CD8+ T cells in breast cancer tumors and draining lymph nodes: PD-1 levels, effector functions and prognostic relevance

Abstract

CD8⁺ T cells shape the antitumor immune response. Here, we evaluated CD8⁺ T cells expressing different levels of PD-1, their functional status, and distribution in different tissues of luminal breast cancer (BC) patients. We characterized the exhaustion stages of CD8⁺ T cells in tumors, juxtatumoral tissues (JTs), and tumor-draining lymph nodes (TDLNs). Terminal exhausted CD8⁺ T cells (PD-1^High CD8⁺) were predominant in tumors and nearly absent in other tissues. However, in all tissues evaluated, most CD8⁺ T cells exhibited a pre-exhausted phenotype (PD-1^Int CD8⁺) or did not express PD-1. PD-1^High and PD-1^Int CD8⁺ T cells from tumors and JTs presented central and effector memory phenotypes, while in TDLNs were primarily central memory. TCR-β sequencing revealed higher clonality among CD8⁺ T cells from tumor than TDLNs, with tumor-enriched clones also detected in TDLNs. Analysis of scRNA-seq datasets from tumors and JTs of colorectal and non-small cell lung cancer patients, identified a CD8⁺ terminal exhaustion and a CD8⁺ pre-exhausted signatures. High expression of exhaustion-associated genes in BC tumors correlated with improved overall survival. Overall, PD-1 expression effectively distinguishes exhaustion stages in CD8⁺ T cells. PD-1^Int cells found in tumors, JTs, and TDLNs represent a promising therapeutic target for cancer immunotherapy.

Keywords: Breast cancer; CD8+ T cells; PD-1; draining lymph nodes; exhaustion.

Figure 1.

CD8⁺ T cells from tumors, JTs and M- and NM-DLNs from BC patients. (a) UMAP projection showing the clustering of 16,000 leukocytes in tumors, JTs, M-DLNs and NM-DLNs from BC patients (N = 4). Each color represents a cluster and is associated with a different immune population defined by phenotypic markers (CD3, CD8, CD4, CD19, FOXP3). (b) Frequencies of CD8⁺ T cells within CD45⁺ cell population in tumors (circle, N = 39), JTs (square, N = 12), M-DLNs (triangle, N = 17) and NM-DLNs (inverted triangle N = 19). (c) Representative contour plots (left panel) and frequencies (right panel) of cells expressing CD45RA and/or CD27 within CD8⁺ T cells in tumors (circle, N = 34), JTs (square, N = 16), M-DLNs (triangle, N = 12) and NM-DLNs (inverted triangle, N = 5). (d) Next-generation-sequencing-based high-throughput TCR-β CDR3 analysis of sorted memory CD8⁺ T cells from matched NM- and M-DLNs and tumors from 3 BC patients. The cumulative frequencies of TCR-β CDR3 clones in memory CD8⁺ T cells (left panel), from NM-DLN (blue), M-DLN (red) and tumor (green). Overlap of the TCR-β CDR3 clones in memory CD8⁺ T cells (right panel), from NM-DLN, M-DLN and tumor. The top 100 TCR-β CDR3s from tumor memory CD8⁺ T cells were identified, and the percentage of these clones found in both NM- and M-DLN are shown. p- values were calculated using Kruskal-Wallis (b) or Friedman test (c). Data presented as mean ± SD, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001. Only significant differences are indicated.

Figure 2.

PD-1 expression on CD8⁺ T cells from different tissues in BC patients. (a) Representative contour plots of PD-1 expression on CD8⁺ T cells from tumors, JTs, M-DLNs and NM-DLNs from BC patients. (b) Frequencies of PD-1^Neg (black), PD-1^Int (blue) or PD-1^High (red) CD8⁺ T cells in tumors (circle, N = 37), JTs (square, N = 12), M-DLNs (triangle, N = 19) and NM-DLNs (inverted triangle, N = 19) – comparison between tissues. (c) Comparison of frequencies of PD-1^Neg (black), PD-1^Int (blue) or PD-1^High (red) CD8⁺ T within tumors (circle, N = 37), JTs (square, N = 12), M-DLNs (triangle, N = 19) and NM-DLNs (inverted triangle, N = 19) (d) Frequency of NAÏVE, central memory (CM), EFFECTOR or effector memory (EM) CD8⁺ T cells within PD-1^Neg (black), PD-1^Int (blue) or PD-1^High (red) CD8⁺ T cells in tumors (circle, N = 20), JTs (square, N = 15), M-DLNs (triangle, N = 12) and NM-DLNs (inverted triangle, N = 5). p-values were calculated using Kruskal-Wallis (b) or Friedman test (c). Data presented as mean ± SD, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001. Only significant differences are indicated.

Figure 3.

Expression of exhaustion markers on CD8⁺ T cells from different tissues in BC patients. (a) Frequency of CD39⁺, Ki-67⁺, TIM-3⁺, TIGIT⁺ or BTLA⁺ within PD-1^Neg (black), PD-1^Int (blue) and PD-1^High (red) tumor-infiltrating CD8⁺ T cells. (b) Representative histograms of TOX expression within tumor-infiltrating CD8⁺ T cells. Fluorescence minus one (FMO) (gray), PD-1^Neg (black), PD-1^Int (blue), PD-1^High (red) (left panel). MFI of TOX in PD-1^Neg (black) PD-1^Int (blue) and PD-1^High (red) tumor-infiltrating CD8⁺ T cells (right panel). (c) Frequencies of CD39⁺, TIGIT⁺, or BTLA⁺ within PD-1^Neg (black), PD-1^Int (blue) and PD-1^High (red) CD8⁺ T cells in JTs samples. (d) Frequencies of CD39⁺, Ki-67⁺, TIGIT⁺, or BTLA⁺ within PD-1^Neg (black), PD-1^Int (blue) and PD-1^High (red) CD8⁺ T cells in M-DLNs samples. p-values were calculated using Friedman test (a) or RM one-way ANOVA Tukey test using Geisser-Greenhouse correction (b). Data presented as mean ± SD, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001, ns: non-significant differences.

Figure 4.

Effector function of CD8⁺ T cells from tumors and M-DLNs in BC patients. (a) Representative contour plots and frequencies of cytokine-producing CD8⁺ T cells (IL-2, TNF and IFN-γ) within PD-1^Neg (black), PD-1^Int (blue) and PD-1^High (red) in tumors (top panel) and M-DLNs (bottom panel) after PMA/Ionomycin stimulation. (b) Representative contour plots and frequencies of Perforin⁺, GrzB⁺ and CD107a⁺, within PD-1^Neg (black), PD-1^Int (blue) and PD-1^High (red) tumor-infiltrating CD8⁺ T cells after PMA/Ionomycin stimulation (left panel) and CD107a⁺ within PD-1^Neg (black), PD-1^Int (blue) and PD-1^High (red) CD8⁺ T cells in M-DLNs (right panel). p-values were calculated using Friedman test. Data presented as mean ± SD, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ns: non-significant differences.

Figure 5.

High expression of exhaustion markers correlates with better overall survival in BC patients. (a) Kaplan-Meier curves comparing the survival probability of BC patients (TCGA cohort) with high expression of PTPRC (CD45) and CD8A genes, stratified by expression levels of TOX. (b) Kaplan-Meier curves comparing the survival probability of the BC patient’s cohort with high expression of PTPRC (CD45), CD8A and TOX genes stratified by expression levels of PDCD1 (PD-1) or, (c) PRF1 (perforin) and GZMB (GrzB). The number of patients in each selected population is indicated in the graph. The survival curves show the 95% confidence interval of the analysis shaded in the plot. The p-value for the survival analysis shown in the graph is derived from the log-rank test.

Figure 6.

scRNA-seq data reveal CD8⁺ T cell clusters exhibiting pre-exhausted and terminally exhausted signatures in tumors and JTs from CRC patients. (a) PDCD1 (PD-1) gene expression in CD8⁺ T cells from tumors and JTs. (b) Distribution of eight clusters of CD8⁺ T cells across tumors and JTs. (c) Analysis of PDCD1 (PD-1), HAVCR2 (TIM-3), ENTPD1 (CD39), TOX, TCF7, PRF1 (perforin), GZMB (GrzB) and IFNG (IFN-γ) gene expression on CD8⁺ T cells clusters of single cells isolated from tumors. Comparison between CD8-LAYN (term-ex) and other clusters. (d) Analysis of PDCD1 (PD-1), TOX, TCF7, PRF1(Perforin), GZMB (GrzB) and IFNG (IFN-γ) gene expression on CD8⁺ T cells clusters of single cells isolated from tumors. Comparison between CD8-GZMK (pre-ex) and other clusters. Clusters were defined in the publication. p-values were calculated using Dunnett’s multiple comparisons test. Data presented as mean ± SD, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001 and ****p ≤ 0.0001. Only significant differences are indicated.