PD-1 and ICOS coexpression identifies tumor-reactive CD4+ T cells in human solid tumors

Abstract

CD4+ Th cells play a key role in orchestrating immune responses, but the identity of the CD4+ Th cells involved in the antitumor immune response remains to be defined. We analyzed the immune cell infiltrates of head and neck squamous cell carcinoma and colorectal cancers and identified a subset of CD4+ Th cells distinct from FOXP3+ Tregs that coexpressed programmed cell death 1 (PD-1) and ICOS. These tumor-infiltrating lymphocyte CD4+ Th cells (CD4+ Th TILs) had a tissue-resident memory phenotype, were present in MHC class II-rich areas, and proliferated in the tumor, suggesting local antigen recognition. The T cell receptor repertoire of the PD-1+ICOS+ CD4+ Th TILs was oligoclonal, with T cell clones expanded in the tumor, but present at low frequencies in the periphery. Finally, these PD-1+ICOS+ CD4+ Th TILs were shown to recognize both tumor-associated antigens and tumor-specific neoantigens. Our findings provide an approach for isolating tumor-reactive CD4+ Th TILs directly ex vivo that will help define their role in the antitumor immune response and potentially improve future adoptive T cell therapy approaches.

Keywords: Cancer; Immunology; Immunotherapy; Oncology; T cells.

Figure 1. PD-1 and ICOS identify distinct subsets of CD4⁺ Th TILs.

(A) t-SNE analysis of tumor-infiltrating CD3⁺CD4⁺ T cells isolated from 22 patients with HNSCC. The gate identified PD-1⁺ cells and gate was applied to all plots. (B) Flow cytometric analysis of CD4⁺ TILs and the gating strategy are shown for a representative patient with HNSCC. Mem, memory. (C) Subsets of CD4⁺ Th cells (DN, SP, and DP) were defined on the basis of differential expression of PD-1 and ICOS. (D) A summary of the frequency (Freq.) of DN, SP, and DP CD4⁺ Th cells in the blood and tumor of patients with HNSCC (left, n = 40) or CRC (right, n = 31) is shown. The DN (teal), SP (gray), and DP (orange) CD4⁺ Th cell subsets for each patient are highlighted on the graphs in C and D. Small horizontal lines indicate the mean ± SEM.

Figure 2. Phenotypic properties of DN, SP, and DP CD4⁺ Th TILs.

(A) Ex vivo phenotypic analysis of the expression of Ki-67, HLA-DR, CD39, CD103, CTLA-4, CCR7, and CD69 on the 3 CD4⁺ TIL subsets of 1 representative patient with HNSCC. (B) Summary of the percentages of the above-mentioned markers among multiple patients with HNSCC (n = 27 for Ki-67, HLA-DR, CD39, and CD103; n = 22 for CTLA-4; n = 34 for CCR7; and n = 11 for CD69). (C) Summary of the percentages of the above-mentioned markers among multiple patients with CRC (n = 26 for Ki-67 and CCR7; n = 15 for HLA-DR; n = 25 for CD39, CD103, CTLA-4; and n = 11 for CD69). (D) Correlation between the percentages of HLA-DR⁺ cells and Ki-67⁺ cells in DP CD4⁺ TILs from patients with HNSCC (left, n = 27) or CRC (right, n = 15). (E) Correlation between the percentages of CD69⁺ cells and CCR7⁺ cells in the different CD4⁺ TIL subsets in patients with HNSCC (n = 11) or CRC (n = 11). Horizontal lines indicate the mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001, by 1-way ANOVA with Tukey’s post hoc test.

Figure 3. Functional properties of DN, SP, and DP CD4⁺ Th TILs.

qPCR analysis of the expression of mRNA transcripts of transcription factors (A), cytokines (B), and CXCL13 (C) by the indicated CD4⁺ Th cell subsets isolated from patients with HNSCC or CRC directly ex vivo. (D) Left: Flow cytometric analysis of CXCL13 production by the different CD4⁺ TIL subsets directly ex vivo from 1 representative patient with HNSCC. Right: Summary of the frequency of CXCL13-producing cells in each CD4⁺ TIL subset in patients with HNSCC (n = 28) or CRC (n = 25). Horizontal lines indicate the mean ± SEM. ***P < 0.001 and ****P < 0.0001, by 1-way ANOVA with Tukey’s post hoc test (D, left and middle graphs) or unpaired t test (D, far right graph).

Figure 4. DP CD4⁺ Th cells are in the tumor stroma proximal to MHC class II⁺ cells.

mIHC for CD3, CD8, FOXP3, ICOS, MHC class II, and cytokeratin on FFPE tissue sections from HNSCC and CRC tumors. Representative ROIs (left) from an HNSCC tumor (A) and a CRC tumor (B) and the corresponding cellular spatial relationship maps (right). Scale bars: 200 μm. The maps show cytokeratin⁺ cells (tumor) in black, MHC class II⁺ cells in blue, and CD3⁺CD8^–FOXP3^–ICOS⁺ (DP CD4⁺ Th cells) in red. (C) Analysis of the distribution of CD3⁺CD8^–FOXP3^–ICOS⁺ (DP CD4⁺ Th cells) in the tumor stroma and tumor epithelium in 14 tumors (n = 8 HNSCC and n = 6 CRC). (D) Neighbor’s cell analysis was performed on 11 tumors (n = 6 HNSCC and n = 5 CRC). The proportion of MHC class II⁺ cells in the stroma and cytokeratin⁺ cells detected within a 15 μm radius from CD3⁺CD8^–FOXP3^–ICOS⁺ (DP CD4⁺ Th) cells is indicated as the frequency of total cells. Circles represent HNSCC tumors, and triangles represent CRC tumors. Horizontal lines indicate the mean ± SEM. Data for patients with MHC class II–expressing tumor cells are shown in red in D.

Figure 5. The presence of DP CD4⁺ Th cells positively correlates with DP CD8⁺ T cells in HNSCC, but not in CRC.

(A) Comparison of the percentage of DP CD4⁺ and DP CD8⁺ cells between HNSCC (n = 49) and CRC (n = 31) tumors. (B) Correlation between the frequencies of DP CD4⁺ and DP CD8⁺ cells in patients with HNSCC (n = 49) or CRC (n = 31). Red circles represent HPV⁺ HNSCC tumors. (C) qPCR analysis of CXCL9, CXCL10, and CXCL11 mRNA expression by the indicated CD4⁺ T cell subsets isolated from patients with HNSCC or CRC directly ex vivo. Horizontal lines indicate the mean ± SEM. *P < 0.05 and **P < 0.01, by unpaired 2-tailed t test (A) or 1-way ANOVA with Tukey’s post hoc test (C).

Figure 6. DP CD4⁺ Th TILs have a unique and oligoclonal TCR repertoire.

(A) Gating strategy for the TCR repertoire analysis. (B) Cumulative frequencies of the top 30 clonotypes for each CD4⁺ Th subset in 9 patients with cancer. (C) Circos plots of unique productive TCRβ nucleotide sequences from each of the indicated cell subsets. Connections highlight sequences from DP CD4⁺ TILs found in the other CD4⁺ Th cell populations. The number of shared sequences is indicated. Circos plots for 2 patients with HNSCC and 1 patient with CRC are shown. (D) Cumulative frequencies of the top 30 clonotypes from DP CD4⁺ TILs (left) and DN CD4⁺ TILs (right) in 9 patients (Pt) with cancer. Their frequencies in blood memory CD4⁺, DN, SP, and DP CD4⁺ TILs are represented. (E) Similarity between the TCR repertoires of the different CD4⁺ T cell subsets was measured using the Morisita-Horn index for 9 patients with cancer. Each color represents a distinct patient connected with a dashed line (D and E). Horizontal lines indicate the mean ± SEM. *P < 0.05, ***P < 0.001, and ****P < 0.0001, by 1-way ANOVA with Tukey’s post hoc test (B).

Figure 7. DP CD4⁺ Th TILs recognize HPV-associated antigens.

(A) In vitro–expanded CD4⁺ TIL subsets (DN, SP, and DP) were cocultured with autologous B cells pulsed with DMSO or the indicated peptide pools. T cell reactivity was assessed by OX40 upregulation after 24 hours of culturing. Data for 1 representative patient with HPV⁺ HNSCC are shown. (B) Summary of the reactivity of CD4⁺ Th subsets to the indicated peptide pools for 6 patients with HPV⁺ HNSCC. (C) Summary of the reactivity of CD8⁺ T cells to HPV16 E6, HPV16 E7, and CEFX peptide pools for the same 6 patients with HPV⁺ HNSCC. CD8⁺ T cell activation was assessed by 4-1BB upregulation. DP CD4⁺ (D) and DP CD8⁺ (E) TILs isolated from 1 patient with HPV⁺ HNSCC were cultured with autologous B cells pulsed with DMSO, the HPV16 E6 peptide pool, or the individual overlapping peptides contained in the HPV16 E6 peptide pool (n = 37). Reactivity was measured by IFN-γ ELISPOT assay. SEB or anti-CD3 antibodies were used as positive controls for CD4⁺ and CD8⁺ T cells, respectively. (F) Summary of the reactivity of DP CD4⁺ and DP CD8⁺ TILs to HPV16 E6 individual peptides for 3 different patients with HPV⁺ HNSCC, as measured by IFN-γ ELISPOT assay. The colors in the heatmap legend represent the number of detected spots/10⁵ cells.

Figure 8. DP CD4⁺ Th TILs recognize tumor-specific neoantigens.

(A) Schematic representation of the methodology used to identify neoantigen-reactive CD4⁺ Th cells. (B) In vitro–expanded CD4⁺ and CD8⁺ T cell subsets (DN, SP, and DP) from patient 1 were cocultured with autologous B cells pulsed with DMSO or the indicated peptide pools. T cell reactivity was measured by IFN-γ ELISPOT assay. (C) Reactivity of DP CD4⁺ and DP CD8⁺ TILs to B cells pulsed with individual 25 mer peptides from pool 1. Flow cytometric analysis of OX40 or 4-1BB upregulation on CD4⁺ or CD8⁺ T cells, respectively is shown. The mutations recognized are highlighted in red. (D) DP CD4⁺ Th TILs were cocultured with autologous B cells pulsed with decreasing concentrations of WT or mutated (Mut) PRKCI^D378N 25 mer peptides. Reactivity was measured by flow cytometric analysis of OX40 upregulation on CD4 cells. (E) DP CD8⁺ TILs were cocultured with autologous B cells pulsed with decreasing concentrations of WT or mutated 8 mer MVK^N190S peptides. Reactivity was measured by flow cytometric analysis of 4-1BB upregulation on CD8⁺ T cells. (F) In vitro–expanded CD4⁺ T cell subsets (DN, SP, and DP) from patient 2 were cocultured with autologous B cells pulsed with DMSO or the indicated peptide pools. T cell reactivity was measured by IFN-γ ELISPOT assay. (G) Reactivity of DP CD4⁺ Th TILs to B cells pulsed with individual 25 mer peptides from peptide pools 2 and 5. Flow cytometric analysis of OX40 upregulation on CD4⁺ cells is shown. The mutations recognized are highlighted in red. (H) DP CD4⁺ Th TILs were cocultured with autologous B cells pulsed with decreasing concentrations of WT or mutated CEP162^E1003K or FBXL3^W410L 25 mer peptides. Reactivity was measured by flow cytometric analysis of OX40 upregulation on CD4⁺ cells.