T-cell receptor (TCR) gene therapy is a promising next-generation antitumor treatment. We previously developed a single–T-cell analysis protocol that allows the rapid capture of paired TCRα and β cDNAs. Here, we applied the protocol to analyze the TCR repertoire of tumor-infiltrating lymphocytes (TIL) of various cancer patients. We found clonally expanded populations of T cells that expressed the same clonotypic TCR in 50% to 70% of CD137+CD8+ TILs, indicating that they responded to certain antigens in the tumor environment. To assess the tumor reactivity of the TCRs derived from those clonally expanded TILs in detail, we then analyzed the CD137+CD8+ TILs from the tumor of B16F10 melanoma cells in six C57BL/6 mice and analyzed their TCR repertoire. We also found clonally expanded T cells in 60% to 90% of CD137+CD8+ TILs. When the tumor reactivity of dominant clonotypic TCRs in each mouse was analyzed, 9 of 13 TCRs induced the secretion of IFNγ in response to, and showed killing of, B16F10 cells in vitro, and 2 of them showed strong antitumor activity in vivo. Concerning their antigen specificity, 7 of them reacted to p15E peptide of endogenous murine leukemia virus-derived envelope glycoprotein 70, and the rest reacted to tumor-associated antigens expressed on EL4 lymphoma as well as B16 melanoma cells. These results show that our strategy enables us to simply and rapidly obtain the tumor-specific TCR repertoire with high fidelity in an antigen- and MHC haplotype–independent manner from primary TILs. Cancer Immunol Res; 6(4); 378–88. ©2018 AACR.
Since the advent of immune checkpoint blockade therapies, tumor immunotherapy has become a promising antitumor approach (1). In this context, T-cell receptor (TCR) gene therapy and T-cell adoptive therapy are also promising next-generation tumor therapies (2, 3). To develop these T cell–related therapies, tumor-specific T cells or TCR genes are required. To obtain tumor-specific T cells or TCR genes, the establishment of tumor-specific T-cell lines is conventionally required, which is time consuming and laborious. Additionally, treating patients using tumor-specific T cells or TCRs derived from the patients themselves is difficult because their cancer may progress rapidly. The subjects of most studies on tumor immunotherapies were skewed toward patients with major human leukocyte antigen (HLA) haplotypes (i.e., HLA-A*2402 for Asian patients and HLA-A*0201 for Caucasian patients; see The Allele Frequency Net Database, http://allelefrequencies.net/default.asp; ref. 4), and thus, those patients with minor HLA haplotypes cannot benefit from results from these studies.
Tumor-infiltrating lymphocytes (TIL) have been used as vehicles for adoptive T-cell therapies and TCR gene therapies (5–7). PD-1 and/or CD137 (4-1BB)-positive T cells in TILs express tumor reactive TCRs against shared tumor antigens and/or neoantigens (8–10). Thus, TILs are an attractive source of tumor-reactive TCRs for individualized cancer immunotherapy.
Various protocols for TCR repertoire analysis have been developed in many laboratories. Comprehensive analyses of the TCR repertoire using next-generation sequencing (NGS) methods have been reported (11–13). However, it is difficult to use conventional NGS analysis to determine the unique TCRα and β pairs from single T cells, although several laboratories developed NGS analyses that are able to determine TCRα and β pairs (14–16). In contrast to NGS, TCR repertoire analyses at the single-cell level (17–19) and their functional determinations (20–23) can determine the unique TCR α and β pairs quite simply. We also established a single–T-cell analysis method and an improved version that enables us to obtain the antigen-specific TCRs from single primary T cells within 10 and 4 days, respectively (24, 25).
In this study, we have tried to obtain tumor-specific TCRs from TILs using our single-cell analysis method in the absence of information on antigen specificity or MHC haplotype. To this end, we first investigated the TCR repertoire of TILs derived from various cancer patients and then analyzed that of TILs derived from tumor-bearing mice. We found clonally expanded populations of CD8+ T cells in human and mouse TILs and demonstrated that many TCRs from the clonally expanded T cells were specific and cytotoxic to tumors both in vivo and in vitro in a mouse model. We show that our single–T cell analysis procedure can identify tumor-reactive TCRs from TILs without knowing their antigen specificity or the haplotype of the MHC-presenting molecule and that tumor-associated antigens can also be good targets for TCRs with tumoricidal activity.
Materials and Methods
The PerCP-Cy5.5–conjugated human CD3 monoclonal antibody (mAb, 45-0037-41), allophycocyanin (APC)-conjugated human CD8 mAb (17-0088-41), APC-Cy7 Fixable Viability Dye (65-0865-14), and APC-conjugated mouse PD-1 mAb (17-9985-82) were purchased from eBioscience. The phycoerythrin (PE)-conjugated human CD137 mAb (309803), fluorescein isothiocyanate (FITC)-conjugated human PD-1 mAb (329903), and PE-conjugated mouse CD137 mAb (106105) were purchased from BioLegend. FITC-conjugated mouse CD8 mAb (MAB 116) was purchased from R&D Systems. APC-conjugated TRP2p/H-2Kb tetramer (TS-M5004-2), APC-conjugated p15Ep/H-2Kb tetramer (TS-M507-2), and p15Ep (TS-M507-P) were purchased from MBL. The pMXs-IRES-GFP retroviral expression vector (RTV-013) was acquired from Cell Biolabs.
B16F10 cells (CRL-6475), B16F0 cells (CRL-6322), and C57BL/6 mouse-derived mouse embryonic fibroblast (SCRC-1008) were provided by the ATCC. 58α−β− T-cell hybridoma-derived cells with no expression of endogenous TCRα and β (26) were generously supplied by B. Malissen, INSERM, France. B16F10 cells, B16F0 cells, and mouse embryonic fibroblast were maintained in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal calf serum (FCS), 50 μmol/L 2-mercaptoethanol, streptomycin (100 μg/mL), and penicillin (100 U/mL). EL4, MC38, and 58α−β− cells were maintained in RPMI 1640 containing 10% FCS, 50 μmol/L 2-mercaptoethanol, streptomycin (100 μg/mL), and penicillin (100 U/mL). To prepare the luciferase-expressing B16F10 cells, luciferase cDNA was excised from pGL3 Luciferase Reporter Vectors (Promega, E1751) and transferred into a PiggyBac single promoter vector containing internal ribosome entry site green fluorescent protein (IRES-GFP; System Biosciences, PB530A-2). The resultant vector was transfected into B16F10 cells using the Super PiggyBac transposase expression vector (System Biosciences, PB200PA-1). To select the luciferase-expressing B16F10 cells (B16F10-Luc), GFP-positive cells were sorted using a FACSAria II (from Becton Dickinson).
Preparation of TILs from human cancer patients
Human experiments were performed with the approval of the Ethical Committee at Osaka University (melanoma), Fukuoka University (lymphoma), and University of Toyama. Informed consent was obtained from the subjects. After surgical resection, the tumor specimens were minced and enzymatically and mechanically digested using gentleMACS (Miltenyi Biotec, 130-093-235) and a tumor dissociation kit according to the manufacturer's instructions. Single-cell suspensions of the tumor tissues were cryopreserved. To prepare the TILs from lymphoma samples, the lymphoma tissue was minced with a scalpel and mashed on a 40-μm cell strainer (Falcon) to prepare a single-cell suspension.
Mouse tumor models and the isolation of TILs
The experiments using mice were approved by the Committee on Animal Experiments at the University of Toyama. Seven- to 8-week-old mice were inoculated subcutaneously in the lower right flank with 5 × 105 B16F10 cells. Seven, 10, or 14 days after inoculation, the mice were sacrificed, and the melanoma tissues were obtained. The tumor specimens were minced and enzymatically and mechanically digested with gentleMACS and a tumor dissociation kit according to the manufacturer's instructions. The tumor single-cell suspensions were cryopreserved.
To sort the mouse TILs, they were stained with FITC-conjugated CD8 mAb, and PE-conjugated CD137 mAb for 20 minutes on ice, followed by staining with APC-Cy7 fixable viability dye for 5 minutes. After the staining, the cells were washed with PBS and analyzed with a FACSAria II. CD137+CD8+ cells were single-cell sorted into a 96-well PCR plate. The data were analyzed using FlowJo 8.4.7 software (Tree Star, Inc.).
Single-cell RT-PCR and sequencing
The TCR cDNAs were amplified from single T cells using one-step multiplex RT-PCR as described previously (25). Briefly, 5 μL of the RT-PCR mix was added to each well containing a single T cell, and one-step RT-PCR was performed. The program for the one-step RT-PCR was as follows: 40 minutes at 45°C for the RT reaction, 98°C for 1 minute and 30 cycles of 98°C for 10 seconds, 52°C for 5 seconds, and 72°C for 1 minute. The resultant one-step RT-PCR products were diluted 10-fold with nuclease-free water and used for a second cycle of PCR. In the second cycle, the cDNAs of TCRα and β were amplified separately. To amplify the cDNA of the TCRα or TCRβ, 2 μL of the diluted one-step RT-PCR products was added to each well of a new 96-well PCR plate containing 18 μL of the 2nd-PCR α mix or the 2nd-PCR β mix, respectively. The PCR program for the second PCR cycle for TCRα and β was as follows: 98°C for 1 minute and 35 cycles of 98°C for 10 seconds, 52°C for 5 seconds, and 72°C for 30 seconds. The DNA sequences of the second PCR products were then determined by direct sequencing. When T cells expressed dual TCRs, the direct sequencing resulted in overlapping sequence signals. In that case, we cloned the PCR products into a plasmid vector using the Mighty TA-cloning Reagent Set for PrimeSTAR (Takara Bio, 6019) and transformed the vector into E. coli JM109 competent cells (Takara, 9052). The dual TCR sequences were analyzed using colony direct PCR and sequencing. The TCR repertoire was analyzed with the IMGT/V-Quest tool (http://www.imgt.org/).
Construction of the TCR expression vectors
The vectors used to express the TCRs were constructed as described previously (27). Briefly, the TCRβ PCR fragment, the gene encoding the mouse TCRβ constant-1 region conjugated with the self-cleaving P2A peptide (mCb1-P2A-fragment), the TCRα PCR fragment and the codon-optimized mouse TCRα constant region gene (mCa-fragment) were assembled together in a linearized pMXs-IRES-GFP retroviral vector using the Gibson Assembly Master Mix (New England Biolabs, E2611) according to the manufacturer's instructions. The constructed plasmid vector, pMXs-TCRβ-P2A-TCRα-IRES-GFP, was used for retrovirus production.
First, 1.25 × 106 Plat-E Cells (generously provided by Professor Toshio Kitamura, University of Tokyo) were cultured in a 6-cm dish with 4 mL of DMEM containing 10% FCS, 50 μmol/L 2-mercaptoethanol, streptomycin (100 μg/mL), and penicillin (100 U/mL) 1 day before the transfection. The pMXs-TCRβ-P2A-TCRα-IRES-GFP vectors were transfected into Plat-E cells using the FuGENE6 transfection reagent (Roche, E2692) according to the manufacturer's instruction. The cells were cultured at 37°C in a 5% CO2 atmosphere. The following day, the cell medium was exchanged. After 2 days, the culture supernatant was harvested, filtered with a 0.45-μm filter (Millipore, SLHV033RS), aliquoted, and stored at −80°C until use.
Retroviral transduction of TCR into splenocytes
For this purpose, 1 × 106 splenocytes from C57BL/6 mice were stimulated with 25 μL of Dynabeads Mouse T-Activator CD3/CD28 (Invitrogen, 11453D) and mouse interleukin (IL)-2 (30 U/mL, PeproTech Inc. 212-12) in 1 mL of RPMI1640 containing 10% FCS, 50 μmol/L 2-mercaptoethanol, streptomycin (100 μg/mL), and penicillin (100 U/mL, RPMI culture medium). Two days later, the splenocytes were harvested and resuspended at 5 × 105 cells/mL in RPMI culture medium in the presence of mouse IL-2 (30 U/mL). To retrovirally transduce the TCRs into the splenocytes, wells in a non–tissue-culture-treated 24-well plate were coated with 0.5 mL of RetroNectin (50 μg/mL, Takara, T100B) at 4°C overnight, and the TCR-encoding retrovirus was spin-loaded into the wells by centrifugation for 2 hours at 1,900 × g at 32°C according to the manufacturer's instructions. The suspended cells were then added to the retrovirus-loaded wells and spun down at 1,000 × g at 32°C for 10 minutes and incubated overnight at 37°C in 5% CO2. The next day, the splenocytes were treated with the retrovirus as described above. The following day, the TCR-transduced cells were expanded in the presence of mouse IL-2 (30 U/mL) for 2 more days and used for the experiments.
IFNγ secretion assessed by enzyme-linked immunosorbent assay (ELISA)
TCR cDNA-transduced spleen cells (1 × 105) were cultured together with 1 × 105 B16F10 cells that had been treated with IFNγ (100 U/mL, PeproTech Inc. 315-05) for 48 hours in 0.2 mL of RPMI culture medium in a 96-well plate. After 16 hours of culture, the supernatants were harvested, and the amount of IFNγ in each supernatant was measured by ELISA (R&D Systems, DY485) according to the manufacturer's instruction.
T-lymphocyte cytotoxicity assessed by Luciferase assay
We analyzed the cytotoxicity against B16F10-Luc cells by measuring the target cell viability based on the luciferase activity (28–30). B16F10-Luc cells (1 × 104) treated with 100 U/mL IFNγ and TCR-transduced spleen cells were cultured together in 96-well plates at the indicated effector-to-target (E/T) ratios and incubated for 48 hours. The viability of the B16F10-Luc cells was assessed by measuring the cell-associated luciferase activity using the Steady-Glo Luciferase Assay System (Promega, E2520).
Analysis of the effect of TCR-transduced T cells on pulmonary metastasis
The experiments using mice were approved by the Committee on Animal Experiments at the University of Toyama. To assess the antitumor effect of TCR-transduced T cells, we used an experimental metastasis assay with B16F10-Luc cells. To this end, 5 × 105 B16F10-Luc cells were i.v. injected into 8-week-old C57BL/6 female mice. On day 1, 2.5 × 107 mock-transduced cells or tumor-specific TCR-transduced T cells were i.v. injected into the mice. On day 4, the lungs were removed from the mice, and bioluminescence imaging was performed using the IVIS Imaging System (PerkinElmer) to monitor the tumor metastasis. The signal intensity of the tumor burdens was expressed in total photons/s/cm2 (p/s/cm2/sr).
Single-cell–based TCR repertoire analysis of “activated” TILs from cancer patients
There are several reports that showed the existence of tumor-specific T cells in CD137+ or PD-1+ TILs (8–10). Thus, we first analyzed the surface expression of CD137 or PD-1 on CD8+ TILs derived from patients with various cancers (melanoma, lymphoma, breast cancer, thyroid cancer and colon cancer) and show data from a representative patient (n = 2 to n = 10) with each cancer is shown in Fig. 1. Flow cytometry analysis revealed that a significant number of CD8+ cells expressed both CD137 and PD-1 in most of tumors (Fig. 1A). We single-cell sorted the CD137+PD-1+CD8+ T cells and analyzed their TCR repertoire. The results showed that the CD137+PD-1+CD8+ TILs could be grouped into populations of T cells expressing the same clonotypic TCRα and β pair (Fig. 1B). The features of TILs in all analyzed tumors are summarized in Supplementary Table S1. Thus, the CD137+PD-1+CD8+ T cells were clonally expanded in the human tumors. We also tried to analyze the TCR repertoire of CD137+CD8+ PBMCs from 2 cancer patients, but we could not because of the small number of CD137+CD8+ T cells in PBMCs. Because tumor cell lines were not available, we did not examine the specificity of the TCRs that were isolated from TILs.
Single-cell–based TCR repertoire analysis of activated TILs from B16F10 tumors
Because we could not determine the human TIL-derived TCR specificity, we next analyzed the TILs of tumor-bearing mouse model. To this end, we subcutaneously inoculated mouse B16F10 melanoma cells in C57BL/6 mice, and after 7, 10, and 14 days, TILs were prepared from the tumors of each mouse. We first analyzed the expression of CD137 and CD8 on the TILs of tumors by flow cytometry. As shown in Fig. 2A and B, approximately 0.6% of TILs expressed CD8 and CD137 on day 7, and 4% and 5% of TILs expressed CD8 and CD137 on days 10 and 14, respectively. We also examined the expression of CD137 on CD8+ T cells in the regional lymph nodes (RLN) of tumor-bearing mice on days 7, 10, and 14 and found that the RLN CD137+CD8+ lymphocyte population was not increased (Fig. 2A and B). The expression of CD137 molecules on CD8+ cells in TILs was higher than that of CD137+CD8+ RLN cells (Fig. 2A). We found virtually no expression of CD137 on CD8+ T cells in the spleens of tumor-bearing mice (Supplementary Fig. S1A). These data show that CD137highCD8+ T cells exist in the TILs of melanoma tumors. Additionally, we analyzed the expression of PD-1 on CD8+ T cells in TILs and found that most CD137+CD8+ T cells expressed PD-1 (Supplementary Fig. S1B).
To analyze the TCR repertoire of CD137+CD8+ T cells in TILs and the RLNs at the single-cell level, we single-cell sorted CD137+CD8+ T cells in the TILs and RLNs of tumor-bearing mice on days 10 and 14 after inoculation of mice with tumor (Supplementary Fig. S2A) and amplified the TCRβ and TCRα cDNA from single cells. We did not sort the TILs on day 7, because the cell number was too small to sort them. We then analyzed their nucleotide sequences via direct sequencing of the TCRβ of CD137+CD8+ T cells in TILs and RLN cells (Fig. 2C). We observed many populations of CD137+CD8+ TILs expressing the same clonotypic TCR, which might be clonally expanded in the tumor but not in the population of CD137+CD8+ T cells in the RLN cells. Supplementary Table S2 shows the genes and amino acid sequences of CDR3 of the TCRβ and α that were identified from clonally expanded T cells in CD137+CD8+ TILs. These results suggest that tumor-specific T cells were clonally expanded in the tumor in response to tumor cells.
We also analyzed the TCR repertoire of CD137−CD8+ T cells in the populations of the TILs, RLN cells, and spleen cells. We observed clonally expanded populations in the CD137−CD8+ TILs of mice #2, #3, and #6 (Supplementary Fig. S2B). Some of the TCR clonotypes were shared between CD137+ and CD137− CD8+ TILs (Supplementary Table S3). These results indicate that the expression of CD137 on CD8+, clonally expanded, TILs fluctuated in response to cell conditions. In this regard, Dawicki and Watts demonstrated the transient expression of CD137 on CD8+ and CD4+ T cells during antigen stimulation in vivo (31). We rarely observed clonally expanded populations in CD137−CD8+ T cells in the RLN and spleen (Supplementary Fig. S2B; Supplementary Table S3).
Cytokine secretion from TIL-derived TCR-transduced T cells
To assess whether the TCRs of clonally expanded populations are tumor reactive, we selected 13 clonotypic TCRs (Supplementary Table S2; Fig. 2C) from clonally expanded populations with high frequencies in CD137+CD8+ T cells in TILs from each mouse, constructed the vectors for their expression, and transduced them into splenic T cells. The transfection efficiency into CD8+ T cells varied from 32% to 51% (Supplementary Fig. S3). We then cocultured TIL-derived TCR-transduced T cells with interferon (IFN)-γ–stimulated B16F10 cells and measured the IFNγ production of the T cells. We stimulated B16F10 cells with IFNγ prior to the coculture because major histocompatibility complex (MHC) class I molecules (H-2Kb and H-2Db) were not expressed on the cells; instead, their expression was induced by stimulation with IFNγ (Supplementary Fig. S4). Nine out of the 13 TIL-derived TCRs from mice #2, #4, #5, and #6 rendered the TCR-transduced T cells responsive to IFNγ-stimulated B16F10 cells and made them secrete IFNγ (Fig. 3). OT-I TCR-expressing T cells that recognized ovalbumin-derived peptides in the context of H-2Kb (32) were used as a negative control. They did not respond to IFNγ-stimulated B16F10 cells. To confirm the TCR reactivity, we sorted TCR-transduced GFP+ T cells (Supplementary Fig. S5A) and examined their IFNγ-secretion in response to IFNγ-stimulated B16F10 cells. As shown in Supplementary Fig. S5B, we obtained a result similar to that shown in Fig. 3.
Cytotoxicity of TIL TCR–transduced T cells in vitro
We next examined whether the TCRs that reacted with B16F10 cells killed B16F10 cells in vitro. To this end, we used the TCR-transduced T cells that had been used for the IFNγ-secretion assay (Supplementary Fig. S3). The TCR-transduced T cells were cultured with luciferase-expressing B16F10-Luc cells for 48 hours, and the CTL activity was assessed by analyzing the viability of the B16F10-Luc cells using the luciferase activity as previously described (28–30). OT-I TCR-expressing T cells were used as a negative control. T cells expressing the TCR that had reacted to B16F10 cells (Fig. 3) showed cytotoxicity against the IFNγ-stimulated B16F10-Luc cells (Fig. 4) but not against the unstimulated cells (Supplementary Fig. S6). Additionally, OT-I TCR-expressing T cells did not cause cytotoxicity to IFNγ-stimulated B16F10-Luc cells (Fig. 4). As expected, the cytotoxicity-inducing activity of the TCRs was correlated with their IFNγ-inducing activity (Supplementary Fig. S7). These data show that TIL-derived TCR-transduced T cells that reacted to B16F10 melanoma cells exhibited cytotoxicity toward the B16F10 melanoma cells in vitro.
Antitumor effect of TIL-derived TCR-transduced T cells in vivo
We then investigated whether TIL-derived TCR-transduced T cells exhibit cytotoxicity toward B16F10 melanoma cells in vivo. To this end, we examined the antitumor effects of TCR 4-3 and TCR 6-1, which induced the highest IFNγ secretion and cytotoxicity against IFNγ-stimulated B16F10 cells among the TCRs obtained from each mouse (Figs. 3 and 4). We first administered B16F10-Luc cells i.v. to five C57BL/6 mice for each TCR, and 1 day later, we i.v. injected the TCR-transduced T cells into the mice (Fig. 5A). The transfection efficiency of the TIL-derived TCR in splenic CD8+ T cells varied from 44% to 56% (Supplementary Fig. S8). On day 4, we sacrificed the mice and analyzed the pulmonary metastasis by monitoring with luciferase imaging. When TCR-untransduced or OT-I TCR-transduced T cells were administered as negative controls, a large number of B16F10 cells exhibited pulmonary metastasis (Fig. 5B). In contrast, when TCR 4-3- and TCR 6-1-transduced T cells were administered 1 day after B16F10-Luc cell administration, pulmonary metastasis was inhibited by 80% and 75%, respectively. These data revealed that the TIL-derived TCR-transduced T cells that had a cytotoxic effect in vitro also exhibited an antitumor effect in vivo on B16F10 melanoma cells, although their efficacy was limited, in spite of early tumor treatment with large numbers of T cells.
Specificity of TIL-derived TCRs
To examine the antigen specificity of TIL-derived TCRs, we first determined whether TIL-derived TCRs were restricted by H-2Kb or Db molecules. We transfected either H-2Kb or H-2Db cDNA in conjunction with EGFP cDNA into B16F10 cells and prepared the cells that constitutively expressed either H-2Kb or H-2Db molecules (Supplementary Fig. S9A). H-2Kb cDNA-transfectants expressed only H-2Kb molecules on the cell surface. In contrast, some H-2Db cDNA transfectants expressed H-2Kb molecules, in addition to H-2Db molecules, in the absence of IFNγ stimulation. When TIL-derived TCR-transduced spleen cells were cocultured with these B16F10 cells, all of them showed strong cytotoxicity to H-2Kb-expressing B16F10 cells, and weak cytotoxicity to H-2Db-expressing B16F10 cells (Supplementary Fig. S9B). The result indicated that our cloned TIL-derived TCRs were specifically restricted to H-2Kb molecules.
We then analyzed the specificity of TIL-derived TCRs. We first examined the reactivity of the TCRs to various cells, including B16F0 melanoma cells, EL4 lymphoma cells, MC38 colon cancer cells, and normal mouse embryonic fibroblast (MEF) cells in addition to B16F10 cells, all of which were derived from C57BL/6 mice. To this end, the cells were stimulated with IFNγ. In the absence of IFNγ stimulation, B16F0 and MEF cells hardly expressed H-2Kb or H-2Db molecules on the cell surface, whereas EL4 cells or MC38 cells expressed them on the cell surface (Supplementary Fig. S10). Their expression on all cell lines was enhanced by stimulation with IFNγ. When the TIL-derived TCR-expressing spleen cells were cocultured with these cells, the cells expressing TCR 4-3 or TCR 6-1 responded to not only B16F10 cells but also B16F0 cells and EL4 cells, and secreted IFNγ (Fig. 6), but they did not respond to MC38 cells or normal MEF cells. In contrast, the spleen cells expressing TCR 4-1, 4-2, 5-1, or 5-2 responded to IFNγ-stimulated B16F0 cells as well as B16F10 cells but neither to EL4 cells, MC38 cells, nor MEF cells. TCR 2-1-expressing T cells, although weakly, responded to only B16F10 cells, but not to the other cells.
We then examined whether those TCRs recognize antigens reported to be expressed in B16F10 cells. Because our cloned TIL-derived TCRs were restricted to H-2Kb molecules (Supplementary Fig. S9), we examined two antigenic peptides: one is tyrosine-related protein 2-drived peptide (TRP2p; ref. 33) and the other is p15E peptide (p15Ep) from envelope protein 70 of endogenous murine-leukemia virus (34), both of which bound to H-2Kb molecules. We stained the TCR-expressing T cells with TRP2p/H-2Kb tetramer or p15Ep/H-2Kb tetramer. Unexpectedly, p15Ep/H-2Kb tetramer, but not TRP2p/H-2Kb tetramer, bound to TCRs 4-1, 4-2, 5-1, 5-2, 5-3, and 5-4 (Fig. 7A). The p15Ep/H-2Kb tetramer very weakly bound to TCR 2-1. Both tetramers did not bind to TCR 4-3 and TCR 6-1. When the TCR-expressing splenic T cells were stimulated with p15E peptide-pulsed EL4 cells, T cells expressing TCRs 2-1, 4-1, 4-2, 5-1, 5-2, 5-3, and 5-4 secreted IFNγ (Fig. 7B). Taken together, these results showed that the TCRs 4-3 and 6-1 recognized certain tumor-associated antigens whose expression was shared with B16 melanoma and EL4 lymphoma cells, whereas TCRs 2-1, 4-1, 4-2, 5-1, 5-2, 5-3, and 5-4 recognized endogenous murine leukemia virus-derived antigens.
In this study, we first analyzed the TILs in various cancer patients. We observed clonally expanded populations in which T cells expressed the same clonotypic TCR in PD-1+CD137+CD8+ TILs in most cancer patients examined. Because CD137 and PD-1 are expressed on activated T cells (35, 36), the results suggest that those clonally expanded populations responded to some tumor components, such as tumor antigens. Because patients' primary culture tumor cells were not fully available to examine the tumor reactivity of the obtained TCRs, we then examined the TILs from a melanoma-bearing mouse model. Similarly, we found clonally expanded populations in CD137+CD8+ TILs from mouse melanoma tissues, yet the regional lymph nodes or spleen in the same mice had few or no clonally expanded populations among the CD137+ or CD137−CD8+ T cells. These results also indicated that tumor-specific T cells were activated and clonally expanded in the tumor. Indeed, T cells that expressed the TCRs obtained from the clonally expanded TILs responded to B16F10 melanoma cells by secretion of IFNγ and exhibited cytotoxicity against B16F10 melanoma cells both in vitro and in vivo. In this regard, Thompson and colleagues also reported that tumors support the infiltration, activation, and effector differentiation of naive CD8+ T cells (37).
We analyzed the TCR repertoire of TILs in six mice and found clonal expansion of CD137+CD8+ T cells in all of the mice. However, some of the TCRs cloned from T cells that showed high clonal expansion in the tumor did not respond to B16F10 cells. The possible reasons for this finding are that (i) the tumor-specific TCRα and β pairs were not efficiently expressed because of a mispairing between the transduced TCRs and endogenous TCRs in splenic T cells; (ii) because of the dual TCR expression in single T cells, a mismatched and tumor–nonreactive TCRα and β pair was cloned and used for functional analysis; and (iii) the tumor antigen–nonspecific T cells were expanded in TILs.
With regard to the specificity of TCRs we obtained from TILs, TCRs 4-3 and 6-1 responded to more than one tumor line (B16F10, B16F0, and EL4 cells) but not to MC38 cells nor normal MEF cells, indicating that they recognized tumor-associated antigen(s) that were shared with melanoma and lymphoma. To our surprise, the other seven TCRs reacted to p15E peptide derived from endogenous murine leukemia virus envelope glycoprotein 70 (gp70; ref. 38). p15E (gp70) is a potent immunogen (34) and highly expressed in B16F10 cells (approximately 105-fold higher than EL4 cells), whereas gp70 expression on normal tissues is nearly undetectable (38). p15E-reactive TCRs induced weaker immune reactions than TCRs 4-3 and 6-1 that recognized shared tumor-associated antigen(s). The results may reflect the central tolerance that depletes T-cell clones with strong reactivity against p15E.
When we analyzed the TCR repertoire of TILs obtained from B16F10 tumors that had developed in C57BL/6 mice, none of the TCRs obtained from different mice showed the same clonotypes, whereas the T cells that expressed the same clonotypic TCR were expanded in the tumor in the same mouse. These results seem to be reasonable because TCR rearrangement occurs among the V, D, and J segments and between the V and J segments in the TCRβ and TCRα genomes, respectively. N sequences are produced in the thymus irrespective of the genome, and this process occurs independently in each mouse. Consequently, T cells produced in the thymus of different mice expressed independent and distinct clonotypic TCRs.
Rosenberg and colleagues have shown that tumor- and neoantigen-reactive TCRs can be identified in TILs by NGS analysis followed by single-cell analysis (6, 37, 39). From their analysis, they discussed the usefulness of single-cell analysis for obtaining and determining TCR α and β pairs (7). In combination with our observations, the single-cell analysis of the TCR repertoire in TILs enabled us to efficiently obtain candidate TCRs for TCR-gene therapy.
In conclusion, the single–T-cell analysis of TILs enabled us to identify tumor-specific T cells and obtain tumor-specific TCRs that exert inhibitory effects against tumor growth, irrespective of the MHC haplotype and antigens. Because our single-cell RT-PCR protocol can obtain TCR cDNAs from primary T cells in a simple, rapid, and high-fidelity manner, we can acquire TCRs from TILs for the treatment of cancer patients. Thus, the single–T-cell analysis of TILs will promote the personalized treatment of cancer patients in the future.
Disclosure of Potential Conflicts of Interest
K. Shitaoka was a senior researcher at SC World, Inc. H. Kishi is director at SC World, Inc. and has ownership interest in patent JP 6126804. K. Tani reports receiving a commercial research grant from NeoPharma Japan Co. Ltd, Shinnihonseiyaku Co. Ltd., and Takara Bio Co. Ltd. No potential conflicts of interest were disclosed by the other authors.
Conception and design: K. Shitaoka, H. Hamana, H. Kishi, Y. Hayakawa, T. Ozawa, A. Muraguchi
Development of methodology: K. Shitaoka, H. Hamana, H. Kishi, Y. Hayakawa, E. Kobayashi1, D. Sugiyama
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K. Shitaoka, H. Hamana, H. Kishi, Y. Hayakawa, K. Sukegawa, X. Piao, F. Lyu, T. Nagata, D. Sugiyama, H. Nishikawa, A. Tanemura, I. Katayama, Y. Takamatsu
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Shitaoka, H. Hamana, H. Kishi, Y. Hayakawa, A. Muraguchi
Writing, review, and/or revision of the manuscript: K. Shitaoka, H. Hamana, H. Kishi, Y. Hayakawa, A. Muraguchi
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Shitaoka, Y. Hayakawa, X. Piao, F. Lyu, M. Murahashi, K. Tani, A. Muraguchi
Study supervision: H. Kishi, A. Muraguchi
This research was supported by MEXT KAKENHI grant numbers JP15H04308 and JP16H06499 (H. Kishi) and JP15K06872 (H. Hamana).
The authors thank Sanae Hirota for providing technical assistance and Kaoru Hata for performing secretarial work. We also thank Professor Toshio Kitamura for generously providing the PLAT-E cells, Professor B. Malissen for kindly suppling the 58α−β− cells and Keisuke Fujii for technical advice.
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.