Age-Associated Contraction of Tumor-Specific T Cells Impairs Antitumor Immunity

Abstract

Progressive decline of the adaptive immune system with increasing age coincides with a sharp increase in cancer incidence. In this study, we set out to understand whether deficits in antitumor immunity with advanced age promote tumor progression and/or drive resistance to immunotherapy. We found that multiple syngeneic cancers grew more rapidly in aged versus young adult mice, driven by dysfunctional CD8+ T-cell responses. By systematically mapping immune cell profiles within tumors, we identified loss of tumor antigen-specific CD8+ T cells as a primary feature accelerating the growth of tumors in aged mice and driving resistance to immunotherapy. When antigen-specific T cells from young adult mice were administered to aged mice, tumor outgrowth was delayed and the aged animals became sensitive to PD-1 blockade. These studies reveal how age-associated CD8+ T-cell dysfunction may license tumorigenesis in elderly patients and have important implications for the use of aged mice as preclinical models of aging and cancer.

Figure 1.

Immunogenic tumors grow more rapidly in aged versus young mice. A, Tumor growth curves in young (2–4 months) vs. aged (20–24 months) WT C57BL/6J mice inoculated with 2.5 × 10⁵ MC38 colorectal cancer (young, n = 17; aged, n = 13), 2.5 × 10⁵ EO771 breast cancer (young, n = 20; aged, n = 17), or 2.5 × 10⁵ B16 melanoma cells (n = 6 per group). Graphs depict mean tumor size. B and C, Tumor growth curves (B) and survivorship (C) of young (2–4 months) vs. aged (20–24 months) WT C57BL/6J mice inoculated with OVA-expressing MC38, EO771, or B16 cancer cells (2.5 × 10⁵ cell per mouse; n = 10 per group). D–E, Tumor growth curves following implantation of MC38 colorectal (D) or EO771 breast (E) tumor cells in young vs. aged WT C57BL/6J mice treated with isotype control or anti-CD8 depleting antibodies (n = 8 per group). Results are representative of at least two independent experiments per tumor model. Graphs depict mean ± SEM. Statistical comparisons between conditions were performed by two-way ANOVA with Bonferroni test to correct for multiple comparisons (A, B, D, and E). Survival statistics were performed by log-rank test (C). *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001.

Figure 2.

Aging reduces the functionality of tumor-infiltrating CD8⁺ T cells. A–F, Flow cytometry analysis of immune cells in tdLN and MC38-OVA, B16-OVA, and EO771-OVA tumors from young (2–4 months) vs. aged (20–24 months) WT C57BL/6J mice on day 12 after tumor implantation [(B16-OVA young, n = 10; aged, n = 9), (MC38-OVA young, n = 8; aged, n = 6), (EO771-OVA young, n = 8; aged, n = 8)]. Quantification of the percentage of CD44 and PD-1 double-positive cells among CD8⁺ intratumoral and tdLN T cells (A). Quantification of the proportions of CD8⁺ T cells with features of progenitor and more differentiated exhausted states (Tim-3⁺/TCF1⁻/CD8⁺/CD44⁺/PD-1⁺) CD8⁺ intratumoral and tdLN T cells (B). Quantification of the percentage of CD44⁺CD62L⁻ effector memory CD8⁺ intratumoral and tdLN T cells (C). Quantification of T-BET expression among intratumoral and tdLN CD8⁺ T cells (D). Quantification of GZMB expression among intratumoral and tdLN CD8⁺ T cells (E). Quantification of MKI-67 expression among intratumoral and tdLN CD8⁺ T cells (F). Results are representative of at least two independent experiments per tumor model. Each dot represents an individual animal. Statistical significance was assessed by Student t test (A–F). Graphs display mean ± SD (A–F). *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001.

Figure 3.

Single-cell analysis reveals a deficit in activated CD8⁺ T-cell states in tumors within aged animals. A, Schematic depicting single-cell RNA-seq experiment and analysis. B, Identification of tumor-infiltrating CD8⁺ T-cell populations. Uniform Manifold Approximation and Projection (UMAP) embeddings of single-cell RNA-seq profiles from 8,081 CD8⁺ T cells isolated from MC38 tumors implanted in young mice and 3,553 CD8⁺ T cells isolated from tumors implanted in aged mice, showing seven clusters, and colored by cluster. Representative of one experiment, n = 8 young mice and n = 7 aged mice, with each mouse barcoded using TotalSeqB hashing antibodies. C, Galaxy plot depicting differences in cluster frequency by age. D, Bar plot depicting proportional differences in CD8⁺ T cell clusters from aged vs. young tumors. E, Bar plot depicting the number of differentially expressed (DE) genes per cluster (top). Heatmap representing stimulation or quiescence gene score per cluster (bottom). F–I, Trajectory analysis by Monocle2. Plot containing all cells colored by cluster (F). Placement of young (G) or aged (H) CD8⁺ T cells along the combined trajectory. Location of cells with gene expression profiles consistent with more stimulated (red) vs. more quiescent (blue) cell states along the pseudotime trajectory (I). Colors in F–H indicate cluster identities as in B. Statistical significance was assessed by unpaired t test (D) and significantly differentially expressed genes were identified by non-parametric Wilcoxon rank sum test (E). *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.

Figure 4.

Aging reduces the proportions and numbers of tumor antigen–specific cytotoxic CD8⁺ T cells. A–C, Representative flow cytometry contour plots depicting staining of intratumoral OVA-specific CD8⁺ T cells from young or aged mice in the MC38-OVA (A), B16-OVA (B), and EO771-OVA (C) tumor models at day 12 after tumor cell implantation. D and E, Quantification of the percentage (D) [(B16-OVA young, n = 15; aged, n = 15), (MC38-OVA young, n = 15; aged, n = 10), (EO771-OVA young, n = 15; aged, n = 14) and absolute numbers (E) [(B16-OVA young, n = 15; aged, n = 14), (MC38-OVA young, n = 15; aged, n = 10), (EO771-OVA young, n = 15; aged, n = 14)] of intratumoral SIINFEKL-reactive CD8⁺ T cells from young or aged mice in the MC38-OVA, B16-OVA, and EO771-OVA tumor models at day 12. F and G, Quantification of the percentage (F) and absolute numbers (G) of melanoma tumor-associated antigen-specific CD8⁺ T cells including MuLV p15E, gp100, and TRP-2 antigen-specific T cells from young or aged mice bearing B16-OVA tumors at day 12 after tumor cell implantation (young, n = 16; aged n = 16). Results are representative of at least two independent experiments per tumor model (A–C). Statistical significance was assessed by Student t test (D–G). Results represent a pool of two independent experiments (D–G). Each dot represents an individual animal. Graphs display mean ± SD (D–G). *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001.

Figure 5.

TCR Vβ usage changes late in mouse lifespan. A–D, Dot plots depicting coefficient of variation (CV) for TCR Vβ usage among naïve CD8⁺ T cells in different age brackets that reside in thymus (A), spleen (B), tdLN (C), and tumor bed of B16-OVA tumors 12 days post-implantation (D). Each dot represents the CV for a single Vβ gene segment among C57BL/6J animals that are 2 months (n = 18), 8 months (n = 18), 12 months (n = 18), 16 months (n = 16), and 20 months (n = 15) of age. E–G, Shannon Diversity Index calculated based on TCR Vβ usage among naïve CD8⁺ T cells as in A–D that reside in thymus (E), spleen (F), tdLN (G), and tumor bed of B16-OVA tumors (H). I–L, Dot plots depicting proportions of SIINFEKL-reactive (I) or TRP-2-reactive (J) CD8⁺ T cells in the tdLN from mice bearing B16-OVA tumors. Dot plots depicting proportions of SIINFEKL-reactive (K) or TRP-2-reactive (L) CD8⁺ T cells in the tumor bed of B16-OVA tumors on day 12 post-implantation. M and N, Dot plots depicting GZMB expression in SIINFEKL-reactive (M) or TRP-2-reactive (N) CD8⁺ T cells isolated from B16-OVA tumors on day 12 post-implantation. O and P, Dot plots depicting the frequencies of PD-1⁺ TIM3⁺ CD8⁺ SIINFEKL-reactive (O) or TRP-2-reactive (P) CD8⁺ T cells isolated from B16-OVA tumors on day 12 post-implantation. Results are pooled from two independent experiments. Each dot represents an individual animal. Significance assessed by one-way ANOVA with Tukey’s correction for multiple hypothesis testing (A–N). Each dot represents an individual animal. Graphs display mean ± SD. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001.

Figure 6.

Restoring tumor-specific T cells enhances tumor control and overcomes resistance to PD-1 blockade. A, Experimental schematic. PBS control or 10,000 naïve young (2–4 months) OT-I CD8⁺ T cells were adoptively transferred into young (2–4 months) or aged (20–24 months) recipients 1 day before tumor implantation (day -1). B16-OVA or MC38-OVA cells (2.5 × 10⁵ cells) were injected subcutaneously in young and aged mice on day 0. B, B16-OVA tumor growth in young (2–4 months) vs. aged (20–24 months) WT C57BL/6J mice injected with mock PBS or young (2–4 months) OT-I CD8⁺ T cells (young, n = 10; young + young OT-I, n = 8; aged, n = 8; aged + young OT-I, n = 8). C, Percent tumor growth inhibition (TGI) following adoptive transfer of young (2–4 months) OT-I T cells in B16-OVA tumor-bearing young or aged mice was quantified using the following formula: % TGI = [(tumor volume of control − tumor volume of OT-I treated)/(tumor volume of control)] × 100. D, Survivorship of tumor-bearing young vs. aged mice injected with PBS or young (2–4 months) OT-I CD8⁺ T cells. E–J, OT-I CD8⁺ T-cell proportion and phenotypes in B16-OVA and MC38-OVA tumor-bearing young or aged mice. % OT-I T cells among total CD8⁺ T cells were quantified in tdLN and tumor of B16-OVA (E) or MC38-OVA (F) tumor-bearing mice [(B16-OVA young, n = 8; aged, n = 8), (MC38-OVA young, n = 5; aged = 7)]. % MKI-67⁺, PD-1⁺, and T-BET⁺ OT-I CD8⁺ T cells in tdLN (G, H) and tumor (I, J) of B16-OVA (G, I) and MC38-OVA (H, J) tumor-bearing. K, Tumor growth curves (n = 10 per group) and survivorship (n = 18 per group) of B16-OVA tumor-bearing young (2–4 months) vs. aged (20–24 months) WT C57BL/6J mice injected with isotype control or anti-PD-1 (29F.1A12). L, Experimental schematic of the combined OT-I T-cell transfers (OT-I donor age of 2–4 months) and PD-1 blockade in young vs. aged recipient mice. M and N, Tumor growth curves (n = 8 per group) and survival (n = 16 per group) of B16-OVA melanoma-bearing young (2–4 months; M) versus aged (20–24 months; N) WT C57BL/6J mice injected with OT-I CD8⁺ T cells and/or anti-PD-1. Results are representative of at least two independent experiments per tumor model. Statistical significance was assessed by Student t test (E–J) and tumor growth curves depict mean ± SEM (B, K, M, and N). Survival statistics were performed by log-rank test. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; and ****, P ≤ 0.0001.

Figure 7.

CD8⁺ T cell–associated immune deficits in human patient tumors during aging. A, Boxplot depicting CD8⁺ T-cell infiltration scores calculated based on gene expression data from patient tumors (n = 9,286) across all tumor types in TCGA, separated by diagnosis age. B and C, Bioinformatics analysis of single cell RNA-sequencing datasets from the Cancer Single-cell Expression Map (CancerSCEM) database of tumor-infiltrating immune cells in human patients. Volcano plot depicting significant correlations between total immune cell frequencies (B) or lymphocyte subset frequencies (C) and patient age. Red dots depict immune populations where cell frequency is significantly positively associated with age across tumor samples. Blue dots depict immune populations where cell frequency is significantly negatively associated with age across tumor samples. D, Scatterplot depicting the significant negative correlation between the proportion of CD8⁺ effector T cells as a fraction of total CD8⁺ T cells in tumors with patient age. E–G, Survival analysis of patient data from TCGA. Progression-free interval in days for young (E) or aged (F) patients with cancer grouped by median T-cell score. Bar plots depicting the percent difference in median progression-free interval (PFI) between patients that score in the (top or bottom) 50th percentile for immune gene signatures, comparing young versus aged patient groups (G), calculated as 100 × (median PFI_{Aged, High} − median PFI_{Aged, Low})/(median PFI_{Aged, Total}) or 100 × (median PFI_{Young, High} − median PFI_{Young, Low})/(median PFI_{Young, Total}). Significance assessed by Kruskal–Wallis one-way ANOVA (A). Correlation coefficients are calculated as Spearman’s rank correlation coefficient (B–D). Significant correlations are calculated as asymptotic P values (B–D). Between-group significance in survival analysis assessed using a log-rank test (E–G). Tcm, central memory T cells.