Dissecting T-cell exhaustion heterogeneity and immune ecosystem dynamics in colorectal cancer through multi-omics machine learning

Dissecting T-cell exhaustion heterogeneity and immune ecosystem dynamics in colorectal cancer through multi-omics machine learning | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Dissecting T-cell exhaustion heterogeneity and immune ecosystem dynamics in colorectal cancer through multi-omics machine learning Zhijing Zhang, Peng Ouyang, Kai Cui, Yixiang Wen, Xin Deng, Wanyu Chen, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8910521/v1 This work is licensed under a CC BY 4.0 License Status: Under Revision Version 1 posted 12 You are reading this latest preprint version Abstract Immunotherapy for colorectal cancer (CRC) currently faces significant dilemmas, but its specific mechanisms remain unclear. T-cell exhaustion (TEX) in the tumor microenvironment has been identified as a pivotal driver of immune evasion and tumor progression. Dissecting its contribution to CRC is essential for the development of rational therapeutic strategies. Here, we integrated scRNA-seq and bulk RNA-seq databases and leveraged pseudotemporal trajectory model to identify core genes. Subsequently, with the help of 10 machine learning models, we constructed a TEX score prognostic model, whose clinical utility was externally validated in independent immunotherapy cohorts, demonstrating intra-tumoral CD8⁺ T cells occupy a continuum of exhaustion states. Besides, the TEX-score model, constructed from five exhaustion-related genes (KLF3, LMNA, SLC2A3, ARL4C, and TIMP1), stably predicted CRC prognosis and immunotherapy responsiveness, validating that patients with low TEX-score exhibited prolonged overall survival (OS), abundant immune infiltrates and better response to immunotherapy. Collectively, our findings elucidate T-cell exhaustion as a central mediator of immunotherapy failure in CRC and provide a clinically actionable guidance for patient stratification and treatment selection. colorectal cancer immunotherapy sensitivity T cell exhaustion (TEX) tumor microenvironment (TME) machine learning predictable model Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Introduction Colorectal cancer (CRC) ranks as the third most prevalent malignancy globally and constitutes the second leading cause of cancer-related mortality worldwide [ 1 ] . Currently, a multidisciplinary paradigm centered on surgical resection remains the cornerstone of curative-intent therapy for CRC patients [ 2 ] . While resection with curative intent successfully eliminates the primary tumor in most patients, a notable proportion experience distant recurrence: roughly 5% of AJCC stage I, 15% of stage II, and 40% of stage III colorectal cancers ultimately develop metastatic disease in the years following surgery [ 3 ] . At the same time, most conventional chemotherapeutics carry a substantial risk of severe adverse effects and often promote the development of resistance mechanisms [ 4 , 5 ] . The emergence of immune-checkpoint blockade has redefined the therapeutic landscape of solid tumors, yet its impact on CRC remains circumscribed. [ 6 ] . However, due to the molecular and clinical heterogeneity in CRC, there was limited efficacy of immune checkpoint inhibitors (ICIs) in most cases [ 5 , 7 ] . Only subgroups with mismatch repair-deficient or microsatellite instability-high (MSI-H) of 5 ~ 15% of all CRC can benefit from immune checkpoint inhibitors [ 8 ] . As a result of this, it is crucial to expand the scope of CRC immunotherapy to benefit a broader patient population or combine other treatments to boost immunotherapy [ 9 ] . CD8 + T cells were always served as a core component of cancer immunotherapy [ 10 ] , however, prolonged antigen exposure and a suppressive tumor microenvironment caused progressive loss of effector function, leading to the exhaustion of CD8 + T cells, a dysfunctional state [ 11 ] . Exhausted T cells are defined as a specific lineage exhibiting progressive and hierarchical loss of effector function that play a crucial role in the therapeutic outcomes of ICIs [ 12 , 13 ] . In order to overcome the dilemma in therapy, we should better understand the hierarchical differentiation trajectory of exhausted T cells and figuring out the core regulatory genes [ 14 ] . Our research aims to dig out the potential molecular regulatory targets and core regulatory genes associated with the complex molecular and clinical heterogeneity in CRC. With the help of systematic bioinformatics analyses using scRNA-seq and bulk RNA-seq data, a multi-biomarker model based on genes linked to T cell exhaustion among CRC patients was constructed to evaluate the tumor microenvironment, predict immunotherapy response, and forecast the prognosis. Overall, our study provides a novel insight into the ICIs of CRC. Method 2.1 Data acquisition The Bulk seq data is derived from the TCGA and GEO databases (GSE39582); Single-cell sequencing data (scRNA seq) are from the GEO database (GSE132465). 2.2 Single-cell data processing and quality control Data analysis and quality evaluation using the R package “Seurat 5.3.0; https://satijalab.org/seurat/ (version). Firstly, cells expressing ≥ 300 genes and each gene being detected in ≥ 5 cells were retained, and the proportions of mitochondrial genes, red blood cell genes, and ribosome genes in each cell were calculated. Cells with a proportion of mitochondrial genes lower than 25%, a proportion of ribosome genes higher than 3%, and a proportion of red blood cell genes lower than 1% were screened. The cells after quality control were standardized: LogNormalize (with a scaling factor of 10,000) was used to correct the differences in sequencing depth, 2000 highly variant genes were screened to capture biological heterogeneity, and Z-score conversion was used to achieve the unification of gene expression level scales. The R package “Harmony” (version 1.2.3; https://www.harmony.r-project.org/ ) was used to eliminate the batch effect. Visualization was carried out using t-SNE. 2.3 Single-cell annotation Cluster with a resolution of 1 to obtain different cell populations. In order to identify and annotate specific cell types, marker genes of multiple cell populations were selected, including: Epithelial cells marker gene EPCAM (Epithelial Cell Adhesion Molecule); CD8 + T cells (CD8 + T cells) marker genes CD3D and CD8A; B cells (B cells) mark genes CD79A and MS4A1 (Membrane Spanning 4-Domains A1); Macrophage marker genes CD68 and CD163; Fibroblasts marker gene COL1A1 (Collagen Type I Alpha 1 Chain); endothelial cells marker gene PECAM1 (Platelet and Endothelial Cell Adhesion Molecule 1). 2.4 Identification and Analysis of CD8 + T Cell Subtypes CD8 + T cells were isolated and re-clustered using the R package “Seurat”. Subsequently, a single-cell pseudo-time trajectory was constructed using the R package “Monocle2”. Next, weighted gene-related network analysis was conducted using the R package “hdWGCNA” to identify the core gene set in the CD8 + T cell population. The R package “Cellchat” was used to explore the intercellular communication among all cell populations. The functional status of the gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways in each CD8 + T cell population was analyzed for differences using the R package “ClusterProfiler”. Furthermore, the GSEA pathway from MSigDB (gsea-msigdb.org) was evaluated using the R package “fgsea”. Finally, gene set variation analysis (GSVA) was conducted through the R package “GSVA” to analyze the differences among populations in the HALLMARK pathway. 2.5 Construction and validation of T-cell exhaustion markers Next, to further screen the core genes, we employed a variety of machine learning methods, including: Partial Least Squares (PLS) [ 15 ] , Random Forest (RF) [ 16 ] , Decision Tree (DTS) [ 17 ] , Support Vector Machine (SVM) [ 18 ] , Logistic Regression (Logistic) [ 19 ] , K-Nearest Neighbor Algorithm (KNN) [ 20 ] , XGBoost [ 21 ] , Gradient Boosting Machine (GBM) [ 22 ] , Neural Network (NeuralNet) [ 23 ] , Generalized Linear Model Boosting (glmBoost) [ 24 ] . Through these methods, we screened out the core genes related to prognosis. To visualize the screened core genes, we utilized the Upset graph to display the intersections among the core genes. Finally, we constructed a prognostic model using the Cox regression model, where the training set was from the TCGA dataset and the validation set was from the GSE39582 dataset. The performance of the prognostic model was evaluated by calculating the area under the curve (AUC) and the receiver operating characteristic curve (ROC), and was visualized. The difference analysis between the two groups was processed using the R packages “survminer”, “survival”, “rms” and “timeROC”. 2.7 Clinical features and nomogram establishment We employed univariate Cox regression and multivariate Cox regression analyses to assess the correlation and independence of risk values and clinical parameters in the TCGA cohort. To depict the differences among patient subgroups, a nomogram was established. This nomogram can accurately predict the probability of an individual experiencing a certain event in a clinical setting, combining independent clinical prognostic factors such as age, gender, and vascular infiltration. Subsequently, the performance of the nomogram in prognosis prediction was evaluated through calibration curves and ROC curves [ 25 ] . 2.8 Evaluation of immune-related characteristics A variety of methods, such as CIBERSORT [ 26 ] , EPIC [ 27 ] , MCP_counter [ 28 ] , Quanti-seq [ 29 ] and xCell [ 30 ] , were adopted for immune infiltration analysis. Based on the CIBERSORT results, we also conducted a correlation analysis between immune cells and genes. Meanwhile, the immune capacity of the tumor microenvironment (TME) was evaluated using seven different steps of the tumor immune cycle (TIP, hrbmu.edu.cn) and various immune indicators calculated through the “easier” package. In addition, we also examined the expression levels of co-stimulatory molecules, co-inhibitory molecules and HLA molecules. We calculated the differences in parameters such as T-cell inflammatory gene expression profile (GEP), cytotoxic activity (CYT), and IFN-γ between the high-risk and low-risk groups. We also conducted an immune checkpoint correlation analysis to further explore the role of the immune system in different risk groups. 2.9 Prediction of Immunotherapy In this study, we conducted immunotherapy validation through the dataset of GSE78002 (Hela cells), GSE126044 (non-small cell lung cancer), and the clinical trial cohort of immunotherapy for urothelial carcinoma led by the MD Anderson Cancer Center in the United States (IMVigor) [ 31 , 32 ] . To evaluate the response to immunotherapy, the proportions of responders and non-responders in the high-risk group and the low-risk group were calculated. In addition, we evaluated the performance of the immunotherapy prediction model by calculating the ROC curve and AUC value. The ROC curve is generated by correlating the immune response with the risk-score and using the AUC value to quantify the accuracy of the predictive model [ 33 , 34 ] . 2.10 Patients and tissue samples After approval by the Institutional Review Board, the studies were performed in accordance with the International Code of Ethics for Biomedical Research involving humans (CIOMS). We obtained written consent from the subjects before the study was performed. Specimens from three colorectal cancer patients who were sensitive to immunotherapy and three patients who were insensitive to immunotherapy were collected from the colorectal cancer database and tissue bank of the First Affiliated Hospital of Gannan Medical University (Approval No. LL8C-2025333). Tissue was taken from patients who underwent bowel cancer surgery between 2025 and 2026. The specimens were properly stored in 4% paraformaldehyde (biosharp, BL539A) or a -80℃ refrigerator after acquisition 2.11 Real-time Quantitative Fluorescence Polymerase Chain Reaction (RT-qPCR) After sample collection, total RNA was extracted using the total RNA Extraction Kit (NCM Biotech, M5105). Subsequently, cDNA was obtained following the steps of the Reverse Transcription Kit (Servicebio, G3331-50). Finally, the qPCR reaction was carried out using Universal Blue SYBR Green qPCR Master Mix (Servicebio, G3326-01). The quantification of relative gene expression levels was conducted using the 2-△△CT method. The primer sequence is shown in Supplemental table 1 . 2.12 Western Blot After sample collection, total protein was extracted using RIPA lysate (Thermo Fisher Scientific) containing PMSF (Servicebio, G2008-1 ml) and phosphatase inhibitors (GLPBIO, GK10011, GK10012). Subsequently, protein quantification was performed using Coomas brilliant Blue (Servicebio, G2039-250ML), followed by SDS-PAGE electrophoresis and electrotransfer onto NC membranes (ShareBio, SB-WB310). After blocking with 5% skimmed milk, the primary antibody was incubated overnight at 4 ℃. Next, after incubating the secondary antibody at room temperature for 1 hour, it was developed in the electrochemiluminescence imaging system (Tanon, Shanghai, China). All the antibodies involved are listed in the Supplemental table 2 . 2.13 Immunofluorescence The specimens were immersed in 4% paraformaldehyde and fixed overnight. Then, the tissues were paraffin-embedded and cut into 6µm sections. After reacting the sections with the specific primary antibody, they were incubated with the corresponding fluorescent secondary antibody, and the nuclei were stained with DAPI (4',6-diamidino-2-phenylindole). The staining results were collected by a fluorescence sectioning scanner, and then the data were processed with CaseViewer to analyze the expression levels of each marker and their spatial correlations. Results Comprehensive single-cell atlas of the CRC immune microenvironment To explore the complete cell atlas of colorectal cancer and its potential cell interactions, we conducted an in-depth analysis of the single-cell database in the GEO database (GSE132465), thereby obtaining 24 cell clusters (Fig. 1 A). Subsequent differential gene analysis and cell-specific markers classified the cells in the tumor microenvironment of colorectal cancer into six types, including B cells, CD8 + T cells, epithelial cells, endothelial cells, fibroblasts, and macrophages (Fig. 1 B, 1 C). Subsequently, further analysis and research on immune cells redefined CD8 + T cells based on their cellular functions (Fig. 1 D). A. Uniform Manifold Approximation and Projection (UMAP) visualization of transcriptionally distinct cell clusters identified from scRNA-seq analysis of CRC samples. B. Annotation of six major cell populations based on established lineage-specific marker genes. Colors denote cell types: T cells (orange), B cells (SaddleBrown), Macrophages (Turquoise), epithelial cells (SteelBlue), fibroblasts (Red), and endothelial cells (Purple). C. Dot plot showing expression profiles of canonical marker genes across the six major cell subsets. Dot size represents the percentage of cells expressing the gene; color intensity indicates average expression level. D. Sub-clustering analysis of CD8⁺ T cells, revealing two distinct subpopulations: Cluster 1 (early activation state, Orange) and Cluster 2 (exhaustion state, Green), as defined by differential expression of activation and exhaustion markers. Integrated analysis of CD8⁺ T cell differentiation trajectory and intercellular communication networks in the CRC microenvironment To further explore the differentiation status of different CD8⁺ T cell subsets and the dynamic rearrangement of cell type composition in the TME of CRC, we used pseudo-time series analysis to find that the trajectory evolution of the two CD8 + T cells was not completely the same and regrouped them according to the differentiation period. This result indicates that in the TME of CRC, CD8 + T cells gradually slide from an early activated state to a more dysfunctional phenotype (Fig. 2 A). Next, we predicted the intercellular interactions based on specific pathways and ligand-receptor pairs, and further constructed the intercellular communication network. The results showed that the interaction weights and intensities received by CD8⁺ T cell cluster 2 were higher than those of cluster 1 (Fig. 2 B, 2 C). Subsequent pathway analysis indicated that in cluster 2 CD8 + T cells, the inflammatory pathway represented by IFN-γ was significantly activated, and the elevated CD99 could endow cancer cells with migration, anti-apoptotic and T cell depletion capabilities, resulting in a “hot but ineffective” inflammatory state (Fig. 2 D). In addition, the visualization of the intensity of ligand-receptor interaction further revealed that the ligand-mediated communication between CD8⁺ T cell cluster 2 and other cell populations was significantly stronger than that of cluster 1 (Fig. 2 E). In conclusion, cluster 2 CD8 + T cells may be an important driver of immune escape and metastasis in colorectal cancer. A. Pseudotime trajectory analysis of CD8⁺ T cell subsets constructed using Monocle2. Cells are ordered along a continuous differentiation trajectory from Cluster 1 (early activation state, purple) to Cluster 2 (exhaustion state, green), with darker colors indicating later pseudotime values. B. Circle plot visualization of intercellular interaction number (left) and weights/strength (right) among major cell populations in the tumor microenvironment. Edge thickness and color intensity represent interaction strength; circle size indicates total interaction number. C. Heatmap depicting the number (left) and weights/strength (right) between cell populations. Rows and columns represent source and target cells, respectively. D. Cluster-cytokine association heatmap showing the activity patterns of cytokine signaling pathways across CD8⁺ T cell Cluster 1 and Cluster 2. Color scale represents normalized pathway enrichment scores or activity indices. E. Bubble plot illustrating cytokine-mediated intercellular communication patterns. Bubble area is proportional to communication probability; color coding distinguishes cytokine families or pathway types. Systematic identification of exhaustion-associated core genes through differential expression and weighted gene co-expression network analysis To identify the key genes that might cause CD8 + T cell exhaustion, we compared the differences in core genes among different clusters and drew volcano plots and heat plots (Fig. 3 A, 3 B). Subsequently, we employed the hdWGCNA (high-dimensional weighted gene co-expression network analysis) algorithm to calculate the gene expression profiles of two CD8⁺ T cell subsets, and divided the core genes among these subsets into different gene modules to identify the core gene set. Ultimately, we verified the correlation between genes and modules in the network. By setting the value (b) to 6, we successfully constructed the scale-free network (Fig. 3 C). The genes were hierarchically clustered into the corresponding modules and labeled with corresponding colors (Fig. 3 D). The most interesting thing is that the core genes are mainly concentrated in the brown module, with a correlation of approximately 47% (Fig. 3 E). A. Volcano plot showing differentially expressed genes between CD8 + T cell Cluster 1 and Cluster 2. Significantly upregulated genes in Cluster 2 (log₂FC > 0.5, adjusted P < 0.05) are highlighted in red. B. Heatmap of top differentially expressed genes across CD8 + T cell clusters. Rows represent genes and columns represent individual cells, with expression values normalized and scaled. C. Scale-free topology fitting index (left y-axis) and mean connectivity (right y-axis) as functions of soft-thresholding power (x-axis) for hdWGCNA network construction. A power of β = 6 was selected to achieve scale-free topology (R² > 0.8). D. Hierarchical clustering dendrogram of genes based on topological overlap matrix dissimilarity measure. Colors below the dendrogram indicate gene modules identified by dynamic tree cutting. The brown module showed the highest correlation with the exhaustion phenotype. E. Module–trait correlation heatmap depicting the relationship between hdWGCNA gene modules (rows) and CD8 + T cell clusters (columns). The brown module exhibited the strongest positive correlation with Cluster 2 (exhaustion state, r = 0.47, P < 0.001) Machine learning-based construction and comprehensive validation of the TEX-signature prognostic model in CRC In the previous study, we initially identified the possible genes for CD8 + T cell depletion with the help of WGNNA. To further search for its core target genes, we used pseudogenes for gene intersection and identified 50 intersection genes (Fig. 4 A). Next, we will utilize ten machine learning algorithms (DTS, GBM, glmBoost, KNN, Logistic, NeuralNet, PLS, RF, SVM, XGBoost) for further gene screening. The C-index of all models was > 0.8, indicating that each algorithm had strong predictive performance (Fig. 4 B, 4 C). Based on the above analysis, we extracted the intersections of the markers jointly screened out by various machine learning models and conducted Cox regression analysis, obtaining five target genes significantly related to survival period: Kruppel-like factor 3 (KLF3), lamin A/C (LMNA), glucose transporter member 3 (SLC2A3), ARF-like 4C (ARL4C) Tissue inhibitor of matrix metalloproteinase 1 (TIMP1) (Fig. 4 D, 4 E). Subsequently, we set TEX-Score as the weighted sum of the expression levels of these five genes and conducted a prognostic analysis based on this to verify our conjecture. In the TCGA and GEO cocohort, Kaplan-Meier analysis showed that the OS and disease-free survival (DFS) of patients in the high-risk group were significantly lower than those in the low-risk group (Fig. 4 F, 4 G). Meanwhile, the ROC curve was used to evaluate the model efficacy. This model has excellent diagnostic efficacy and predictive ability in 1, 3 and 5 years (Fig. 4 H). A. Venn diagram showing the intersection of pseudotime-derived differential genes and brown module core genes identified by hdWGCNA, yielding 50 candidate genes for downstream machine learning analysis. B. Importance ranking of genes selected by ten machine learning algorithms (DTS, GBM, glmBoost, KNN, Logistic, NeuralNet, PLS, RF, SVM, and XGBoost) in bubble plot. Bubble size indicates relative feature importance; consistency across algorithms highlights robust candidate biomarkers. C. ROC curves comparing the predictive performance of ten machine learning models. AUC values demonstrate model discrimination capability, with PLS and XGBoost showing optimal performance. D. UpSet plot illustrating the intersection of biomarkers selected by each machine learning algorithm. E. Forest plot of multivariate Cox regression analysis for the five core genes in the TCGA cohort. Hazard ratios (HR) and 95% confidence intervals indicate the independent prognostic value of each gene. F. Kaplan-Meier survival curves comparing OS between high-risk and low-risk patients stratified by TEX-signature in the TCGA training cohort. P-value calculated by log-rank test. G. From top to bottom, the point plot shows high- and low‐risk patients groups divided by the cutoff values (top). The distribution plot of survival time and survival status of high- and low-risk patient of TCGA dataset (bottom). H. Time-dependent ROC curves for 1-, 3-, and 5-year survival predictions in the TCGA cohort, demonstrating the prognostic accuracy of the TEX-signature model across different time horizons. Clinically applicable nomogram for individualized survival prediction and subgroup validation To better conduct clinical prediction, we constructed a nomogram integrating gender, age and TEX score (Fig. 5 A). The subsequent calibration curve indicated that there was a high degree of consistency between the predicted values of this model and the actual observed values, suggesting that it had good predictive efficacy (Fig. 5 B). To further test the predictive accuracy of TEX score for colorectal cancer, we conducted a subgroup analysis based on the clinical characteristics of the GEO database. It can be seen from the Kaplan-Meier survival analysis that the prognosis of the low TEX score group was consistently better than that of the high TEX group in subgroups of different genders and ages (Fig. 5 C, 5 D), confirming the feasibility of this model. A. Prognostic nomogram integrating clinical variables (gender, age) with the molecular TEX-signature for predicting 1-year, 3-year, and 5-year OS probability in colorectal cancer patients. Each variable axis is scaled according to its prognostic contribution; the sum of individual points yields total points, which correspond to specific survival probabilities on the bottom scale. B. Calibration curves assessing the agreement between nomogram-predicted survival probabilities and observed outcomes at 1, 3, and 5 years. The diagonal dashed line represents perfect calibration; the solid lines indicate the actual performance of the nomogram. C-D. Subgroup survival analysis stratified by clinical characteristics. Kaplan-Meier curves comparing TEX-signature high-risk versus low-risk patients in (C) age subgroups (≤ 65 years versus > 65 years) and (D) gender subgroups (female versus male). Immune landscape and modulator associations of the TEX-signature Considering the significance of CD8 + T cells in the immune system, we delved deeply into the relationship between TEX scores and immune cell infiltration as well as immune regulatory factors to assess the impact of TEX scores on the immune microenvironment of CRC. By analyzing with five independent algorithms (CIBERSORT, EPIC, MCP_counter, Quanti-seq, xCell), we found that the low TEX score group had infiltration of various immune cells including CD8 + , CD4, macrophages and NK cells (Fig. 6 A), and at the same time, Combined with the expression of immunomodulatory factors, the low TEX score group had higher expression of immunomodulatory factors, such as CD274, PDCD1, CTLA4, etc. (Fig. 6 B). Not only that, the expression of core genes is also correlated with various immunomodulatory factors and immune-infiltrating cells (Fig. 6 C, 6 D), among which SLC2A3 and ARL4C have the most significant correlations with immunity. In conclusion, the TEX score and the expression of core genes can serve as important target cues for immunotherapy, providing guidance for clinical decision-making in immunotherapy. A. Correlation between TEX-signature risk score and immune cell infiltration levels inferred by five deconvolution algorithms: CIBERSORT, MCP-counter, xCell, EPIC, and quanTIseq. Heatmap displays Pearson correlation coefficients; asterisks indicate statistical significance. B. Correlation between TEX-signature and immune modulators categorized into seven functional groups: chemokines, receptors, MHC molecules, immune-inhibitors, immune-stimulators, cell adhesion molecules, and others. Color intensity represents correlation strength; circle size indicates significance level. C. Core gene-specific immune modulator associations. Heatmap displays the correlation matrix between individual core genes (rows) and immune modulator categories (columns), revealing distinct immune-modulatory patterns for KLF3, LMNA, SLC2A3, ARL4C, and TIMP1. D. Pearson correlation analysis between core genes and significantly differentially infiltrated immune cells. Bar length represents correlation coefficient magnitude; color indicates positive (red) or negative (blue) associations. All data were shown in mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001. TEX-signature association with cancer immunity cycles and immune-related biological processes Previous studies have confirmed the correlation between TEX scores and tumor immunity. Therefore, we wish to continue exploring what immunological biological processes are influenced by this correlation. Analysis of multiple steps in the immune cycle suggests that low TEX score groups can activate multiple steps in the immune cycle. Including Release of cancer cell antigens, Priming and activation, Infiltration of immune cells into tumors and many immune cell recruiting, including T cell, CD4 T cell, CD8 + T cell, Th1 T cell. Dendritic cell, Th22 T cell, Macrophage, Neutrophil, NK T cell, Eosinophil, Basophil, Th17 cell. (Fig. 7 A) Analysis of immune-related responses suggested that multiple immune function-related responses related to T-cell activity and inflammation produced by IFN-γ were significantly active in the low TEX score group (Fig. 7 B). The immune resistance programs (RESF-Down, RESF-UP and resF) represent the effectiveness of immune resistance in the tumor microenvironment. Patients in the high TEX score group showed stronger immune resistance (resF, ResF-up). However, the low TEX score group had a lower level of immune resistance (resF-down) (Fig. 7 B). Subsequently, we comprehensively analyzed five databases, namely LM22, HALLMARK, IEGS, ANGELOVA, and REACTOME, to assess the infiltration scores of immune cells in different groups. The results showed that multiple immune cell infiltrations were observed in the low TEX score group (Fig. 7 C). In addition, the correlation analysis between transcription factors and TEX scores suggested that tumor-promoting transcription factors were highly expressed in the high TEX group, while tumor suppressor factors were present in higher amounts in the low TEX group (Fig. 7 D). Finally, with the help of ssGSEA (single-sample gene set enrichment analysis) assessment, we can find that in the high TEX score group, the expressions of the three inflammatory pathways GEP, INF, and CYT are all increased, and there is a significant correlation (Fig. 7 E- 7 G). All the above results indicate that CRC in the low-TEX group may have more immune cell infiltration and better immunotherapy efficacy. A. Stepwise analysis of the cancer-immunity cycle comparing high-risk versus low-risk TEX-signature patients. The seven-step cycle includes: (1) release of cancer cell antigens, (2) cancer antigen presentation, (3) priming and activation of T cells, (4) trafficking of immune cells to tumors, (5) infiltration of immune cells into tumors, (6) recognition of cancer cells by T cells, and (7) killing of cancer cells. Bar plots display normalized activity scores with statistical comparisons between risk groups. B. Comparison of immune-related indices between TEX-signature risk groups. Indices were calculated using the "easier" package, including T cell inflamed score, interferon-gamma response, and cytolytic activity. Box plots display distribution differences with statistical significance indicated. C. Multi-database immune signature enrichment analysis. Box-and-whisker plots comparing scores for immune cell populations and functional pathways derived from five reference collections: LM22 (CIBERSORT), HALLMARK, IEGS (innate immune gene sets), Angelova (immune checkpoint), and REACTOME. Each panel represents signatures from one database; y-axis indicates normalized enrichment score. D. Transcriptional regulatory network analysis. Heatmap visualization of correlations between TEX-signature risk score and activity scores of immune-related transcription factors (TFs). Color scale indicates Pearson correlation coefficients with hierarchical clustering of transcription factors. E-G. Pathway-specific immune activity correlations. Scatter plots with fitted regression lines showing the relationship between TEX-signature and ssGSEA enrichment scores for: (E) GEP signature representing antigen presentation machinery, (F) IFN response pathway, and (G) CYT score reflecting granzyme and perforin expression. Enrichment analysis of functional pathways for TEX scoring After confirming the correlation between TEX score and tumor-related immunity, we then investigated the relationship between TEX score and pathway enrichment. GO analysis demonstrated pathways enriched in high TEX scores (Fig. 8 A, 8 B), and combined with GSVA analysis of different TEX scores (Fig. 9 C), it can be confirmed that CRC with high TEX scores can promote chemotaxis of immune cells. Subsequently, KEGG and HALLMARK analyses of core genes suggested that the WNT, TGF-β, NOTCH, MTOR, p53, and HEDGEHOG pathways were significantly associated with core genes (Fig. 8 D, 8 E). A. GO-BP enrichment analysis of high-risk TEX-signature patients. Bar plot shows significantly enriched biological processes ranked by normalized enrichment score (NES); bar color indicates statistical significance. B. Tripartite GO landscape of high-risk patients. Cellular Component (GO-CC) showing subcellular localization of enriched genes; Biological Process (GO-BP) highlighting dysregulated pathways; Molecular Function (GO-MF) describing altered protein activities. Dot size corresponds to gene ratio (enriched genes/background genes); color intensity indicates -log10(FDR). C. GSVA comparing high-risk versus low-risk TEX-signature groups. Bar plot displays differential enrichment scores for hallmark gene sets; bar height indicates normalized enrichment score, with positive values (red) indicating up-regulated and negative values (blue) indicating down-regulated in high-risk group. D. KEGG pathway enrichment analysis of five core genes (KLF3, LMNA, SLC2A3, ARL4C, TIMP1). Heatmap displays enrichment significance (-log10(FDR)) across pathways; color intensity indicates statistical significance. E. MSigDB HALLMARK gene set analysis of core gene functions. Heatmap of normalized enrichment scores (NES) across 50 hallmark gene sets. Rows: individual gene sets; columns: sample groups or core genes. Red: pathway activation; blue: pathway suppression; hierarchical clustering applied to both dimensions. Clinical utility of TEX-signature for immunotherapy stratification and mechanistic validation To deeply explore the potential value of TEX scores in immunotherapy, we verified their predictive efficacy in multiple published treatment cohorts. The GSE78002, IMvigor, and GSE126044 datasets all suggest that high TEX scores are often associated with no response to immunotherapy, and the ROC curve validates the efficacy of the predictive model (Fig. 9 A- 9 C). Subsequently, to verify the expression of TEX core genes in tumor tissues of CRC patients with different sensitivities to immunotherapy, we detected the gene expression of 6 immunotherapy-sensitive/resistant patients by qPCR and Western Blot. It could be seen that KLF3 was lowly expressed in tumor tissues of the immunotherapy-sensitive group. LMNA, SLC2A3, ARL4C, and TIMP1 were highly expressed in tumor tissues of the immunotherapy-sensitive group (Fig. 9 D, 9 E). Further immunofluorescence techniques suggested that the core gene was co-localized with CD8 + (Fig. 9 F). Based on this, it is speculated that the abnormal expression of core genes may induce T cell exhaustion, reduce the efficacy of anti-tumor immunotherapy, and thereby promote the progression of CRC. A-C. Immunotherapy response prediction by TEX-signature across three independent cohorts. (A) GSE78002 Hela cell cohort, (B) IMvigor urothelial cancer cohort, and (C) GSE126044 non-small lung cancer cohort. Left panels: distribution of TEX-scores in responders versus non-responders; right panels: ROC curves evaluating predictive accuracy with AUC values. D-E. Experimental validation of core gene expression in immunotherapy-sensitive versus resistant tumor tissues. (D) RT-qPCR analysis of KLF3, LMNA, SLC2A3, ARL4C, and TIMP1 mRNA levels normalized to GAPDH. (E) Western blot analysis of core protein expression with β-actin as loading control. Data are presented as mean ± SD; statistical significance determined by Student's t-test. *P < 0.05, **P < 0.01, ***P < 0.001. F. Spatial validation by multiplex immunofluorescence. Confocal microscopy images (40× objective) demonstrating co-expression patterns of CD8⁺ T cell marker (green) with individual core genes (red) in formalin-fixed paraffin-embedded tumor sections. Nuclear counterstain with DAPI (blue). Insets show magnified regions of interest. Scale bar represents 20 micrometers. Discussion CRC remains a formidable health challenge globally, ranking as the third most common cancer and the second leading cause of cancer-related deaths [ 5 ] . Although significant progress has been made in the treatment of CRC in recent years, chemotherapy resistance and toxic side effects remain important factors affecting the prognosis of patients with CRC [ 35 ] . Despite immunotherapy has got great success in treating various cancers, the limited responsiveness of CRC to ICIs underscores the need for deeper understanding of them to broaden the population that can benefit from these treatments [ 9 ] . CD8 + T cells, originating from CD34 hematopoietic stem cells located in the bone marrow, are the main effector cells of antitumor immunity therapy, and CD8 + T cell exhaustion often leads to tumor deregulation and progression [ 36 ] . Tumor-induced T-cell exhaustion may be more complicated due to the tumor microenvironment [ 37 ] . Recently, the proteome of exhausted CD8 + T cell rose a mechanistic vulnerability and a new therapeutic target to improve cancer immunotherapies [ 38 ] , which reminded us further refine and scrutinize exhaustion models, seeking additional insights to advance immunotherapy for CRC. Through an exploration of the molecular and functional attributes of distinct CD8 + T cell subgroups in CRC, we figure out the intercellular communication and functional pathway enrichment of different clusters. It helps us better understand the microenvironment of CRC and visualize the function of different T cell cluster. We realized that the cluster 2 demonstrated more pronounced advantages across most inflammatory, immune activation pathways, and the ligand interaction. Summing up, we propose a process of functional exhaustion in the differentiation of CD8 + T cells between the two clusters. Subsequently, with the help of machine–learning and the identification of the core gene associated with exhaustion, five core genes, KLF3, LMNA, SLC2A3, ARL4C, and TIMP1 was figured out and the TEX-signature was built and verified with nomogram. Besides, the five core genes, all of which have been previously implicated in colorectal cancer prognosis [ 39 – 42 ] , have not been systematically examined in relation to the tumor immune microenvironment. By defining TEX-Score as the weighted sum of their expression levels, we uncover a significant association with immune infiltration, thereby establishing a previously unrecognized link between these prognostic genes and antitumor immunity. Plenty of learning model have confirmed that the low TEX-signature symbolize the better survival, more immune infiltration, and stronger immunity cycle. Finally, with the aid of GEO datasets and clinical cohort of immune therapy in Hela cells, non-small cell lung cancer, and urothelial carcinoma, we validate the TEX-signature model again. At the same time, the expression of core genes in CRC proves our conjectures. All in all, our research sets a promising foundation for future research and potential advancements in the treatment of CRC. It should be mentioned that there are a few of restrictions. Firstly, basing on single-cell sequencing data from a relatively small sample number, our research may not fully capture the heterogeneity, which limits efficacy of ICIs in CRC most importantly. Further validation in larger cohorts would provide more robust and generalizable results. Secondly, despite the advantages of large sample sizes and robust statistical power afforded by publicly available datasets, the integration of multi-source scRNA-seq and bulk RNA-seq data inevitably introduced technical heterogeneity attributable to disparities in sample processing protocols, platform-specific batch effects, and variable detection sensitivities. Additionally, the lack of comprehensive clinical annotations—including ethnic background, MSI/MSS status, specific chemotherapeutic regimens, and objective immunotherapy response criteria (e.g., RECIST)—in certain public cohorts constrained our capacity to delineate the precise clinical correlates of T-cell exhaustion states. To address these limitations, future studies will incorporate prospectively collected, clinically annotated biospecimens with standardized procurement protocols and harmonized analytical workflows, thereby ensuring robust cross-validation and clinical translatability of the TEX-score model. Thirdly, further molecular experiments are necessary to elucidate the functional roles of core genes and understand the underlying molecular mechanisms of TEX-signature. It is imperative to acknowledge that the majority of our findings were predicated upon correlative bioinformatic analyses, which, while robust for delineating transcriptional regulatory networks underlying T-cell exhaustion, inherently preclude definitive conclusions regarding protein-level expression dynamics and spatial heterogeneity within the colorectal tumor microenvironment. Specifically, our reliance on scRNA-seq and bulk RNA-seq datasets, integrated through computational deconvolution algorithms, may not fully resolve the intricate cellular crosstalk or the spatially restricted niches that govern T-cell dysfunction. This limitation is particularly pertinent given our observation that cluster 2 CD8 + T cells exhibited enhanced intercellular communication and ligand-receptor interactions, which necessitate spatial contextualization. To circumvent these constraints, future investigations should incorporate orthogonal validation strategies, including multiparameter flow cytometry for quantitative assessment of exhaustion marker co-expression (e.g., PD-1, TIM-3, LAG-3), multiplex immunofluorescence (mIF) or imaging mass cytometry (IMC) for spatially resolved profiling of immune cell infiltration patterns, and RNAscope or in situ hybridization for direct visualization of core gene transcripts (KLF3, LMNA, SLC2A3, ARL4C, TIMP1) within tissue architecture. These approaches will be instrumental in bridging the gap between transcriptional signatures and functional phenotypes, thereby strengthening the mechanistic underpinnings of the TEX-score model. Finally, the clinical research with only six samples is too limited. The need for further validation and clinical implementation is fundamental for digging out the potential clinical utility of the TEX-signature in guiding treatment decisions. In conclusion, our research aims to establish a novel assessment system, distinct from MSS/MSI-H, for personalized evaluation of immunotherapy sensitivity in colorectal cancer patients, thereby providing more individualized immunotherapy regimens and shedding light on new research directions for colorectal cancer immunotherapy. In conclusion, our research aims to establish a novel assessment system, distinct from MSS/MSI-H, for personalized evaluation of immunotherapy sensitivity in colorectal cancer patients, thereby providing more individualized immunotherapy regimens and shedding light on new research directions for colorectal cancer immunotherapy. Conclusion By integrating single-cell and bulk RNA-seq data via multiple machine-learning frameworks, we developed a prognostic model that simultaneously predicts OS and T-cell infiltration in CRC. The model estimates both patient survival probability and immunotherapy response, offering a robust biomarker for therapeutic efficacy and informing precision targeting strategies in CRC. Abbreviations CRC Colorectal cancer TEX T cell exhaustion scRNA-seq single cell RNA sequencing KLF3 Kruppel-like factor 3 LMNA Lamin A/C SLC2A3 Glucose transporter member 3 ARL4C ARF-like 4C TIMP1 Tissue inhibitor of matrix metalloproteinase 1 OS Overall survival DFS Disease-free survival HR Hazard ratios TME Tumor microenvironment AJCC American Joint Committee on Cancer Declarations Ethics approval and consent to participate Informed consent was obtained from all participants, and the study was approved by the Ethics Committee of First Affiliated Hospital of Gannan Medical University (approval number: LL8C-2025333). This study adhered to the principles of the Declaration of Helsinki. Data and material availability All data will be available on request. Consent for publication Not applicable. This study utilized anonymized data from public repositories (TCGA, GEO) and de-identified clinical samples. No individual patient data or identifiable information are presented Competing interests The authors declare that they have no competing interests. Funding This work was supported by the Guangdong Association for Science and Technology Youth Science and Technology Talent Development Program (NO. SKXRC2025117), National Natural Science Foundation of China (NO. 82504048), the Bethune Charitable Foundation (NO. 803292), Guangzhou Basic Research Program City - University (Institute) - Enterprise Joint Funding Project (NO. SL2024A03J01364), Medical Scientific Research Foundation of Guangdong Province of China (NO. A2023398), Fundamental Research Funds for the Central Universities (NO. 11624305). Author Contribution Zhijing Zhang, Peng Ouyang, and Kai Cui contributed equally to this work. Zhijing Zhang conceived the study, designed and conducted the experiments, and wrote the initial draft of the manuscript. Peng Ouyang and Kai Cui performed data analysis, interpreted the results, and contributed to manuscript writing. Xin Deng assisted with data interpretation and manuscript revision and obtained funding. Yixiang Wen and Wanyu Chen assisted with experimental work and data collection. Zhenhong Xian assisted with specimen and data collection. Qi Qi, Zhen Bao, Jin Gong and Xiao He share corresponding authorship. They supervised the project, provided critical feedback, obtained funding and were responsible for the final approval of the manuscript. All authors agree to be accountable for all aspects of work ensuring integrity and accuracy. Acknowledgement We extend our sincere gratitude to all researchers who have generously shared their datasets in public repositories, enabling this work. We also thank our colleagues and collaborators whose contributions were instrumental to the research cited herein. Finally, we are deeply indebted to the patients and their families who agreed to participate in this study. Data Availability All data will be available on request.Publicly available datasets analyzed during this study can be found in the following repositories. [https://portal.gdc.cancer.gov/](https:/portal.gdc.cancer.gov) (TCGA Colorectal Adenocarcinoma dataset); GEO: GSE39582 (bulk RNA-seq), GSE132465 (scRNA-seq), GSE78002 and GSE126044 (immunotherapy cohorts), and IMvigor210 (urothelial cancer immunotherapy cohort). The clinical experimental data and analysis scripts used during the current study are available from the corresponding author on reasonable request. References Bray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. Cancer J Clin. 2024;74:229–63. Lai JQ, Zhao LL, Hong C, Zou QM, Su JX, Li SJ, et al. Baicalein triggers ferroptosis in colorectal cancer cells via blocking the JAK2/STAT3/GPX4 axis. Acta Pharmacol Sin. 2024;45:1715–26. Cañellas-Socias A, Cortina C, Hernando-Momblona X, Palomo-Ponce S, Mulholland EJ, Turon G, et al. Metastatic recurrence in colorectal cancer arises from residual EMP1(+) cells. Nature. 2022;611:603–13. Su JX, Li SJ, Zhou XF, Zhang ZJ, Yan Y, Liu SL, et al. Chemotherapy-induced metastasis: molecular mechanisms and clinical therapies. Acta Pharmacol Sin. 2023;44:1725–36. Dekker E, Tanis PJ, Vleugels JLA, Kasi PM, Wallace MB. Colorectal cancer. Lancet (London England). 2019;394:1467–80. Page DB, Postow MA, Callahan MK, Allison JP, Wolchok JD. Immune modulation in cancer with antibodies. Annu Rev Med. 2014;65:185–202. Arnold M, Sierra MS, Laversanne M, Soerjomataram I, Jemal A, Bray F. Global patterns and trends in colorectal cancer incidence and mortality. Gut. 2017;66:683–91. Chen EX, Jonker DJ, Loree JM, Kennecke HF, Berry SR, Couture F, et al. Effect of Combined Immune Checkpoint Inhibition vs Best Supportive Care Alone in Patients With Advanced Colorectal Cancer: The Canadian Cancer Trials Group CO.26 Study. JAMA Oncol. 2020;6:831–8. Liu Y, Liu Z, Hu Y, Ling Y, Qing S, Liu Y, et al. Caffeine enhances antitumor T-cell activity by suppressing kynurenine pathway in colorectal cancer. Nat Commun. 2025;16:5906. Sun L, Su Y, Jiao A, Wang X, Zhang B. T cells in health and disease. Signal Transduct Target therapy. 2023;8:235. Wherry EJ. T cell exhaustion. Nat Immunol. 2011;12:492–9. Pauken KE, Sammons MA, Odorizzi PM, Manne S, Godec J, Khan O, et al. Epigenetic stability of exhausted T cells limits durability of reinvigoration by PD-1 blockade. Volume 354. New York, N.Y.): Science; 2016. pp. 1160–5. McLane LM, Abdel-Hakeem MS, Wherry EJ. CD8 T Cell Exhaustion During Chronic Viral Infection and Cancer. Annu Rev Immunol. 2019;37:457–95. Wang H, Zhou Y, Tu Y, You J, Gao B, Yuan K, et al. Targeting hepatocytic TβRI ameliorates liver metastatic outcomes by revitalizing stem-like CD8(+) Tex subsets. Nat Commun. 2025;16:10627. Ge J, Ge J, Tang G, Xiong D, Zhu D, Ding X, et al. Machine learning-based identification of biomarkers and drugs in immunologically cold and hot pancreatic adenocarcinomas. J translational Med. 2024;22:775. Qi X, Wang S, Fang C, Jia J, Lin L, Yuan T. Machine learning and SHAP value interpretation for predicting comorbidity of cardiovascular disease and cancer with dietary antioxidants. Redox Biol. 2025;79:103470. Zuo D, Yang L, Jin Y, Qi H, Liu Y, Ren L. Machine learning-based models for the prediction of breast cancer recurrence risk. BMC Med Inf Decis Mak. 2023;23:276. Huang S, Cai N, Pacheco PP, Narrandes S, Wang Y, Xu W. Applications of Support Vector Machine (SVM) Learning in Cancer Genomics. Cancer Genomics Proteomics. 2018;15:41–51. Qi W, Wang Y, Wang Y, Huang S, Li C, Jin H, et al. Prediction of postpartum depression in women: development and validation of multiple machine learning models. J translational Med. 2025;23:291. Li B, Kugeratski FG, Kalluri R. A novel machine learning algorithm selects proteome signature to specifically identify cancer exosomes. eLife 2024; 12. Yang X, Qiu H, Wang L, Wang X. Predicting Colorectal Cancer Survival Using Time-to-Event Machine Learning: Retrospective Cohort Study. J Med Internet Res. 2023;25:e44417. Vaziri-Moghadam A, Foroughmand-Araabi MH. Integrating machine learning and bioinformatics approaches for identifying novel diagnostic gene biomarkers in colorectal cancer. Sci Rep. 2024;14:24786. Yuan B, Yang D, Rothberg BEG, Chang H, Xu T. Unsupervised and supervised learning with neural network for human transcriptome analysis and cancer diagnosis. Sci Rep. 2020;10:19106. Wu J, Zhou X, Ren J, Zhang Z, Ju H, Diao X, et al. Glycosyltransferase-related prognostic and diagnostic biomarkers of uterine corpus endometrial carcinoma. Comput Biol Med. 2023;163:107164. Fan Q, Wang Y, Cheng J, Pan B, Zang X, Liu R, et al. Single-cell RNA-seq reveals T cell exhaustion and immune response landscape in osteosarcoma. Front Immunol. 2024;15:1362970. She H, Tan L, Wang Y, Du Y, Zhou Y, Zhang J, et al. Integrative single-cell RNA sequencing and metabolomics decipher the imbalanced lipid-metabolism in maladaptive immune responses during sepsis. Front Immunol. 2023;14:1181697. Yu T, Fang Z, Cheng Y, Zhou Y, Ji Y, Liu C. A novel palmitoylation-based molecular signature reveals COX6A1 as a key regulator in metabolic dysfunction-associated steatotic liver disease. J translational Med. 2025;23:1212. Shu J, Jiang J, Zhao G. Identification of novel gene signature for lung adenocarcinoma by machine learning to predict immunotherapy and prognosis. Front Immunol. 2023;14:1177847. Fan D, Yang M, Lee HJ, Lee JH, Kim HS. AVEN: a novel oncogenic biomarker with prognostic significance and implications of AVEN-associated immunophenotypes in lung adenocarcinoma. Front Mol Biosci. 2023;10:1265359. Fu S, Cheng Y, Wang X, Huang J, Su S, Wu H, et al. Identification of diagnostic gene biomarkers and immune infiltration in patients with diabetic kidney disease using machine learning strategies and bioinformatic analysis. Front Med. 2022;9:918657. Zhu W, Zeng H, Huang J, Wu J, Wang Y, Wang Z, et al. Integrated machine learning identifies epithelial cell marker genes for improving outcomes and immunotherapy in prostate cancer. J translational Med. 2023;21:782. Powles T, Bellmunt J, Comperat E, De Santis M, Huddart R, Loriot Y, et al. Bladder cancer: ESMO Clinical Practice Guideline for diagnosis, treatment and follow-up. Annals oncology: official J Eur Soc Med Oncol. 2022;33:244–58. Yang L, Wei S, Zhang J, Hu Q, Hu W, Cao M, et al. Construction of a predictive model for immunotherapy efficacy in lung squamous cell carcinoma based on the degree of tumor-infiltrating immune cells and molecular typing. J translational Med. 2022;20:364. Baty F, Joerger M, Früh M, Klingbiel D, Zappa F, Brutsche M. 24h-gene variation effect of combined bevacizumab/erlotinib in advanced non-squamous non-small cell lung cancer using exon array blood profiling. J translational Med. 2017;15:66. Yu T, Guo F, Yu Y, Sun T, Ma D, Han J et al. Fusobacterium nucleatum Promotes Chemoresistance to Colorectal Cancer by Modulating Autophagy. Cell 2017; 170: 548 – 63.e16. Wang Q, Qin Y, Li B. CD8(+) T cell exhaustion and cancer immunotherapy. Cancer Lett. 2023;559:216043. McKinney EF, Smith KGC. Metabolic exhaustion in infection, cancer and autoimmunity. Nat Immunol. 2018;19:213–21. Wang Y, Ma A, Song NJ, Shannon AE, Amankwah YS, Chen X, et al. Proteotoxic stress response drives T cell exhaustion and immune evasion. Nature. 2025;647:1025–35. Shen W, Yuan L, Hao B, Xiang J, Cheng F, Wu Z, et al. KLF3 promotes colorectal cancer growth by activating WNT1. Aging. 2024;16:2475–93. Dai W, Xu Y, Mo S, Li Q, Yu J, Wang R, et al. GLUT3 induced by AMPK/CREB1 axis is key for withstanding energy stress and augments the efficacy of current colorectal cancer therapies. Signal Transduct Target therapy. 2020;5:177. Wang Y, Chang Z, Sun Z, Wang L, Li J, Li G, et al. Targeting the ARL4C/RAP1/PI3K-Akt-mTOR signaling loop promotes ARL4C ubiquitination and reverses oxaliplatin resistance in colorectal cancer. Theranostics. 2026;16:99–122. Ben-Ami R, Wang QL, Zhang J, Supplee JG, Fahrmann JF, Lehmann-Werman R, et al. Protein biomarkers and alternatively methylated cell-free DNA detect early stage pancreatic cancer. Gut. 2024;73:639–48. Additional Declarations No competing interests reported. Supplementary Files Supplementaltable1Primer.xlsx Supplementaltable2antibody.xlsx rawdataofWBinfig.9.zip graphicalabstract.tif Cite Share Download PDF Status: Under Revision Version 1 posted Editorial decision: Revision requested 23 Mar, 2026 Reviews received at journal 18 Mar, 2026 Reviews received at journal 08 Mar, 2026 Reviews received at journal 07 Mar, 2026 Reviewers agreed at journal 28 Feb, 2026 Reviewers agreed at journal 25 Feb, 2026 Reviewers agreed at journal 23 Feb, 2026 Reviewers invited by journal 23 Feb, 2026 Editor invited by journal 20 Feb, 2026 Editor assigned by journal 20 Feb, 2026 Submission checks completed at journal 19 Feb, 2026 First submitted to journal 19 Feb, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8910521","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":597060944,"identity":"cae067dd-c7ca-4550-b02f-49c1ec980029","order_by":0,"name":"Zhijing Zhang","email":"","orcid":"","institution":"The First Affiliated Hospital of Jinan University","correspondingAuthor":false,"prefix":"","firstName":"Zhijing","middleName":"","lastName":"Zhang","suffix":""},{"id":597060945,"identity":"10487d52-8cba-4f18-8c6b-6c7b7f28d40b","order_by":1,"name":"Peng Ouyang","email":"","orcid":"","institution":"First Affiliated Hospital of Gannan Medical University","correspondingAuthor":false,"prefix":"","firstName":"Peng","middleName":"","lastName":"Ouyang","suffix":""},{"id":597060946,"identity":"ccbc6ab5-97f0-4ae2-91ba-8f57b5b4dc75","order_by":2,"name":"Kai Cui","email":"","orcid":"","institution":"Jinan University","correspondingAuthor":false,"prefix":"","firstName":"Kai","middleName":"","lastName":"Cui","suffix":""},{"id":597060947,"identity":"2df34847-76a0-470e-892f-27bed9aa6480","order_by":3,"name":"Yixiang Wen","email":"","orcid":"","institution":"The First Affiliated Hospital of Jinan University","correspondingAuthor":false,"prefix":"","firstName":"Yixiang","middleName":"","lastName":"Wen","suffix":""},{"id":597060948,"identity":"d68fcdb2-88eb-46a2-90e2-4b469fe8096a","order_by":4,"name":"Xin Deng","email":"","orcid":"","institution":"The First Affiliated Hospital of Jinan University","correspondingAuthor":false,"prefix":"","firstName":"Xin","middleName":"","lastName":"Deng","suffix":""},{"id":597060949,"identity":"6ff5dc4a-e244-41cb-97a6-7c88134b8a75","order_by":5,"name":"Wanyu Chen","email":"","orcid":"","institution":"The First Affiliated Hospital of Jinan University","correspondingAuthor":false,"prefix":"","firstName":"Wanyu","middleName":"","lastName":"Chen","suffix":""},{"id":597060950,"identity":"7a288795-aba5-4576-9d5b-fa4a8c76eab3","order_by":6,"name":"Zhenhong Xian","email":"","orcid":"","institution":"The First Affiliated Hospital of Jinan University","correspondingAuthor":false,"prefix":"","firstName":"Zhenhong","middleName":"","lastName":"Xian","suffix":""},{"id":597060951,"identity":"38c655cf-4af7-4a75-a84a-3e2e0e866fe8","order_by":7,"name":"Qi Qi","email":"","orcid":"","institution":"Jinan University","correspondingAuthor":false,"prefix":"","firstName":"Qi","middleName":"","lastName":"Qi","suffix":""},{"id":597060952,"identity":"5537d16f-3608-4f74-aef8-ff0d6c452088","order_by":8,"name":"Zhen Bao","email":"","orcid":"","institution":"The First Affiliated Hospital of Jinan University","correspondingAuthor":false,"prefix":"","firstName":"Zhen","middleName":"","lastName":"Bao","suffix":""},{"id":597060953,"identity":"b5486234-2d3d-47b5-b663-f7353403939f","order_by":9,"name":"Jin Gong","email":"","orcid":"","institution":"The First Affiliated Hospital of Jinan University","correspondingAuthor":false,"prefix":"","firstName":"Jin","middleName":"","lastName":"Gong","suffix":""},{"id":597060954,"identity":"83efcec6-f752-4b44-bedf-ee3ca9590f50","order_by":10,"name":"Xiao He","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA3ElEQVRIie3RsQrCMBCA4ZNAXSpxrCA+Q6WDCD5MgpBNcMzgUFDaweruWzg6Rg7qEnHtZovg5OIiLoLZlbZuDvngtvyQSwAs6w/5ZhSA51ICPGdyVjsZ9ToxXPxcp/USQwT+qXHtFAtSnQyaSV89JfItEiF56ACNl6w0GSYp2yca+QYdkfFdFzx93JZfLBsr1YqQr9E1iXbA9yYVybkI9y+ThEgfUx6RGklGFLYiEbQRBNRLtGDY1eaR5zD2mE7d6l0OOrjfpPlKqvj9KWc9Gq/Kkw/ub8cty7Ksr95B0lF/U4aI2AAAAABJRU5ErkJggg==","orcid":"","institution":"First Affiliated Hospital of Gannan Medical University","correspondingAuthor":true,"prefix":"","firstName":"Xiao","middleName":"","lastName":"He","suffix":""}],"badges":[],"createdAt":"2026-02-18 15:08:19","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8910521/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8910521/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":103567980,"identity":"3d033810-8a74-47e9-857f-c33ceb9dbf6a","added_by":"auto","created_at":"2026-02-27 07:36:25","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":6362135,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eComprehensive single-cell atlas of the CRC immune microenvironment\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA. Uniform Manifold Approximation and Projection (UMAP) visualization of transcriptionally distinct cell clusters identified from scRNA-seq analysis of CRC samples.\u003c/p\u003e\n\u003cp\u003eB. Annotation of six major cell populations based on established lineage-specific marker genes. Colors denote cell types: T cells (orange), B cells (SaddleBrown), Macrophages (Turquoise), epithelial cells (SteelBlue), fibroblasts (Red), and endothelial cells (Purple).\u003c/p\u003e\n\u003cp\u003eC. Dot plot showing expression profiles of canonical marker genes across the six major cell subsets. Dot size represents the percentage of cells expressing the gene; color intensity indicates average expression level.\u003c/p\u003e\n\u003cp\u003eD. Sub-clustering analysis of CD8⁺ T cells, revealing two distinct subpopulations: Cluster 1 (early activation state, Orange) and Cluster 2 (exhaustion state, Green), as defined by differential expression of activation and exhaustion markers.\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-8910521/v1/593d6197efa1d6ce50ec9608.png"},{"id":104398568,"identity":"237ff23c-01d2-4801-b24a-ce214f120408","added_by":"auto","created_at":"2026-03-11 12:02:58","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":5257219,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTrajectory and intercellular communication analysis of CD8⁺ T cells\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA. Pseudotime trajectory analysis of CD8⁺ T cell subsets constructed using Monocle2. Cells are ordered along a continuous differentiation trajectory from Cluster 1 (early activation state, purple) to Cluster 2 (exhaustion state, green), with darker colors indicating later pseudotime values.\u003c/p\u003e\n\u003cp\u003eB. Circle plot visualization of intercellular interaction number (left) and weights/strength (right) among major cell populations in the tumor microenvironment. Edge thickness and color intensity represent interaction strength; circle size indicates total interaction number.\u003c/p\u003e\n\u003cp\u003eC. Heatmap depicting the number (left) and weights/strength (right) between cell populations. Rows and columns represent source and target cells, respectively.\u003c/p\u003e\n\u003cp\u003eD. Cluster-cytokine association heatmap showing the activity patterns of cytokine signaling pathways across CD8⁺ T cell Cluster 1 and Cluster 2. Color scale represents normalized pathway enrichment scores or activity indices.\u003c/p\u003e\n\u003cp\u003eE. Bubble plot illustrating cytokine-mediated intercellular communication patterns. Bubble area is proportional to communication probability; color coding distinguishes cytokine families or pathway types.\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-8910521/v1/1f4f8bd1aa13c45c3d076bd7.png"},{"id":103567975,"identity":"9675170f-fabe-4cd4-a2e2-1d529c3d22e6","added_by":"auto","created_at":"2026-02-27 07:36:24","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":3557411,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIdentification of exhaustion-associated core genes via hdWGCNA\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA. Volcano plot showing differentially expressed genes between CD8+ T cell Cluster 1 and Cluster 2. Significantly upregulated genes in Cluster 2 (log₂FC \u0026gt; 0.5, adjusted P \u0026lt; 0.05) are highlighted in red.\u003c/p\u003e\n\u003cp\u003eB. Heatmap of top differentially expressed genes across CD8+ T cell clusters. Rows represent genes and columns represent individual cells, with expression values normalized and scaled.\u003c/p\u003e\n\u003cp\u003eC. Scale-free topology fitting index (left y-axis) and mean connectivity (right y-axis) as functions of soft-thresholding power (x-axis) for hdWGCNA network construction. A power of β = 6 was selected to achieve scale-free topology (R² \u0026gt; 0.8).\u003c/p\u003e\n\u003cp\u003eD. Hierarchical clustering dendrogram of genes based on topological overlap matrix dissimilarity measure. Colors below the dendrogram indicate gene modules identified by dynamic tree cutting. The brown module showed the highest correlation with the exhaustion phenotype.\u003c/p\u003e\n\u003cp\u003eE. Module–trait correlation heatmap depicting the relationship between hdWGCNA gene modules (rows) and CD8\u003csup\u003e+\u003c/sup\u003e T cell clusters (columns). The brown module exhibited the strongest positive correlation with Cluster 2 (exhaustion state, r = 0.47, P \u0026lt; 0.001)\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-8910521/v1/705313808fc91a56833e0243.png"},{"id":103567984,"identity":"7ac367be-2032-4c76-af59-e386f1417986","added_by":"auto","created_at":"2026-02-27 07:36:25","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":4634120,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMachine learning-based construction of TEX-signature prognostic model\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA. Venn diagram showing the intersection of pseudotime-derived differential genes and brown module core genes identified by hdWGCNA, yielding 50 candidate genes for downstream machine learning analysis.\u003c/p\u003e\n\u003cp\u003eB. Importance ranking of genes selected by ten machine learning algorithms (DTS, GBM, glmBoost, KNN, Logistic, NeuralNet, PLS, RF, SVM, and XGBoost) in bubble plot. Bubble size indicates relative feature importance; consistency across algorithms highlights robust candidate biomarkers.\u003c/p\u003e\n\u003cp\u003eC. ROC curves comparing the predictive performance of ten machine learning models. AUC values demonstrate model discrimination capability, with PLS and XGBoost showing optimal performance.\u003c/p\u003e\n\u003cp\u003eD. UpSet plot illustrating the intersection of biomarkers selected by each machine learning algorithm.\u003c/p\u003e\n\u003cp\u003eE. Forest plot of multivariate Cox regression analysis for the five core genes in the TCGA cohort. Hazard ratios (HR) and 95% confidence intervals indicate the independent prognostic value of each gene.\u003c/p\u003e\n\u003cp\u003eF. Kaplan-Meier survival curves comparing OS between high-risk and low-risk patients stratified by TEX-signature in the TCGA training cohort. P-value calculated by log-rank test.\u003c/p\u003e\n\u003cp\u003eG. From top to bottom, the point plot shows high‐ and low‐risk patients groups divided by the cutoff values (top). The distribution plot of survival time and survival status of high- and low-risk patient of TCGA dataset (bottom).\u003c/p\u003e\n\u003cp\u003eH. Time-dependent ROC curves for 1-, 3-, and 5-year survival predictions in the TCGA cohort, demonstrating the prognostic accuracy of the TEX-signature model across different time horizons.\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-8910521/v1/f8d4df4730c397747a40deed.png"},{"id":103567985,"identity":"5c7a31c2-4e2c-411b-b035-93308cdb0745","added_by":"auto","created_at":"2026-02-27 07:36:25","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":3304698,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eClinical nomogram construction and subgroup validation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA. Prognostic nomogram integrating clinical variables (gender, age) with the molecular TEX-signature for predicting 1-year, 3-year, and 5-year OS probability in colorectal cancer patients. Each variable axis is scaled according to its prognostic contribution; the sum of individual points yields total points, which correspond to specific survival probabilities on the bottom scale.\u003c/p\u003e\n\u003cp\u003eB. Calibration curves assessing the agreement between nomogram-predicted survival probabilities and observed outcomes at 1, 3, and 5 years. The diagonal dashed line represents perfect calibration; the solid lines indicate the actual performance of the nomogram.\u003c/p\u003e\n\u003cp\u003eC-D. Subgroup survival analysis stratified by clinical characteristics. Kaplan-Meier curves comparing TEX-signature high-risk versus low-risk patients in (C) age subgroups ( ≤ 65 years versus \u0026gt; 65 years) and (D) gender subgroups (female versus male).\u003c/p\u003e","description":"","filename":"Figure5.png","url":"https://assets-eu.researchsquare.com/files/rs-8910521/v1/1001de80d2edda82e55ba56a.png"},{"id":103567976,"identity":"baccb881-036a-4428-b706-37139809bac3","added_by":"auto","created_at":"2026-02-27 07:36:24","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":10107131,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eImmune landscape and modulator associations of the TEX-signature\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA. Correlation between TEX-signature risk score and immune cell infiltration levels inferred by five deconvolution algorithms: CIBERSORT, MCP-counter, xCell, EPIC, and quanTIseq. Heatmap displays Pearson correlation coefficients; asterisks indicate statistical significance.\u003c/p\u003e\n\u003cp\u003eB. Correlation between TEX-signature and immune modulators categorized into seven functional groups: chemokines, receptors, MHC molecules, immune-inhibitors, immune-stimulators, cell adhesion molecules, and others. Color intensity represents correlation strength; circle size indicates significance level.\u003c/p\u003e\n\u003cp\u003eC. Core gene-specific immune modulator associations. Heatmap displays the correlation matrix between individual core genes (rows) and immune modulator categories (columns), revealing distinct immune-modulatory patterns for KLF3, LMNA, SLC2A3, ARL4C, and TIMP1.\u003c/p\u003e\n\u003cp\u003eD. Pearson correlation analysis between core genes and significantly differentially infiltrated immune cells. Bar length represents correlation coefficient magnitude; color indicates positive (red) or negative (blue) associations.\u003c/p\u003e\n\u003cp\u003eAll data were shown in mean ± SD. *P \u0026lt; 0.05, **P \u0026lt; 0.01, ***P \u0026lt; 0.001.\u003c/p\u003e","description":"","filename":"Figure6.png","url":"https://assets-eu.researchsquare.com/files/rs-8910521/v1/d1e4b9644cb0214456ba0f5f.png"},{"id":103567979,"identity":"5f33db80-9894-423e-bed5-0e5ec916b171","added_by":"auto","created_at":"2026-02-27 07:36:25","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":5898681,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTEX-signature association with cancer immunity cycles\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA. Stepwise analysis of the cancer-immunity cycle comparing high-risk versus low-risk TEX-signature patients. The seven-step cycle includes: (1) release of cancer cell antigens, (2) cancer antigen presentation, (3) priming and activation of T cells, (4) trafficking of immune cells to tumors, (5) infiltration of immune cells into tumors, (6) recognition of cancer cells by T cells, and (7) killing of cancer cells. Bar plots display normalized activity scores with statistical comparisons between risk groups.\u003c/p\u003e\n\u003cp\u003eB. Comparison of immune-related indices between TEX-signature risk groups. Indices were calculated using the \"easier\" package, including T cell inflamed score, interferon-gamma response, and cytolytic activity. Box plots display distribution differences with statistical significance indicated.\u003c/p\u003e\n\u003cp\u003eC. Multi-database immune signature enrichment analysis. Box-and-whisker plots comparing scores for immune cell populations and functional pathways derived from five reference collections: LM22 (CIBERSORT), HALLMARK, IEGS (innate immune gene sets), Angelova (immune checkpoint), and REACTOME. Each panel represents signatures from one database; y-axis indicates normalized enrichment score.\u003c/p\u003e\n\u003cp\u003eD. Transcriptional regulatory network analysis. Heatmap visualization of correlations between TEX-signature risk score and activity scores of immune-related transcription factors (TFs). Color scale indicates Pearson correlation coefficients with hierarchical clustering of transcription factors.\u003c/p\u003e\n\u003cp\u003eE-G. Pathway-specific immune activity correlations. Scatter plots with fitted regression lines showing the relationship between TEX-signature and ssGSEA enrichment scores for: (E) GEP signature representing antigen presentation machinery, (F) IFN response pathway, and (G) CYT score reflecting granzyme and perforin expression.\u003c/p\u003e","description":"","filename":"Figure7.png","url":"https://assets-eu.researchsquare.com/files/rs-8910521/v1/070d42de9ca57ca793a3a078.png"},{"id":103567986,"identity":"6b382dc3-f6b8-4fc0-b689-1e9d2fead2b4","added_by":"auto","created_at":"2026-02-27 07:36:25","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":5309851,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFunction enrichment of the TEX-signature and core genes\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA. GO-BP enrichment analysis of high-risk TEX-signature patients. Bar plot shows significantly enriched biological processes ranked by normalized enrichment score (NES); bar color indicates statistical significance.\u003c/p\u003e\n\u003cp\u003eB. Tripartite GO landscape of high-risk patients. Cellular Component (GO-CC) showing subcellular localization of enriched genes; Biological Process (GO-BP) highlighting dysregulated pathways; Molecular Function (GO-MF) describing altered protein activities. Dot size corresponds to gene ratio (enriched genes/background genes); color intensity indicates -log10(FDR).\u003c/p\u003e\n\u003cp\u003eC. GSVA comparing high-risk versus low-risk TEX-signature groups. Bar plot displays differential enrichment scores for hallmark gene sets; bar height indicates normalized enrichment score, with positive values (red) indicating up-regulated and negative values (blue) indicating down-regulated in high-risk group.\u003c/p\u003e\n\u003cp\u003eD. KEGG pathway enrichment analysis of five core genes (KLF3, LMNA, SLC2A3, ARL4C, TIMP1). Heatmap displays enrichment significance (-log10(FDR)) across pathways; color intensity indicates statistical significance.\u003c/p\u003e\n\u003cp\u003eE. MSigDB HALLMARK gene set analysis of core gene functions. Heatmap of normalized enrichment scores (NES) across 50 hallmark gene sets. Rows: individual gene sets; columns: sample groups or core genes. Red: pathway activation; blue: pathway suppression; hierarchical clustering applied to both dimensions.\u003c/p\u003e","description":"","filename":"Figure8.png","url":"https://assets-eu.researchsquare.com/files/rs-8910521/v1/119e8db55ce8e6f7a6796f33.png"},{"id":103567988,"identity":"03a80ee6-f5f6-4d99-8248-942b110290e2","added_by":"auto","created_at":"2026-02-27 07:36:25","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":14521694,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eClinical utility and mechanistic validation of TEX-signature\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA-C. Immunotherapy response prediction by TEX-signature across three independent cohorts. (A) GSE78002 Hela cell cohort, (B) IMvigor urothelial cancer cohort, and (C) GSE126044 non-small lung cancer cohort. Left panels: distribution of TEX-scores in responders versus non-responders; right panels: ROC curves evaluating predictive accuracy with AUC values.\u003c/p\u003e\n\u003cp\u003eD-E. Experimental validation of core gene expression in immunotherapy-sensitive versus resistant tumor tissues. (D) RT-qPCR analysis of KLF3, LMNA, SLC2A3, ARL4C, and TIMP1 mRNA levels normalized to GAPDH. (E) Western blot analysis of core protein expression with β-actin as loading control. Data are presented as mean ± SD; statistical significance determined by Student's t-test. *P \u0026lt; 0.05, **P \u0026lt; 0.01, ***P \u0026lt; 0.001.\u003c/p\u003e\n\u003cp\u003eF. Spatial validation by multiplex immunofluorescence. Confocal microscopy images (40× objective) demonstrating co-expression patterns of CD8⁺ T cell marker (green) with individual core genes (red) in formalin-fixed paraffin-embedded tumor sections. Nuclear counterstain with DAPI (blue). Insets show magnified regions of interest. Scale bar represents 20 micrometers.\u003c/p\u003e","description":"","filename":"Figure9.png","url":"https://assets-eu.researchsquare.com/files/rs-8910521/v1/cbca13ee935aabf3f753575f.png"},{"id":104407641,"identity":"ce199449-d1e4-45ef-8f3e-a29902a45db5","added_by":"auto","created_at":"2026-03-11 12:39:18","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":58857195,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8910521/v1/cb9d4461-287f-42d1-b7d9-5d6a581c907f.pdf"},{"id":103567981,"identity":"b3b99c68-c630-494c-8a5b-cdf1d50a574d","added_by":"auto","created_at":"2026-02-27 07:36:25","extension":"xlsx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":9400,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaltable1Primer.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-8910521/v1/a5b294aa7ddef1b69554e814.xlsx"},{"id":104398089,"identity":"2f052e79-216f-489a-a52e-bd006ffd302c","added_by":"auto","created_at":"2026-03-11 11:59:38","extension":"xlsx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":10337,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaltable2antibody.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-8910521/v1/27fb4c8d6649cf540ba07b41.xlsx"},{"id":103567992,"identity":"bf5b6379-482f-4365-a207-b19378621522","added_by":"auto","created_at":"2026-02-27 07:36:26","extension":"zip","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":4386525,"visible":true,"origin":"","legend":"","description":"","filename":"rawdataofWBinfig.9.zip","url":"https://assets-eu.researchsquare.com/files/rs-8910521/v1/d1bdf4ceb36670008f1c3a65.zip"},{"id":103567982,"identity":"d5ddce74-2313-412a-a85f-45c24a03a3ec","added_by":"auto","created_at":"2026-02-27 07:36:25","extension":"tif","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":6898256,"visible":true,"origin":"","legend":"","description":"","filename":"graphicalabstract.tif","url":"https://assets-eu.researchsquare.com/files/rs-8910521/v1/517fe9f75e90358e40320324.tif"}],"financialInterests":"No competing interests reported.","formattedTitle":"Dissecting T-cell exhaustion heterogeneity and immune ecosystem dynamics in colorectal cancer through multi-omics machine learning","fulltext":[{"header":"Introduction","content":"\u003cp\u003eColorectal cancer (CRC) ranks as the third most prevalent malignancy globally and constitutes the second leading cause of cancer-related mortality worldwide\u003csup\u003e[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]\u003c/sup\u003e. Currently, a multidisciplinary paradigm centered on surgical resection remains the cornerstone of curative-intent therapy for CRC patients\u003csup\u003e[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]\u003c/sup\u003e. While resection with curative intent successfully eliminates the primary tumor in most patients, a notable proportion experience distant recurrence: roughly 5% of AJCC stage I, 15% of stage II, and 40% of stage III colorectal cancers ultimately develop metastatic disease in the years following surgery\u003csup\u003e[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]\u003c/sup\u003e. At the same time, most conventional chemotherapeutics carry a substantial risk of severe adverse effects and often promote the development of resistance mechanisms\u003csup\u003e[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eThe emergence of immune-checkpoint blockade has redefined the therapeutic landscape of solid tumors, yet its impact on CRC remains circumscribed.\u003csup\u003e[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]\u003c/sup\u003e. However, due to the molecular and clinical heterogeneity in CRC, there was limited efficacy of immune checkpoint inhibitors (ICIs) in most cases\u003csup\u003e[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]\u003c/sup\u003e. Only subgroups with mismatch repair-deficient or microsatellite instability-high (MSI-H) of 5\u0026thinsp;~\u0026thinsp;15% of all CRC can benefit from immune checkpoint inhibitors\u003csup\u003e[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]\u003c/sup\u003e. As a result of this, it is crucial to expand the scope of CRC immunotherapy to benefit a broader patient population or combine other treatments to boost immunotherapy\u003csup\u003e[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eCD8\u003csup\u003e+\u003c/sup\u003e T cells were always served as a core component of cancer immunotherapy\u003csup\u003e[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]\u003c/sup\u003e, however, prolonged antigen exposure and a suppressive tumor microenvironment caused progressive loss of effector function, leading to the exhaustion of CD8\u003csup\u003e+\u003c/sup\u003e T cells, a dysfunctional state\u003csup\u003e[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]\u003c/sup\u003e. Exhausted T cells are defined as a specific lineage exhibiting progressive and hierarchical loss of effector function that play a crucial role in the therapeutic outcomes of ICIs\u003csup\u003e[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]\u003c/sup\u003e. In order to overcome the dilemma in therapy, we should better understand the hierarchical differentiation trajectory of exhausted T cells and figuring out the core regulatory genes\u003csup\u003e[\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eOur research aims to dig out the potential molecular regulatory targets and core regulatory genes associated with the complex molecular and clinical heterogeneity in CRC. With the help of systematic bioinformatics analyses using scRNA-seq and bulk RNA-seq data, a multi-biomarker model based on genes linked to T cell exhaustion among CRC patients was constructed to evaluate the tumor microenvironment, predict immunotherapy response, and forecast the prognosis. Overall, our study provides a novel insight into the ICIs of CRC.\u003c/p\u003e"},{"header":"Method","content":"\u003cp\u003e \u003cb\u003e2.1 Data acquisition\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThe Bulk seq data is derived from the TCGA and GEO databases (GSE39582); Single-cell sequencing data (scRNA seq) are from the GEO database (GSE132465).\u003c/p\u003e \u003cp\u003e \u003cb\u003e2.2 Single-cell data processing and quality control\u003c/b\u003e \u003c/p\u003e \u003cp\u003eData analysis and quality evaluation using the R package \u0026ldquo;Seurat 5.3.0;\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://satijalab.org/seurat/\u003c/span\u003e\u003cspan address=\"https://satijalab.org/seurat/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (version). Firstly, cells expressing\u0026thinsp;\u0026ge;\u0026thinsp;300 genes and each gene being detected in \u0026ge;\u0026thinsp;5 cells were retained, and the proportions of mitochondrial genes, red blood cell genes, and ribosome genes in each cell were calculated. Cells with a proportion of mitochondrial genes lower than 25%, a proportion of ribosome genes higher than 3%, and a proportion of red blood cell genes lower than 1% were screened. The cells after quality control were standardized: LogNormalize (with a scaling factor of 10,000) was used to correct the differences in sequencing depth, 2000 highly variant genes were screened to capture biological heterogeneity, and Z-score conversion was used to achieve the unification of gene expression level scales. The R package \u0026ldquo;Harmony\u0026rdquo; (version 1.2.3; \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.harmony.r-project.org/\u003c/span\u003e\u003cspan address=\"https://www.harmony.r-project.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) was used to eliminate the batch effect. Visualization was carried out using t-SNE.\u003c/p\u003e \u003cp\u003e \u003cb\u003e2.3 Single-cell annotation\u003c/b\u003e \u003c/p\u003e \u003cp\u003eCluster with a resolution of 1 to obtain different cell populations. In order to identify and annotate specific cell types, marker genes of multiple cell populations were selected, including: Epithelial cells marker gene EPCAM (Epithelial Cell Adhesion Molecule); CD8\u003csup\u003e+\u003c/sup\u003e T cells (CD8\u003csup\u003e+\u003c/sup\u003e T cells) marker genes CD3D and CD8A; B cells (B cells) mark genes CD79A and MS4A1 (Membrane Spanning 4-Domains A1); Macrophage marker genes CD68 and CD163; Fibroblasts marker gene COL1A1 (Collagen Type I Alpha 1 Chain); endothelial cells marker gene PECAM1 (Platelet and Endothelial Cell Adhesion Molecule 1).\u003c/p\u003e \u003cp\u003e \u003cb\u003e2.4 Identification and Analysis of CD8\u003c/b\u003e \u003csup\u003e \u003cb\u003e+\u003c/b\u003e \u003c/sup\u003e \u003cb\u003eT Cell Subtypes\u003c/b\u003e\u003c/p\u003e \u003cp\u003eCD8\u003csup\u003e+\u003c/sup\u003e T cells were isolated and re-clustered using the R package \u0026ldquo;Seurat\u0026rdquo;. Subsequently, a single-cell pseudo-time trajectory was constructed using the R package \u0026ldquo;Monocle2\u0026rdquo;. Next, weighted gene-related network analysis was conducted using the R package \u0026ldquo;hdWGCNA\u0026rdquo; to identify the core gene set in the CD8\u003csup\u003e+\u003c/sup\u003e T cell population. The R package \u0026ldquo;Cellchat\u0026rdquo; was used to explore the intercellular communication among all cell populations. The functional status of the gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways in each CD8\u003csup\u003e+\u003c/sup\u003e T cell population was analyzed for differences using the R package \u0026ldquo;ClusterProfiler\u0026rdquo;. Furthermore, the GSEA pathway from MSigDB (gsea-msigdb.org) was evaluated using the R package \u0026ldquo;fgsea\u0026rdquo;. Finally, gene set variation analysis (GSVA) was conducted through the R package \u0026ldquo;GSVA\u0026rdquo; to analyze the differences among populations in the HALLMARK pathway.\u003c/p\u003e \u003cp\u003e \u003cb\u003e2.5 Construction and validation of T-cell exhaustion markers\u003c/b\u003e \u003c/p\u003e \u003cp\u003eNext, to further screen the core genes, we employed a variety of machine learning methods, including: Partial Least Squares (PLS)\u003csup\u003e[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]\u003c/sup\u003e, Random Forest (RF)\u003csup\u003e[\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]\u003c/sup\u003e, Decision Tree (DTS)\u003csup\u003e[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]\u003c/sup\u003e, Support Vector Machine (SVM)\u003csup\u003e[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]\u003c/sup\u003e, Logistic Regression (Logistic)\u003csup\u003e[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]\u003c/sup\u003e, K-Nearest Neighbor Algorithm (KNN)\u003csup\u003e[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]\u003c/sup\u003e, XGBoost\u003csup\u003e[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]\u003c/sup\u003e, Gradient Boosting Machine (GBM)\u003csup\u003e[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]\u003c/sup\u003e, Neural Network (NeuralNet)\u003csup\u003e[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]\u003c/sup\u003e, Generalized Linear Model Boosting (glmBoost)\u003csup\u003e[\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]\u003c/sup\u003e. Through these methods, we screened out the core genes related to prognosis. To visualize the screened core genes, we utilized the Upset graph to display the intersections among the core genes. Finally, we constructed a prognostic model using the Cox regression model, where the training set was from the TCGA dataset and the validation set was from the GSE39582 dataset. The performance of the prognostic model was evaluated by calculating the area under the curve (AUC) and the receiver operating characteristic curve (ROC), and was visualized. The difference analysis between the two groups was processed using the R packages \u0026ldquo;survminer\u0026rdquo;, \u0026ldquo;survival\u0026rdquo;, \u0026ldquo;rms\u0026rdquo; and \u0026ldquo;timeROC\u0026rdquo;.\u003c/p\u003e \u003cp\u003e \u003cb\u003e2.7 Clinical features and nomogram establishment\u003c/b\u003e \u003c/p\u003e \u003cp\u003eWe employed univariate Cox regression and multivariate Cox regression analyses to assess the correlation and independence of risk values and clinical parameters in the TCGA cohort. To depict the differences among patient subgroups, a nomogram was established. This nomogram can accurately predict the probability of an individual experiencing a certain event in a clinical setting, combining independent clinical prognostic factors such as age, gender, and vascular infiltration. Subsequently, the performance of the nomogram in prognosis prediction was evaluated through calibration curves and ROC curves\u003csup\u003e[\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003e \u003cb\u003e2.8 Evaluation of immune-related characteristics\u003c/b\u003e \u003c/p\u003e \u003cp\u003eA variety of methods, such as CIBERSORT\u003csup\u003e[\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]\u003c/sup\u003e, EPIC\u003csup\u003e[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]\u003c/sup\u003e, MCP_counter\u003csup\u003e[\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]\u003c/sup\u003e, Quanti-seq\u003csup\u003e[\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]\u003c/sup\u003e and xCell\u003csup\u003e[\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]\u003c/sup\u003e, were adopted for immune infiltration analysis. Based on the CIBERSORT results, we also conducted a correlation analysis between immune cells and genes. Meanwhile, the immune capacity of the tumor microenvironment (TME) was evaluated using seven different steps of the tumor immune cycle (TIP, hrbmu.edu.cn) and various immune indicators calculated through the \u0026ldquo;easier\u0026rdquo; package. In addition, we also examined the expression levels of co-stimulatory molecules, co-inhibitory molecules and HLA molecules. We calculated the differences in parameters such as T-cell inflammatory gene expression profile (GEP), cytotoxic activity (CYT), and IFN-γ between the high-risk and low-risk groups. We also conducted an immune checkpoint correlation analysis to further explore the role of the immune system in different risk groups.\u003c/p\u003e \u003cp\u003e \u003cb\u003e2.9 Prediction of Immunotherapy\u003c/b\u003e \u003c/p\u003e \u003cp\u003eIn this study, we conducted immunotherapy validation through the dataset of GSE78002 (Hela cells), GSE126044 (non-small cell lung cancer), and the clinical trial cohort of immunotherapy for urothelial carcinoma led by the MD Anderson Cancer Center in the United States (IMVigor)\u003csup\u003e[\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]\u003c/sup\u003e. To evaluate the response to immunotherapy, the proportions of responders and non-responders in the high-risk group and the low-risk group were calculated. In addition, we evaluated the performance of the immunotherapy prediction model by calculating the ROC curve and AUC value. The ROC curve is generated by correlating the immune response with the risk-score and using the AUC value to quantify the accuracy of the predictive model\u003csup\u003e[\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003e \u003cb\u003e2.10 Patients and tissue samples\u003c/b\u003e \u003c/p\u003e \u003cp\u003eAfter approval by the Institutional Review Board, the studies were performed in accordance with the International Code of Ethics for Biomedical Research involving humans (CIOMS). We obtained written consent from the subjects before the study was performed.\u003c/p\u003e \u003cp\u003eSpecimens from three colorectal cancer patients who were sensitive to immunotherapy and three patients who were insensitive to immunotherapy were collected from the colorectal cancer database and tissue bank of the First Affiliated Hospital of Gannan Medical University (Approval No. LL8C-2025333). Tissue was taken from patients who underwent bowel cancer surgery between 2025 and 2026. The specimens were properly stored in 4% paraformaldehyde (biosharp, BL539A) or a -80℃ refrigerator after acquisition\u003c/p\u003e \u003cp\u003e \u003cb\u003e2.11 Real-time Quantitative Fluorescence Polymerase Chain Reaction (RT-qPCR)\u003c/b\u003e \u003c/p\u003e \u003cp\u003eAfter sample collection, total RNA was extracted using the total RNA Extraction Kit (NCM Biotech, M5105). Subsequently, cDNA was obtained following the steps of the Reverse Transcription Kit (Servicebio, G3331-50). Finally, the qPCR reaction was carried out using Universal Blue SYBR Green qPCR Master Mix (Servicebio, G3326-01). The quantification of relative gene expression levels was conducted using the 2-△△CT method. The primer sequence is shown in Supplemental table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cb\u003e2.12 Western Blot\u003c/b\u003e \u003c/p\u003e \u003cp\u003eAfter sample collection, total protein was extracted using RIPA lysate (Thermo Fisher Scientific) containing PMSF (Servicebio, G2008-1 ml) and phosphatase inhibitors (GLPBIO, GK10011, GK10012). Subsequently, protein quantification was performed using Coomas brilliant Blue (Servicebio, G2039-250ML), followed by SDS-PAGE electrophoresis and electrotransfer onto NC membranes (ShareBio, SB-WB310). After blocking with 5% skimmed milk, the primary antibody was incubated overnight at 4 ℃. Next, after incubating the secondary antibody at room temperature for 1 hour, it was developed in the electrochemiluminescence imaging system (Tanon, Shanghai, China). All the antibodies involved are listed in the Supplemental table \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cb\u003e2.13 Immunofluorescence\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThe specimens were immersed in 4% paraformaldehyde and fixed overnight. Then, the tissues were paraffin-embedded and cut into 6\u0026micro;m sections. After reacting the sections with the specific primary antibody, they were incubated with the corresponding fluorescent secondary antibody, and the nuclei were stained with DAPI (4',6-diamidino-2-phenylindole). The staining results were collected by a fluorescence sectioning scanner, and then the data were processed with CaseViewer to analyze the expression levels of each marker and their spatial correlations.\u003c/p\u003e"},{"header":"Results","content":" \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003cdiv id=\"Sec4\" class=\"Section3\"\u003e \u003ch2\u003eComprehensive single-cell atlas of the CRC immune microenvironment\u003c/h2\u003e \u003cp\u003eTo explore the complete cell atlas of colorectal cancer and its potential cell interactions, we conducted an in-depth analysis of the single-cell database in the GEO database (GSE132465), thereby obtaining 24 cell clusters (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eA). Subsequent differential gene analysis and cell-specific markers classified the cells in the tumor microenvironment of colorectal cancer into six types, including B cells, CD8\u003csup\u003e+\u003c/sup\u003e T cells, epithelial cells, endothelial cells, fibroblasts, and macrophages (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eB, \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eC). Subsequently, further analysis and research on immune cells redefined CD8\u003csup\u003e+\u003c/sup\u003e T cells based on their cellular functions (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eD).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eA. Uniform Manifold Approximation and Projection (UMAP) visualization of transcriptionally distinct cell clusters identified from scRNA-seq analysis of CRC samples.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eB. Annotation of six major cell populations based on established lineage-specific marker genes. Colors denote cell types: T cells (orange), B cells (SaddleBrown), Macrophages (Turquoise), epithelial cells (SteelBlue), fibroblasts (Red), and endothelial cells (Purple).\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eC. Dot plot showing expression profiles of canonical marker genes across the six major cell subsets. Dot size represents the percentage of cells expressing the gene; color intensity indicates average expression level.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eD. Sub-clustering analysis of CD8⁺ T cells, revealing two distinct subpopulations: Cluster 1 (early activation state, Orange) and Cluster 2 (exhaustion state, Green), as defined by differential expression of activation and exhaustion markers.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eIntegrated analysis of CD8⁺ T cell differentiation trajectory and intercellular communication networks in the CRC microenvironment\u003c/b\u003e \u003c/p\u003e \u003cp\u003eTo further explore the differentiation status of different CD8⁺ T cell subsets and the dynamic rearrangement of cell type composition in the TME of CRC, we used pseudo-time series analysis to find that the trajectory evolution of the two CD8\u003csup\u003e+\u003c/sup\u003e T cells was not completely the same and regrouped them according to the differentiation period. This result indicates that in the TME of CRC, CD8\u003csup\u003e+\u003c/sup\u003e T cells gradually slide from an early activated state to a more dysfunctional phenotype (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). Next, we predicted the intercellular interactions based on specific pathways and ligand-receptor pairs, and further constructed the intercellular communication network. The results showed that the interaction weights and intensities received by CD8⁺ T cell cluster 2 were higher than those of cluster 1 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB, \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC).\u003c/p\u003e \u003cp\u003eSubsequent pathway analysis indicated that in cluster 2 CD8\u003csup\u003e+\u003c/sup\u003e T cells, the inflammatory pathway represented by IFN-γ was significantly activated, and the elevated CD99 could endow cancer cells with migration, anti-apoptotic and T cell depletion capabilities, resulting in a \u0026ldquo;hot but ineffective\u0026rdquo; inflammatory state (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eD). In addition, the visualization of the intensity of ligand-receptor interaction further revealed that the ligand-mediated communication between CD8⁺ T cell cluster 2 and other cell populations was significantly stronger than that of cluster 1 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eE). In conclusion, cluster 2 CD8\u003csup\u003e+\u003c/sup\u003e T cells may be an important driver of immune escape and metastasis in colorectal cancer.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eA. Pseudotime trajectory analysis of CD8⁺ T cell subsets constructed using Monocle2. Cells are ordered along a continuous differentiation trajectory from Cluster 1 (early activation state, purple) to Cluster 2 (exhaustion state, green), with darker colors indicating later pseudotime values.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eB. Circle plot visualization of intercellular interaction number (left) and weights/strength (right) among major cell populations in the tumor microenvironment. Edge thickness and color intensity represent interaction strength; circle size indicates total interaction number.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eC. Heatmap depicting the number (left) and weights/strength (right) between cell populations. Rows and columns represent source and target cells, respectively.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eD. Cluster-cytokine association heatmap showing the activity patterns of cytokine signaling pathways across CD8⁺ T cell Cluster 1 and Cluster 2. Color scale represents normalized pathway enrichment scores or activity indices.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eE. Bubble plot illustrating cytokine-mediated intercellular communication patterns. Bubble area is proportional to communication probability; color coding distinguishes cytokine families or pathway types.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003c/div\u003e \u003c/div\u003e\n\u003ch3\u003eSystematic identification of exhaustion-associated core genes through differential expression and weighted gene co-expression network analysis\u003c/h3\u003e\n\u003cp\u003eTo identify the key genes that might cause CD8\u003csup\u003e+\u003c/sup\u003e T cell exhaustion, we compared the differences in core genes among different clusters and drew volcano plots and heat plots (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA, \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB). Subsequently, we employed the hdWGCNA (high-dimensional weighted gene co-expression network analysis) algorithm to calculate the gene expression profiles of two CD8⁺ T cell subsets, and divided the core genes among these subsets into different gene modules to identify the core gene set. Ultimately, we verified the correlation between genes and modules in the network. By setting the value (b) to 6, we successfully constructed the scale-free network (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC). The genes were hierarchically clustered into the corresponding modules and labeled with corresponding colors (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eD). The most interesting thing is that the core genes are mainly concentrated in the brown module, with a correlation of approximately 47% (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eE).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eA. Volcano plot showing differentially expressed genes between CD8\u0026thinsp;+\u0026thinsp;T cell Cluster 1 and Cluster 2. Significantly upregulated genes in Cluster 2 (log₂FC\u0026thinsp;\u0026gt;\u0026thinsp;0.5, adjusted P\u0026thinsp;\u0026lt;\u0026thinsp;0.05) are highlighted in red.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eB. Heatmap of top differentially expressed genes across CD8\u0026thinsp;+\u0026thinsp;T cell clusters. Rows represent genes and columns represent individual cells, with expression values normalized and scaled.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eC. Scale-free topology fitting index (left y-axis) and mean connectivity (right y-axis) as functions of soft-thresholding power (x-axis) for hdWGCNA network construction. A power of β\u0026thinsp;=\u0026thinsp;6 was selected to achieve scale-free topology (R\u0026sup2; \u0026gt; 0.8).\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eD. Hierarchical clustering dendrogram of genes based on topological overlap matrix dissimilarity measure. Colors below the dendrogram indicate gene modules identified by dynamic tree cutting. The brown module showed the highest correlation with the exhaustion phenotype.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eE. Module\u0026ndash;trait correlation heatmap depicting the relationship between hdWGCNA gene modules (rows) and CD8\u003csup\u003e+\u003c/sup\u003e T cell clusters (columns). The brown module exhibited the strongest positive correlation with Cluster 2 (exhaustion state, r\u0026thinsp;=\u0026thinsp;0.47, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001)\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e\n\u003ch3\u003eMachine learning-based construction and comprehensive validation of the TEX-signature prognostic model in CRC\u003c/h3\u003e\n\u003cp\u003eIn the previous study, we initially identified the possible genes for CD8\u003csup\u003e+\u003c/sup\u003e T cell depletion with the help of WGNNA. To further search for its core target genes, we used pseudogenes for gene intersection and identified 50 intersection genes (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA). Next, we will utilize ten machine learning algorithms (DTS, GBM, glmBoost, KNN, Logistic, NeuralNet, PLS, RF, SVM, XGBoost) for further gene screening. The C-index of all models was \u0026gt;\u0026thinsp;0.8, indicating that each algorithm had strong predictive performance (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eB, \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eC). Based on the above analysis, we extracted the intersections of the markers jointly screened out by various machine learning models and conducted Cox regression analysis, obtaining five target genes significantly related to survival period: Kruppel-like factor 3 (KLF3), lamin A/C (LMNA), glucose transporter member 3 (SLC2A3), ARF-like 4C (ARL4C) Tissue inhibitor of matrix metalloproteinase 1 (TIMP1) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eD, \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eE). Subsequently, we set TEX-Score as the weighted sum of the expression levels of these five genes and conducted a prognostic analysis based on this to verify our conjecture. In the TCGA and GEO cocohort, Kaplan-Meier analysis showed that the OS and disease-free survival (DFS) of patients in the high-risk group were significantly lower than those in the low-risk group (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eF, \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eG). Meanwhile, the ROC curve was used to evaluate the model efficacy. This model has excellent diagnostic efficacy and predictive ability in 1, 3 and 5 years (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eH).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eA. Venn diagram showing the intersection of pseudotime-derived differential genes and brown module core genes identified by hdWGCNA, yielding 50 candidate genes for downstream machine learning analysis.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eB. Importance ranking of genes selected by ten machine learning algorithms (DTS, GBM, glmBoost, KNN, Logistic, NeuralNet, PLS, RF, SVM, and XGBoost) in bubble plot. Bubble size indicates relative feature importance; consistency across algorithms highlights robust candidate biomarkers.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eC. ROC curves comparing the predictive performance of ten machine learning models. AUC values demonstrate model discrimination capability, with PLS and XGBoost showing optimal performance.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eD. UpSet plot illustrating the intersection of biomarkers selected by each machine learning algorithm.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eE. Forest plot of multivariate Cox regression analysis for the five core genes in the TCGA cohort. Hazard ratios (HR) and 95% confidence intervals indicate the independent prognostic value of each gene.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eF. Kaplan-Meier survival curves comparing OS between high-risk and low-risk patients stratified by TEX-signature in the TCGA training cohort. P-value calculated by log-rank test.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eG. From top to bottom, the point plot shows high- and low‐risk patients groups divided by the cutoff values (top). The distribution plot of survival time and survival status of high- and low-risk patient of TCGA dataset (bottom).\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eH. Time-dependent ROC curves for 1-, 3-, and 5-year survival predictions in the TCGA cohort, demonstrating the prognostic accuracy of the TEX-signature model across different time horizons.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e\n\u003ch3\u003eClinically applicable nomogram for individualized survival prediction and subgroup validation\u003c/h3\u003e\n\u003cp\u003eTo better conduct clinical prediction, we constructed a nomogram integrating gender, age and TEX score (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA). The subsequent calibration curve indicated that there was a high degree of consistency between the predicted values of this model and the actual observed values, suggesting that it had good predictive efficacy (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB). To further test the predictive accuracy of TEX score for colorectal cancer, we conducted a subgroup analysis based on the clinical characteristics of the GEO database. It can be seen from the Kaplan-Meier survival analysis that the prognosis of the low TEX score group was consistently better than that of the high TEX group in subgroups of different genders and ages (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eC, \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eD), confirming the feasibility of this model.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eA. Prognostic nomogram integrating clinical variables (gender, age) with the molecular TEX-signature for predicting 1-year, 3-year, and 5-year OS probability in colorectal cancer patients. Each variable axis is scaled according to its prognostic contribution; the sum of individual points yields total points, which correspond to specific survival probabilities on the bottom scale.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eB. Calibration curves assessing the agreement between nomogram-predicted survival probabilities and observed outcomes at 1, 3, and 5 years. The diagonal dashed line represents perfect calibration; the solid lines indicate the actual performance of the nomogram.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003eC-D. Subgroup survival analysis stratified by clinical characteristics. Kaplan-Meier curves comparing TEX-signature high-risk versus low-risk patients in (C) age subgroups (\u0026le;\u0026thinsp;65 years versus \u0026gt;\u0026thinsp;65 years) and (D) gender subgroups (female versus male).\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eImmune landscape and modulator associations of the TEX-signature\u003c/h2\u003e \u003cp\u003eConsidering the significance of CD8\u003csup\u003e+\u003c/sup\u003e T cells in the immune system, we delved deeply into the relationship between TEX scores and immune cell infiltration as well as immune regulatory factors to assess the impact of TEX scores on the immune microenvironment of CRC. By analyzing with five independent algorithms (CIBERSORT, EPIC, MCP_counter, Quanti-seq, xCell), we found that the low TEX score group had infiltration of various immune cells including CD8\u003csup\u003e+\u003c/sup\u003e, CD4, macrophages and NK cells (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eA), and at the same time, Combined with the expression of immunomodulatory factors, the low TEX score group had higher expression of immunomodulatory factors, such as CD274, PDCD1, CTLA4, etc. (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eB). Not only that, the expression of core genes is also correlated with various immunomodulatory factors and immune-infiltrating cells (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eC, \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eD), among which SLC2A3 and ARL4C have the most significant correlations with immunity. In conclusion, the TEX score and the expression of core genes can serve as important target cues for immunotherapy, providing guidance for clinical decision-making in immunotherapy.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eA. Correlation between TEX-signature risk score and immune cell infiltration levels inferred by five deconvolution algorithms: CIBERSORT, MCP-counter, xCell, EPIC, and quanTIseq.\u0026nbsp;Heatmap displays Pearson correlation coefficients; asterisks indicate statistical significance.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eB. Correlation between TEX-signature and immune modulators categorized into seven functional groups: chemokines, receptors, MHC molecules, immune-inhibitors, immune-stimulators, cell adhesion molecules, and others. Color intensity represents correlation strength; circle size indicates significance level.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eC. Core gene-specific immune modulator associations. Heatmap displays the correlation matrix between individual core genes (rows) and immune modulator categories (columns), revealing distinct immune-modulatory patterns for KLF3, LMNA, SLC2A3, ARL4C, and TIMP1.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eD. Pearson correlation analysis between core genes and significantly differentially infiltrated immune cells. Bar length represents correlation coefficient magnitude; color indicates positive (red) or negative (blue) associations.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003eAll data were shown in mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD. *P\u0026thinsp;\u0026lt;\u0026thinsp;0.05, **P\u0026thinsp;\u0026lt;\u0026thinsp;0.01, ***P\u0026thinsp;\u0026lt;\u0026thinsp;0.001.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eTEX-signature association with cancer immunity cycles and immune-related biological processes\u003c/h3\u003e\n\u003cp\u003ePrevious studies have confirmed the correlation between TEX scores and tumor immunity. Therefore, we wish to continue exploring what immunological biological processes are influenced by this correlation. Analysis of multiple steps in the immune cycle suggests that low TEX score groups can activate multiple steps in the immune cycle. Including Release of cancer cell antigens, Priming and activation, Infiltration of immune cells into tumors and many immune cell recruiting, including T cell, CD4 T cell, CD8\u003csup\u003e+\u003c/sup\u003e T cell, Th1 T cell. Dendritic cell, Th22 T cell, Macrophage, Neutrophil, NK T cell, Eosinophil, Basophil, Th17 cell. (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eA)\u003c/p\u003e \u003cp\u003eAnalysis of immune-related responses suggested that multiple immune function-related responses related to T-cell activity and inflammation produced by IFN-γ were significantly active in the low TEX score group (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eB). The immune resistance programs (RESF-Down, RESF-UP and resF) represent the effectiveness of immune resistance in the tumor microenvironment. Patients in the high TEX score group showed stronger immune resistance (resF, ResF-up). However, the low TEX score group had a lower level of immune resistance (resF-down) (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eB). Subsequently, we comprehensively analyzed five databases, namely LM22, HALLMARK, IEGS, ANGELOVA, and REACTOME, to assess the infiltration scores of immune cells in different groups. The results showed that multiple immune cell infiltrations were observed in the low TEX score group (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eC). In addition, the correlation analysis between transcription factors and TEX scores suggested that tumor-promoting transcription factors were highly expressed in the high TEX group, while tumor suppressor factors were present in higher amounts in the low TEX group (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eD). Finally, with the help of ssGSEA (single-sample gene set enrichment analysis) assessment, we can find that in the high TEX score group, the expressions of the three inflammatory pathways GEP, INF, and CYT are all increased, and there is a significant correlation (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eE-\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eG). All the above results indicate that CRC in the low-TEX group may have more immune cell infiltration and better immunotherapy efficacy.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eA. Stepwise analysis of the cancer-immunity cycle comparing high-risk versus low-risk TEX-signature patients. The seven-step cycle includes: (1) release of cancer cell antigens, (2) cancer antigen presentation, (3) priming and activation of T cells, (4) trafficking of immune cells to tumors, (5) infiltration of immune cells into tumors, (6) recognition of cancer cells by T cells, and (7) killing of cancer cells. Bar plots display normalized activity scores with statistical comparisons between risk groups.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eB. Comparison of immune-related indices between TEX-signature risk groups. Indices were calculated using the \"easier\" package, including T cell inflamed score, interferon-gamma response, and cytolytic activity. Box plots display distribution differences with statistical significance indicated.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eC. Multi-database immune signature enrichment analysis. Box-and-whisker plots comparing scores for immune cell populations and functional pathways derived from five reference collections: LM22 (CIBERSORT), HALLMARK, IEGS (innate immune gene sets), Angelova (immune checkpoint), and REACTOME. Each panel represents signatures from one database; y-axis indicates normalized enrichment score.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eD. Transcriptional regulatory network analysis. Heatmap visualization of correlations between TEX-signature risk score and activity scores of immune-related transcription factors (TFs). Color scale indicates Pearson correlation coefficients with hierarchical clustering of transcription factors.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003eE-G. Pathway-specific immune activity correlations. Scatter plots with fitted regression lines showing the relationship between TEX-signature and ssGSEA enrichment scores for: (E) GEP signature representing antigen presentation machinery, (F) IFN response pathway, and (G) CYT score reflecting granzyme and perforin expression.\u003c/p\u003e\n\u003ch3\u003eEnrichment analysis of functional pathways for TEX scoring\u003c/h3\u003e\n\u003cp\u003eAfter confirming the correlation between TEX score and tumor-related immunity, we then investigated the relationship between TEX score and pathway enrichment. GO analysis demonstrated pathways enriched in high TEX scores (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eA, \u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eB), and combined with GSVA analysis of different TEX scores (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003eC), it can be confirmed that CRC with high TEX scores can promote chemotaxis of immune cells. Subsequently, KEGG and HALLMARK analyses of core genes suggested that the WNT, TGF-β, NOTCH, MTOR, p53, and HEDGEHOG pathways were significantly associated with core genes (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eD, \u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eE).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eA. GO-BP enrichment analysis of high-risk TEX-signature patients. Bar plot shows significantly enriched biological processes ranked by normalized enrichment score (NES); bar color indicates statistical significance.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eB. Tripartite GO landscape of high-risk patients. Cellular Component (GO-CC) showing subcellular localization of enriched genes; Biological Process (GO-BP) highlighting dysregulated pathways; Molecular Function (GO-MF) describing altered protein activities. Dot size corresponds to gene ratio (enriched genes/background genes); color intensity indicates -log10(FDR).\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eC. GSVA comparing high-risk versus low-risk TEX-signature groups. Bar plot displays differential enrichment scores for hallmark gene sets; bar height indicates normalized enrichment score, with positive values (red) indicating up-regulated and negative values (blue) indicating down-regulated in high-risk group.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eD. KEGG pathway enrichment analysis of five core genes (KLF3, LMNA, SLC2A3, ARL4C, TIMP1). Heatmap displays enrichment significance (-log10(FDR)) across pathways; color intensity indicates statistical significance.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eE. MSigDB HALLMARK gene set analysis of core gene functions. Heatmap of normalized enrichment scores (NES) across 50 hallmark gene sets. Rows: individual gene sets; columns: sample groups or core genes. Red: pathway activation; blue: pathway suppression; hierarchical clustering applied to both dimensions.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eClinical utility of TEX-signature for immunotherapy stratification and mechanistic validation\u003c/h2\u003e \u003cp\u003eTo deeply explore the potential value of TEX scores in immunotherapy, we verified their predictive efficacy in multiple published treatment cohorts. The GSE78002, IMvigor, and GSE126044 datasets all suggest that high TEX scores are often associated with no response to immunotherapy, and the ROC curve validates the efficacy of the predictive model (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003eA-\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003eC).\u003c/p\u003e \u003cp\u003eSubsequently, to verify the expression of TEX core genes in tumor tissues of CRC patients with different sensitivities to immunotherapy, we detected the gene expression of 6 immunotherapy-sensitive/resistant patients by qPCR and Western Blot. It could be seen that KLF3 was lowly expressed in tumor tissues of the immunotherapy-sensitive group. LMNA, SLC2A3, ARL4C, and TIMP1 were highly expressed in tumor tissues of the immunotherapy-sensitive group (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003eD, \u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003eE). Further immunofluorescence techniques suggested that the core gene was co-localized with CD8\u003csup\u003e+\u003c/sup\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003eF). Based on this, it is speculated that the abnormal expression of core genes may induce T cell exhaustion, reduce the efficacy of anti-tumor immunotherapy, and thereby promote the progression of CRC.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eA-C. Immunotherapy response prediction by TEX-signature across three independent cohorts. (A) GSE78002 Hela cell cohort, (B) IMvigor urothelial cancer cohort, and (C) GSE126044 non-small lung cancer cohort. Left panels: distribution of TEX-scores in responders versus non-responders; right panels: ROC curves evaluating predictive accuracy with AUC values.\u003c/p\u003e \u003cp\u003eD-E. Experimental validation of core gene expression in immunotherapy-sensitive versus resistant tumor tissues. (D) RT-qPCR analysis of KLF3, LMNA, SLC2A3, ARL4C, and TIMP1 mRNA levels normalized to GAPDH. (E) Western blot analysis of core protein expression with β-actin as loading control. Data are presented as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD; statistical significance determined by Student's t-test. *P\u0026thinsp;\u0026lt;\u0026thinsp;0.05, **P\u0026thinsp;\u0026lt;\u0026thinsp;0.01, ***P\u0026thinsp;\u0026lt;\u0026thinsp;0.001.\u003c/p\u003e \u003cp\u003eF. Spatial validation by multiplex immunofluorescence. Confocal microscopy images (40\u0026times; objective) demonstrating co-expression patterns of CD8⁺ T cell marker (green) with individual core genes (red) in formalin-fixed paraffin-embedded tumor sections. Nuclear counterstain with DAPI (blue). Insets show magnified regions of interest. Scale bar represents 20 micrometers.\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eCRC remains a formidable health challenge globally, ranking as the third most common cancer and the second leading cause of cancer-related deaths\u003csup\u003e[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]\u003c/sup\u003e. Although significant progress has been made in the treatment of CRC in recent years, chemotherapy resistance and toxic side effects remain important factors affecting the prognosis of patients with CRC\u003csup\u003e[\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]\u003c/sup\u003e. Despite immunotherapy has got great success in treating various cancers, the limited responsiveness of CRC to ICIs underscores the need for deeper understanding of them to broaden the population that can benefit from these treatments\u003csup\u003e[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eCD8\u003csup\u003e+\u003c/sup\u003e T cells, originating from CD34 hematopoietic stem cells located in the bone marrow, are the main effector cells of antitumor immunity therapy, and CD8\u003csup\u003e+\u003c/sup\u003e T cell exhaustion often leads to tumor deregulation and progression\u003csup\u003e[\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]\u003c/sup\u003e. Tumor-induced T-cell exhaustion may be more complicated due to the tumor microenvironment\u003csup\u003e[\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e]\u003c/sup\u003e. Recently, the proteome of exhausted CD8\u003csup\u003e+\u003c/sup\u003e T cell rose a mechanistic vulnerability and a new therapeutic target to improve cancer immunotherapies\u003csup\u003e[\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e]\u003c/sup\u003e, which reminded us further refine and scrutinize exhaustion models, seeking additional insights to advance immunotherapy for CRC.\u003c/p\u003e \u003cp\u003eThrough an exploration of the molecular and functional attributes of distinct CD8\u003csup\u003e+\u003c/sup\u003e T cell subgroups in CRC, we figure out the intercellular communication and functional pathway enrichment of different clusters. It helps us better understand the microenvironment of CRC and visualize the function of different T cell cluster. We realized that the cluster 2 demonstrated more pronounced advantages across most inflammatory, immune activation pathways, and the ligand interaction. Summing up, we propose a process of functional exhaustion in the differentiation of CD8\u003csup\u003e+\u003c/sup\u003e T cells between the two clusters.\u003c/p\u003e \u003cp\u003eSubsequently, with the help of machine\u0026ndash;learning and the identification of the core gene associated with exhaustion, five core genes, KLF3, LMNA, SLC2A3, ARL4C, and TIMP1 was figured out and the TEX-signature was built and verified with nomogram. Besides, the five core genes, all of which have been previously implicated in colorectal cancer prognosis\u003csup\u003e[\u003cspan additionalcitationids=\"CR40 CR41\" citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e]\u003c/sup\u003e, have not been systematically examined in relation to the tumor immune microenvironment. By defining TEX-Score as the weighted sum of their expression levels, we uncover a significant association with immune infiltration, thereby establishing a previously unrecognized link between these prognostic genes and antitumor immunity. Plenty of learning model have confirmed that the low TEX-signature symbolize the better survival, more immune infiltration, and stronger immunity cycle.\u003c/p\u003e \u003cp\u003eFinally, with the aid of GEO datasets and clinical cohort of immune therapy in Hela cells, non-small cell lung cancer, and urothelial carcinoma, we validate the TEX-signature model again. At the same time, the expression of core genes in CRC proves our conjectures. All in all, our research sets a promising foundation for future research and potential advancements in the treatment of CRC.\u003c/p\u003e \u003cp\u003eIt should be mentioned that there are a few of restrictions. Firstly, basing on single-cell sequencing data from a relatively small sample number, our research may not fully capture the heterogeneity, which limits efficacy of ICIs in CRC most importantly. Further validation in larger cohorts would provide more robust and generalizable results. Secondly, despite the advantages of large sample sizes and robust statistical power afforded by publicly available datasets, the integration of multi-source scRNA-seq and bulk RNA-seq data inevitably introduced technical heterogeneity attributable to disparities in sample processing protocols, platform-specific batch effects, and variable detection sensitivities. Additionally, the lack of comprehensive clinical annotations\u0026mdash;including ethnic background, MSI/MSS status, specific chemotherapeutic regimens, and objective immunotherapy response criteria (e.g., RECIST)\u0026mdash;in certain public cohorts constrained our capacity to delineate the precise clinical correlates of T-cell exhaustion states. To address these limitations, future studies will incorporate prospectively collected, clinically annotated biospecimens with standardized procurement protocols and harmonized analytical workflows, thereby ensuring robust cross-validation and clinical translatability of the TEX-score model. Thirdly, further molecular experiments are necessary to elucidate the functional roles of core genes and understand the underlying molecular mechanisms of TEX-signature. It is imperative to acknowledge that the majority of our findings were predicated upon correlative bioinformatic analyses, which, while robust for delineating transcriptional regulatory networks underlying T-cell exhaustion, inherently preclude definitive conclusions regarding protein-level expression dynamics and spatial heterogeneity within the colorectal tumor microenvironment. Specifically, our reliance on scRNA-seq and bulk RNA-seq datasets, integrated through computational deconvolution algorithms, may not fully resolve the intricate cellular crosstalk or the spatially restricted niches that govern T-cell dysfunction. This limitation is particularly pertinent given our observation that cluster 2 CD8\u003csup\u003e+\u003c/sup\u003e T cells exhibited enhanced intercellular communication and ligand-receptor interactions, which necessitate spatial contextualization. To circumvent these constraints, future investigations should incorporate orthogonal validation strategies, including multiparameter flow cytometry for quantitative assessment of exhaustion marker co-expression (e.g., PD-1, TIM-3, LAG-3), multiplex immunofluorescence (mIF) or imaging mass cytometry (IMC) for spatially resolved profiling of immune cell infiltration patterns, and RNAscope or in situ hybridization for direct visualization of core gene transcripts (KLF3, LMNA, SLC2A3, ARL4C, TIMP1) within tissue architecture. These approaches will be instrumental in bridging the gap between transcriptional signatures and functional phenotypes, thereby strengthening the mechanistic underpinnings of the TEX-score model. Finally, the clinical research with only six samples is too limited. The need for further validation and clinical implementation is fundamental for digging out the potential clinical utility of the TEX-signature in guiding treatment decisions.\u003c/p\u003e \u003cp\u003eIn conclusion, our research aims to establish a novel assessment system, distinct from MSS/MSI-H, for personalized evaluation of immunotherapy sensitivity in colorectal cancer patients, thereby providing more individualized immunotherapy regimens and shedding light on new research directions for colorectal cancer immunotherapy. In conclusion, our research aims to establish a novel assessment system, distinct from MSS/MSI-H, for personalized evaluation of immunotherapy sensitivity in colorectal cancer patients, thereby providing more individualized immunotherapy regimens and shedding light on new research directions for colorectal cancer immunotherapy.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eBy integrating single-cell and bulk RNA-seq data via multiple machine-learning frameworks, we developed a prognostic model that simultaneously predicts OS and T-cell infiltration in CRC. The model estimates both patient survival probability and immunotherapy response, offering a robust biomarker for therapeutic efficacy and informing precision targeting strategies in CRC.\u003c/p\u003e "},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eCRC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eColorectal cancer\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eTEX\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eT cell exhaustion\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003escRNA-seq\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003esingle cell RNA sequencing\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eKLF3\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eKruppel-like factor 3\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eLMNA\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eLamin A/C\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eSLC2A3\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eGlucose transporter member 3\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eARL4C\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eARF-like 4C\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eTIMP1\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eTissue inhibitor of matrix metalloproteinase 1\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eOS\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eOverall survival\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eDFS\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eDisease-free survival\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eHR\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eHazard ratios\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eTME\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eTumor microenvironment\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eAJCC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eAmerican Joint Committee on Cancer\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e \u003ch2\u003eEthics approval and consent to participate\u003c/h2\u003e \u003cp\u003eInformed consent was obtained from all participants, and the study was approved by the Ethics Committee of First Affiliated Hospital of Gannan Medical University (approval number: LL8C-2025333). This study adhered to the principles of the Declaration of Helsinki.\u003c/p\u003e \u003c/p\u003e\u003cp\u003e \u003ch2\u003eData and material availability\u003c/h2\u003e \u003cp\u003eAll data will be available on request.\u003c/p\u003e \u003c/p\u003e\u003cp\u003e \u003ch2\u003eConsent for publication\u003c/h2\u003e \u003cp\u003eNot applicable. This study utilized anonymized data from public repositories (TCGA, GEO) and de-identified clinical samples. No individual patient data or identifiable information are presented\u003c/p\u003e \u003c/p\u003e\u003cp\u003e \u003ch2\u003eCompeting interests\u003c/h2\u003e \u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eFunding\u003c/h2\u003e \u003cp\u003eThis work was supported by the Guangdong Association for Science and Technology Youth Science and Technology Talent Development Program (NO. SKXRC2025117), National Natural Science Foundation of China (NO. 82504048), the Bethune Charitable Foundation (NO. 803292), Guangzhou Basic Research Program City - University (Institute) - Enterprise Joint Funding Project (NO. SL2024A03J01364), Medical Scientific Research Foundation of Guangdong Province of China (NO. A2023398), Fundamental Research Funds for the Central Universities (NO. 11624305).\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eZhijing Zhang, Peng Ouyang, and Kai Cui contributed equally to this work. Zhijing Zhang conceived the study, designed and conducted the experiments, and wrote the initial draft of the manuscript. Peng Ouyang and Kai Cui performed data analysis, interpreted the results, and contributed to manuscript writing. Xin Deng assisted with data interpretation and manuscript revision and obtained funding. Yixiang Wen and Wanyu Chen assisted with experimental work and data collection. Zhenhong Xian assisted with specimen and data collection. Qi Qi, Zhen Bao, Jin Gong and Xiao He share corresponding authorship. They supervised the project, provided critical feedback, obtained funding and were responsible for the final approval of the manuscript. All authors agree to be accountable for all aspects of work ensuring integrity and accuracy.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eWe extend our sincere gratitude to all researchers who have generously shared their datasets in public repositories, enabling this work. We also thank our colleagues and collaborators whose contributions were instrumental to the research cited herein. Finally, we are deeply indebted to the patients and their families who agreed to participate in this study.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eAll data will be available on request.Publicly available datasets analyzed during this study can be found in the following repositories. [https://portal.gdc.cancer.gov/](https:/portal.gdc.cancer.gov) (TCGA Colorectal Adenocarcinoma dataset); GEO: GSE39582 (bulk RNA-seq), GSE132465 (scRNA-seq), GSE78002 and GSE126044 (immunotherapy cohorts), and IMvigor210 (urothelial cancer immunotherapy cohort). The clinical experimental data and analysis scripts used during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eBray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. Cancer J Clin. 2024;74:229\u0026ndash;63.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLai JQ, Zhao LL, Hong C, Zou QM, Su JX, Li SJ, et al. Baicalein triggers ferroptosis in colorectal cancer cells via blocking the JAK2/STAT3/GPX4 axis. Acta Pharmacol Sin. 2024;45:1715\u0026ndash;26.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCa\u0026ntilde;ellas-Socias A, Cortina C, Hernando-Momblona X, Palomo-Ponce S, Mulholland EJ, Turon G, et al. Metastatic recurrence in colorectal cancer arises from residual EMP1(+) cells. Nature. 2022;611:603\u0026ndash;13.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSu JX, Li SJ, Zhou XF, Zhang ZJ, Yan Y, Liu SL, et al. Chemotherapy-induced metastasis: molecular mechanisms and clinical therapies. Acta Pharmacol Sin. 2023;44:1725\u0026ndash;36.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDekker E, Tanis PJ, Vleugels JLA, Kasi PM, Wallace MB. Colorectal cancer. Lancet (London England). 2019;394:1467\u0026ndash;80.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePage DB, Postow MA, Callahan MK, Allison JP, Wolchok JD. Immune modulation in cancer with antibodies. Annu Rev Med. 2014;65:185\u0026ndash;202.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eArnold M, Sierra MS, Laversanne M, Soerjomataram I, Jemal A, Bray F. Global patterns and trends in colorectal cancer incidence and mortality. Gut. 2017;66:683\u0026ndash;91.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen EX, Jonker DJ, Loree JM, Kennecke HF, Berry SR, Couture F, et al. Effect of Combined Immune Checkpoint Inhibition vs Best Supportive Care Alone in Patients With Advanced Colorectal Cancer: The Canadian Cancer Trials Group CO.26 Study. JAMA Oncol. 2020;6:831\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu Y, Liu Z, Hu Y, Ling Y, Qing S, Liu Y, et al. Caffeine enhances antitumor T-cell activity by suppressing kynurenine pathway in colorectal cancer. Nat Commun. 2025;16:5906.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSun L, Su Y, Jiao A, Wang X, Zhang B. T cells in health and disease. Signal Transduct Target therapy. 2023;8:235.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWherry EJ. T cell exhaustion. Nat Immunol. 2011;12:492\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePauken KE, Sammons MA, Odorizzi PM, Manne S, Godec J, Khan O, et al. Epigenetic stability of exhausted T cells limits durability of reinvigoration by PD-1 blockade. Volume 354. New York, N.Y.): Science; 2016. pp. 1160\u0026ndash;5.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcLane LM, Abdel-Hakeem MS, Wherry EJ. CD8 T Cell Exhaustion During Chronic Viral Infection and Cancer. Annu Rev Immunol. 2019;37:457\u0026ndash;95.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang H, Zhou Y, Tu Y, You J, Gao B, Yuan K, et al. Targeting hepatocytic TβRI ameliorates liver metastatic outcomes by revitalizing stem-like CD8(+) Tex subsets. Nat Commun. 2025;16:10627.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGe J, Ge J, Tang G, Xiong D, Zhu D, Ding X, et al. Machine learning-based identification of biomarkers and drugs in immunologically cold and hot pancreatic adenocarcinomas. J translational Med. 2024;22:775.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eQi X, Wang S, Fang C, Jia J, Lin L, Yuan T. Machine learning and SHAP value interpretation for predicting comorbidity of cardiovascular disease and cancer with dietary antioxidants. Redox Biol. 2025;79:103470.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZuo D, Yang L, Jin Y, Qi H, Liu Y, Ren L. Machine learning-based models for the prediction of breast cancer recurrence risk. BMC Med Inf Decis Mak. 2023;23:276.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang S, Cai N, Pacheco PP, Narrandes S, Wang Y, Xu W. Applications of Support Vector Machine (SVM) Learning in Cancer Genomics. Cancer Genomics Proteomics. 2018;15:41\u0026ndash;51.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eQi W, Wang Y, Wang Y, Huang S, Li C, Jin H, et al. Prediction of postpartum depression in women: development and validation of multiple machine learning models. J translational Med. 2025;23:291.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi B, Kugeratski FG, Kalluri R. A novel machine learning algorithm selects proteome signature to specifically identify cancer exosomes. eLife 2024; 12.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang X, Qiu H, Wang L, Wang X. Predicting Colorectal Cancer Survival Using Time-to-Event Machine Learning: Retrospective Cohort Study. J Med Internet Res. 2023;25:e44417.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVaziri-Moghadam A, Foroughmand-Araabi MH. Integrating machine learning and bioinformatics approaches for identifying novel diagnostic gene biomarkers in colorectal cancer. Sci Rep. 2024;14:24786.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYuan B, Yang D, Rothberg BEG, Chang H, Xu T. Unsupervised and supervised learning with neural network for human transcriptome analysis and cancer diagnosis. Sci Rep. 2020;10:19106.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWu J, Zhou X, Ren J, Zhang Z, Ju H, Diao X, et al. Glycosyltransferase-related prognostic and diagnostic biomarkers of uterine corpus endometrial carcinoma. Comput Biol Med. 2023;163:107164.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFan Q, Wang Y, Cheng J, Pan B, Zang X, Liu R, et al. Single-cell RNA-seq reveals T cell exhaustion and immune response landscape in osteosarcoma. Front Immunol. 2024;15:1362970.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShe H, Tan L, Wang Y, Du Y, Zhou Y, Zhang J, et al. Integrative single-cell RNA sequencing and metabolomics decipher the imbalanced lipid-metabolism in maladaptive immune responses during sepsis. Front Immunol. 2023;14:1181697.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYu T, Fang Z, Cheng Y, Zhou Y, Ji Y, Liu C. A novel palmitoylation-based molecular signature reveals COX6A1 as a key regulator in metabolic dysfunction-associated steatotic liver disease. J translational Med. 2025;23:1212.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShu J, Jiang J, Zhao G. Identification of novel gene signature for lung adenocarcinoma by machine learning to predict immunotherapy and prognosis. Front Immunol. 2023;14:1177847.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFan D, Yang M, Lee HJ, Lee JH, Kim HS. AVEN: a novel oncogenic biomarker with prognostic significance and implications of AVEN-associated immunophenotypes in lung adenocarcinoma. Front Mol Biosci. 2023;10:1265359.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFu S, Cheng Y, Wang X, Huang J, Su S, Wu H, et al. Identification of diagnostic gene biomarkers and immune infiltration in patients with diabetic kidney disease using machine learning strategies and bioinformatic analysis. Front Med. 2022;9:918657.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhu W, Zeng H, Huang J, Wu J, Wang Y, Wang Z, et al. Integrated machine learning identifies epithelial cell marker genes for improving outcomes and immunotherapy in prostate cancer. J translational Med. 2023;21:782.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePowles T, Bellmunt J, Comperat E, De Santis M, Huddart R, Loriot Y, et al. Bladder cancer: ESMO Clinical Practice Guideline for diagnosis, treatment and follow-up. Annals oncology: official J Eur Soc Med Oncol. 2022;33:244\u0026ndash;58.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang L, Wei S, Zhang J, Hu Q, Hu W, Cao M, et al. Construction of a predictive model for immunotherapy efficacy in lung squamous cell carcinoma based on the degree of tumor-infiltrating immune cells and molecular typing. J translational Med. 2022;20:364.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBaty F, Joerger M, Fr\u0026uuml;h M, Klingbiel D, Zappa F, Brutsche M. 24h-gene variation effect of combined bevacizumab/erlotinib in advanced non-squamous non-small cell lung cancer using exon array blood profiling. J translational Med. 2017;15:66.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYu T, Guo F, Yu Y, Sun T, Ma D, Han J et al. Fusobacterium nucleatum Promotes Chemoresistance to Colorectal Cancer by Modulating Autophagy. Cell 2017; 170: 548\u0026thinsp;\u0026ndash;\u0026thinsp;63.e16.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang Q, Qin Y, Li B. CD8(+) T cell exhaustion and cancer immunotherapy. Cancer Lett. 2023;559:216043.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcKinney EF, Smith KGC. Metabolic exhaustion in infection, cancer and autoimmunity. Nat Immunol. 2018;19:213\u0026ndash;21.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang Y, Ma A, Song NJ, Shannon AE, Amankwah YS, Chen X, et al. Proteotoxic stress response drives T cell exhaustion and immune evasion. Nature. 2025;647:1025\u0026ndash;35.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShen W, Yuan L, Hao B, Xiang J, Cheng F, Wu Z, et al. KLF3 promotes colorectal cancer growth by activating WNT1. Aging. 2024;16:2475\u0026ndash;93.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDai W, Xu Y, Mo S, Li Q, Yu J, Wang R, et al. GLUT3 induced by AMPK/CREB1 axis is key for withstanding energy stress and augments the efficacy of current colorectal cancer therapies. Signal Transduct Target therapy. 2020;5:177.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang Y, Chang Z, Sun Z, Wang L, Li J, Li G, et al. Targeting the ARL4C/RAP1/PI3K-Akt-mTOR signaling loop promotes ARL4C ubiquitination and reverses oxaliplatin resistance in colorectal cancer. Theranostics. 2026;16:99\u0026ndash;122.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBen-Ami R, Wang QL, Zhang J, Supplee JG, Fahrmann JF, Lehmann-Werman R, et al. Protein biomarkers and alternatively methylated cell-free DNA detect early stage pancreatic cancer. Gut. 2024;73:639\u0026ndash;48.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-cancer","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bcan","sideBox":"Learn more about [BMC Cancer](http://bmccancer.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bcan/default.aspx","title":"BMC Cancer","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"colorectal cancer, immunotherapy sensitivity, T cell exhaustion (TEX), tumor microenvironment (TME), machine learning, predictable model","lastPublishedDoi":"10.21203/rs.3.rs-8910521/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8910521/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eImmunotherapy for colorectal cancer (CRC) currently faces significant dilemmas, but its specific mechanisms remain unclear. T-cell exhaustion (TEX) in the tumor microenvironment has been identified as a pivotal driver of immune evasion and tumor progression. Dissecting its contribution to CRC is essential for the development of rational therapeutic strategies. Here, we integrated scRNA-seq and bulk RNA-seq databases and leveraged pseudotemporal trajectory model to identify core genes. Subsequently, with the help of 10 machine learning models, we constructed a TEX score prognostic model, whose clinical utility was externally validated in independent immunotherapy cohorts, demonstrating intra-tumoral CD8⁺ T cells occupy a continuum of exhaustion states. Besides, the TEX-score model, constructed from five exhaustion-related genes (KLF3, LMNA, SLC2A3, ARL4C, and TIMP1), stably predicted CRC prognosis and immunotherapy responsiveness, validating that patients with low TEX-score exhibited prolonged overall survival (OS), abundant immune infiltrates and better response to immunotherapy. Collectively, our findings elucidate T-cell exhaustion as a central mediator of immunotherapy failure in CRC and provide a clinically actionable guidance for patient stratification and treatment selection.\u003c/p\u003e","manuscriptTitle":"Dissecting T-cell exhaustion heterogeneity and immune ecosystem dynamics in colorectal cancer through multi-omics machine learning","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-27 07:36:15","doi":"10.21203/rs.3.rs-8910521/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-03-23T11:35:36+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-03-18T21:05:36+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-03-09T03:42:12+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-03-08T04:43:00+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"24666438493212308712138219435160666030","date":"2026-02-28T18:58:09+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"317936110155125936993061657142514563314","date":"2026-02-25T16:24:36+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"3093385494823924036837154922241201326","date":"2026-02-23T22:07:07+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-02-23T16:14:52+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-02-20T13:51:44+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-02-20T08:13:17+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-02-20T04:23:03+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Cancer","date":"2026-02-20T04:16:41+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-cancer","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bcan","sideBox":"Learn more about [BMC Cancer](http://bmccancer.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bcan/default.aspx","title":"BMC Cancer","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"81f1cecb-a14e-42d6-ad66-6018d686f970","owner":[],"postedDate":"February 27th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"in-revision","subjectAreas":[],"tags":[],"updatedAt":"2026-05-14T15:09:50+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-27 07:36:15","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8910521","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8910521","identity":"rs-8910521","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}