MC38 cells, a murine colon adenocarcinoma cell line, were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA) and cultured in accordance with ATCC-recommended protocols.

For in vitro T cell activation, 48-well culture plates were coated with anti-mouse CD3 monoclonal antibody (5 μg/mL; BioLegend, #100201) and incubated overnight at 4°C (Day 1). On Day 2, spleens were harvested from naïve C57BL/6 mice, and single-cell suspensions were prepared via mechanical dissociation, followed by red blood cell (RBC) lysis using RBC lysis buffer. CD8⁺ T cells were subsequently isolated using magnetic CD8a⁺ T cell isolation beads (Miltenyi Biotec, #130-104-075) according to the manufacturer’s instructions. Purified CD8⁺ T cells were resuspended at a density of 5 × 10⁵ cells/mL in complete RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS), 1% L-glutamine, 1% Penicillin–Streptomycin, 1% non-essential amino acids (NEAA), 1% HEPES buffer, 1% 2-Mercaptoethanol, and an additional 1% GlutaMAX. Anti-mouse CD28 monoclonal antibody (2 μg/mL; BioLegend, #102102) was added to provide co-stimulatory signaling.

On Day 3, recombinant mouse IL-2 (10 ng/mL; Thermo Fisher, #212-12-100UG) and IL-7 (10 ng/mL; Thermo Fisher, #217-17-100UG) were added to further support T cell activation and expansion.

On Day 4, MC38 tumor cells were added to the activated T cells at a 1:10 ratio (tumor cell: T cell), and co-culture was maintained for 48 hours under standard culture conditions (37°C, 5% CO₂). After incubation, cells were harvested for flow cytometric analysis to evaluate the expression of immune checkpoint molecules, including TIGIT and PD-1, on CD8⁺ T cells.

Following the 48-hour co-culture of the in vitro activation T cells and MC38, cells were harvested and washed twice with phosphate-buffered saline (PBS). Surface staining was performed in FACS buffer (PBS supplemented with 2% fetal bovine serum) containing the viability dye Propidium Iodide (PI; Thermo Fisher Scientific, #R37169) and the following fluorochrome-conjugated monoclonal antibodies: anti-mouse CD8-BV421(Biolegend#100737), anti-mouse TIGIT–PE (BioLegend, #142101) and anti-mouse PD-1–APC (BioLegend, #135209). In addition, levels of granzyme B (Biolegend, #396409), and tumor necrosis factor-alpha (TNF-α) (BioLegend, #506315) were measured to evaluate the cytotoxic effects of in vitro T cell stimulation, both in the presence and absence of tumor cell co-culture. Surface staining was carried out for 30 minutes on ice in the dark. After washing, Intracellular staining was carried out according to the manufacturer's protocol (Invitrogen, #00-5523-00) cells were resuspended in a FACS buffer and analyzed using a BD LSRFortessa™ flow cytometer (BD Biosciences). Data acquisition was followed by analysis using FlowJo software (version 10.8.1, BD Biosciences). The isotype controls for TIGIT-PE and PD-1-APC are PE Mouse IgG1, κ Isotype Control (BioLegend #981804) and APC Rat IgG2a, κ Isotype Control Antibody (BioLegend #400511), respectively.

Publicly available Sc sequencing datasets from colorectal cancer patients and (GSE1467711) were obtained from the Gene Expression Omnibus (GEO) database. Data preprocessing, including quality control, normalization, and scaling, was performed using the Seurat package (version 4.0) in R. Principal component analysis (PCA) and Uniform Manifold Approximation and Projection (UMAP) were applied for dimensionality reduction and clustering. The expression patterns and co-localization of TIGIT, PDCD1, and CXCL13 within CD8⁺ T cell subsets were visualized using feature and box plots in the TISCH 2.0 (Tumor Immune Single-cell Hub, http://tisch.comp-genomics.org) database. Additionally, the GSE178341 from Single cell portal (Tumor Immune Single-cell Hub, https://singlecell.broadinstitute.org/single_cell) database was used to validate co-expression patterns in multiple CRC scRNA-seq datasets.

To evaluate tumor immune microenvironment composition, bulk RNA-seq data from CRC patient cohorts were analyzed using CIBERSORT-ABS and QUANTISEQ algorithms via the TIMER2.0 platform. CIBERSORT-ABS estimates absolute immune cell fractions, while QUANTISEQ offers robust profiling of immune cell types in solid tumors. Correlation analysis between TIGIT expression and the abundance of tumor-infiltrating immune cells—particularly CD8⁺ T cells—was performed using Pearson correlation coefficients. Results were visualized via dot plots. Receiver operating characteristic (ROC) analysis was performed using the pROC package in R to assess the prognostic value of TIGIT expression.

Gene co-expression analysis was conducted using both the Single Cell Portal and TISCH2.0 database to explore genes significantly correlated with TIGIT in CRC-derived CD8⁺ T cell subsets. Co-expression networks were constructed focusing on genes involved in T cell exhaustion, chemokine signaling, and immune metabolism. Specific emphasis was placed on the identification of functional gene modules co-expressed with TIGIT, including PDCD1 and CXCL13, across healthy donors and CRC patients. Bubble plots representing gene expression differences between conditions were obtained from TISCH 2.0.

The relationship between TIGIT expression and immune cell infiltration was further validated using the Tumor Immune Estimation Resource (TIMER) and GEPIA2 (http://gepia2.cancer-pku.cn) platforms. The correlation between TIGIT and gene signatures indicative of CD8⁺ T, CD4+ T and nature killer (NK) cell functional states—such as effector, memory, and exhausted phenotypes—was evaluated. Log₂-transformed TPM (transcripts per million) values were used for statistical correlation analyses.

Intercellular communication networks within the CRC tumor microenvironment were reconstructed using the TISCH2.0 database and the CellChat R package. Ligand–receptor interaction analyses focused on the inhibitory axes involving TIGIT–PVR(CD155) and TIGIT–NECTIN-2 (CD112). Heatmaps, circle plots, and bubble charts were generated to illustrate cell-type-specific interaction strengths. Special emphasis was placed on the interaction between exhausted CD8⁺ T cells and tumor/stromal cells, and the relative contribution of each population to inhibitory signaling pathways was quantitatively inferred.

All in vitro experiments were performed in biological triplicates unless otherwise indicated. Data are presented as mean ± standard error of the mean (SEM). Two-group comparisons were conducted using unpaired two-tailed Student’s t-tests. P-values < 0.05 were considered statistically significant (*p < 0.05; **p < 0.01; ***p < 0.001). Statistical analyses were carried out using GraphPad Prism (version 9.0, GraphPad Software, CA, USA).

TIGIT was significantly upregulated in tumor-infiltrating immune cells in CRC, with predominant expression observed in exhausted CD8⁺ T cells (CD8⁺ Tex) (19). scRNA-seq data from CRC patients (GSE146771) revealed a strong co-localization of TIGIT and PD-1 (PDCD1) within the CD8⁺ Tex subset ( Figures 1A–C ), indicative of a terminally exhausted phenotype (20). Analysis of the overall immune landscape in CRC samples revealed that the predominant immune cell types included CD4⁺ conventional T cells, Tregs, CD8⁺ T cells, CD8⁺ Tex cells, monocytes/macrophages, and natural killer (NK) cells ( Supplementary Figures S1A, B ). Violin plots further demonstrated that TIGIT expression was largely restricted to immune cell populations, with the highest levels detected in regulatory T cells (Tregs) and CD8⁺ Tex cells ( Supplementary Figures S1C, D ) (21, 22).

TIGIT is known to exert its immunosuppressive effects via binding to its ligands, PVR (CD155) and NECTIN-2 (CD112), which are commonly expressed on antigen-presenting cells and tumor cells, thereby attenuating cytotoxic T cell responses (16).Cell–cell interaction (CCI) analysis revealed extensive crosstalk between CD8⁺ Tex cells and other immune populations, including dendritic cells, neutrophils, and Tregs ( Figures 1D, E ). Notably, strong interaction scores were observed between CD8⁺ Tex cells and Tregs via the TIGIT–PVR axis, suggesting that this pathway plays a pivotal role in reinforcing immunosuppressive signaling within the tumor microenvironment. These findings highlight the immunoregulatory centrality of CD8⁺ Tex cells in shaping the TIME through TIGIT-mediated inhibitory interactions (5, 18, 23).

To further explore the functional impact of TIGIT expression, pathway analysis of CD8+ Tex cells demonstrated downregulation of cytotoxic cytokine production, mRNA processing, and antigen receptor signaling, alongside altered oxidative phosphorylation (OXPHOS) ( Figure 1F ) (24, 25). Interestingly, while many functional pathways were suppressed, OXPHOS-related genes were paradoxically upregulated, supporting the notion that exhausted T cells attempt to maintain mitochondrial metabolism despite diminished effector function ( Figure 1F ) (26). Further CCI network visualization demonstrated substantial overlap and connectivity among TIGIT⁺ CD8⁺ Tex cells and other major immune populations in CRC, including Tregs and NK cells, through shared inhibitory circuits ( Figure 1E ). Additionally, analysis of the source–target relationship between CD8⁺ exhausted cells (source cluster) and CD4⁺ regulatory cells (target cluster) reveals a prominent TIGIT–PVR interaction. This source–target mapping highlights exhausted CD8⁺ T cells as major initiators of cellular communication, with the TIGIT–PVR axis playing a crucial role in their interaction ( Figure 1G ).

Given TIGIT’s role in T cell inhibition, we next investigated the cellular sources of its ligands. Using ScRNA seq data (GSE146771) of CRC pre-clinical model, we found that NECTIN-2 and PVR, the two primary ligands of TIGIT, are predominantly expressed by stromal cells, especially endothelial and fibroblast populations (16, 20, 23), particularly endothelial and fibroblast cells ( Figures 2A, B ). Malignant epithelial cells also contributed to ligand expression, albeit to a lesser extent ( Figures 2A, B ). Stromal cells enhance immunosuppression through multiple mechanisms, including physical barrier formation, immune cell reprogramming, and the secretion of immunosuppressive cytokines (27). These functions highlight stromal components as potential therapeutic targets for future research.

Importantly, the expression pattern of TIGIT ligands suggests that the immunosuppressive TME is actively maintained by non-immune stromal components, not just by cancer cells (16, 18). Stromal cells such as fibroblasts and endothelial cells are often spatially proximal to infiltrating T cells and may thus serve as persistent, non-mutated sources of inhibitory ligand signals, contributing to chronic TIGIT engagement and sustained T cell exhaustion (28, 29).

Supporting this, CCI analysis revealed a recurring inhibitory loop in which TIGIT+ CD8+ Tex interact with fibroblasts, endothelial cells, and malignant epithelial cells via NECTIN-2/PVR signaling ( Figure 2C ). These findings highlight a multi-cellular suppressive network that is not confined to tumor–immune cell crosstalk but involves broader tissue architecture within the TME. This CCI cycle likely facilitates the maintenance of T cell dysfunction and represents a barrier to effective anti-tumor immunity.

Analysis of RNA-sequencing data revealed that TIGIT mRNA expression was significantly elevated in tumor tissues compared to matched adjacent normal tissues in various of tumors ( Supplementary Figure S2A ) (30). In addition, genomic profiling demonstrated that TIGIT alterations, predominantly in the form of mRNA upregulation and copy number gains, were common among CRC patients, and these alterations were associated with higher TIGIT expression levels ( Figure 2D ) (30). Although TIGIT expression was not significantly different between normal and tumor sites in microsatellite instability (MSI)-low CRC patients, a relatively lower expression was still observed in normal tissues ( Supplementary Figure S2B ).

Survival analysis indicated that high TIGIT expression was significantly correlated with reduced overall survival (OS). CRC patients with high TIGIT expression exhibited markedly poorer OS compared to those with low TIGIT expression ( Figure 2E ).

Interestingly, TIGIT expression appeared to be independent of tumor stage, as no significant differences were observed between early- and late-stage tumors ( Figure 2F ). This suggests that TIGIT upregulation is an intrinsic immunoregulatory feature of the tumor microenvironment, rather than a reflection of tumor burden or progression stage. While TIGIT expression does not vary significantly across tumor stages, its consistent upregulation in the tumor microenvironment suggests its utility as a predictive rather than prognostic biomarker (29). TIGIT⁺ exhausted T cells may indicate a dysfunctional immune contexture that correlates with response to immunotherapy. Further studies are warranted to validate TIGIT as part of a combinatorial marker panel to guide personalized treatment decisions (23).

Receiver operating characteristic (ROC) curve analysis, performed using the R statistical package (31), yielded an area under the curve (AUC) value of 0.61, indicating a moderate predictive potential of TIGIT expression for clinical prognosis in CRC ( Supplementary Figure S2C ) (32). Generally, AUC of 0.5 indicates no discriminatory power, whereas an AUC of 1.0 reflects perfect prediction; thus, an AUC of 0.61 denotes modest discriminatory ability (32). We examined the expression of the TIGIT ligands, PVR and NECTIN-2 in CRC patients. The results demonstrated that there were no significant differences in the expression levels of PVR and NECTIN-2 among CRC patients ( Supplementary Figure S2D ). These findings suggest that while TIGIT expression shows modest prognostic value in CRC, its ligands PVR and NECTIN-2 are not differentially expressed among CRC patients.

Collectively, these findings support the notion that TIGIT expression serves as a negative prognostic biomarker in CRC. Elevated TIGIT expression may contribute to an immunosuppressive tumor microenvironment by interacting with its ligands (e.g., PVR, NECTIN-2) expressed by stromal and malignant cells, thereby promoting T cell exhaustion and enabling tumor immune evasion.

To elucidate the broader CCI network of TIGIT⁺ CD8⁺ Tex cells, we performed CCI analysis using the Smart-seq2 dataset (GSE146771) (20). This analysis identified TIGIT–NECTIN-2 and TIGIT–PVR as dominant inhibitory ligand–receptor axes connecting CD8⁺ Tex cells with both stromal and malignant compartments within the CRC TME (23, 25). The TME comprises a heterogeneous mix of immune cell types, including NK cells, Tregs, CD4⁺ T cells, and CD8⁺ Tex cells ( Supplementary Figures S1A–D ).

Bubble plots and pie charts revealed that fibroblasts, endothelial cells, and malignant epithelial cells were the primary stromal interactors of CD8⁺ Tex cells, with strong engagement via TIGIT–NECTIN-2/PVR signaling ( Supplementary Figures S1A–D ; Figures 3A, B ). These findings underscore the coordinated inhibitory interactions between exhausted T cells and non-immune cellular components of the TME, reinforcing the role of these ligand–receptor pathways in facilitating immune suppression and T cell exhaustion (29).

Notably, fibroblasts and endothelial cells emerged as key mediators of T cell dysfunction, predominantly signaling through TIGIT–NECTIN-2/PVR axes ( Figures 3A, B ). These interactions highlight the contribution of both stromal and malignant cells to immune evasion mechanisms, supporting the hypothesis that TIGIT engagement plays a central role in shaping the exhausted T cell phenotype in the CRC microenvironment ( Figures 3A, B ; Supplementary Figures S5A, B ).

To validate the immune cell populations associated with TIGIT expression in CRC, we performed immune deconvolution using two complementary algorithms—CIBERSORT-ABS and QUANTISEQ—on publicly available bulk RNA-seq datasets Timer 2.0 (33). While CIBERSORT-ABS estimates absolute immune cell fractions, QUANTISEQ is specifically optimized for solid tumors with heterogeneous cell populations, thereby providing robust and complementary insights into the tumor immune microenvironment (33, 34).

Across both analytical approaches, TIGIT expression showed the strongest positive correlation with CD8⁺ T cell abundance, exceeding correlations observed with other immune cell subsets in the dominant immune population ( Figures 4A–D ). This association supports the hypothesis that TIGIT⁺ CD8⁺ T cells represent a prominent immune phenotype in CRC and reinforces their potential role in mediating T cell exhaustion within the tumor microenvironment. The correlation coefficient (Rho) between TIGIT and CD8⁺ T cells was significantly higher than that of other immune subsets, indicating a preferential enrichment pattern (35).

It is important to acknowledge that the accuracy and resolution of deconvolution algorithms are contingent upon the quality of reference gene signatures, which may vary between datasets and may not fully distinguish exhausted T cell phenotypes. Nevertheless, the consistent results observed across two independent computational frameworks enhance the reliability of our findings and further substantiate the central involvement of TIGIT⁺ CD8⁺ T cells in CRC pathophysiology.

To further dissect the immunoregulatory role of TIGIT in the tumor microenvironment, we analyzed scRNA-seq data from CRC patients (GSE178341), analysis revealed that TIGIT expression was predominantly localized to TNKILCC cells, a T cell population enriched in exhausted and cytotoxic features ( Figure 5A ) (36).

Since the expression of TIGIT and PDCD1 on the CD8 T cells are widely used to identify dysfunction or early activated effector T cells in several cancer types such as bladder cancer, these markers alone are insufficient to define terminally exhausted CD8⁺ T cell in CRC (37). Their expression encompasses a heterogeneous population that includes partially exhausted, activated and potentially bystander T cells, limiting their utility as precise prognostic or immunosuppressive markers in CRC TIME (38). CXCL13 has emerged as the critical marker that more specifically delineates the effector-exhausted subset within the PD-1⁺TIGIT⁺CD8 T cell population (39). These CXCL13⁺CD8⁺PD-1⁺TIGIT⁺ T cells, referred to as effector-like CD8⁺ Tex cells, are characterized by robust cytotoxicity, secretion of inflammatory cytokines and chemokines, and strong association with antitumor immune responses, suggesting their involvement in effective tumor control cells (40, 41). In contrast, the CXCL13^-CD8⁺PD-1⁺TIGIT⁺ T cells exhibit features indicative of terminal exhaustion phenotypes, including diminished functional capacity and reduced responsiveness to tumor antigens, aligning more closely with terminally exhausted CD8⁺ Tex cells.

Therefore, accessing CXCL13 expression within PD-1⁺TIGIT⁺CD8 ⁺ T cells provide a more precise stratification of T cell exhaustion states in the CRC TIME and may serve as a valuable prognostic biomarker. In the ScRNA-seq dataset (GSE1665555), CXCL13 expression was highly enriched in subclusters co-expression TIGIT and PDCD1, particularly in patients with MSI CRC ( Supplementary Figures S4A, B ) (42). Compared to microsatellite stable (MSS) CRC patients, MSI patients(Patient 25) exhibited higher co-expression levels of CXCL13 with TIGIT and PDCD1 in CD8⁺ T cells (14, 43)( Supplementary Figures S4A, B ). This co-expression pattern suggests a coordinated transcriptional program driving T cell exhaustion in the CRC microenvironment indicate a potential responsiveness to ICB and predict favorable antitumor outcomes.

Upon reclustering this population using refined marker sets, we identified distinct CD8⁺ T cell subsets. Notably, TIGIT, PDCD1, and CXCL13 were co-expressed in the same subclusters, corresponding to CD8⁺ Tex ( Figures 5A, B ). This co-localization suggests a coordinated transcriptional program that defines a dysfunctional T cell phenotype in the CRC microenvironment (44).

Moreover, in the healthy donor samples, it showed markedly lower expression of TIGIT, PDCD1, and CXCL13, indicating that this specific co-expression pattern is tumor-specific ( Figure 5C ). These findings were further supported by t-SNE projections, which revealed overlapping expression domains of TIGIT, PDCD1, and CXCL13 in CRC patients but not in healthy controls ( Supplementary Figures S3A, B ). Boxplot analysis of expression across different cell clusters confirmed that these three genes share similar expression patterns within CRC-infiltrating CD8⁺ T cells ( Supplementary Figures S3C–E ).

This coordinated signaling network, enriched in exhausted CD8⁺ T cells and sustained by signals from both tumor and stromal compartments, underscores a multicellular inhibitory circuit that perpetuates T cell dysfunction (36). These observations support the notion that TIGIT⁺ CD8⁺ T cells function within a suppressive communication loop, positioning TIGIT as a central node in CRC immune evasion (45).

Furthermore, correlation analysis across CRC patient samples demonstrated that TIGIT expression positively correlates with both CXCL13 and PDCD1, reinforcing the existence of a shared regulatory axis in T cell exhaustion ( Figures 5D, E ). In contrast, healthy donors exhibited minimal expression of these genes, further emphasizing their pathological relevance in the tumor context.

To experimentally validate the transcriptomic observations, murine CD8⁺ T cells were isolated from the spleens of naive mice, activated in vitro, and subsequently co-cultured with MC38 colon adenocarcinoma cells (46). Correlation analyses (Pearson and Spearman) demonstrated a strong association between CXCL13 and TIGIT expression in CD8⁺ T cells ( Figure 6A ). Flow cytometric analysis further revealed a significant upregulation of TIGIT and PD-1 in CD8⁺ T cells after co-culture, compared to activated CD8⁺ T cells cultured alone ( Figures 6B–F ). To confirm the exhausted status of CD8⁺ T cells following co-culture with MC38 tumor cells, we also assessed the expression of cytotoxic molecules and cytokines, including granzyme B, and tumor necrosis factor-alpha (TNF-α) ( Figure 6F ). The results showed a significant decrease in granzyme B and TNF-α levels after co-culture, indicating reduced cytotoxic activity.

Notably, the proportion of TIGIT⁺PD-1⁺ double-positive CD8⁺ T cells was significantly elevated under co-culture conditions, indicating the acquisition of an exhaustion-like phenotype upon direct interaction with tumor cells (25). These results strongly support the hypothesis that tumor-derived factors contribute to the induction of immune checkpoint expression, thereby promoting CD8⁺ T cell dysfunction in the colorectal cancer microenvironment.

This in vitro evidence aligns with our computational findings and underscores the functional relevance of tumor–T cell interactions in shaping T cell exhaustion. Together, these data highlight a potential mechanistic basis for the observed enrichment of TIGIT⁺PD-1⁺ CD8⁺ T cells in CRC patients and provide a rationale for targeting these pathways therapeutically.