The accumulation of tumor-specific CD4+ and CD8+ effector T cells is key to an effective antitumor response. Locally, CD4+ T cells promote the recruitment and effector function of tumor-specific CD8+ T cells and activate innate killer cells in the tumor. Here, we show that tumor-specific CD4+ T cells were predominantly present in the CD39+ subset of tumor-infiltrating lymphocytes (TIL). The CD39+ CD4+ and CD8+ TILs were detected in three different tumor types, and displayed an activated (PD-1+, HLA-DR+) effector memory phenotype. CD4+CD39+ single-cell RNA-sequenced TILs shared similar well-known activation, tissue residency, and effector cell–associated genes with CD8+CD39+CD103+ TILs. Finally, analysis of directly ex vivo cell-sorted and in vitro expanded pure populations of CD39-positive and negative CD4+ and CD8+ TILs revealed that tumor-specific antigen reactivity was almost exclusively detected among CD39+ cells. Immunotherapy of cancer is based on the activation of tumor-reactive CD4+ and CD8+ T cells. We show that the expression of CD39 can be used to identify, isolate, and expand tumor-reactive T-cell populations in cancers.

In many cancers, the concerted action of high numbers of effector memory cytotoxic CD8+ T cells and type-I CD4+ T cells associates with better prognosis (1, 2) and response to immunotherapy (3). CD4+ T cells can activate tumor-infiltrating tumoricidal eosinophils, macrophages, and natural killer (NK) cells (4–7) and promote the recruitment and cytolytic function of CD8+ T cells (8). In the tumor microenvironment (TME), CD8+ T cells can specifically recognize and kill tumor cells (9). IFNγ produced by both subpopulations of T cells is known to enhance MHC class I and II antigen presentation in tumor cells, thereby making them more vulnerable to CD4+ and CD8+ T-cell attack. Furthermore, IFNγ exerts direct antiproliferative and proapoptotic antitumor effects (10).

Not all intratumoral T cells respond to tumor antigens (11, 12), because ongoing inflammation also attracts so-called activated bystander T cells, which recognize antigens unrelated to cancer (11). A more accurate identification of the actual tumor-reactive T-cell population would allow for better interpretation of the potential interactions within the TME, but it may also help study current and novel immunotherapies. For instance, this could aid in higher percentages yields of tumor-specific T cells for adoptive cell transfer approaches, either by direct expansion and infusion of isolated T cells or by using gene-modified T cells expressing the T-cell receptor (TCR) of the isolated tumor-reactive T cells. A number of groups isolated tumor-reactive T cells directly from tumor-infiltrating lymphocytes (TIL) based on the transient expression of CD137 from recently activated T cells (13–15) or on the expression of PD-1 (16) as this was predominantly present on highly expanded TCRβ clonotypes (17). The expression of PD-1 and many other coinhibitory, costimulatory, and activation markers is not exclusive for tumor-reactive T cells among T cells in the TME (13, 18) and therefore, not directly useful for the isolation of tumor-reactive T cells. The use of enhanced phenotype definition may overcome such limitations (19). CD39 is proposed as a marker to identify tumor-reactive CD8+ T cells (11, 18, 20). CD39 expression is detected on CD8+ T cells with hallmarks of chronic antigenic stimulation at the tumor site (11) and correlate with clinical beneficial parameters in colorectal cancer, lung cancer, and head and neck cancer (11, 18).

Tumor-reactive CD4+ T cells are required for the control of tumor progression in mice (2, 4, 21) and mediate direct effects in patients with cancer (22–24). CD39 is an ectonucleotidase that becomes expressed at the cell surface of not only activated CD8+ T cells but also on activated CD4+ T cells, and it is retained long after other markers of activation (e.g., CD25, HLA-DR) have perished (25). As such, the possibility exists that CD39 may also identify tumor-reactive CD4+ TILs. To answer this question, we studied their presence in squamous cell carcinomas of the vulva, oropharynx, and cervix. This provided us with the advantage that a substantial fraction of these tumors were induced by high-risk human papillomavirus (HPV), which expresses the known tumor-specific proteins E6 and E7, and were infiltrated by CD4+ and CD8+ TILs responding to these antigens (26, 27). The detection of HPV-reactive CD4+ T cells among TIL associates with significantly improved clinical benefit in these cancers (26). Here, we showed that a significant fraction of the CD4+ and CD8+ TILs expressed CD39. The CD4+CD39+ cells within these TILs exhibited a tissue-resident/activation-like phenotype similar to that of CD8+CD39+CD103+ TILs and the CD39+ TILs almost exclusively comprised the tumor-specific T-cell fraction. These results provided a second rationale to focus on CD39 as a marker for the identification and isolation of tumor-specific T cells and may aid in the development of effective T-cell–based immunotherapies.

Patients

In total, 63 patients [13 cervical squamous cell carcinoma (CxCa), 18 vulvar squamous cell carcinoma (VSCC), and 32 oropharyngeal squamous cell carcinoma (OPSCC)] were included in this study as part of two larger observational studies focusing on circulating and local immune responses in either anogenital lesions (P08.197) or head and neck cancer (P07.112; ref. 26). Patients were included when histopathology confirmed the presence of a carcinoma and after signing informed consent. The study was conducted in accordance with the Declaration of Helsinki and approved by the local medical ethical committee of Leiden University Medical Center and in agreement with the Dutch law. Blood and tumor tissue was collected and these materials were used according to the Dutch Federal of Medical Research Association guidelines. All patients received standard-of-care treatment. HPV typing and p16ink4a-IHC was performed on formalin-fixed paraffin-embedded (FFPE) material as described previously (27).

Peripheral blood mononuclear cells

Venous (60 mL) blood samples were drawn and collected in sodium-heparin collection tubes (Vacuette; Greiner, Alphen a/d Rijn, the Netherlands) and 8 mL in a clotting tube (Vacuette). Peripheral blood mononuclear cells (PBMC) were isolated by Ficoll density-gradient centrifugation of the heparinized blood, washed in PBS (B. Braun Melsungen), resuspended in cold FCS (PAA Laboratories), stored on ice for 15 minutes before drop-wise the freezing medium [80% FCS and 20% dimethyl sulfoxide (DMSO; Sigma)] was added to cryopreserve the PBMCs using a controlled freezing machine (IceCube 1810, Cryo Solutions) according to standard operating procedures. PBMCs were stored in equal aliquots at 1 × 107 cells per vial in the vapor phase of liquid nitrogen until use. The blood in clotting tube was taken along in the centrifugation step during the Ficoll step after which the serum could be harvested and aliquoted (1 mL/vial) and stored at −20°C until use.

Single-cell tumor cell digests

Tumor material was obtained after surgery and handled as described previously (26). The tumor was cut in small pieces, and subsequently incubated for 15 minutes at 37°C in IMDM dissociation mixture containing 10% human AB serum, gentamycin (50 μg/mL), fungizone (25 μg/mL), penicillin (100 IU/mL), streptomycin (100 μg/mL), and 0.38 mg/mL of Liberase (Liberase TL, research grade, Roche). After 15 minutes, the cell suspension was flushed through a 70-μm cell strainer (Falcon) to acquire a single-cell suspension, counted using Trypan blue exclusion (Sigma), and cryopreserved at approximately 2 million cells/vial. All cells were stored in the vapor phase of liquid nitrogen until further use.

Flow cytometry of ex vivo tumors

Cryopreserved single-cell tumor samples (n = 30; 10 CxCa, 9 VSCC, and 11 OPSCC) were thawed and assessed by flow cytometry as described previously (28). Briefly dead cells were identified using the LIVE-DEAD Fixable Yellow Dead Cell Stain Kit (Thermo Fisher Scientific) for 20 minutes at room temperature. After incubation, cells were washed and incubated with PBS/0.5%BSA/10%FCS for 15 minutes on ice for FC receptor blockade. Following washing, the cells were stained for 30 minutes on ice in the dark with fluorochrome-conjugated antibodies (Supplementary Table S1). Acquisition of stained cells was performed with a BD LSR Fortessa. Data was analyzed by high-dimensional single-cell data analysis using Hierarchical Stochastical Neighbor Embedding (HSNE) in Cytosplore (29), by clustering using a self-organizing map (FlowSOM) in Cytobank or by manual gating using DIVA software (version 8.02; BD Biosciences).

Single-cell RNA sequencing and data analysis

Live, CD3+ T cells were isolated from thawed single-cell tumor samples using the Dead Cell Removal Kit (Miltenyi Biotec; #130-090-101) followed by CD3-guided magnetic cell sorting using CD3-microbeads (Miltenyi Biotec; #130-050-101) according to manufacturer's instructions. Postisolation viability was >70%. Single-cell RNA-sequencing was performed as described previously with some modifications (30). Briefly, between 2,108 and 6,107 sorted cells per sample were loaded on a Chromium Single Cell Controller (10x Genomics). Lysis and barcoded reverse transcription of polyadenylated mRNA from single cells were performed inside each gel bead emulsion using the 5′ gene expression pipeline. Next-generation sequencing libraries were prepared in a single bulk reaction, and transcripts were sequenced using a HiSeq4000 System (Illumina). Single-cell transcriptome sequencing data were preprocessed with cell ranger v3.0.0 (10x Genomics) using the GRCh38 reference genome. Downstream analysis was performed using scanpy v1.4.4 (31) following the best practice recommendations (32). All scripts used to perform the analysis are wrapped into a fully reproducible Nextflow pipeline [10.1038/nbt.3820] and publicly available from GitHub (https://github.com/icbi-lab/kortekaas2020_paper). The data analysis reports, including the full list of differentially expressed genes, are available on https://icbi-lab.github.io/kortekaas2020_paper). In brief, low quality cells were excluded by retaining only cells with (i) >700 detected genes, (ii) >200 detected reads, and (iii) <11% mitochondrial reads. In silico doublet detection was performed with scrublet v0.2.1 (33). Finally, the 3,000 most highly variable genes (HVG) were selected and clustered unsupervised using the Leiden algorithm (34). Embeddings were visualized using UMAP (35). Differential gene expression between clusters was computed using edgeR (36). The whole single-cell RNA-sequencing dataset is available upon reasonable written request without restrictions.

Cell sorting and T-cell expansion

Cryopreserved single-cell tumor samples were thawed, and stained for dead cells and surface markers CD4, CD8, CD39, and CD103 (Supplementary Table S1) as described above. Antibodies to CD3 were not used for cell sorting (18). CD4+CD39, CD4+CD39+, CD8+CD39CD103+, and CD8+CD39+CD103+ populations were sorted using a BD FACSAria III. When sufficient number of cells were obtained, a purity check was performed.

Next, the sorted cells were cultured in a 96-well plate (20,000 TILs/well), with irradiated (3500 RAD, Gammacell 1000 Elite) PBMCs of 5 different donors (100,000 cells/96 well total) and irradiated (7500 RAD) Epstein–Barr virus–transformed B-lymphoblast cell lines (established >10 years ago, cultured for maximal 9 weeks, and not authenticated in the past year, Mycoplasma tests were performed every month for these cell lines by PCR and found negative) of 3 different donors (10,000 cells/96 well total) in IMDM (Gibco by Life Technologies, Thermo Fisher Scientific, Lonza) with 10% FCS (PAA Laboratories) and 100 IU/mL penicillin, 100 μg/mL streptomycin, 2 mmol/L l-glutamine (Gibco by Life Technologies, Thermo Fisher Scientific) supplemented with 10% TCG-F (ZeptoMetrix Corporation), 5 ng/mL human IL7 (PeproTech), 5 ng/mL human IL15 (Invitrogen by Thermo Fisher Scientific), 5 ng/mL human IL21 (generously provided by AIMM Therapeutics), and anti-CD3/CD28 Dynabeads (Gibco by Life Technologies, Thermo Fisher Scientific) at a PBMC:bead ratio of 3:1. Every 2 to 3 days, the cultures were replenished with IMDM complete medium supplemented with 10% TCG-F, 5 ng/mL of IL7, IL15, and IL21. After 3 weeks, the cultures were harvested, anti-CD3/CD28 Dynabeads were removed, and cells were counted using Trypan blue exclusion. Per condition, 400,000 cells were stimulated for a second expansion round as described above. The remaining cells were stored in the vapor phase of liquid nitrogen until further use (expansion round 1). Three weeks later, the cultures were harvested, Dynabeads were removed, and cells were stored in the vapor phase of liquid nitrogen until further use (expansion round 2).

Tumor-specific T-cell reactivity analysis

The presence of HPV-specific T cells was tested by assessment of antigen-specific cytokine production as described previously (26). T-cell responses against autologous HPV16 or HPV18 E6/E7 synthetic long peptide (SLP; 22-mers with 14 amino acids overlap; produced at the peptide facility of the LUMC, Leiden, the Netherlands)–loaded monocytes were tested in triplicate. The following peptides were used for HPV16 E6: 1–22 MHQKRTAMFQDPQERPRKLPQL; 11–32 DPQERPRKLPQLCTELQTTIHD; 21–42 QLCTELQTTIHDIILECVYCKQ; 31–52 HDIILECVYCKQQLLRREVYDF; 41–61 KQQLLRREVYDFAFRD LCIVYR; 51–72 DFAFRDLCIVYRDGNPYAVCDK; 61–82 YRDGNPYAVCDKCLKFYSKISE; 71–92 DKCLKFYSK ISEYRHYCYSLYG; 81–102 SEYRHYCYSLYGTTLEQQYNKP; 91–112 YGTTLEQQYNKPLCDLLIRCIN; 101–122 KPLCDLLIRCINCQKPLCPEEK; 111–132 INCQKPLCPEEKQRHL DKKQRF; 121–142 EKQRHLDKKQRFHNIR GRWTGR; 131–152 RFHNIRGRWTGRCMSCCRSSRT; 137–158 GRWTGRCMSCC RSSRTRRETQL. For HPV16 E7:1–21 MHGDTPTLHEYMLDLQPETTDL; 11–32 YMLDLQPETTDLYCYEQLNDSS; 21–42 DLYCYEQ LNDSSEEEDEIDGPA; 31–52 SSEEEDEIDGPAGQAEPDRAHY; 41–62 PAGQAEPDRAHYNIVTFCCKCD; 51–72 HYN IVTFCCKCDTLRLCVQST; 61–82 CDSTLRLCVQSTHVDIRTLEDL; 71–92 STHVDIRTLEDLLMGTLGIVCP; 77–98 RTLEDLL MGTLGIVCPICSQKP. For HPV18 E6: 1–22 MARFEDPTRRPYKLPDLCTELN; 11–32 PYKLPDLCT ELNTSLQDIEITC; 21–42 LNTSLQDIEITCVYCKTVLELT; 31–52 TCVYCKTVLELTEVFEFAFKDL; 41–61 LTEV FEFAFKDLFVVYRDSIPH; 51–72 DLFVV YRDSIPHAACHKCIDFY; 61–82 PHAACHKCIDFYSRIRELRHYS; 71–92 FYSRIRELRHYSDSVYGDTLEK; 81–102 YSDSVYGDTLEKLTNTGLYNLL; 91–112 EKLTNTGLYNLLIRCLRCQKPL; 101–122 LLIRCLRCQKPLNPAEKLRHLN; 111–132 PLNPAEKLRHLNEKRRFHNIAG; 121–142 LNEKRRFHNIA GHYRGQCHSCC; 131–152 AGHYRGQCHSCCNRARQERLQR; 137–158 QCHSCCN RARQERLQRRRETQV. For HPV18 E7: 1–21 MHGPKATLQDIVLHLEPQNEIP; 11–32 IVLHLEPQNEIPVDLLCHEQLS; 21–42 IPVDLLCH EQLSDSEEENDEID; 31–52 LSDSEEENDEIDGV NHQHLPAR; 41–62 IDGVNHQHLPARRAEPQRHTML; 51–72 ARRAEPQRHTMLCMCCKCEARI; 61–82 MLCMCCKCEARIELVVESSADD; 71–92 RIELVVESSADDLRAFQQLF LN; 77–98 DDLRA FQQLFLNTLSFVCPWCA; 80–105 RAFQQLFLNTLSFVCPWCASQQ. PHA (0.5 μg/mL; HA16 Remel; Thermo Fisher Scientific) served as a positive control, while unloaded monocytes served as negative controls. At day 1.5 and 4, supernatant (50 μL/well) was harvested for cytokine analyses. The antigen-specific cytokine production in the supernatants was determined by cytometric bead array (CBA, Th1/Th2 kit, BD Biosciences) according to the manufacturer's instructions. The cutoff value for cytokine production was 20 pg/mL, except for IFNγ for which it was 40 pg/mL. Antigen-specific cytokine production was defined as at least twice above that of the unstimulated cells.

Statistical analysis

A repeated measure one-way ANOVA with Tukey multiple comparison was used to determine significant difference in numerical data between four subpopulations. Differences were considered significant when P < 0.05 as indicated with asterisks (*, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001). Statistical analysis was performed using GraphPad Prism 8.3.1.

CD39 was expressed by CD8+ and CD4+ TILs in different cancer types

To explore the expression of CD39 on CD4+ T cells, a total of 30 cryopreserved single-cell tumor digests (9 CxCa, 10 VSCC, and 11 OPSCC) were stained with a panel of antibodies to phenotype the lineage (CD3, CD4, CD8), central and effector memory type (CCR7, CD45RA), tissue resident (CD39, CD103), and activation or cytolytic capacity (CD38, PD-1, HLA-DR, NKG2A, CD161; Supplementary Table S1). HSNE (29) analysis revealed 15 subpopulations of T cells, which were present at similar percentages in all three types of cancer (Fig. 1A; Supplementary Fig. S1). About half of the CD8+ population displayed CD39, often in coexpression with CD103, PD-1, HLA-DR, NKG2A, and CD161, whereas the CD39-negative population rarely expressed PD-1 or CD45RA (Fig. 1BD). The CD4+ T-cell population tended to comprise more CD39+ cells coexpressing CD38, PD-1, and HLA-DR, whereas a small percentage also coexpressed CD103 and CD161 (Fig. 1BD; Supplementary Fig. S2A). The percentage of CD39-expressing CD4+ T cells was much higher in tumors (∼25%) than in the blood (∼1%), suggesting their tumor-specific accumulation (Supplementary Fig. S2B). Analysis of the TILs according to the gating strategy depicted in Supplementary Fig. S3A and S3B displayed that the great majority of CD8+ TILs were either double-positive (DP) or double-negative (DN) for CD39 and CD103. There was a small population of single-positive (SP) CD103+ and a rare population of SP CD39+ CD8+ T cells (Fig. 1D), confirming previous publications (18). In contrast, the SP CD39+ and the DN populations were most dominant among CD4+ TILs, whereas the DP population was much smaller and the SP CD103+ populations was rare among CD4+ TIL (Fig. 1D). The distribution of these four subpopulations was similar among all three cancer types.

Figure 1.

CD39 was expressed by CD8+ and CD4+ TILs in different types of cancer. Cryopreserved single-cell tumor digests of patients (n = 30) with CxCa, VSCC, and OPSCC were analyzed by 13-parameter flow cytometry analysis. A, An HSNE density plot (left) and corresponding cluster point plot (middle) and sample distribution plot (right) visualizing the high-dimensional flow cytometry data in two dimensions for the collective immune cells derived from 30 patients (n = 9 CxCa, n = 10 VSCC, and n = 11 OPSCC). The identified cell subsets are indicated in the density and cluster point plots by numbers. The contribution of the different tumors to the clustering is depicted in the sample distribution plot by the different colors (CxCa in green, VSCC in red, and OPSCC in blue). B, HSNE dot plots of the immune cells at the single-cell level. Colors represent arcsin150-transformed marker expression as indicated. C, Minimum Spanning Tree displaying star plots of the clusters identified by FlowSOM analysis using the Cytobank platform. The mean intensities of the clustering channels are depicted by the different colors in the pie chart. The total T-cell populations separate into CD4+CD39+, CD4+CD39, and CD4+CD39+CD103+ populations (left; in green) and CD8+CD39+CD103+ and CD8+CD39CD103+ populations (right; purple). D, Graphs depicting the mean (line) and individual percentage of CD39+CD103, CD39+CD103+, CD39CD103, and CD39CD103+ CD4+ (left) and CD8+ (right) T cells within CxCa (green), VSCC (red), and OPSCC (blue) patients. Patients used in expansion and HPV reactivity assays are depicted by the open circles (*, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001).

Figure 1.

CD39 was expressed by CD8+ and CD4+ TILs in different types of cancer. Cryopreserved single-cell tumor digests of patients (n = 30) with CxCa, VSCC, and OPSCC were analyzed by 13-parameter flow cytometry analysis. A, An HSNE density plot (left) and corresponding cluster point plot (middle) and sample distribution plot (right) visualizing the high-dimensional flow cytometry data in two dimensions for the collective immune cells derived from 30 patients (n = 9 CxCa, n = 10 VSCC, and n = 11 OPSCC). The identified cell subsets are indicated in the density and cluster point plots by numbers. The contribution of the different tumors to the clustering is depicted in the sample distribution plot by the different colors (CxCa in green, VSCC in red, and OPSCC in blue). B, HSNE dot plots of the immune cells at the single-cell level. Colors represent arcsin150-transformed marker expression as indicated. C, Minimum Spanning Tree displaying star plots of the clusters identified by FlowSOM analysis using the Cytobank platform. The mean intensities of the clustering channels are depicted by the different colors in the pie chart. The total T-cell populations separate into CD4+CD39+, CD4+CD39, and CD4+CD39+CD103+ populations (left; in green) and CD8+CD39+CD103+ and CD8+CD39CD103+ populations (right; purple). D, Graphs depicting the mean (line) and individual percentage of CD39+CD103, CD39+CD103+, CD39CD103, and CD39CD103+ CD4+ (left) and CD8+ (right) T cells within CxCa (green), VSCC (red), and OPSCC (blue) patients. Patients used in expansion and HPV reactivity assays are depicted by the open circles (*, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001).

Close modal

Intratumoral CD39+ CD4+ and CD8+ TILs shared markers of activation and tissue residency

The DP CD4+ and CD8+ populations exclusively comprised memory T cells, most prominently of an effector memory phenotype (Fig. 2A and B). Whereas the larger CD4+CD39+ SP population exhibited a similar phenotype as the DP population, the DN CD4+ TIL population was more of a central memory phenotype (Fig. 2A). In addition, the DP CD4+ and CD8+ populations comprised the highest percentage of PD-1–, HLA-DR–, CD38-, CD161-, and NKG2A (predominantly CD8+ TILs)-expressing cells (Fig. 2C and D; Supplementary Fig. S4A and S4B) and the highest expression of PD-1 and HLA-DR (Supplementary Fig. S4C and S4D), emphasizing that they were highly activated. The percentage of HLA-DR–expressing cells was also higher in the SP CD39+ population when compared with the CD4+ and CD8+ DN or SP CD103+ TIL populations (Fig. 2C and D), suggesting that also the SP CD39+ T cells may have been locally activated. A high percentage of CD4+ T cells expressing NKG2A+ was found in 2 of 9 patients with VSCC and none of the patients with CxCa and OPSCC. Apparently, this is a VSCC-specific feature as a reanalysis of nine VSCC samples from a previous study (27) and seven additional OPSCC TILs revealed that high percentages of NKG2A+ CD4+ TILs were found in five of 18 VSCC, but not in any of the 27 OPSCC or CxCa (Supplementary Fig. S4E).

Figure 2.

Intratumoral CD39+ CD4+ and CD8+ TILs were highly activated. Cryopreserved single-cell tumor digests of patients (n = 30) with CxCa, VSCC, and OPSCC were analyzed by 13-parameter flow cytometry analysis. Stacked bar graphs showing the subdivision of the CD39+CD103, CD39+CD103+, CD39CD103, and CD39CD103+ CD4+ (A) and CD8+ (B) T cells into naïve T cells (Tn; white; CCR7+CD45RA+), central memory T cells (Tcm; light gray; CCR7+CD45RA), effector memory T cells (Tem; dark gray; CCR7CD45RA), and CD45RA+ effector memory T cells (Temra; black; CCR7CD45RA+) based on the mean percentage. Graphs depicting the mean (line) and individual percentage of PD-1 (top) and HLA-DR (bottom) within CD39+CD103, CD39+CD103+, CD39CD103, and CD39CD103+ CD4+ (C) and CD8+ (D) T cells within patients with CxCa (green, n = 9), VSCC (red, n = 10), and OPSCC (blue, n = 11). Patients used in expansion and HPV reactivity assays are depicted by the open circles (*, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001).

Figure 2.

Intratumoral CD39+ CD4+ and CD8+ TILs were highly activated. Cryopreserved single-cell tumor digests of patients (n = 30) with CxCa, VSCC, and OPSCC were analyzed by 13-parameter flow cytometry analysis. Stacked bar graphs showing the subdivision of the CD39+CD103, CD39+CD103+, CD39CD103, and CD39CD103+ CD4+ (A) and CD8+ (B) T cells into naïve T cells (Tn; white; CCR7+CD45RA+), central memory T cells (Tcm; light gray; CCR7+CD45RA), effector memory T cells (Tem; dark gray; CCR7CD45RA), and CD45RA+ effector memory T cells (Temra; black; CCR7CD45RA+) based on the mean percentage. Graphs depicting the mean (line) and individual percentage of PD-1 (top) and HLA-DR (bottom) within CD39+CD103, CD39+CD103+, CD39CD103, and CD39CD103+ CD4+ (C) and CD8+ (D) T cells within patients with CxCa (green, n = 9), VSCC (red, n = 10), and OPSCC (blue, n = 11). Patients used in expansion and HPV reactivity assays are depicted by the open circles (*, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001).

Close modal

CD4+ T cells expressing CD39 have mainly been classified as regulatory T cells (Treg) based on the coexpression of Foxp3 and CTLA-4 or by the large production of IL10 (37). However, CD4+CD39+ effector memory T cells secreting IFNγ and IL17 have been detected in the blood of patients with renal allograft rejection (38), indicating that CD39 expression is not always a marker of suppressive T cells. To address whether the CD4+CD39+ TILs here were Tregs or activated effector T cells, we analyzed the CD4+CD39+ T cells both in PBMCs and in 4 single-cell tumor digests (n = 2 VSCC and n = 2 CxCa) by flow cytometry using our previously described essential Treg marker set (28). Analysis showed that on average 78.8% (69.4–94.9; n = 4) of the CD4+CD39+ TILs did not fall into the category of CD25+Foxp3+ Tregs (Supplementary Fig. S4F). A similar percentage of CD4+CD39+ T cells expressing CD25 and Foxp3 was also found in the blood (Supplementary Fig. S4F), but here the CD39+ T-cell population was much smaller (Supplementary Fig. S2B). A larger proportion of the CD4+CD39+ TILs exhibited an CD25+Foxp3activated non-Treg phenotype, illustrating that the CD4+CD39+ T cells in tumors or tissues, which cannot be excluded as we had no healthy tissue to compare with. have a more activated profile (Supplementary Fig. S4F). In addition, we also performed single-cell RNA-sequencing analysis on magnetic bead–sorted CD3+ T cells from 13 OPSCC samples, yielding data for 20,199 T cells. Unsupervised clustering using the Leiden algorithm (34) as implemented in the scanpy package (31), resulted in 23 different CD4+ and CD8+ TIL clusters (Fig. 3A). Within these 23 clusters, three populations of CD39 (ENTPD1)-expressing cells existed, being CD4+ T cells, CD8+ T cells and CD4+Foxp3+ Tregs. The CD8+CD39+ TIL cluster also expressed the tissue-residency markers ALOX5AP, ITGAE (CD103), and ITGA1 (CD49a; ref. 39), and genes associated with exhaustion/activation [PDCD1 (PD-1), LAG-3, CXCL13] and their function to control tumor cell growth [IFNγ, GNLY (granulysin), GZMB (granzyme B), PRF1 (perforin); refs. 40, 41]. One of the two identified CD4+CD39+ TIL clusters clearly expressed Foxp3 and IL2RA (CD25), marking them as Tregs (Fig. 3). The second CD4+CD39+ TIL cluster did not express these Treg markers, but showed a gene expression profile more closely related to the CD8+CD39+ TILs, with high expression of ALOX5AP, CXCL13, and PDCD1, and with substantial expression of LAG-3, IFNγ, GNLY, and GZMB (Fig. 3B). These data indicated that CD8+CD39+ TILs were more likely to represent proinflammatory effector cells and, as such, may contain tumor-reactive T cells. The CD4+CD39+ and CD8+CD39+ TILs also expressed several genes associated with exhaustion/coinhibition and a substantial number of genes associated with activation/costimulation. The expression of genes associated with activation was more pronounced in CD4+CD39+ TILs, and included CD40LG, suggesting that this CD39+CD4+ T-cell population displayed a more activated rather than exhausted T-cell profile (Supplementary Fig. S5). We further illustrated the similarity between the tissue-resident memory T-cell (Trm) and proinflammatory effector T-cell profile for CD8+CD39+ and CD4+CD39+ T cells by pairwise comparisons of differentially expressed genes (DEG) between these T cells and the cluster of CD4+CD39+ Tregs. This showed an overlap in 15 of the 50 most significant DEGs (Fig. 3C).

Figure 3.

Intratumoral CD39+ CD4+ and CD8+ TIL shared markers of activation and tissue residency. Single-cell RNA-sequencing analysis was performed on magnetic bead–sorted CD3+ T cells from 13 OPSCC samples, yielding data for 20,199 T cells. A, Two-dimensional UMAP plots visualizing the 23 clusters identified using the Leiden algorithm (top UMAP), and three CD39+ populations comprising CD4+ T cells, CD8+ T cells, and CD4+Foxp3+ Tregs (middle and bottom UMAP). Expression of CD4, CD8A, Foxp3, and ENTPD1 (CD39) at the single-cell level is depicted in color code. B, Stacked violin plot depicting expression of ENTPD1 (CD39), ITGAE (CD103), ITGA1 (CD49a), ALOX5AP, PDCD1, CXCL13, LAG3, IFNG, GNLY (granulysin), GZMB (granzyme B), PRF1 (perforin), Foxp3, and IL2RA (CD25) for the indicated three populations. C, Graph displaying the top 50 significantly (P < 9.2 × 10e-20) differentially expressed genes (DEG) between CD4+CD39+ (left) and CD8+CD39+ (right) and CD39+ Tregs. Values depicted are log2 fold-change (log2FC) values. Overlap in DEG expression in the pairwise comparison between CD4+CD39+ and CD8+CD39+ and Tregs is depicted by the green circles (up) and red circles (down) in CD4+CD39+ and CD8+CD39+ cells compared to Tregs.

Figure 3.

Intratumoral CD39+ CD4+ and CD8+ TIL shared markers of activation and tissue residency. Single-cell RNA-sequencing analysis was performed on magnetic bead–sorted CD3+ T cells from 13 OPSCC samples, yielding data for 20,199 T cells. A, Two-dimensional UMAP plots visualizing the 23 clusters identified using the Leiden algorithm (top UMAP), and three CD39+ populations comprising CD4+ T cells, CD8+ T cells, and CD4+Foxp3+ Tregs (middle and bottom UMAP). Expression of CD4, CD8A, Foxp3, and ENTPD1 (CD39) at the single-cell level is depicted in color code. B, Stacked violin plot depicting expression of ENTPD1 (CD39), ITGAE (CD103), ITGA1 (CD49a), ALOX5AP, PDCD1, CXCL13, LAG3, IFNG, GNLY (granulysin), GZMB (granzyme B), PRF1 (perforin), Foxp3, and IL2RA (CD25) for the indicated three populations. C, Graph displaying the top 50 significantly (P < 9.2 × 10e-20) differentially expressed genes (DEG) between CD4+CD39+ (left) and CD8+CD39+ (right) and CD39+ Tregs. Values depicted are log2 fold-change (log2FC) values. Overlap in DEG expression in the pairwise comparison between CD4+CD39+ and CD8+CD39+ and Tregs is depicted by the green circles (up) and red circles (down) in CD4+CD39+ and CD8+CD39+ cells compared to Tregs.

Close modal

CD39+ TILs were readily expanded in vitro and were the main population responding to tumor antigens

Taking advantage of the fact that a number of these tumors were caused by high-risk HPV and as a consequence express the highly immunogenic tumor-specific antigens E6 and E7 (26, 42), we sorted cryopreserved single-cell tumor digests from seven HPV-induced tumors (n = 2 CxCa, n = 3 OPSCC, n = 2 VSCC) into CD4+CD39, CD4+CD39+, CD8+CD39CD103+, and CD8+CD39+CD103+ TIL populations, according to a predefined gating strategy (Supplementary Fig. S6) yielding highly pure populations of cells (Fig. 4A and B). In all cases, sufficient cells were obtained to allow for two rounds of expansion using CD3/CD28 beads and a mix of TCG-F, IL7, IL15, and IL21. Cell fractions obtained at cell counts of less than 1,000 cells were expanded to a few million and then to several million cells (Supplementary Table S2), delivering enough cells of each population to assess their specificity (Fig. 4C and D). Phenotypic analysis of the expanded T-cell populations revealed that the expression of the cell surface markers CD39 and CD103 may have changed after in vitro culture outside the context of the tumor microenvironment (Supplementary Fig. S7A and S7B), indicating that cell sorting for tumor-reactive T cells has to be performed directly on single tumor cell digests. No expansion of CD39+Foxp3+ Tregs was observed (Supplementary Fig. S7C and S7D).

Figure 4.

Schematic workflow of CD39+ T-cell sorting, expansion, and functional testing. A, HPV-induced CxCa, VSCC, or OPSCC tumors freshly obtained from surgery were cut into small pieces, digested through enzymatic dissociation into single cells, and stored in the vapor phase of liquid nitrogen and then were thawed and stained with fluorochrome-labeled antibodies directed against CD4, CD8, CD39, and CD103. Next, the cells were sorted using a FACS ARIAIII into CD4+CD39, CD4+CD39+, CD8+CD39CD103+, and CD8+CD39+CD103+ populations and subsequently expanded. B, The mean and SD as well as individual data points for the purity of the sorted populations (n = 23). C, The median (line), minimum, and maximum number of cells after expansion round 1 (day 23; left) and expansion round 2 (day 42; right) of these 23 populations. Populations are grouped on the basis of the depicted numbers obtained at the start of expansion round 1 (day 0; left) or round 2 (and day 23; right). D, The cytokine production of CD4+CD39, CD4+CD39+, CD8+CD39CD103+, and CD8+CD39+CD103+ cells in response to HPV18 E6 peptide pool (pool 1+2 and 3+4)– and E7 peptide pool (pool 1+2)–loaded autologous monocytes for a representative VSCC patient. Cytokines are measured by Th1/Th2 CBA as a single measurement and depicted in pg/mL.

Figure 4.

Schematic workflow of CD39+ T-cell sorting, expansion, and functional testing. A, HPV-induced CxCa, VSCC, or OPSCC tumors freshly obtained from surgery were cut into small pieces, digested through enzymatic dissociation into single cells, and stored in the vapor phase of liquid nitrogen and then were thawed and stained with fluorochrome-labeled antibodies directed against CD4, CD8, CD39, and CD103. Next, the cells were sorted using a FACS ARIAIII into CD4+CD39, CD4+CD39+, CD8+CD39CD103+, and CD8+CD39+CD103+ populations and subsequently expanded. B, The mean and SD as well as individual data points for the purity of the sorted populations (n = 23). C, The median (line), minimum, and maximum number of cells after expansion round 1 (day 23; left) and expansion round 2 (day 42; right) of these 23 populations. Populations are grouped on the basis of the depicted numbers obtained at the start of expansion round 1 (day 0; left) or round 2 (and day 23; right). D, The cytokine production of CD4+CD39, CD4+CD39+, CD8+CD39CD103+, and CD8+CD39+CD103+ cells in response to HPV18 E6 peptide pool (pool 1+2 and 3+4)– and E7 peptide pool (pool 1+2)–loaded autologous monocytes for a representative VSCC patient. Cytokines are measured by Th1/Th2 CBA as a single measurement and depicted in pg/mL.

Close modal

In 6 of the 7 patients, a response to E6 and/or E7 was detected (Fig. 5). Two of the patients displayed a tumor-specific CD8+ T-cell response, in both cases only detected in the CD8+CD39+CD103+ fraction. Six of the patients showed CD4+ T-cell reactivity against E6 and/or E7 by specifically producing IFNγ. In five cases, the response was measured in the CD39+ sorted cell fraction of CD4+ T cells. In two cases, a weak response was found in the CD39-negative fraction, one of which was also detected at much higher levels in the CD39+ cell population (Fig. 5). In all cases, the production of the type I cytokine IFNγ dominated the response, but occasionally also low amounts of the type II cytokines IL10 and IL5 were measured (Supplementary Fig. S8). These data clearly showed that tumor-specific antigen reactivity were almost exclusively detected among CD39+ TILs.

Figure 5.

Predominant detection of HPV-specific reactivity amongst CD4+CD39+ and CD8+CD39+CD103+ TILs. The IFNγ production of CD4+CD39 and CD4+CD39+ TILs (left) and CD8+CD39CD103+ and CD8+CD39+CD103+ TILs (right) in response to medium (control) or the HPV16 or HPV18 E6 peptide pool (pool 1+2 and 3+4)– and E7 peptide pool (pool 1+2)–loaded autologous monocytes for 7 HPV-induced SCC (n = 2 VSCC, n = 3 OPSCC, n = 2 CxCa from top to bottom). IFNγ production in response to PHA, which served as a positive control, is depicted with the gray triangles. Positive cytokine production, which is defined as at least twice above that of the medium stimulated cells and above the assay cutoff, is indicated by the red asterisks.

Figure 5.

Predominant detection of HPV-specific reactivity amongst CD4+CD39+ and CD8+CD39+CD103+ TILs. The IFNγ production of CD4+CD39 and CD4+CD39+ TILs (left) and CD8+CD39CD103+ and CD8+CD39+CD103+ TILs (right) in response to medium (control) or the HPV16 or HPV18 E6 peptide pool (pool 1+2 and 3+4)– and E7 peptide pool (pool 1+2)–loaded autologous monocytes for 7 HPV-induced SCC (n = 2 VSCC, n = 3 OPSCC, n = 2 CxCa from top to bottom). IFNγ production in response to PHA, which served as a positive control, is depicted with the gray triangles. Positive cytokine production, which is defined as at least twice above that of the medium stimulated cells and above the assay cutoff, is indicated by the red asterisks.

Close modal

In this study, we showed that tumors contain two populations of CD4+ T cells expressing CD39. One of these populations displayed markers of Tregs, whereas the other exhibited an activated tissue-resident nonregulatory profile similar to that of the CD39+CD103+CD8+ TIL population; this CD4+CD39+ T-cell fraction could be readily expanded and was enriched with tumor-specific T cells.

The search for an optimal marker to dissect bystander activated T cells that coinfiltrate tumor tissue from tumor-reactive T cells, led to the realization that TILs expressing the tissue-resident marker CD103 (43) and the chronic local antigen stimulation marker CD39 (11) may be enriched for tumor-reactive T cells. Indeed, CD8+CD39+CD103+ TILs from melanoma (18), colorectal cancer (20), VSCC, and CxCa (this study), were highly enriched for tumor-reactive T cells. Importantly, whereas the SP CD8+CD39+ TILs were rare and the DP CD8+ TILs were frequently detected by flow cytometry, the opposite held true for CD4+ TILs. The DP tissue-resident TILs among both CD4+ and CD8+ TILs most likely were locally stimulated highly active T cells, based on the highest expression of PD-1, HLA-DR, and CD38. However, we showed that also the SP CD39+ T cells display such an activated profile, hence CD39 as a marker for local chronic activation may by itself identify both CD8+ and CD4+ tumor-reactive T cells when markers for Tregs were also applied.

This notion was corroborated by our single-cell RNA analyses of the three cell clusters of T cells expressing CD39. Not only did the CD4+CD39+ non-Treg cluster express genes associate with a functional antitumor response but it also expressed genes associated with cell retention in the tumor (e.g., ITGAE, ITGA1) or associated with tumor-residency (e.g., ALOX5AP, CXCL13). In comparison with the CD39+ Tregs, the CD4+CD39+ effector T-cell population had 15 of the top 50 differentially expressed genes in common with the CD8+CD39+ tumor-resident population. Finally, although the expression of CD39 often was associated with T cells displaying an exhausted phenotype, the CD4+CD39+ non-Treg cluster also displayed a substantial number of activation/costimulatory genes arguing against exhaustion and fitting with the fact that they readily expanded and responded to cognate antigen in vitro.

It is well established that when naïve CD4+ T cells encounter their cognate antigen presented by dendritic cells, they become activated and can migrate to tumors. Less is known about the fate of these cells in the TME, in particular how their activation is maintained. The expression of CD39 and CD103 in combination with other markers such as CD38, HLA-DR, and PD-1 on these CD4+ T cells sheds some light on this mechanism as it implies that CD4+ T cells are retained in the tumor to locally receive their stimulus. Indeed, chronic TCR stimulation, a process that must frequently occur in the TME, and the presence of certain cytokines (e.g., TGFβ, IL6, IL27) drives the expression of CD39 on T cells (18, 44). The signals for cell retention in the TME are most likely mediated by local macrophages (45) but potentially also via interaction with MHC class II–positive tumor cells (46, 47). CD39, which is an ectonucleotidase that regulates extracellular ATP/adenosine levels, potentially also plays a role in the retention of these cells in the TME. Extracellular ATP allows trafficking of immune cells and this is blocked by adenosine (48–50), effectively keeping the T cells in the tumor. CD39 may also function to counterbalance the ongoing immune response in tumors by suppressing local T-cell proliferation and effector function (50, 51), perhaps to prevent immune pathology.

The isolation and expansion of TILs enriched for tumor-reactive T cells may aid the identification of new tumor antigens and improve the development of clinically oriented applications such as the adoptive transfer of either TCR transgenic T cells or bulk T-cell cultures, which are effective in a number of cancers (46, 52–58). The isolated and expanded DP CD8+ TILs in our study responded to the E6 and E7 tumor-specific antigens in 2 of 7 cases, whereas such a response was observed in 5 of the 7 CD4+CD39+ TIL cultures. The low response rate to E6 and E7 by CD8+ T cells is in line with earlier observations showing the presence of much higher frequencies of E6/E7-specific CD4+ TILs than E6/E7-specific CD8+ TILs in these types of tumors (26) and observations that neoantigens are likely to be targeted by CD8+ T cells (59). In addition, only limited amounts of single-cell tumor digest material, stored in liquid nitrogen, was available which further restricted the chance to isolate the scarcer E6/E7-specific CD8+ T cells. Clinical oriented application of isolated CD39+ T cells, therefore, may benefit from more and freshly isolated single-cell digests. The fact that also regulatory CD4+ T cells express CD39 (37) provides the theoretical problem that Tregs are coamplified and/or may prevent the outgrowth of the CD39+ tumor-reactive effector cells. Postexpansion analyses of the CD4+CD39+ T-cell cultures showed that this was not the case, with less than 1.8% of CD25+Foxp3+ Tregs having been present in the cultures after expansion.

In conclusion, we propose that primed tumor-specific T cells migrating to the tumor receive a series of antigenic stimuli following cognate interactions with tumor cells or local APC, resulting in CD39 upregulation on both CD8+ and CD4+ TILs. The identification of CD39 as a marker for both CD4+ and CD8+ tumor-reactive T cells will be highly useful with respect to studies on the susceptibility or resistance of several immunotherapeutic approaches for cancer, including checkpoint blockade, adoptive cell transfer and therapeutic vaccines. CD39 blockade, in a setting where human T cells were the only cells expressing CD39, stimulated their trafficking, proliferation, and IFNγ production (50). In combination with our data, this drives the expectation that tumors with a high CD39+ T-cell count are the best to respond to CD39 blockade therapy.

K.E. Kortekaas reports grants from KWF during the conduct of the study. G. Sturm reports personal fees from Pieris Pharmaceuticals GmbH outside the submitted work. S.H. van der Burg reports grants from Dutch Cancer Society (2016-10168) and Oncode Institute during the conduct of the study, as well as personal fees from ISA Pharmaceuticals, PCI Biotech, and DC Prime outside the submitted work. No potential conflicts of interest were disclosed by the other authors.

K.E. Kortekaas: Data curation, formal analysis, investigation, methodology, writing–original draft, writing–review and editing. S.J. Santegoets: Conceptualization, data curation, formal analysis, supervision, writing–original draft, writing–review and editing. G. Sturm: Data curation, formal analysis, writing–review and editing. I. Ehsan: Data curation, writing–review and editing. S.L. van Egmond: Data curation, writing–review and editing. F. Finotello: Data curation, formal analysis, writing–review and editing. Z. Trajanoski: Formal analysis, writing–review and editing. M.J.P. Welters: Data curation, formal analysis, writing–review and editing. M.I.E. van Poelgeest: Resources, writing–review and editing. S.H. van der Burg: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, writing–original draft, writing–review and editing.

The authors gratefully thank all the patients and healthy individuals who participated in this study. They also thank Sandra van der Broek-Veldstra and Lena van Doorn for including the patients in the Circle Study. K.E. Kortekaas, S.J. Santegoets, and I. Ehsan are supported by a grant from the Dutch Cancer Society 2016–10168 to M.I.E. van Poelgeest and S.H. van der Burg. S.H. van der Burg is supported by a base fund from the Oncode Institute. F. Finotello is supported by the Austrian Science Fund (FWF; project no. T 974-B30). Z. Trajanoski is supported by the European Research Council (grant agreement no. 786295) and by the Austrian Science Fund (FWF; project no. I 3978). Z. Trajanoski is a member of the German Research Foundation (DFG) project TRR 241(INF).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Fridman
WH
,
Pages
F
,
Sautes-Fridman
C
,
Galon
J
. 
The immune contexture in human tumours: impact on clinical outcome
.
Nat Rev Cancer
2012
;
12
:
298
306
.
2.
Shankaran
V
,
Ikeda
H
,
Bruce
AT
,
White
JM
,
Swanson
PE
,
Old
LJ
, et al
IFNgamma and lymphocytes prevent primary tumour development and shape tumour immunogenicity
.
Nature
2001
;
410
:
1107
11
.
3.
Galon
J
,
Bruni
D
. 
Approaches to treat immune hot, altered and cold tumours with combination immunotherapies
.
Nat Rev Drug Discov
2019
;
18
:
197
218
.
4.
Hung
K
,
Hayashi
R
,
Lafond-Walker
A
,
Lowenstein
C
,
Pardoll
D
,
Levitsky
H
. 
The central role of CD4(+) T cells in the antitumor immune response
.
J Exp Med
1998
;
188
:
2357
68
.
5.
van den Broeke
LT
,
Daschbach
E
,
Thomas
EK
,
Andringa
G
,
Berzofsky
JA
. 
Dendritic cell-induced activation of adaptive and innate antitumor immunity
.
J Immunol
2003
;
171
:
5842
52
.
6.
Doorduijn
EM
,
Sluijter
M
,
Salvatori
DC
,
Silvestri
S
,
Maas
S
,
Arens
R
, et al
CD4(+) T cell and NK cell interplay key to regression of MHC class I(low) tumors upon TLR7/8 agonist therapy
.
Cancer Immunol Res
2017
;
5
:
642
53
.
7.
Kang
TW
,
Yevsa
T
,
Woller
N
,
Hoenicke
L
,
Wuestefeld
T
,
Dauch
D
, et al
Senescence surveillance of pre-malignant hepatocytes limits liver cancer development
.
Nature
2011
;
479
:
547
51
.
8.
Bos
R
,
Sherman
LA
. 
CD4+ T-cell help in the tumor milieu is required for recruitment and cytolytic function of CD8+ T lymphocytes
.
Cancer Res
2010
;
70
:
8368
77
.
9.
Klebanoff
CA
,
Gattinoni
L
,
Restifo
NP
. 
CD8+ T-cell memory in tumor immunology and immunotherapy
.
Immunol Rev
2006
;
211
:
214
24
.
10.
Cheon
H
,
Borden
EC
,
Stark
GR
. 
Interferons and their stimulated genes in the tumor microenvironment
.
Semin Oncol
2014
;
41
:
156
73
.
11.
Simoni
Y
,
Becht
E
,
Fehlings
M
,
Loh
CY
,
Koo
SL
,
Teng
KWW
, et al
Bystander CD8(+) T cells are abundant and phenotypically distinct in human tumour infiltrates
.
Nature
2018
;
557
:
575
9
.
12.
Scheper
W
,
Kelderman
S
,
Fanchi
LF
,
Linnemann
C
,
Bendle
G
,
de Rooij
MAJ
, et al
Low and variable tumor reactivity of the intratumoral TCR repertoire in human cancers
.
Nat Med
2019
;
25
:
89
94
.
13.
Ye
Q
,
Song
D-G
,
Poussin
M
,
Yamamoto
T
,
Best
A
,
Li
C
, et al
CD137 accurately identifies and enriches for naturally occurring tumor-reactive T cells in tumor
.
Clin Cancer Res
2014
;
20
:
44
55
.
14.
Parkhurst
M
,
Gros
A
,
Pasetto
A
,
Prickett
T
,
Crystal
JS
,
Robbins
P
, et al
Isolation of T-cell receptors specifically reactive with mutated tumor-associated antigens from tumor-infiltrating lymphocytes based on CD137 expression
.
Clin Cancer Res
2017
;
23
:
2491
505
.
15.
Choi
BK
,
Lee
SC
,
Lee
MJ
,
Kim
YH
,
Kim
YW
,
Ryu
KW
, et al
4–1BB-based isolation and expansion of CD8+ T cells specific for self-tumor and non-self-tumor antigens for adoptive T-cell therapy
.
J Immunother
2014
;
37
:
225
36
.
16.
Fernandez-Poma
SM
,
Salas-Benito
D
,
Lozano
T
,
Casares
N
,
Riezu-Boj
JI
,
Mancheño
U
, et al
Expansion of tumor-infiltrating CD8(+) T cells expressing PD-1 improves the efficacy of adoptive T-cell therapy
.
Cancer Res
2017
;
77
:
3672
84
.
17.
Gros
A
,
Robbins
PF
,
Yao
X
,
Li
YF
,
Turcotte
S
,
Tran
E
, et al
PD-1 identifies the patient-specific CD8(+) tumor-reactive repertoire infiltrating human tumors
.
J Clin Invest
2014
;
124
:
2246
59
.
18.
Duhen
T
,
Duhen
R
,
Montler
R
,
Moses
J
,
Moudgil
T
,
de Miranda
NF
, et al
Co-expression of CD39 and CD103 identifies tumor-reactive CD8 T cells in human solid tumors
.
Nat Commun
2018
;
9
:
2724
.
19.
Martinez-Usatorre
A
,
Carmona
SJ
,
Godfroid
C
,
Yacoub Maroun
C
,
Labiano
S
,
Romero
P
. 
Enhanced phenotype definition for precision isolation of precursor exhausted tumor-infiltrating CD8 T cells
.
Front Immunol
2020
;
11
:
340
.
20.
van den Bulk
J
,
Verdegaal
EME
,
Ruano
D
,
Ijsselsteijn
ME
,
Visser
M
,
van der Breggen
R
, et al
Neoantigen-specific immunity in low mutation burden colorectal cancers of the consensus molecular subtype 4
.
Genome Med
2019
;
11
:
87
.
21.
Levitsky
HI
,
Lazenby
A
,
Hayashi
RJ
,
Pardoll
DM
. 
In vivo priming of two distinct antitumor effector populations: the role of MHC class I expression
.
J Exp Med
1994
;
179
:
1215
24
.
22.
Quezada
SA
,
Simpson
TR
,
Peggs
KS
,
Merghoub
T
,
Vider
J
,
Fan
X
, et al
Tumor-reactive CD4(+) T cells develop cytotoxic activity and eradicate large established melanoma after transfer into lymphopenic hosts
.
J Exp Med
2010
;
207
:
637
50
.
23.
Tran
E
,
Turcotte
S
,
Gros
A
,
Robbins
PF
,
Lu
YC
,
Dudley
ME
, et al
Cancer immunotherapy based on mutation-specific CD4+ T cells in a patient with epithelial cancer
.
Science
2014
;
344
:
641
5
.
24.
Hunder
NN
,
Wallen
H
,
Cao
J
,
Hendricks
DW
,
Reilly
JZ
,
Rodmyre
R
, et al
Treatment of metastatic melanoma with autologous CD4+ T cells against NY-ESO-1
.
N Engl J Med
2008
;
358
:
2698
703
.
25.
Kansas
GS
,
Wood
GS
,
Tedder
TF
. 
Expression, distribution, and biochemistry of human CD39. Role in activation-associated homotypic adhesion of lymphocytes
. J Immunol
1991
;
146
:
2235
44
.
26.
Welters
MJP
,
Ma
W
,
Santegoets
SJAM
,
Goedemans
R
,
Ehsan
I
,
Jordanova
ES
, et al
Intratumoral HPV16-specific T cells constitute a type I-oriented tumor microenvironment to improve survival in HPV16-driven oropharyngeal cancer
.
Clin Cancer Res
2018
;
24
:
634
47
.
27.
Kortekaas
KE
,
Santegoets
SJ
,
Abdulrahman
Z
,
van Ham
VJ
,
van der Tol
M
,
Ehsan
I
, et al
High numbers of activated helper T cells are associated with better clinical outcome in early stage vulvar cancer, irrespective of HPV or p53 status
.
J Immunother Cancer
2019
;
7
:
236
.
28.
Santegoets
SJAM
,
Dijkgraaf
EM
,
Battaglia
A
,
Beckhove
P
,
Britten
CM
,
Gallimore
A
, et al
Monitoring regulatory T cells in clinical samples: consensus on an essential marker set and gating strategy for regulatory T cell analysis by flow cytometry
.
Cancer Immunol Immunother
2015
;
64
:
1271
86
.
29.
van Unen
V
,
Höllt
T
,
Pezzotti
N
,
Li
Na
,
Reinders
MJT
,
Eisemann
E
, et al
Visual analysis of mass cytometry data by hierarchical stochastic neighbour embedding reveals rare cell types
.
Nat Commun
2017
;
8
:
1740
.
30.
Zheng
GXY
,
Terry
JM
,
Belgrader
P
,
Ryvkin
P
,
Bent
ZW
,
Wilson
R
, et al
Massively parallel digital transcriptional profiling of single cells
.
Nat Commun
2017
;
8
:
14049
.
31.
Wolf
FA
,
Angerer
P
,
Theis
FJ
. 
SCANPY: large-scale single-cell gene expression data analysis
.
Genome Biol
2018
;
19
:
15
.
32.
Luecken
MD
,
Theis
FJ
. 
Current best practices in single-cell RNA-seq analysis: a tutorial
.
Mol Syst Biol
2019
;
15
:
e8746
.
33.
Wolock
SL
,
Lopez
R
,
Klein
AM
. 
Scrublet: computational identification of cell doublets in single-cell transcriptomic data
.
Cell Syst
2019
;
8
:
281
91
.
34.
Traag
VA
,
Waltman
L
,
van Eck
NJ
. 
From Louvain to Leiden: guaranteeing well-connected communities
.
Sci Rep
2019
;
9
:
5233
.
35.
McInnes
L
,
Healy
JU
. 
UMAP: uniform manifold approximation and projection for dimension reduction
2018
.
36.
Robinson
MD
,
McCarthy
DJ
,
Smyth
GK
. 
edgeR: a Bioconductor package for differential expression analysis of digital gene expression data
.
Bioinformatics
2010
;
26
:
139
40
.
37.
Allard
B
,
Longhi
MS
,
Robson
SC
,
Stagg
J
. 
The ectonucleotidases CD39 and CD73: novel checkpoint inhibitor targets
.
Immunol Rev
2017
;
276
:
121
44
.
38.
Dwyer
KM
,
Hanidziar
D
,
Putheti
P
,
Hill
PA
,
Pommey
S
,
McRae
JL
, et al
Expression of CD39 by human peripheral blood CD4+ CD25+ T cells denotes a regulatory memory phenotype
.
Am J Transplant
2010
;
10
:
2410
20
.
39.
Kumar
BV
,
Ma
W
,
Miron
M
,
Granot
T
,
Guyer
RS
,
Carpenter
DJ
, et al
Human tissue-resident memory T cells are defined by core transcriptional and functional signatures in lymphoid and mucosal sites
.
Cell Rep
2017
;
20
:
2921
34
.
40.
Li
H
,
van der Leun
AM
,
Yofe
I
,
Lubling
Y
,
Gelbard-Solodkin
D
,
van Akkooi
ACJ
, et al
Dysfunctional CD8 T cells form a proliferative, dynamically regulated compartment within human melanoma
.
Cell
2019
;
176
:
775
89
.
41.
Guo
X
,
Zhang
Y
,
Zheng
L
,
Zheng
C
,
Song
J
,
Zhang
Q
, et al
Global characterization of T cells in non-small-cell lung cancer by single-cell sequencing
.
Nat Med
2018
;
24
:
978
85
.
42.
Piersma
SJ
,
Welters
MJP
,
van der Hulst
JM
,
Kloth
JN
,
Kwappenberg
KMC
,
Trimbos
BJ
, et al
Human papilloma virus specific T cells infiltrating cervical cancer and draining lymph nodes show remarkably frequent use of HLA-DQ and -DP as a restriction element
.
Int J Cancer
2008
;
122
:
486
94
.
43.
Djenidi
F
,
Adam
J
,
Goubar
A
,
Durgeau
A
,
Meurice
G
,
de Montpréville
V
, et al
CD8+CD103+ tumor-infiltrating lymphocytes are tumor-specific tissue-resident memory T cells and a prognostic factor for survival in lung cancer patients
.
J Immunol
2015
;
194
:
3475
86
.
44.
Canale
FP
,
Ramello
MC
,
Núñez
N
,
Furlan
CLA
,
Bossio
SN
,
Serrán
MG
, et al
CD39 expression defines cell exhaustion in tumor-infiltrating CD8(+) T cells
.
Cancer Res
2018
;
78
:
115
28
.
45.
Haabeth
OAW
,
Tveita
AA
,
Fauskanger
M
,
Schjesvold
F
,
Lorvik
KB
,
Hofgaard
PO
, et al
How do CD4(+) T cells detect and eliminate tumor cells that either lack or express MHC class II molecules?
Front Immunol
2014
;
5
:
174
.
46.
Linnemann
C
,
van Buuren
MM
,
Bies
L
,
Verdegaal
EME
,
Schotte
R
,
Calis
JJA
, et al
High-throughput epitope discovery reveals frequent recognition of neo-antigens by CD4+ T cells in human melanoma
.
Nat Med
2015
;
21
:
81
5
.
47.
Cioni
B
,
Jordanova
ES
,
Hooijberg
E
,
van der Linden
R
,
de Menezes
RX
, et al
HLA class II expression on tumor cells and low numbers of tumor-associated macrophages predict clinical outcome in oropharyngeal cancer
.
Head Neck
2019
;
41
:
463
78
.
48.
Mahnke
K
,
Useliene
J
,
Ring
S
,
Kage
P
,
Jendrossek
V
,
Robson
SC
, et al
Down-regulation of CD62L shedding in T cells by CD39(+) regulatory T cells leads to defective sensitization in contact hypersensitivity reactions
.
J Invest Dermatol
2017
;
137
:
106
14
.
49.
Adamiak
M
,
Bujko
K
,
Brzezniakiewicz-Janus
K
,
Kucia
M
,
Ratajczak
J
,
Ratajczak
MZ
. 
The inhibition of CD39 and CD73 cell surface ectonucleotidases by small molecular inhibitors enhances the mobilization of bone marrow residing stem cells by decreasing the extracellular level of adenosine
.
Stem Cell Rev Rep
2019
;
15
:
892
9
.
50.
Li
X-Y
,
Moesta
AK
,
Xiao
C
,
Nakamura
K
,
Casey
M
,
Zhang
H
, et al
Targeting CD39 in cancer reveals an extracellular ATP- and inflammasome-driven tumor immunity
.
Cancer Discov
2019
;
9
:
1754
73
.
51.
Noble
A
,
Mehta
H
,
Lovell
A
,
Papaioannou
E
,
Fairbanks
L
. 
IL-12 and IL-4 activate a CD39-dependent intrinsic peripheral tolerance mechanism in CD8(+) T cells
.
Eur J Immunol
2016
;
46
:
1438
48
.
52.
Doran
SL
,
Stevanović
S
,
Adhikary
S
,
Gartner
JJ
,
Jia
Li
,
Kwong
MLiM
, et al
T-cell receptor gene therapy for human papillomavirus-associated epithelial cancers: a first-in-human, phase I/II study
.
J Clin Oncol
2019
;
37
:
2759
68
.
53.
Verdegaal
EME
,
de Miranda
NFCC
,
Visser
M
,
Harryvan
T
,
van Buuren
MM
,
Andersen
RS
, et al
Neoantigen landscape dynamics during human melanoma-T cell interactions
.
Nature
2016
;
536
:
91
5
.
54.
Ramachandran
I
,
Lowther
DE
,
Dryer-Minnerly
R
,
Wang
R
,
Fayngerts
S
,
Nunez
D
, et al
Systemic and local immunity following adoptive transfer of NY-ESO-1 SPEAR T cells in synovial sarcoma
.
J Immunother Cancer
2019
;
7
:
276
.
55.
Chapuis
AG
,
Afanasiev
OK
,
Iyer
JG
,
Paulson
KG
,
Parvathaneni
U
,
Hwang
JH
, et al
Regression of metastatic Merkel cell carcinoma following transfer of polyomavirus-specific T cells and therapies capable of re-inducing HLA class-I
.
Cancer Immunol Res
2014
;
2
:
27
36
.
56.
Morgan
RA
,
Dudley
ME
,
Wunderlich
JR
,
Hughes
MS
,
Yang
JC
,
Sherry
RM
, et al
Cancer regression in patients after transfer of genetically engineered lymphocytes
.
Science
2006
;
314
:
126
9
.
57.
Stevanović
S
,
Helman
SR
,
Wunderlich
JR
,
Langhan
MM
,
Doran
SL
,
Kwong
MLM
, et al
A phase II study of tumor-infiltrating lymphocyte therapy for human papillomavirus-associated epithelial cancers
.
Clin Cancer Res
2019
;
25
:
1486
93
.
58.
Verdegaal
E
,
van der Kooij
MK
,
Visser
M
,
van der Minne
C
,
de Bruin
L
,
Meij
P
, et al
Low-dose interferon-alpha preconditioning and adoptive cell therapy in patients with metastatic melanoma refractory to standard (immune) therapies: a phase I/II study
.
J Immunother Cancer
2020
;
8
:
e000166
.
59.
Yang
W
,
Lee
K-W
,
Srivastava
RM
,
Kuo
F
,
Krishna
C
,
Chowell
D
, et al
Immunogenic neoantigens derived from gene fusions stimulate T cell responses
.
Nat Med
2019
;
25
:
767
75
.

Supplementary data