Tumor-Specific T Cell Dysfunction Is a Dynamic Antigen-Driven Differentiation Program Initiated Early during Tumorigenesis

Abstract

CD8(+) T cells recognizing tumor-specific antigens are detected in cancer patients but are dysfunctional. Here we developed a tamoxifen-inducible liver cancer mouse model with a defined oncogenic driver antigen (SV40 large T-antigen) to follow the activation and differentiation of naive tumor-specific CD8(+) T (TST) cells after tumor initiation. Early during the pre-malignant phase of tumorigenesis, TST cells became dysfunctional, exhibiting phenotypic, functional, and transcriptional features similar to dysfunctional T cells isolated from late-stage human tumors. Thus, T cell dysfunction seen in advanced human cancers may already be established early during tumorigenesis. Although the TST cell dysfunctional state was initially therapeutically reversible, it ultimately evolved into a fixed state. Persistent antigen exposure rather than factors associated with the tumor microenvironment drove dysfunction. Moreover, the TST cell differentiation and dysfunction program exhibited features distinct from T cell exhaustion in chronic infections. Strategies to overcome this antigen-driven, cell-intrinsic dysfunction may be required to improve cancer immunotherapy.

Figure 1. Rapid Induction of Dysfunction in Naive Tumor-Specific CD8⁺ T Cells Encountering an Oncogenic Driver Neoantigen in Pre-malignant Lesions

(A) Tamoxifen (Tam)-inducible ASTxCre-ER^T2 tumor model. Tam-induced Cre-mediated excision of the flox-stop cassette leads to SV40 large T antigen expression. Peptide sequence in red indicates epitope I recognized by transgenic TCR_SV40-I CD8⁺ T cells. (B) Liver carcinogenesis in ASTxCre-ER^T2 mice. Hematoxylin and eosin (H&E) staining of liver sections collected at D10, D34, and D97 after Tam treatment. See also Figure S1. (C) Flow cytometric analysis of CD8⁺ splenocytes from TCR_SV40-I transgenic mice using Db/Tag-I tetramers. (D) Differentiation of naive TCR_SV40-I CD8⁺ T cells into functional effector and memory T cells. Left: Expression levels of CD44 and CD62L. Middle: 1 day after adoptive transfer of naive TCR_SV40-I (Thy1.1⁺) into B6 (Thy1.2⁺) hosts, mice were immunized with 5 × 10⁶ cfu of LM-Tag-I or control LM-Ø. At 7 days after infection, peripheral blood was analyzed for expansion of donor TCR_SV40-I. Each symbol represents an individual mouse; data show mean ± SEM. *p < 0.0001. Results are representative of at least ten independent experiments. Right: Intracellular IFN-γ and TNF-α production for naive, effector (8 days after LM-Tag-I), and memory (2–3 months LM-Tag-I) T cells. FACS plots are gated on CD8⁺Thy1.1⁺ cells. (E) Experimental scheme. (F) CD44 expression and carboxy-fluorescein-succimidyl ester (CFSE) dilution of transferred, naive TCR_SV40-I isolated from spleens of ASTxCre-ER^T2 1–2 days after transfer (left) and isolated from livers 3 days after Tam (right). (G) CD44, CD62L, Ki67, PD1, LAG3, 2B4, and TIM3 expression of D8 TCR_SV40-I (blue) and D30 TCR_SV40-I (red) isolated from livers of Tam-treated ASTxCre-ER^T2 mice, and intracellular IFN-γ and TNF-α production. Naive T cells are shown as controls (gray). Results are representative of at least six independent experiments; data show mean ± SEM (*p = 0.0001, **p < 0.0001; ns, not statistically significant) using unpaired, two-tailed Student’s t test (for analysis between groups ~D8 and >D30 from 2–3 independent experiments).

Figure 2. Tumor-Specific CD8⁺ T Cells in Pre-malignant Lesions Enter a Fixed State of Dysfunction Not Reliant on External Cues

(A) Experimental scheme. D8 and D30 TCR_SV40-I were isolated from livers and transferred into normal B6 mice and immunized with LM-Tag-I either 1 day or 3–4 weeks after transfer. (B) 1 day after second transfer, mice were immunized with 5 × 10⁶ cfu of LM-Tag-I, and 7 days later, spleens were analyzed for expansion of donor T cells by flow cytometry. *p = 0.00042 using Mann-Whitney U-test. (C) D8 TCR_SV40-I that expanded after immunization as described in (B) were isolated from spleens 1 month later and evaluated for CD44, CD62L, PD1, LAG3, CCR7, and CD127 expression, and for IFN-γ and TNF-α production in response to antigen ex vivo. D8 TCR_SV40-I (blue); naive (gray) and memory (solid black) are shown as controls. (D) 3–4 weeks after second transfer mice were immunized and analyzed as described in (B). Each symbol represents an individual mouse with the mean shown. Data are pooled from two independent experiments, with n = 8–9 for D8 TCR_SV40-I and n = 7–9 for D30 TCR_SV40-I. *p = 0.0015 using Mann-Whitney U-test. (E) D8 and D35 TCR_SV40-I were cultured in vitro for 72 hr in the presence of IL-2 with or without anti-PD1 antibody (IL-2 [20 U/mL]; anti-PD1 [10 μg/mL; clone RMP1-14]). IFN-γ and TNF-α production was determined after 4.5 hr peptide stimulation. (F) Phenotypic and functional analyses of D35 TCR_SV40-I after PD1 and PD-L1 blockade in vivo. Top: Experimental scheme. Bottom: Expression levels of TBET, Ki67, PD1, TIM3, GZMB, and IFN-γ and TNF-α production (after 4.5 hr peptide stimulation). Each symbol represents an individual mouse. Data show mean ± SEM (p values are shown; ns, not statistically significant) using unpaired, two-tailed Student’s t test (for analysis between isotype and treatment group).

Figure 3. Genome-wide Transcriptome Analysis of Tumor-Specific TCR_SV40-I CD8⁺ T Cells in Pre-malignant Lesions

(A) Principal component analysis of D8–12 TCR_SV40-I (n = 3) and D34 TCR_SV40-I (n = 4) isolated from pre-malignant liver lesions, with naive (N; n = 3) and effector (Eff; n = 3) TCR_SV40-I as controls. (B) K-means clustering; heatmap of transcript levels for all 14 clusters show log₂-tranformed expression intensities that were mean-centered at the probe level. (C) Selected K-means clusters. Lack of effector genes (blue, top); progressively up- or downregulated genes (green, middle); early, D8-upregulated genes (red, bottom). (D) Top 10 biological processes (BP) (Gene Ontology [GO] terms) enriched in clusters 5 and 10. Numbers in parentheses indicate the numbers of genes within each GO term. See also Figure S3 and Table S1; for (A) and (C), FDR q value ≤ 0.05.

Figure 4. T Cells Isolated from Pre- and Early Malignant Lesions Exhibit the Molecular Hallmarks of T Cells from Late-Stage Human and Mouse Tumors

(A) Enrichment of genes sets in D34 TCR_SV40-I from pre-malignant lesions described for T cells from late-stage and metastatic human tumors (GEO: GSE24536) (top) and mouse tumors (GEO: GSE42824) (lower). 200 most differentially expressed genes in T cells from human or mouse late-stage tumors (compared to naive T cells) were used for GSEA analyses of D34 TCR_SV40-I. NES, normalized enrichment score. (B) Examples of genes enriched in D34 TCR_SV40-I. See also Figure S4.

Figure 5. Tumor-Specific TCR_SV40-I T Cells and Exhausted Virus-Specific T Cells Share a Conserved Core Program of Dysfunction

(A) Gene set enrichment analysis (GSEA) in D8 TCR_SV40-I of “D8 exhaustion in chronic viral infection” genes (GEO: GSE30962). GSEA plot of D8 TCR_SV40-I (versus D8 Effectors) of genes previously identified to be upregulated (left) or downregulated (right) in exhausted, virus-specific T cells 8 days after chronic viral infection (compared to Effectors, 8 days after acute infection). NES, normalized enrichment score. See also Figure S5. FDR q value ≤ 0.001. (B) Venn diagram of D8 TCR_SV40-I (red) and ~D30 TCR_SV40-I (green) from pre- and early malignant lesions and of ~D30 exhausted T cells during chronic viral LCMV infection (clone 13) (GEO: GSE9650). See Table S2 for full gene list. (C) TBET and EOMES expression in naive (N), effectors from spleens and livers 8 days after Listeria immunization (E_S and E_L), and TCR_SV40-I isolated from pre- and early malignant lesions at D8 and D35 after tumor initiation (D8, D35+). Each symbol represents an individual mouse; data show mean ± SEM. p values are shown. See also Figure S5C.

Figure 6. Persistent Antigen, Not the Microenvironment, Drives T Cell Dysfunction in Established, Solid Tumors

(A) Experimental scheme of co-transfer experiment. (B) H&E staining of liver sections from ASTxAlb:Cre mice, showing hepatocellular carcinoma. Scale bar (top) represents 2 mm; scale bar (bottom) represents 50 μm. (C) 1 week after infection with 5 × 10⁶ cfu of LM-Tag-I-OVA, livers from ASTxAlb:Cre and spleens from B6 mice were analyzed for expansion of donor TCR_SV40-I (Thy1.1⁺) and TCR_OT-I (Ly5.1⁺) by FACS. (D) Flow cytometric analysis of TCR_SV40-I and TCR_OT-I from livers (ASTxAlb:Cre) or spleens (B6) 1 week (w1) and 2 weeks (w2) after transfer and immunization. Histograms are gated on CD8⁺ Thy1.1⁺ and Ly5.1+ cells, respectively. Right: TCR_SV40-I (red) and TCR_OT-I (blue) isolated from B6 or ASTxAlb:Cre host mice 2–3 weeks after transfer; naive TCR_SV40-I are shown as control (gray); *p < 0.0001 using unpaired, two-tailed Student’s t test (for analysis between TCR_SV40-I andTCR_OT-I, from 3–4 independent experiments, with n = 5–8). (E) Intracellular IFN-γ and TNF-α production by TCR_SV40-I and TCR_OT-I isolated from livers (ASTxAlb:Cre) or spleens (B6) 1 (w1), 2.5 (w2.5), or 7 (w7) weeks after transfer. Results are representative of four independent experiments. Right: IFN-γ⁺ and IFN-γ⁺/TNF-α⁺ TCR_SV40-I (red) and TCR_OT-I (blue) 2–3 weeks after transfer; *p < 0.0001 using unpaired, two-tailed Student’s t test (for analysis between SV40-1 and TCR_OT-I, from 2 independent experiments, with n = 2–4).