CD69 Imposes Tumor-Specific CD8⁺ T-cell Fate in Tumor-Draining Lymph Nodes

There are two major subpopulations of intratumoral CD8⁺ T cells in human and murine cancers (1–3). One is noncytotoxic, stemlike PD-1⁺Tcf1⁺Tim-3⁻ CD8⁺ T-cells that persist long-term and importantly serve as a source of the other subpopulation, terminally differentiated PD-1⁺Tcf1⁻Tim-3⁺ CD8⁺ T cells, which have reduced longevity. Because terminally differentiated CD8⁺ T cells are the primary cytotoxic populations in the tumor microenvironment (TME), the efficient generation of terminally differentiated CD8⁺ T cells is critical for antitumor immune responses (4, 5). Recent reports have shown that tumor-draining lymph nodes (TDLN), initial sites of the T-cell immune response, are enriched with tumor-specific stemlike CD8⁺ T cells and are crucial for the antitumor effect of anti–programmed cell death protein 1 (PD-1)/ programmed death ligand 1 (PD-L1) blockade (6–9), raising the importance of TDLNs for an efficient antitumor immune response. However, the molecular mechanisms underlying how and where tumor-specific stemlike CD8⁺ T cells give rise to terminally differentiated CD8⁺ T cells has not been elucidated.

CD69, a type-2 transmembrane glycoprotein, has been known as an activation marker on leukocytes, and recent studies have shown that CD69 is a functional molecule rather than a simple activation marker (10). CD69 suppresses the function of S1P₁, thereby inhibiting lymphocyte egress from lymphoid organs (10–12). Furthermore, CD69 interacts with myosin light chain 9/12 (Myl9/12), playing a role in recruiting T cells to inflamed tissues (13). CD69 is also known to be expressed on CD8⁺ T cells in TDLNs and the TME in human patients with cancer, suggesting its role in antitumor immunity (14). Indeed, we and other groups have independently demonstrated that anti-CD69 treatment enhances antitumor immunity, with similar effects observed in CD69-deficient mice (15–18). These data suggest that CD69 negatively regulates antitumor immunity; however, the detailed molecular mechanisms of CD69-mediated regulation of antitumor immunity have not been elucidated. In this study, our single-cell transcriptomics defined CD69 as an important regulator for differentiation of tumor-specific CD8⁺ T cells. CD69 deficiency resulted in decreased expression of the transcription factor TOX in tumor-specific CD8⁺ T cells in TDLNs, promoting differentiation of stemlike CD8⁺ T cells into the functional terminally differentiated CD8⁺ T cells in TDLNs. Our results suggest that CD69 imposes tumor-specific CD8⁺ T-cell fate within TDLNs.

BALB/c mice and C57BL/6 mice were purchased from Charles River Laboratories Japan or CLEA Japan, respectively. OT-I transgenic mice expressing TCR specific for H-2K^b/OVA^257−264 was purchased from The Jackson Laboratory. CD69-deficient mice were established in our laboratory (19) and backcrossed more than 15 times. All mice were maintained under specific pathogen-free conditions and used at 6 to 16 weeks of age. All animal experiments were approved by the Chiba University Review Board for Animal Care.

The murine CT26 colon carcinoma cell line was purchased from ATCC. The murine MC38 colon carcinoma cell line was purchased from Kerafast. Both cell lines have been verified by short tandem repeat (STR) analysis by the providers and frozen in multiple cryovials immediately upon receipt. MC38-OVA cells were made in our laboratory. In brief, chicken OVA cDNA was retrovirally introduced into MC38 cells using pMSCV-OVA-IRES-EGFP vector (VectorBuilder). The murine B16 melanoma expressing chicken OVA was kindly provided by Dr. Nakatsura (National Cancer Center, Tokyo, Japan). CT26 cells were cultured in RPMI 1640 (Sigma) supplemented with 10% FBS (Hyclone), and MC38-OVA cells and B16-OVA cells were cultured in DMEM (Thermo Fisher Scientific) supplemented with 10% FBS. After thawing, cells were passaged a minimum of three times before use in experiments and maintained for no longer than 3 months. All cells were regularly tested for Mycoplasma contamination by PCR.

CT26 (1 × 10⁶) cells were subcutaneously inoculated into the left flank of BALB/c mice. MC38-OVA (1 × 10⁶), and B16-OVA (5 × 10⁵) were subcutaneously inoculated into the left flank of C57BL/6 mice. Tumor growth was monitored by periodic caliper measurements, as previously described (20). FTY720 (20 μg; Cayman chemical) was intraperitoneally administered every other day. Monoclonal anti-mouse CD69 (H1.2F3), anti-mouse PD-1 (RMP1–14, Bio X Cell), or isotype control antibody (each 200 μg) were intraperitoneally administered twice a week.

Tumor tissues from all tumor models were minced into small pieces and dissociated using a mouse Tumor Dissociation Kit (Miltenyi Biotec) and Gentle MACS Octo Dissociator with Heaters (Miltenyi Biotec), as described in the manufacturer's protocol. Dissociated cells were filtered through a 70-μm mesh and separated by Percoll (Cytiva) density gradient centrifugation to enrich tumor-infiltrating lymphocytes. ACK lysis buffer (Thermo Fisher Scientific) was used for lysis of red blood cells. Cells from lymph nodes were isolated by gently teasing between frosted glass slides and filtered through a 70-μm mesh. Because tumor cells were subcutaneously inoculated into the left flank of mice, the left axillary lymph node was considered as the TDLN and the right axillary lymph node was considered as the non-TDLN. Naïve CD8⁺ T cells were isolated from spleen by negative selection using MojoSort Mouse CD8 Naive T-cell Isolation Kit (BioLegend) with MACS column (Miltenyi Biotec). More than 95% of the isolated cells were CD62L⁺CD44^lo CD8⁺ T cells, determined by flow cytometry.

CD8⁺ T cells or DCs in TDLN or TME were stained with following fluorochrome-conjugated monoclonal antibodies. CD8α (53–6.7), CD11b (M1/70), CD11c (N418), CD24 (M1/69), CD44 (IM7), CD45.2 (104), CD62 L (MEL-14), CD69 (H1.2F3), CD80 (16–10A1), CD86 (GL1), CD94 (18d3), Lag-3 (C9B7W), Ly6c (HK1.4), Ly108 (330-AJ), MHC Class II (I-A/I-E) (M5/114.15.2), NFAT2 (7A6), PD-1 (29F.1A12), Tigit (1G9), Tim-3 (RMT3–23) were obtained from Biolegend; CD45 (30–F11), CD45.1 (A20) were obtained from BD Biosciences; Eomes (Dan11mag), Granzyme B (GB11) were obtained from Thermo Fisher Scientific; Tcf1 (C63D9) was obtained from Cell Signaling Technology; TOX (REA473) was obtained from Miltenyi Biotec. Propidium Iodide (PI; Fujifilm) or Fixable Viability Dye eFluor 506 (Thermo Fisher Scientific) were used to exclude dead cells.

For intracellular staining, cells were fixed/permeabilized with the FOXP3 Staining Buffer Set (Thermo Fisher Scientific) according to the manufacturer's instructions. AH1-specific CD8⁺ T cells were detected with an APC-conjugated MuLV gp70^423–431 H-2L^d tetramer (NIH Tetramer Core Facility), and OVA-specific CD8⁺ T cells were detected with an APC-conjugated chicken OVA^257–264 (SIINFEKL) H-2K^b tetramer (NIH Tetramer Core Facility). The following gating strategies were used: Tcf1⁺ stemlike CD8⁺ T cells (CD8α⁺tetramer⁺PD-1⁺Tcf1⁺Tim-3⁻); Tim-3⁺ terminally differentiated CD8⁺ T cells (CD8α⁺tetramer⁺PD-1⁺Tcf1⁻Tim-3⁺). Data were acquired on a BD FACSCanto II flow cytometer (BD Biosciencies) or LSRFortessa flow cytometer (BD Biosciences) and analyzed with the FlowJo software program (BD Biosciences). For live cell sorting of Tcf1⁺ stemlike or Tim-3⁺ terminally differentiated CD8⁺ T cells, the cell surface marker Ly108 was used instead of Tcf1 (1). A FACSAria III cell sorter (BD Biosciences) was used for live cell sorting.

Naïve CD8⁺ T cells were isolated from spleen of CD69-sufficient (CD45.2⁺) or -deficient (CD45.1⁺) OT-I mice. An equal mixture of naïve OT-I CD8⁺ T cells (5×10⁴ each) were intravenously injected into the lateral tail vein of C57BL/6 mice (CD45.1⁺/CD45.2⁺). MC38-OVA cells were subcutaneously inoculated on the next day, and OT-I CD8⁺ T cells in TDLN were analyzed by flow cytometry.

CD8α⁺PD-1⁺ cells were sorted from TDLNs of CT26-bearing BALB/c mice using a BD FACS Aria III. Nuclei of T cells were isolated using a protocol adapted from a previous report (21). Briefly, cells were resuspended in buffer containing 320 mmol/L sucrose, 10 mmol/L HEPES (Thermo Fisher Scientific), 8 mmol/L MgCl₂, and 0.1% (vol/vol) Triton X-100 (Sigma) for 10 minutes. The lysed cells were centrifuged to collect the nuclei, and the nuclear pellet was fixed in 3% paraformaldehyde (PFA) for 15 minutes. The fixed nuclei were permeabilized with 0.3% Triton X-100 in PBS plus 2% FBS for 10 minutes and stained with a fluorochrome-conjugated anti-NFAT2 (7A6; Biolegend). The nuclei were counterstained with To-Pro-3 (Thermo Fisher Scientific).

TDLN cells were isolated from CT26-bearing BALB/c mice and cultured in RPMI1640 supplemented with 10% FBS, 5 ng/mL human IL2 (Peprotech), 10 mmol/L HEPES, 1 mmol/L sodium pyruvate (Thermo Fisher Scientific), 1% MEM nonessential amino acids (Thermo Fisher Scientific), 55 μmol/L 2-mercaptoethanol (Thermo Fisher Scientific) in the presence or absence of the calcineurin inhibitor cyclosporin A (10 nmol/L; Nacalai) for 4 hours. CD8α⁺AH1 tetramer⁺ cells in TDLNs were analyzed by flow cytometry.

CD8α⁺AH1 tetramer⁺ cells sorted from TDLN or TME of CT26-bearing BALB/c mice using a BD FACS Aria III (10,000 cells each) were encapsulated into droplets, and libraries were prepared using Chromium Single-Cell 3′ Reagent Kits v3.1 according to manufacturer's protocol (10× Genomics). The generated single-cell RNA sequencing (scRNA-seq) libraries were sequenced on a NovaSeq 6000 (Illumina) with 28 cycles for Read 1, 8 cycles for i7 index, and 91 cycles for Read 2. Alternatively, 50 ng of scRNA-seq libraries were converted into a compatible library with MGI sequencer using the MGIEasy Universal Library Conversion kit (MGI). The converted libraries were sequenced on a DNBSEQ-400RS (MGI) with 28 cycles for Read 1, 8 cycles for i7 index, and 91 cycles for Read 2.

Sequence reads from all samples were processed and aggregated using Cell Ranger v.6.0.0 (10× Genomics) based on mouse genome assembly mm10. Aggregated data were further analyzed using Seurat v.4.0.1 (22). Specifically, the expression matrix was log-normalized, and the data were regressed against the total number of unique molecular identifiers (UMI) detected per cell. Principal component analysis (PCA) was performed, and PCA dimensions 1 to 50 were used to find clusters. Finally, marker genes for these clusters were identified. Clusters expressing two or more canonical cell-type markers were classified as doublet cells and excluded from further analyses. The single-cell gene expression was visualized as Uniform Manifold Approximation and Projection (UMAP) overlays, marker gene heat maps, and violin plots using Seurat. The cell-cycle phase score was calculated using Seurat based on the gene list described in Tirosh and colleagues (23). The single-cell gene expression was also visualized as density plots using the Nebulosa R software package v.1.0.1 algorithm (24). The activated pathways were detected by a single-sample gene set variation analysis (ssGSVA; ref. 25). The pathway genes of CA-RIT-NFAT1_up and CA-RIT-NFAT1_down were obtained from (GSE64409; ref. 26). The top 500 differentially expressed genes were determined. The ssGSVA scores were visualized with UMAP overlays and box plots using ggplot2 v.3.3.5.

RNA-seq was carried out as previously described (27). Total cellular RNA of stemlike CD8⁺ T cells (CD8α⁺AH1 tetramer⁺PD-1⁺Ly108⁺Tim-3⁻) or terminally differentiated CD8⁺ T cells (CD8α⁺AH1 tetramer⁺PD-1⁺Ly108⁻Tim-3⁺) sorted from TME of CT26-bearing BALB/c mice was extracted with TRIzol reagent (Thermo Fisher Scientific). For cDNA library construction, a SMARTer Stranded Total RNA-Seq Kit v2 - Pico Input Mammalian (Takara) was used in accordance with the manufacturer's protocol. 20 ng total RNA was used for each sample. Sequencing of library fragments was performed with 76 cycles for Read 1, 8 cycles for index1, and 8 cycles for index2 using an Illumina HiSeq 1500 System. Bcl2fastq v.2.20 was used to convert sequence data to fastq format. FastQC v.0.1.18 was used to check the quality of sequence. For the data analyses, read sequences (50 bp) were aligned to the mm10 mouse reference genome (University of California, Santa Cruz, December 2011) using Bowtie v.2.3.4 and Hisat v.2.1.0. Fragments per kilobase of exon per million mapped reads for each gene were calculated using Cufflinks v.2.2.1, Picard v.2.18.21 and Samtools v.1.9 were used as command line tools to manipulate the SAM/BAM format.

Genes in which maximum FPKM value of each gene was ≥ 1.0 were used for further analyses. Gene clustering was performed on the basis of Pearson correlation coefficient using APCluster (an R Package for Affinity Propagation Clustering). PCA was perform using R. A GSEA was performed using the Python package (28). The MA plots depicts was constructed using the ggplot2 v.3.3.5.

To compare our data with other RNA-seq datasets from the relevant literature, the top 250 differentially expressed genes were determined between the two indicated states for each generated sample. The gene sets of B16-OVA melanoma stemlike and B16-OVA melanoma terminal differentiation were determined using the gene expression profiles of CD8⁺ T-cell subpopulations in B16-OVA melanoma-bearing mice (GSE122713; ref. 1). The gene sets of LCMV C13 stemlike and LCMV C13 terminal differentiation were determined using the gene expression profiles of CD8⁺ T-cell subpopulations in LCMV clone-13–infected mice (GSE122713). The gene sets of Tox^WT LCMV C13 Tim-3⁻ (stemlike) and Tox^KO LCMV C13 Tim-3⁻ (stemlike) were determined using the gene expression profiles of Tim-3⁻ CD8⁺ T cells in LCMV clone-13–infected mice (GSE131643; ref. 29). The gene sets of Tox^WT LCMV C13 Tim-3⁺ (term diff) and Tox^KO LCMV C13 Tim-3⁺ (term diff) were determined using the gene expression profiles of Tim-3⁺ CD8⁺ T cells in LCMV clone-13-infected mice (GSE131643).

Total cellular RNA was extracted with TRIzol reagent, and cDNA was synthesized with ReverTra Ace qPCR RT Master Mix (Toyobo). 10 ng total RNA was used for each sample. Quantitative RT-PCR was performed with a StepOnePlus Real-Time PCR System (Thermo Fisher Scientific) and TaqMan Fast Advanced Master Mix (Thermo Fisher Scientific). The TaqMan probes were purchased from Roche. Relative gene expression was calculated from CT values using a ΔΔ^C_t method. The expression of the target genes was normalized to the expression of Actb. The following probes and primers were used to measure the expression of each gene. Actb, probe (#64, catalog no. 04688635001); forward primer 5′-CTAAGGCCAACCGTGAAAAG-3′; and reverse primer 5′-ACCAGAGGCATACAGGGACA-3′. Tox, probe (#42, catalog no. 04685016001); forward primer 5′-GCAAGTCTGCAACCCCTTC-3′; and reverse primer 5′-CGCTTCTCTCCTCCATTGAT-3′.

scRNA-seq data of patients with treatment-naïve lung adenocarcinoma were obtained from GSE131907 (30). Data processing, analysis, and visualization were conducted using R program with package Seurat. Only CD8⁺ T cells (original labels: CD8 low T, cytotoxic CD8⁺ T, naïve CD8⁺ T, and exhausted CD8⁺ T) with tissues from 11 tumor-containing lung tissues and 10 nonmetastatic lung-draining lymph nodes (original labels: tLung and nLN, respectively) were extracted and used for the analysis. CD8⁺ T cells were grouped into four clusters on the basis of the marker gene expressions: stemlike (TCF7⁺GZMB⁻PDCD1⁺); terminally differentiated (TCF7⁻GZMB⁺PDCD1⁺); naïve (TCF7⁺GZMB⁻PDCD1⁻); effector (TCF7⁻GZMB⁺PDCD1⁻) and visualized with UMAP.

The data that support the findings of this study are available from the corresponding author upon reasonable request. The scRNA-seq and RNA-seq data reported in this paper have been deposited in Gene Expression Omnibus (GEO) under accession number GSE228386, GSE228387, and GSE228388.

Data were analyzed using the Graphpad Prism v.8.4.3 (GraphPad Software). The results of statistical analyses are indicated as the mean ± SD, unless otherwise indicated. Intergroup variance was assessed using the F-test. For normally distributed data, statistical significance was determined using an unpaired two-sided Student t test. An unpaired two-sided Welch t test was used to analyze groups that showed differing variance. For multiple comparisons, statistical significance was determined using a one-way ANOVA, followed by Dunnett test or Tukey test. P values of < 0.05 were considered statistically significant. Gene clustering was performed on the basis of Pearson correlation coefficient.

To investigate the mechanism of enhanced antitumor immunity in Cd69^KO mice, wherein tumor growth was significantly diminished (ref. 17; Fig. 1A), we utilized CT26 colorectal carcinoma cells that expressed AH1 antigen so that tumor-specific CD8⁺ T cells could be detected using an H-2L^d-AH1 tetramer (Supplementary Fig. S1A; ref. 31). We first analyzed tumor-infiltrating lymphocytes 14 days after tumor inoculation and found that CD69 deficiency increased the frequency of total CD8⁺ T cells (Fig. 1B; Supplementary Fig. S1B). Moreover, the frequency of AH1 tetramer⁺ tumor-specific CD8⁺ T cells was increased in Cd69^KO mice (Fig. 1C). Almost all tumor-infiltrating AH1 tetramer⁺ CD8⁺ T cells in both Cd69^WT and Cd69^KO mice expressed PD-1 (Fig. 1D), whereas the frequency of granzyme B-expressing AH1 tetramer⁺ CD8⁺ T cells was significantly increased in Cd69^KO mice (Fig. 1E). These data demonstrate that CD69 deficiency increases tumor-specific cytotoxic CD8⁺ T cells in the TME.

To understand the impact of CD69 deficiency on intratumoral CD8⁺ T cells at a single-cell level, we performed an unbiased analysis of intratumoral tumor-specific CD8⁺ T cells using scRNA-seq and analyzed pooled scRNA-seq data of AH1-specific CD8⁺ T cells from Cd69^WT and Cd69^KO mice. UMAP and clustering analysis assigned cells into four clusters based on their transcriptomes (Fig. 1F), and the top 10 differentially expressed genes across the clusters were examined (Supplementary Fig. S2A–S2C). Cells in cluster 0 highly expressed Gzmb and Havcr2 (encoding Tim-3), representing the terminally differentiated CD8⁺ T-cell population. Cells in cluster 1 expressed high Tcf7 (encoding Tcf1) and Slamf6 (encoding Ly108), which represent the stemlike CD8⁺ T-cell population. Cells in clusters 2 and 3 exhibited the expression of proliferation markers, such as Birc5 (for G₂–M phase) and Mcm6 (for S phase), which were confirmed by cell-cycle scoring (Supplementary Fig. S2D; ref. 23). Cd69^WT and Cd69^KO cells occupied distinct spaces in the UMAP plots; more terminally differentiated CD8⁺ T cells and fewer stemlike CD8⁺ T cells were detected in Cd69^KO mice than in Cd69^WT mice (Fig. 1G). Flow cytometry confirmed that the frequency of AH1-specific, Tim-3⁺ terminally differentiated CD8⁺ T cells was increased in Cd69^KO mice, whereas Tcf1⁺ stemlike CD8⁺ T cells were decreased (Fig. 1H). To estimate the absolute cell number, we calculated the frequency of each cell population among CD45⁺ cells; Tim-3⁺ terminally differentiated CD8⁺ T cells were significantly increased in Cd69^KO mice, whereas Tcf1⁺ stemlike CD8⁺ T cells did not change (Fig. 1I). In addition, we found that CD69 was expressed on both AH1-specific stemlike and terminally differentiated CD8⁺ T cells (Supplementary Fig. S2E). These data indicate that the number of tumor-specific terminally differentiated CD8⁺ T cells is increased in Cd69^KO mice, whereas the proportion of stemlike CD8⁺ T cells is comparable between Cd69^WT and Cd69^KO mice.

To investigate the mechanism behind the impact of CD69 deficiency on tumor-specific CD8⁺ T cells, we next assessed the gene expression profiles of individual AH1-specific, Ly108⁺ stemlike and Tim-3⁺ terminally differentiated CD8⁺ T-cells isolated from Cd69^WT and Cd69^KO mice (Supplementary Fig. S3A and S3B). Data were first compared with the datasets of antigen-specific CD8⁺ T-cell subpopulations (i.e., stemlike and terminally differentiated CD8⁺ T cells) obtained from B16-OVA melanoma-bearing mice and LCMV clone-13-infected mice, which were previously reported (1). A gene set enrichment analysis (GSEA) demonstrated significant transcriptional similarity between the corresponding subpopulations in both Cd69^WT and Cd69^KO mice (Supplementary Fig. S3C and S3D; Supplementary Table S1). However, the Cd69^WT and Cd69^KO CD8⁺ T-cell subpopulations exhibited a distinct transcriptome profile (Fig. 1J and K). Furthermore, Cd69^KO Ly108⁺ stemlike CD8⁺ T-cells showed a gene expression profile with more signatures of terminally differentiated CD8⁺ T cells than Cd69^WT Ly108⁺ stemlike CD8⁺ T cells (Fig. 1L and M). These data raise the possibility that CD69 suppresses the transcriptional program toward terminal differentiation in stemlike CD8⁺ T cells.

To test whether CD69 regulated the transcriptional program for differentiation of terminally differentiated CD8⁺ T cells, gene expression profiles of individual Ly108⁺ stemlike and Tim-3⁺ terminally differentiated CD8⁺ T cells were analyzed, and several differentially expressed genes between Cd69^WT and Cd69^KO mice were identified (Fig. 2A and B; Supplementary Fig. S4A and S4B; Supplementary Tables S2–S3). Three genes encoding transcription factors (i.e., Tox, Crem, and Irf8) were downregulated, whereas one gene (i.e., Satb1) was upregulated in both stemlike and terminally differentiated Cd69^KO CD8⁺ T cells (Supplementary Fig. S4A and S4B; Supplementary Tables S2–S3). Among them, we focused on Tox because TOX has been reported to be important for the CD8⁺ T-cell function under conditions associated with chronic antigen stimulation, such as cancer or chronic viral infection (29, 32–34), and to be crucial for maintaining Tcf1⁺ stemlike CD8⁺ T cells (35). qRT-PCR (Supplementary Fig. S4C) and flow cytometry (Fig. 2C) confirmed low expression of TOX mRNA and protein in both Ly108⁺ stemlike and Tim-3⁺ terminally differentiated CD8⁺ T cells in the TME of Cd69^KO mice. We also found that both stemlike and terminally differentiated CD8⁺ T cells in Cd69^KO mice were significantly enriched in gene signatures from the corresponding subpopulations in TOX-deficient mice infected with LCMV clone-13 (Supplementary Fig. S4D and S4E; ref. 29). Furthermore, expression of inhibitory receptors, such as PD-1, Lag-3, and TIGIT, known as downstream molecules of TOX (29, 32, 33, 34), was significantly diminished in intratumoral CD8⁺ T-cell subpopulations in Cd69^KO mice (Fig. 2D). These data demonstrate that the TOX-mediated transcriptional program is repressed in intratumoral CD8⁺ T cells in Cd69^KO mice.

It has been reported that TDLNs play a pivotal role in initiating antitumor CD8⁺ T-cell responses (6–9) and that the expression of TOX can be induced by T-cell receptor (TCR) stimulation (29, 32, 33). These facts prompted us to investigate the TOX levels in tumor-specific CD8⁺ T cells in TDLNs (Fig. 2E; Supplementary Fig. S5A). We first found that the number of AH1-specific CD8⁺ T cells in TDLNs from Cd69^KO mice was significantly higher compared with Cd69^WT mice, whereas AH1-specific CD8⁺ T cells were nearly absent in the non-TDLNs in both Cd69^WT and Cd69^KO mice (Fig. 2F). Expression of TOX and PD-1 in AH1 tetramer⁺ tumor-specific CD8⁺ T cells in TDLNs was significantly decreased in Cd69^KO mice (Fig. 2G), whereas the expression of Lag-3 and TIGIT were very low on CD8⁺ T cells, even in Cd69^WT mice, and there were no significant differences between Cd69^WT and Cd69^KO mice (Supplementary Fig. S5B). In addition, we confirmed that CD69 was expressed on AH1-specific CD8⁺ T cells in TDLNs (Supplementary Fig. S5C). These results demonstrate that CD69 deficiency diminishes the induction of TOX in TDLNs.

Because TOX expression in Cd69^KO CD8⁺ T cells was already diminished in TDLNs led us to examine the characteristics of tumor-specific CD8⁺ T cells in TDLNs. We performed an unbiased analysis of AH1-specific CD8⁺ T cells in TDLNs by conducting scRNA-seq and analyzed pooled scRNA-seq data from Cd69^WT and Cd69^KO mice (Fig. 3A). Five clusters were identified on the basis of their transcriptomes (Fig. 3B), and the top 10 differentially expressed genes across the clusters were examined (Supplementary Fig. S6A and S6B). Cells in both clusters 0 and 1 expressed high Tcf7 and Slamf6, markers for the stemlike CD8⁺ T-cell population, indicating that there were two types of stemlike CD8⁺ T-cell subsets. Cells in cluster 2 expressed Gzmb and Havcr2, representing the terminally differentiated CD8⁺ T-cell population. Cells in cluster 3 represented the proliferating cell population because they exhibited the expression of proliferation markers (e.g., Birc5 and Mcm6), which is consistent with the cell-cycle scoring (Supplementary Fig. S6C). Cells in cluster 4 expressed Cd69, Ccr9, and Sell (encoding CD62L), representing that the early-activated T-cell population still maintained some naïve T-cell signatures. Importantly, the proportion of Cd69^KO cells in each cluster was altered compared to Cd69^WT cells; more terminally differentiated CD8⁺ T cells and fewer stemlike CD8⁺ T cells were detected in Cd69^KO mice than in Cd69^WT mice (Fig. 3C). Flow cytometry confirmed that AH1-specific, Tim-3⁺ terminally differentiated CD8⁺ T cells were present and that the frequency of AH1-specific, Tim-3⁺ terminally differentiated CD8⁺ T cells was significantly increased in Cd69^KO mice (Fig. 3D). A time-course analysis revealed that Tim-3⁺ terminally differentiated CD8⁺ T cells were barely detectable eight days after tumor inoculation, although TOX expression was already decreased in Cd69^KO mice, suggesting that the enhanced generation of Tim-3⁺ terminally differentiated CD8⁺ T cells was preceded by TOX downregulation in TDLNs (Supplementary Fig. S6D–S6F). Furthermore, we found that the downregulation of TOX was observed in both Eomes⁺ stemlike 1 and CD94⁺ stemlike 2 CD8⁺ T cells, as well as in Tim-3⁺ terminally differentiated CD8⁺ T cells in Cd69^KO mice (Fig. 3E; Supplementary Fig. S5A), suggesting that CD69 controls the induction of TOX in the early-phases of tumor-specific CD8⁺ T-cell differentiation in TDLNs.

We further confirmed this using MC38 colorectal carcinoma cells expressing ovalbumin (MC38-OVA) by detecting the tumor-specific CD8⁺ T cells using H-2K^b-OVA tetramer (Supplementary Fig. S7A and S7B). OVA-specific CD8⁺ T-cell number in TDLNs from Cd69^KO mice was significantly higher than that in Cd69^WT mice (Supplementary Fig. S7C). CD69 deficiency increased the frequency of Tim-3⁺ terminally differentiated CD8⁺ T cells and decreased the frequency of Tcf1⁺ stemlike CD8⁺ T cells within TDLNs (Supplementary Fig. S7D). Moreover, reduced TOX expression was observed in both CD8⁺ T-cell subpopulations in Cd69^KO mice (Supplementary Fig. S7E).

To determine the cell-intrinsic role of CD69 in CD8⁺ T-cell differentiation, we inoculated MC38-OVA into wild-type (WT) mice that had been previously treated with equal numbers of Cd69^WT and Cd69^KO naïve CD8⁺ T cells from OT-I TCR transgenic mice expressing OVA-specific TCR. We then analyzed the infused CD8⁺ T cells in TDLNs at the indicated times (Fig. 3F). Even in the same TDLN, the frequency of Cd69^KO Tcf1⁺ stemlike OT-I CD8⁺ T cells was significantly decreased eight days after tumor inoculation, whereas the frequency of Cd69^KO Tim-3⁺ terminally differentiated OT-I CD8⁺ T cells was inversely increased compared to Cd69^WT cells (Fig. 3G). OVA-specific CD8⁺ T cells were likely more rapidly activated than AH1-specific CD8⁺ T cells in TDLNs (Supplementary Fig. S6E). Furthermore, the TOX expression in Cd69^KO PD-1⁺ OT-I CD8⁺ T cells was also significantly decreased 8 days after tumor inoculation (Fig. 3H). These data indicate that CD69 regulates the induction of TOX in a cell-intrinsic manner, thereby controlling tumor-specific CD8⁺ T-cell differentiation in TDLNs.

We next investigated whether CD8⁺ T cells in TDLNs from patients with cancer expressed CD69 and TOX. To this end, we utilized a published scRNA-seq dataset from a study that included tumor-containing lung tissue (tLung) and nonmetastatic lung-draining lymph nodes (nLNs) collected from treatment-naïve patients with lung adenocarcinoma (30). CD8⁺ T cells from both tissues were grouped into four clusters (terminally differentiated, stemlike, effector, and naïve) and visualized by UMAP (Supplementary Fig. S8A and S8B). Importantly, both stemlike and terminally differentiated CD8⁺ T cells were present in nLNs of patients with cancer and expressed substantial levels of CD69 and TOX (Supplementary Fig. S8C and S8D).

Because dendritic cells (DC) play a central role in the priming of tumor-specific CD8⁺ T cells and because CD69 is known to control antigen presentation from monocytes (36, 37), we next tested whether CD69 played a role in the major DC populations in the TME and TDLNs (Supplementary Fig. S9A and S9B). We first observed that CD69 was expressed in all major DC populations in the TME and that the highest expression of CD69 was detected in monocyte-derived DCs (mo-DC; Supplementary Fig. S9C). In contrast, relatively low CD69 expression was observed in conventional DC1 (cDC1) and conventional DC2 (cDC2) in TDLNs (Supplementary Fig. S9C), and mo-DCs were not detected in TDLNs (Supplementary Fig. S9E and S9G).

We next evaluated whether CD69 deficiency influenced DCs (Supplementary Fig. S9D and S9E). The frequency of mo-DCs in CD45⁺ cells in the TME was increased in Cd69^KO mice, suggesting that CD69 may control the number of mo-DCs (Supplementary Fig. S9D). Because the frequency of mo-DCs in TDLNs was very low in both Cd69^WT and Cd69^KO mice (Supplementary Fig. S9E), this increase in mo-DCs in Cd69^KO mice may have occurred within the TME. We observed that the number of cDC1 and cDC2 was not changed between Cd69^WT and Cd69^KO mice in both the TME and TDLNs (Supplementary Fig. S9D and S9E), and no significant differences in the expression of costimulatory molecules, such as CD80 and CD86, were observed between Cd69^WT and Cd69^KO mice in DCs in the TME or TDLNs (Supplementary Fig. S9D and S9E). We next evaluated the effects of anti-CD69 administration on DCs in the TME and TDLNs. No significant differences in the frequency or phenotype of DCs were observed between anti-CD69–treated mice and control IgG-treated mice (Supplementary Fig. S9F and S9G).

These results suggest that CD69 deficiency and anti-CD69 treatment do not impact cDC1 and cDC2 in the TME and TDLNs. Because cDC1 and cDC2 are known to be major DC subsets activating tumor-specific CD8⁺ T cells, we conclude that antitumor DC function in Cd69^KO mice was nearly intact.

We next asker how CD69 deficiency resulted in the low expression of TOX in TDLNs. We hypothesized that CD69 deficiency induced the surface expression of S1P₁ so that T-cells leave early from the TDLN before upregulating the high levels of TOX, thus resulting in the low expression of TOX and efficient differentiation to terminally differentiated CD8⁺ T cells in the TME. However, this was unlikely based on the data. First, the frequency of AH1 tetramer⁺ CD8⁺ T cells was high, rather than low, in CD69^KO mice (Fig. 2F), which was contrary to our expectation. Second, the frequency of Tim-3⁺ terminally differentiated CD8⁺ T cells was already increased in TDLNs in Cd69^KO mice (Fig. 3C and D). Nevertheless, we administered an S1P receptor agonist, FTY720, during tumor development to examine the impact of the abrogation of T-cell exit from lymphoid organs (ref. 11; Fig. 4A). We first confirmed that the administration of FTY720 resulted in an increase of AH1-specific CD8⁺ T cells in TDLNs in both Cd69^WT and Cd69^KO mice (Fig. 4B), indicating that exit from the TDLNs required S1P₁, as previously reported (8, 9). More Cd69^KO cells egressed from the TDLNs than Cd69^WT cells, as evidenced by cell accumulation in TDLNs with FTY720 treatment, which was more dramatic in Cd69^KO mice than in Cd69^WT mice. Nevertheless, the expression of TOX was again low in both stemlike and terminally differentiated CD8⁺ T cells in Cd69^KO mice, even after FTY720 treatment (Fig. 4C), and the frequency of Tim-3⁺ terminally differentiated CD8⁺ T cells in Cd69^KO mice was significantly higher than that in Cd69^WT mice (Fig. 4D). These data demonstrate that the low expression of TOX in Cd69^KO CD8⁺ T cells is not due to early exit from the TDLNs, and rather it is because of the loss of cell-intrinsic regulation by CD69.

It has been reported that the induction of TOX in CD8⁺ T-cells requires calcineurin/NFAT signaling (29, 32–35). We observed that NFAT2 (Nfatc1), but not NFAT1 (Nfatc2), was substantially expressed in AH1-specific CD8⁺ T cells in TDLNs (Supplementary Fig. S10A), which is consistent with the notion that NFAT2 mainly controls TOX expression in antigen-specific CD8⁺ T cells (32). Furthermore, nuclear NFAT2 was found to be decreased in PD-1⁺ CD8⁺ T cells in TDLNs in Cd69^KO mice (Fig. 4E and F). We also confirmed that treatment with Cyclosporin A, an inhibitor of the calcineurin/NFAT pathway, decreased the TOX expression in tumor-specific CD69⁺CD8⁺ T cells in TDLNs (Supplementary Fig. S10B), suggesting that in vivo calcineurin/NFAT signaling indeed required TOX expression. We next examined whether CD69 deficiency influenced calcineurin/NFAT signaling using single-sample gene set variation analysis (ssGSVA). Constitutive active NFAT1 lacking AP-1-binding sites (CA-RIT-NFAT1) reportedly promotes Tox expression in CD8⁺ T cells (26, 34). Therefore, we used the gene set related to CA-RIT-NFAT1 (26) and found that it was depleted in Cd69^KO AH1-specific CD8⁺ T cells in TDLNs (Supplementary Fig. S10C).

These data suggest that CD69-deficient CD8⁺ T cells reduce NFAT-dependent transactivation and leads to diminished TOX expression. We conclude that CD69 deficiency attenuates the expression of TOX in tumor-specific CD8⁺ T cells in TDLNs, most likely through an NFAT-dependent pathway, and thereby promotes the initiation of CD8⁺ T-cell differentiation in TDLNs, leading to enhanced antitumor immune responses.

With the aim of applying anti-CD69 and anti–PD-1 in the clinical setting, we first examined the effect of anti-CD69 administration on tumor-specific CD8⁺ T cells (Supplementary Fig. S11A). We confirmed that anti-CD69 resulted in reduced TOX expression inAH1-specific CD8⁺ T cells in TDLNs (Supplementary Fig. S11B). Furthermore, the generation of Tim-3⁺ terminally differentiated CD8⁺ T cells was enhanced by anti-CD69 treatment (Supplementary Fig. S11C). These results suggest that anti-CD69 treatment and CD69 deficiency exert similar effects on tumor-specific CD8⁺ T cells. We therefore next examined the differentiation of intratumoral CD8⁺ T cells (Fig. 5A). We pooled scRNA-seq data from intratumoral AH1-specific CD8⁺ T cells in mice treated with the indicated antibodies (Fig. 5B; Supplementary Fig. S11D–S11F). Anti-CD69 treatment increased Tim-3⁺ terminally differentiated CD8⁺ T cells, whereas anti–PD-1 treatment mainly increased Tcf1⁺ stemlike CD8⁺ T cells, which is consistent with the notion that anti–PD-1 therapy preferentially expands the stemlike subpopulation (refs. 1, 2; Fig. 5B). The combined use of anti-CD69 and anti–PD-1 increased both stemlike and terminally differentiated subpopulations (Fig. 5B). As a result, combined treatment with anti-CD69 and anti–PD-1 had an efficient therapeutic effect, whereas anti-CD69 or anti–PD-1 administration alone showed significant retardation of tumor growth (Fig. 5C). We also confirmed that anti-CD69 treatment resulted in the decreased expression of Tox in AH1-specific CD8⁺ T cells in the TME (Supplementary Fig. S11G).

We further examined the therapeutic effect of anti-CD69 administration on the immunorefractory B16 melanoma tumor model (38). To monitor tumor-specific CD8⁺ T cells, we utilized B16-OVA cells so that the tumor-specific CD8⁺ T cells could be detected by H-2K^b-OVA tetramer (Supplementary Fig. S11H). Treatment with either anti-CD69 or anti–PD-1 administration alone was not associated with retardation of tumor growth, whereas combined treatment with anti-CD69 and anti–PD-1 had an enhanced therapeutic effect on melanoma (Fig. 5D).

To investigate the mechanism of enhanced antitumor immunity by combination treatment, we analyzed tumor-specific CD8⁺ T cells in both TDLNs and the TME (Fig. 5E). A flow cytometry analysis showed that the combined use of anti-CD69 and anti–PD-1 led to an increase in the number of OVA-specific CD8⁺ T cells in TDLNs (Fig. 5F). Consistent with this result, the number of tumor-specific CD8⁺ T cells in the TME was significantly increased by combination treatment (Fig. 5G). The frequency and number of Tim-3⁺ terminally differentiated CD8⁺ T cells were increased in mice treated with anti-CD69 and anti–PD-1 (Fig. 5H). The number of Tcf1⁺ stemlike CD8⁺ T cells did not change significantly, although the frequency of Tcf1⁺ stemlike CD8⁺ T cells was decreased by combination treatment (Fig. 5H). The frequency and number of granzyme B-expressing cells were also significantly increased by combination treatment (Fig. 5I). Taken together, these data demonstrate that anti-CD69 therapy promotes the generation of terminally differentiated CD8⁺ T cells and enhances antitumor immune responses that synergize with anti–PD-1 therapy.

The present study significantly enhances our understanding of how tumor-specific CD8⁺ T-cells differentiate into functional, Tim-3⁺ terminally differentiated CD8⁺ T cells. Our single-cell transcriptomics defined CD69 as an important regulator for the differentiation of tumor-specific CD8⁺ T cells. CD69 deficiency resulted in decreased expression of the transcription factor TOX in tumor-specific CD8⁺ T cells not only in the TME but also in TDLNs. Importantly, CD69-deficient mice showed enhanced antitumor immunity, with a higher number of functional Tim-3⁺ terminally differentiated CD8⁺ T cells in the TME. Furthermore, anti-CD69 treatment enhanced the generation of Tim-3⁺ terminally differentiated CD8⁺ T cells with enhanced antitumor function and also synergized with anti–PD-1 treatment to render immunorefractory B16 melanoma susceptible to immunotherapy.

An important finding in this study is that CD69 controls TOX expression in tumor-specific CD8⁺ T cells. Our data showed that CD69-deficient tumor-specific CD8⁺ T cells had reduced levels of TOX and enhanced antitumor immunity. Interestingly, Khan, and colleagues shows that tumor-specific CD8⁺ T cells with heterozygous deletion in the Tox gene have enhanced antitumor effects compared with WT CD8⁺ T cells that recognize same antigen (32). This data suggests that partial downregulation of TOX has a beneficial effect on antitumor immunity, although complete deletion of the Tox gene leads to a defect in the persistence of antigen-specific CD8⁺ T cells (29, 32–35). Furthermore, TOX is known to be critical for the maintenance of Tcf1⁺ stemlike CD8⁺ T cells, therefore, reduced TOX expression may render Tcf1⁺ stemlike CD8⁺ T cells susceptible to differentiation. Indeed, we observed that the generation of Tim-3⁺ terminally differentiated CD8⁺ T cells was increased in Cd69^KO mice, which may account for the augmentation of antitumor immunity. Although the molecular mechanism underlying CD69-mediated TOX induction remains unclear, one possibility is that CD69 contributes to signaling through the TCR because nuclear NFAT2 was found to be decreased in Cd69^KO mice. Consistent with this result, our ssGSVA suggested that CD69-deficient CD8⁺ T cells had reduced NFAT-dependent transactivation in TDLNs. Nevertheless, further study is needed to elucidate the detailed molecular mechanisms.

CD69 has a large extracellular domain that is said to interact with ligand molecules (10). Indeed, Myl9/12, Galectin-1, S100A8/A9, and oxidized low-density lipoprotein have been identified as potential CD69 ligands (13, 39–41). However, their contribution to the CD69 function in antitumor immunity remains unknown. The fact that anti-CD69 exerts antitumor effect suggests the involvement of the ligand-receptor interaction. Because we observed the phenotype of CD69 deficiency in tumor-specific CD8⁺ T cells in TDLNs, we think that some CD69 ligands may be supplied in cells localized in TDLNs. Nevertheless, screening of known and unknown ligands that regulate the CD69-mediated induction of TOX should be performed in the future, which may reveal the detailed molecular mechanisms underlying the CD69 function.

Another important finding in this study is that Tim-3⁺ terminally differentiated CD8⁺ T cells could be generated within TDLNs and that CD69 regulated this differentiation process by controlling TOX expression, which contrasts with recent studies reporting that Tim-3⁺ terminally differentiated CD8⁺ T cells are generated from Tcf1⁺ stemlike CD8⁺ T cells within the TME (1, 2, 4, 5). Our data support two nonmutually exclusive models of the generation of intratumoral Tim-3⁺ terminally differentiated CD8⁺ T cells: Tim-3⁺ terminally differentiated CD8⁺ T cells generated from Tcf1⁺ stemlike CD8⁺ T cells in TDLNs directly migrate into the TME and Tcf1⁺ stemlike CD8⁺ T cells in TDLNs directly migrate into the TME and then efficiently give rise to Tim-3⁺ terminally differentiated CD8⁺ T cells. In both cases, the fact that CD69 deficiency attenuated the expression of TOX in tumor-specific CD8⁺ T cells in TDLNs suggests that CD69 imposed tumor-specific CD8⁺ T-cell fate in TDLNs by regulating TOX expression.

One of the most significant findings of this study is that anti-CD69 treatment increased the generation of Tim-3⁺ terminally differentiated CD8⁺ T cells and thereby enhanced antitumor immune responses. Because anti–PD-1 therapy preferentially increased the frequency of Tcf1⁺ stemlike CD8⁺ T cells (1, 2), the combination therapies of anti-CD69 and anti–PD-1 are compatible and showed efficient therapeutic effect even on immunorefractory B16 melanoma. Because CD69 is expressed on not only T cells but also almost every leukocyte, side effects from anti-CD69 administration may be unavoidable (10). However, mice lacking whole-body CD69 are healthy and show no severe defects. We therefore predict that anti-CD69–associated adverse events will be tolerable. In conclusion, we showed that CD69 plays a crucial role in antitumor immunity by controlling CD8⁺ T-cell differentiation. Because CD69 is frequently expressed on the surface of CD8⁺ T cells in TDLNs in human cancers (14), our study provides the rationale for clinical evaluation of humanized monoclonal antibody to human CD69 as a new and feasible strategy for cancer immunotherapy.

A. Onodera reports grants from Japan Society for the Promotion of Science (JPSP) during the conduct of the study; grants from JSPS and grants from Astellas Foundation for Research on Metabolic Disorders (AFRMD) outside the submitted work. No disclosures were reported by the other authors.

R. Koyama-Nasu: Conceptualization, data curation, supervision, funding acquisition, investigation, writing–review, and editing. M.Y. Kimura: Conceptualization, data curation, supervision, funding acquisition, investigation, writing–review, and editing. M. Kiuchi: Investigation. A. Aoki: Investigation. Y. Wang: Investigation. Y. Mita: Investigation. I. Hasegawa: Investigation. Y. Endo: Investigation. A. Onodera: Investigation. K. Hirahara: Investigation. S. Motohashi: Investigation. T. Nakayama: Supervision.

This work was supported by the following grants: Ministry of Education, Culture, Sports, Science, and Technology (MEXT Japan) Grants-in-Aid for Scientific Research (S) 19H05650, (B) 20H03464, 18H02892, (C) 21K07234, 20K07652, 18K07257, Challenging Research (Pioneering) 18H05375, (Exploratory) 20K21537, 18K19466, Innovative Areas 19H04800; Grants-in-Aid for Transformative Research Areas (A) JP22H05189; Project for Cancer Research and Therapeutic Evolution (P-CREATE) from the Japan Agency for Medical Research and development (AMED) JP18cm0106339, JP20cm01106372; AMED-PRIME JP21gm6310024; AMED JP223fa627003. We thank Drs. Alfred Singer and Damon Tumes for their critical reading of the manuscript; Dr. T. Nakatsura (National Cancer Center Japan) for the B16-OVA cells; the NIH Tetramer facility for the tetramer reagents; and K. Sugaya for their expert aid in flow cytometry.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.