Low TCR Binding Strength Results in Increased Progenitor-like CD8+ Tumor-Infiltrating Lymphocytes

Abstract

T-cell receptor (TCR) binding strength to peptide-MHC antigen complex influences numerous T-cell functions. However, the vast diversity of a polyclonal T-cell repertoire for even a single antigen greatly increases the complexity of studying the impact of TCR affinity on T-cell function. Here, we determined how TCR binding strength affected the protein and transcriptional profile of an endogenous, polyclonal T-cell response to a known tumor-associated antigen (TAA) within the tumor microenvironment (TME). We confirmed that the staining intensity by flow cytometry and the counts by sequencing from MHC-tetramer labeling were reliable surrogates for the TCR-peptide-MHC steady-state binding affinity. We further demonstrated by single-cell RNA sequencing that tumor-infiltrating lymphocytes (TIL) with high and low binding affinity for a TAA can differentiate into cells with many antigen-specific transcriptional profiles within an established TME. However, more progenitor-like phenotypes were significantly biased towards lower affinity T cells, and proliferating phenotypes showed significant bias towards high-affinity TILs. In addition, we found that higher affinity T cells advanced more rapidly to terminal phases of T-cell exhaustion and exhibited better tumor control. We confirmed the polyclonal TIL results using a TCR transgenic mouse possessing a single low-affinity TCR targeting the same TAA. These T cells maintained a progenitor-exhausted phenotype and exhibited impaired tumor control. We propose that high-affinity TCR interactions drive T-cell fate decisions more rapidly than low-affinity interactions and that these cells differentiate faster. These findings illustrate divergent forms of T-cell dysfunction based on TCR affinity which may impact TIL therapies and antitumor responses.

Figure 1.

Tetramer staining distinguishes CD8⁺ TILs with different antigen binding strengths. A, Representative histograms showing CD8⁺ TIL tetramer staining directly after sorting into Tet-High (red), Tet-Low (blue), and Tet-negative (Tet-Neg; grey) populations (left), and after the same sorted populations were cultured ex vivo for 7 days and re-stained with tetramer (right). Representative plot of 3 independent experiments. B, The fold change in tetramer geometric mean fluorescent intensity (gMFI) between Tet-High and Tet-Low T cells after 7 days in culture was determined. Differences between the gMFI of Tet-High– and Tet-Low–sorted populations were determined by a paired t test. Error bars show standard deviation of the mean (n = 3). C, Histograms showing tetramer staining of CD8⁺ TILs from 5 individual mice after gating into Tet-High, Tet-Low, or Tet-Neg populations 14 days post tumor challenge. D, Tetramer fluorescence intensity was divided by CD3 fluorescence intensity for each cell independently, to account for the influence of CD3 surface expression on tetramer binding. E, Tetramer:CD3 fluorescent intensity ratio was used to calculate the gMFI ratio for each sample. Data were compared by paired ANOVA. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001. All flow cytometry data displayed as a histogram is normalized to the mean.

Figure 2.

CD8⁺ TILs with stronger tetramer binding upregulate inhibitory receptor expression more rapidly than those with lower tetramer binding. A, The frequency of Tet-Low and Tet-High cells is shown as a percentage of total tetramer binding cells over time. The frequency of PD-1– and TIM3-expressing cells from Tet-Low CD8⁺ (B) TILs and Tet-High CD8⁺ TILs (C) is shown. D, Comparison of the frequency of PD-1– and TIM3-expressing cells of Tet-Low and Tet-High CD8⁺ TILs on day 7, 10, and 14. E, Frequency of PD-1⁺ TIM3⁻ cells between Tet-Low or Tet-High CD8⁺ TILs over time. F, The frequency of PD-1⁻ TIM3⁻ CD8⁺ TILs between Tet-Low and Tet-High groups over time. Each dot represents TIL sample from one mouse. A paired two-way ANOVA was used to determine statistical significance (n = 10 at Days 7 and 14, n = 9 at Day 10 from 2 independent experiments). ns, not significant; *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001.

Figure 3.

The binding strength of individual CD8 TILs is distinguished by barcoded MHC-tetramers and scRNA-seq. A, Purity checks were performed directly after FACS on CD8⁺ TILs into Tet-High (red), Tet-Low (blue) and Tet-negative (Tet-Neg; grey) populations to confirm separation of different tetramer binding populations for two experiments. B, Histograms showing normalized tetramer-barcode counts of the same tetramer-sorted CD8⁺ TIL groups as measured by CITE-seq for two experiments. C, scRNA-seq data from both experiments were used to generate a UMAP of 15 transcriptionally unique clusters of CD8⁺ TILs (clusters 14 was excluded from all visualization due to low cell numbers). The list on the right shows the immunophenotype of each cluster. Tetramer binding was determined by either tetramer barcode counts (D) or by the flow-sorted Tet-High, Tet-Low, and Tet-Neg samples (E) and were visualized on a UMAP. F, Tet-High and Tet-Low cell counts within each tetramer positive cluster are shown. G, Percentage of total Tet-High or total Tet-Low cells distributed within each tetramer positive cluster is shown. A two-proportion z-test with FDR correction was used to test differences in cluster contribution from Tet-High and Tet-Low cells within each cluster and determine statistical significance (*, q-value ≤ 0.05). For scRNA-seq, 2 independent experiments were performed on CT26 TILs isolated 14 days posttumor challenge pooled from 15 mice.

Figure 4.

Differential gene expression of scRNA-seq clusters defines unique gene signatures. A–D, Blended UMAPs displaying the dual expression of the indicated genes or proteins in red or green if singly expressed or in yellow if coexpressed. A, Cell surface expression of CD44 and CD62 L proteins determined by ADT, and the corresponding genes (B) Cd44 and Sell (encodes CD62L) via scRNA-seq. Blended UMAPs are shown for Tox and Tcf7 (C) and Mki67 and Btg2 (D). E, Heat map showing the 10 most positively expressed genes from each cluster based on the fold change within a single cluster relative to all other clusters. F–H, Selected gene expression patterns visualized on the UMAP for transcription factors (F), functional markers (G), and exhaustion markers (H). For scRNA-seq, 2 independent experiments were performed on CT26 TILs isolated 14 days post tumor challenge pooled from 15 mice.

Figure 5.

Clustering of the CITE-seq data corroborates the scRNA-seq clustering and informs on cluster functions. A, UMAP based on 14 CITE-seq markers (Supplementary Table S3). Clusters determined from the scRNA-seq data were visualized on the CITE-seq UMAP (B), and clusters determined by the CITE-seq markers were visualized on the scRNA-seq UMAP (C). The average position of a cluster within the UMAP space is marked with the cluster name on the UMAP. D, Pearson correlation r values between all of the CITE-seq markers are shown in a heat map. Violin plots showing the CITE-seq expression of PD-1 (E), TIM3 (F), LAG-3 (G), and CD73 (H) in Tet-Low (blue) and Tet-High (red) cells within each scRNA-seq tetramer-positive cluster. For all violin plots statistical significance as determined by Seurat “FindMarkers” with default settings (Wilcoxon Rank Sum test) and FDR corrected q-values. The mean of the data in the violin plot is represented with a dot, SD is shown with lines. A single “*” was placed above clusters with significant differences between Tet-High and Tet-Low cells for a given gene and cluster names below the x-axis were bolded (*, q ≤ 0.05). For scRNA-seq and CITE-seq, 2 independent experiments were performed on CT26 TILs isolated 14 days posttumor challenge pooled from 15 mice.

Figure 6.

Clusters 6 and 12, enriched for Tet-Low cells, are similar based on their expression of surface proteins and retain a more progenitor-like phenotype. Differences in gene expression between Tet-High and Tet-Low cells are displayed as a volcano plot within clusters 6 and 12 (A), within Cluster 0 (B), and within all tetramer positive clusters (C). Genes enriched in Tet-Low cells are highlighted in blue, while genes enriched in Tet-High cells are highlighted in red. The number of genes found significantly altered in Tet-High or Tet-Low cells is displayed in the top corners of each volcano plot. Thresholds of FDR 0.05 (horizontal line) and a fold change ± 1.2 (vertical lines) were used to identify significantly altered genes. Violin plots of tetramer-positive RNA clusters subdivided by Tet-High and Tet-Low cells and expression of quiescence markers (D), effector molecules (E), and exhaustion and activation markers was assessed (F). G, Violin plots combining all tetramer positive clusters subdivided by Tet-High and Tet-Low cells are shown for selected genes. For all violin plots statistical significance was determined by Wilcoxon Rank Sum test and FDR corrected q-values. The mean of the data in the violin plot is represented with a dot, SD is shown with lines. A single “*” was placed above clusters with significant differences between Tet-High and Tet-Low cells for a given gene and cluster names below the x-axis were bolded (*, q ≤ 0.05). For scRNA-seq and CITE-seq, 2 independent experiments were performed on CT26 TILs isolated 14 days posttumor challenge pooled from 15 mice.

Figure 7.

TCR-transgenic mice with exclusively low-affinity TCR for the tumor antigen have impaired tumor control and fewer terminally exhausted TIL. The weight (A) and total viable cell count (B) from tumors 14 days posttumor challenge from 1D4 (blue), gp70-deficient (red), or WT BALB/c, gp70-sufficient (black) mice is shown. C, The frequency of CD8⁺ TILs as a percentage of total viable cells is shown. The frequency of CD8⁺ TILs that were PD-1⁺TIM3⁻ (D) and PD-1⁺TIM3⁻SLAMF6⁺ (E) is shown. One-way ANOVA was used to determine statistical significance between tumor sizes and TIL profiles (n = 5 for all groups, error bars show SD of the mean). ns = not significant; *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001.