U test, using the Benjamini–Hochberg false discovery rate (FDR) procedure in multiple groups across multiple conditions. The one-way analysis of variance (ANOVA) was applied for comparison among multi-groups, followed by post-hoc Tukey’s test as needed. The hypergeometric test was used to evaluate statistical significance in the functional enrichment analysis of gene lists. To evaluate the potential influence of age variation on 5g- or 3g-Tregfci, we re-evaluated our own and public datasets and categorized them into young (