Activation of NKT Cells in an Anti-PD-1 Resistant Tumor Model … · Tumor Biology and Immunology Activation of NKT Cells in an Anti-PD-1–Resistant Tumor Model Enhances Antitumor

Tumor Biology and Immunology

Activation of NKT Cells in an Anti-PD-1–ResistantTumor Model Enhances Antitumor Immunityby Reinvigorating Exhausted CD8 T CellsEun-Ah Bae1, Hyungseok Seo1,2, Byung-Seok Kim3, Jeongwon Choi1, Insu Jeon1,Kwang-Soo Shin2, Choong-Hyun Koh2, Boyeong Song1, Il-Kyu Kim1,2,Byung Soh Min4, Yoon Dae Han4, Sang Joon Shin5, and Chang-Yuil Kang1,2

Abstract

PD-1–based cancer immunotherapy is a successful exampleof immune checkpoint blockade that provides long-termdurable therapeutic effects in patients with cancer across awide spectrum of cancer types. Accumulating evidence sug-gests that anti-PD-1 therapy enhances antitumor immunity byreversing the function of exhausted T cells in the tumorenvironment. However, the responsiveness rate of patientswith cancer to anti-PD-1 therapy remains low, providing anurgent need for optimization and improvement. In this study,we designed an anti-PD-1–resistant mouse tumor model andshowed that unresponsiveness to anti-PD-1 is associated witha gradual increase in CD8 T-cell exhaustion. We also foundthat invariant natural killer T cell stimulation by the syntheticligand a-galactosylceramide (aGC) can enhance the antitu-mor effect in anti-PD-1–resistant tumors by restoring the

effector function of tumor antigen–specific exhausted CD8T cells. IL2 and IL12 were among the cytokines produced byaGC stimulation critical for reinvigorating exhausted CD8T cells in tumor-bearing mice and patients with cancer.Furthermore, we observed a synergistic increase in theantitumor effect between aGC-loaded antigen-presentingcells and PD-1 blockade in a therapeutic murine tumormodel. Our study suggests NKT cell stimulation as a prom-ising therapeutic strategy for the treatment of patients withanti-PD-1–resistant cancer.

Significance: These findings provide mechanistic insightsinto the application of NKT cell stimulation as a potentadjuvant for immunotherapy against advanced cancer.Cancer Res; 78(18); 5315–26. �2018 AACR.

IntroductionCD8 T cells play a crucial role in eliminating abnormal cells in

the body, including cancer cells. However, prolonged exposure toantigen in the tumor microenvironment results in functionalexhaustion of CD8 T cells, which is associated with the highexpression levels of coinhibitory receptors such as programmeddeath-1 (PD-1, CD279), LAG-3, and Tim-3 (1, 2). To reinvigoratethe antitumor effector functionof exhausted/dysfunctional CD8Tcells, immune checkpoint inhibitors targeting the coinhibitory

receptors have been developed (3–5) and have produced prom-ising outcomes in clinical trials treating patients with melanoma,renal cell cancer, and non–small cell lung cancer (6, 7). Inaddition, the combination of ipilimumab (anti-CTLA4) withnivolumab (anti-PD-1) for the treatment of patients withadvanced melanoma has resulted in more rapid and durableclinical responses with a manageable safety profile (8). However,not all patients with cancer are readily responsive to immunecheckpoint therapy, demonstrating an unmet need to optimizeimmune checkpoint blockade and combination therapies (6, 9).

To guarantee optimal activation of CD8 T cells, three indepen-dent signals are needed: TCR engagement (signal 1), costimula-tion (signal 2), and cytokines (signal 3) (10). It has been suggestedthat these three different signals are also required for the func-tional recovery of exhausted T cells. For example, the CD28/B7costimulatory pathway is essential for the functional restorationof exhausted CD8 T cells by anti-PD-1 therapy in the tumor orduring chronic viral infection (11). In addition, several studieshave shown that various cytokines, such as IL2 or IL12, areimportant for restoring the effector functions of exhausted CD8T cells (12, 13). IL21 is also required for CD8 T-cell effectorfunctions during chronic viral infection (14, 15). Furthermore,tolerant CD8 T cells can be rescued and expanded by exogenousIL15 (16).

Natural killer T (NKT) cells are a unique subset of T lympho-cytes positioned at the border between the innate and adaptiveimmune system. Invariant NKT (iNKT) cells, which express aninvariant TCR composed of Va14-Ja18 chains in mice (Va24-

1Laboratory of Immunology, Department of Molecular Medicine and Biophar-maceutical Sciences, Graduate School of Convergence Science and Technology,and College of Pharmacy, Seoul National University, Seoul, Republic of Korea.2Laboratory of Immunology, Research Institute of Pharmaceutical Sciences,College of Pharmacy, Seoul National University, Seoul, Republic of Korea.3Laboratory of Immune Regulation, Research Institute of PharmaceuticalSciences, College of Pharmacy, Seoul National University, Seoul, Republic ofKorea. 4Department of Surgery, Yonsei University College of Medicine, Seoul,Republic of Korea. 5Department of Internal Medicine, Yonsei University Collegeof Medicine, Seoul, Republic of Korea.

Note: Supplementary data for this article are available at Cancer ResearchOnline (http://cancerres.aacrjournals.org/).

Corresponding Author: Chang-Yuil Kang, Laboratory of Immunology, Collegeof Pharmacy, Seoul National University, 1, Gwanak-ro, Gwanak-gu, Seoul 08826,Republic of Korea. Phone: 82-2-880-7860; Fax: 82-2-885-1373; E-mail:[email protected]

doi: 10.1158/0008-5472.CAN-18-0734

�2018 American Association for Cancer Research.

CancerResearch

www.aacrjournals.org 5315

on November 3, 2020. © 2018 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst July 16, 2018; DOI: 10.1158/0008-5472.CAN-18-0734

http://crossmark.crossref.org/dialog/?doi=10.1158/0008-5472.CAN-18-0734&domain=pdf&date_stamp=2018-8-31

http://cancerres.aacrjournals.org/

Ja18 inhumans), constitute amajor portion ofNKT cells (17, 18).Alpha-galactosylceramide (aGC), which is loaded onto a non-classical class I-likemolecule, CD1d, is a synthetic ligand for iNKTcells (18). Upon in vivo administration, aGCs loaded onto CD1dexpressed in antigen-presenting cells activates iNKT cells to rap-idly produce various cytokines, including IFNg , IL4, IL2, IL10, andIL21 (19–22), and also transactivates dendritic cells (DC) tosecrete IL12 (23, 24). As iNKT cells can induce both innate andadaptive antitumor immunity, iNKT cell activation is one of thepromising therapeutic approaches for cancer immunotherapy(25, 26). While the direct administration of soluble aGC inducesanergy of iNKT cells after a rapid production of cytokines, cell-associated aGC can extend the duration of NKT cell responses(27). In this regard,wehave previously reported thataGC-loaded,antigenic peptide-pulsed B cells elicit potent and long-lastingantitumor immunity (28, 29).

As B cells and monocytes (Mo) could also become immuno-genic antigen-presenting cells (APC) in response to activated iNKTcells in our previous study (28, 30),we established the B/Mo/aGCvaccine platform in which B cells and monocytes serve as profes-sional APCs, and aGC acts as an iNKT cell stimulator (31). Werecently demonstrated that iNKT cell stimulation by aGC-loadedAPCs can eradicate advanced tumors containing MHC classI–deficient tumor cells by reinvigorating the effector function ofNK cells (31). We have also found that IL21 produced by iNKTcells is responsible for the functional recovery of exhausted NKcells (31, 32). However, whether cytokines induced by the admin-istration of aGC-loaded APCs can restore the function ofexhausted CD8 T cells in the tumor environment remains elusive.In this study, we evaluated the effects of cytokines induced byaGC-loaded APCs on exhausted CD8 T cells in the tumors. Wefound that cytokines induced by iNKT cell activation, especiallyIL2 and IL12, could reverse the effector functions of exhaustedCD8 T cells in the tumor microenvironment. In addition, weshowed that the combination of aGC-loaded APCs and PD-1blockade synergistically augmented the antitumor effects in amouse therapeutic tumor model. Collectively, we provide mech-anistic insight into the efficacy of the combined treatment withaGC-loaded APCs and PD-1 blockades by showing that iNKT cellactivation can overcome the exhausted phenotype of tumor-associated CD8 T cells via IL2 and IL12.

Materials and MethodsMice and human samples

Female C57BL/6 and BALB/c mice were purchased fromCharles River Laboratories.OVA-specific T-cell receptor transgenicOT-I mice (C57BL/6-Tg(TcraTcrb)1100Mjb/J) and CD45.1 con-genic mice (B6.SJL-Ptprca Pepcb/BoyJ) were purchased from theJackson Laboratory. OT-I mice were crossed with CD45.1 mice togenerate CD45.1þOT- I mice. All mice were used at 6 to 12 weeksof age andwere bred andmaintained in the specific pathogen-freevivarium of Seoul National University. All animal experimentswere approved by the Institutional Animal Care and Use Com-mittee (IACUC) at Seoul National University.

Human tumor tissue specimens from 14 patients with colo-rectal cancer were collected at the Department of Surgery,Shinchon Severance Hospital, Yonsei University College of Med-icine (Seoul, Republic of Korea). Written informed consent wasobtained from the patients or their guardians. Human peripheralblood mononuclear cells (PBMC) were obtained from 2 healthy

donors in compliance with Institutional Review Board protocols,and written informed consent was granted from all donors. Thecollection of human samples and all human experiments wereperformed in accordance with the principles of the HelsinkiDeclaration and approved by the ethical committee of SeoulNational University and Shinchon Severance Hospital, YonseiUniversity College of Medicine.

The tumor tissues were placed in in RPMI1640 medium(Gibco) supplemented with 10% FBS (Gibco), 1% penicillin/streptomycin (Lonza), 1 mmol/L sodium pyruvate (Gibco), 0.1mmol/L NEAA (Gibco), 55 mmol/L 2-mercaptoethanol (Gibco),and 25mmol/L HEPES (Gibco) on ice before tumor dissociation.Single-cell suspensions from tumors were rested and stimulatedwith recombinant human IL2 (10 ng/mL; R&D Systems, catalogno.: 202-IL) and IL12 (10 ng/mL; R&D Systems, catalog no.: 219-IL) overnight. They were further restimulated with soluble anti-CD3 (1 mg/mL,OKT3; BioLegend) in the presence of cytokines for5 hours.

Tumor cell linesB16F10-OVA cells (kindly provided by Dr. K. Rock, University

of Massachusetts Medical School, Boston, MA) were cultured inDMEM (Gibco) supplemented with 10% FBS, 1% penicillin–streptomycin, 200 mg/mL of geneticin (Gibco), and 60 mg/mL ofhygromycin (Invitrogen). CT26 cells and MC38 cells (purchasedfrom ATCC; CT26 was purchased in 2002, and MC38 was pur-chased in 2006) were cultured in DMEM with 10% FBS and 1%penicillin–streptomycin. Tumor cell lines were validated by mor-phology, growth kinetics, and antigen expression. The cells weredistributed in several vials (1.0 � 106/vial) with culture mediasupplemented with 10% DMSO (Sigma-Aldrich) and stored in aliquid nitrogen tank. All cell lines were found to be negative forMycoplasma contamination using e-Myco Mycoplasma PCRDetection Kit (iNtRON) in 2016, and these cell lines were usedfrom passages 3 to 7 for all experiments.

For the in vivo transplant tumor models, B16F10-OVA cells(1.0 � 106 per mouse), MC38 (3.0 � 105 per mouse), and CT26cells (3.0 � 105 per mouse) were subcutaneously (s.c.) injectedinto the left flank of female C57BL/6 mice and BALB/c mice,respectively. Tumor growth was measured using a metric caliper2–3 times a week. Tumor volume was calculated as 0.5236 �length � width � height as described previously (33). Mousesurvivalwas examinedby actual survival, andmousewhole tumorweight was measured by microbalance upon sacrifice on day 22after tumor inoculation.

Preparation of aGC-cultured supernatantSplenocytes isolated from C57BL/6 mice were homogenized

using a 70-mm cell strainer (BD FALCON), and red blood cells(RBC) were lysed using ACK lysing buffer (Gibco). Isolated totalsplenocytes (3.3 � 106/mL) were stimulated with aGC (200ng/mL; KRN7000, Enzo Life Sciences) or their vehicle DMSO asa control for 24 hours. The aGC-cultured supernatant (aGC Sup)was collected and stored at�20�C. For aGC-loaded APC (B/Mo/aGC)-stimulated culture supernatants, total splenocytes werecultured with aGC-loaded B/Mo or their vehicle-loaded B/Moat 5:1 for 24 hours. B/Mo/aGC Sup was also stored at �20�Cbefore use.

To make human aGC-cultured supernatant, PBMCs were pre-pared by density gradient centrifugation using Ficoll-Histopaque-1077Hybri-Max (Sigma-Aldrich) as described previously (34). To

Bae et al.

Cancer Res; 78(18) September 15, 2018 Cancer Research5316




enrich for iNKT cells, PBMCs were cultured with aGC for 7 daysand then cultured for an additional 7–10 days in the presenceof recombinant human IL2 (200 U/mL). Cultured iNKT cellswere purified with anti-human iNKT microbeads (MiltenyiBiotec). To prepare aGC-loaded APCs in humans, T-cell–depleted PBMCs were incubated with aGC (1 mg/mL) orDMSO overnight. In a 96-well plate, 2.0 � 105 iNKT cellswere stimulated with 1.0 � 106 aGC-loaded APCs for 24 hoursas previously demonstrated with some modifications (35). Theresultant human aGC Sup was stored at �20�C before use.

Preparation of aGC-loaded APCSplenocytes were isolated from C57BL/6 or BALB/c mice. After

eliminating RBCs using ACK lysing buffer, the CD11bþ cells andB220þ cells were purified using anti-mouse CD11b and anti-mouse B220 microbeads (all from Miltenyi Biotec). Isolated Bcells andmonocytes (Mo)were incubatedwithaGC (1 mg/mL) ortheir vehicle DMSO for 18 hours based on our previous studies(28–31, 36–38).

Isolation and activation of Ag-specific CD8 T cellsTotal CD8 T cells isolated from the spleen and lymph nodes of

OT-I mice by positive selection using anti-mouse CD8 microbe-ads (Miltenyi Biotec) were activated by anti-CD3 (clone 145-2C11) and anti-CD28 (clone 37.51; BioLegend) at 1 mg/mL onprecoated plates based on previous studies (39). On day 2, T cellswere harvested from the antibody-coated plate and amplifiedwith low-dose recombinant IL2 (PeproTech, catalog no.: 212-12)until day 5, when the cells were adoptively transferred into tumor-bearing mice.

Adoptive T-cell transfer and antibody treatments in the B16F10-OVA tumor model

CD45.1þOT-I CD8 T cells generated in vitrowere transferred byintravenous injection (2–3 � 106 per mouse) into the tumor-bearingmice 12 days after B16F10-OVA tumor inoculation. For exvivo analyses, mice were sacrificed on day 8 after adoptive T-celltransfer, and T cells in the tumor were isolated as described in thesection on tumor-infiltrating lymphocyte preparation. To identifyin vivo effects of aGC, 8 days after adoptive T-cell transfer, 2.0 �106 B cells andmonocytes loadedwithaGC (B/Mo/aGC) or theircontrol-loaded B/Mo were transferred into the tumor-bearingmice by intravenous injection. The mice were sacrificed 2 daysafter B/Mo/aGC transfer.

In selected experiments, to deplete CD8 T cells, 300 mg ofanti-CD8 (2.43, catalog no.: BE0061) or control antibody wasinjected intraperitoneally (i.p.) every 3 days into the mice,beginning on the day before B/Mo/aGC transfer. To neutral-ize IL2 and IL12, anti-IL2 (JES6-5H4, catalog no.: BE0042),anti-IL12p40 (C17.8, catalog no.: BE0051), or both wereinjected intraperitoneally (1 mg/mouse) 2 hours before theB/Mo/aGC transfer, and then 24, 48 hours after the firstinjection (500 mg/mouse).

In combined therapy experiments, B/Mo/aGC was injectedintravenously when the tumor size was greater than 200 mm3

and 300 mg of anti-PD-1 (RMP1-14, catalog no.: BE0146) treat-ment was administered intraperitoneally every 3 days beginningon the same day as B/Mo/aGC. The antibodies for in vivo deple-tion (CD8, IL2, IL12p40, PD-1, and control antibodies) werepurchased from BioXcell.

Preparation of tumor-infiltrating lymphocyteTumor tissues frommice andpatientswere cut into small pieces

(approximately 2–5 mm) and dissociated in C tubes using thegentleMACS Dissociator (Miltenyi Biotec). Single-cell suspen-sions from dissociated mouse tumor samples were prepared aspreviously described with minor modifications (32, 33). Briefly,tumors were digested in 2% FBS with RPMI medium containing1 mg/mL collagenase D (Roche), 100 mg/mL hyaluronidase(Sigma Aldrich), and 100 mg/mL DNase I (Sigma Aldrich) at37�C for 30 minutes, followed by lymphocyte isolation usinglymphocyte separation medium (MP Biomedicals, catalog no.:0850494). The patient tumor samples were digested with thetumor dissociation kit containing enzyme A, H, and R (MiltenyiBiotec, catalog no.: 130-095-929) at 37�C using the manufac-turer's recommended instructions. After dissociation, single-cellsuspensions were filtered through a 70-mm cell strainer and usedfor the experiments.

Antibodies and flow cytometry analysisCD8 T cells were isolated from tumor-infiltrating lymphocytes

using anti-mouseCD8microbeads (Miltenyi Biotec)with aMACSLS column (Miltenyi Biotec). Isolated tumor-infiltrating CD8 Tcells were stimulatedwithaGC Sup (containing 50%) and T-cell–depleted APCs loaded with OT-I peptide (SIINFEKL, 257-264;AnaSPEC) for 3 days. T-cell–depleted APCs were generated byCD3 depletion as described previously (40). Briefly, splenocyteswere stained with biotinylated anti-CD3 mAb (BioLegend) andanti-biotin microbeads (Miltenyi Biotec), after which a MACS LDcolumn (Miltenyi Biotec) was used. In a selected experiment,CD38hiCD101hi and CD38lowCD101low CD8 T cells were furthersorted for CD45.1þusing a FACSAIRA II at theNational Center forInter-University Research Facilities (NCIRF) at Seoul NationalUniversity (Seoul, Republic of Korea).

The cells (mouse and human primary cells: 1� 105–106) werestained with the specified antibodies in FACS buffer (PBSNþ1%FBS). The antibodies for flow cytometry were purchased fromBioLegend and eBioscience and BD Biosciences. For intracellularcytokine staining, CD8 T cells were restimulated with brefeldinA (GolgiPlug; BD Biosciences) and monensin (GolgiStop; BDBiosciences) for the last 4 hours, followed by fixation andpermeabilization using the Cytofix-Cytoperm kit (BD Bio-sciences). A phycoerythrin-conjugated CD107a antibody(1D4B; BioLegend, catalog no.:1216) was added during cellrestimulation. For transcription factor staining, cells were per-meabilized with the Foxp3 staining kit (eBioscience) accordingto the manufacturer's instructions.

The antibodies against mouse PD-1 (RMP1-30, catalog no.:1091), Tim3 (RMT3-23, catalog no.: 1197), LAG3 (C9B7W,catalog no.: 1252), CD3e (145-2C11, catalog no.: 1003), CD8a(53-6.7, catalog no.: 1007), TNFa (MP6-XT22, catalog no.: 5063),granzyme B (GB11, catalog no.: 5154), CD44 (IM7, catalog no.:1030), CD28 (37.51, catalog no.: 1021), CD38 (90, catalog no.:1027), Ki67 (16A8, catalog no.: 6524), CD45.2 (104, catalog no.:1098), CD45.1 (A20, catalog no.: 1107), CD69 (H1.2F3, catalogno.: 1045), B220 (RA3-6B2, catalog no.: 1032) CD11b (M1/70,catalog no.: 1012), CD11c (N418, catalog no.: 1173), and IL2(JES6-5H4, catalog no.: 5038) were purchased from BioLegend.The antibody against mouse IL12 (p40/p70; C45.6, catalog no.:554479) was purchased from BD Biosciences. The antibodiesagainst mouse IFNg (XMG1.2, catalog no.: 7311), PD-1 (J43,catalog no.: 9985), CD101 (Moushi101, catalog no.: 1011), T-bet

NKT Activation Reinvigorates Exhausted CD8 T Cells in Tumors

www.aacrjournals.org Cancer Res; 78(18) September 15, 2018 5317




(4B10, catalog no.: 5825), and human IFNg (4S.B3, catalog no.:7319) were purchased from eBioscience. The antibodies againsthuman CD45 (2D1, catalog no.: 3685), CD3 (SK7, catalog no.:3448), CD8a (RPA-T8, catalog no.: 3010), and TNFa (MAB11,catalog no.: 5029)were purchased fromBioLegend. All antibodieswere used at a concentration of 2 mg/mL except for the FITC-conjugated antibodies, whichwere used at 5mg/mL. The cellswereincubated at 4�C in the dark for 15minutes and thenwerewashedwith FACS buffer. Dead cells were excluded by staining with theFixable Violet Dead cell kit (Molecular Probes, Invitrogen) orFixable Viability Dye (eBioscience) following the manufacturers'instructions. Intracellular staining was performed for 30 minutesafter fixation and permeabilization according to the manufac-turer's instructions. The samples were analyzed with a FACSAriaIII or LSRFortessa X-20 instrument (BD Biosciences), and the datawere analyzed with FlowJo software (Tree Star). Tumor-infiltrat-ing CD45.1þOT-I cells were calculated as follows:

¼ % of CD45:1þOT � I cells�No:of tumor � infiltrating lymphocytesTumor weight ðmgÞ

� �:

In vitro cytotoxicity assayEffector CD8 T cells were prepared from the tumor-bearing

mice that hadbeen transferredwithCD45.1þOT-I cells andB/Mo/aGC. CD8 T cells were isolated from tumor-infiltrating lympho-cytes using anti-mouse CD8 microbeads (Miltenyi Biotec) andused to further sort CD45.1þ populations using a FACSARIA III(BD Biosciences). Target cells were B16F10-OVA cells that hadbeen cultured in the presence of murine recombinant IFNg (10ng/mL, R&DSystems) for 24 hours to induce increased expressionof MHC class I. CD45.1þCD8 T cells were cocultured with 51Cr-labeled B16F10-OVA tumor cells for 4–5 hours. The effector-to-target cell ratios (E:T) were 3:1 and 20:1, and the CTL activity wascalculated by the release of 51Cr into the culture supernatants viathe specific lysis of B16F10-OVA target cells, which was measuredusing a Wallac 1470 Wizard automatic g-counter (PerkinElmer).Specific lysis (%) was calculated as follows: % ¼ [(experimentalrelease-spontaneous release)/(maximum release–spontaneousrelease)] � 100 as described previously (37).

Statistical analysisStatistical comparisons were performed using Prism 6.0

software (GraphPad Software). In experiments with mice, anunpaired two-tailed Student t test was used to compare twogroups (Figs. 2C–F and 3B–D; Supplementary Figs. S4A andS4B and S5B). Two-way ANOVA with Bonferroni multiplecomparisons test was used for multiple comparisons (Figs.1D and G, 2B, 3E, 4A and C, and 6B; Supplementary Figs.S1A–S1C, S2, and S6). A log-rank (Mantel–Cox) test (conser-vative) was used to analyze tumor survival (Figs. 1E and Hand 6C). For human samples, the paired Student t test was used(Fig. 5B, D, and E). P values <0.05 were considered statisticallysignificant.

ResultsAdministration of aGC-loaded APCs inhibits tumor growth inan anti-PD-1–resistant tumor model

To investigate the effects of PD-1 blockade on exhausted CD8 Tcells in tumors, we subcutaneously inoculated C57BL/6 micewith B16F10-OVA, a murine melanoma cell line expressing the

exogenous antigen ovalbumin (OVA). As previously shown byothers, the cell surface expression of PD-1 in tumor-infiltratingCD8 T cells gradually increased with tumor progression, whereasthat in tumor-draining lymph nodes was not upregulated (Fig. 1Aand B; refs. 1, 39, 41). Of note, anti-PD-1 antibody treatmenteradicated tumors if administered at the early stage of tumorprogression when the tumor was palpable (anti-PD-1 workingmodel); however, tumor growth was not controlled by the treat-ment of anti-PD-1 at the later timepointswhen the tumor sizewasapproximately 200mm3 (anti-PD-1–resistant model; Fig. 1C–E).Previously, we showed that the aGC-loaded, tumor antigen–expressing APC-based vaccine could eradicate advanced tumors(31). Thus, we examined whether treatment with aGC-loaded Bcells and monocytes (aGC-loaded APCs) could induce antitu-mor responses in the anti-PD-1–resistant model. The admin-istration of aGC-loaded APCs significantly inhibited tumorgrowth, which was accompanied by the increase in overallsurvival rates (Fig. 1F–H). We also observed comparable iNKTcell activation and tumor regression with intravenous injectionof aGC (Supplementary Fig. S1A and S1B). The activationstatus of iNKT cells in the tumor site was not changed withaGC treatment regardless of the delivery system, althoughintravenous injection of aGC more effectively expanded iNKTcells in the spleen than aGC-loaded APC administration (Sup-plementary Fig. S1B and S1C). Tumor growth in another anti-PD-1-resistant tumor model was also inhibited by the treat-ment of aGC-loaded APCs (Supplementary Fig. S2). Further-more, we found that depletion of CD8 T cells significantlyreversed the inhibition of tumor growth, implying that antitu-mor immunity mediated by the treatment of aGC-loaded APCswas dependent on CD8 T cells (Fig. 1G and H).

aGC-loaded APCs augment antitumor immunity of exhaustedCD8 T cells in vivo

To explorewhether the enhanced antitumor activity induced byaGC-loaded APCs in the anti-PD-1–resistant tumor model wasdue to the reinvigoration of tumor antigen–specific exhaustedCD8 T cells, we utilized theOT-I TCR transgenicmouse system, inwhich most CD8 T cells recognize OVA antigenic peptide loadedonto MHC class I. A recent study demonstrated that 8 days aftertransfer of in vitro–generated OT-I effector CD8 T cells (OT-I cells)into mice bearing B16-OVA, OT-I cells in the tumor environmentproduced few cytokines and were found to be exhausted (39).CD45.1þOT-I cells that were activated with anti-CD3/anti-CD28for 2 days and then expanded with low-dose IL2 were transferredinto B16F10-OVA–bearing C57BL/6 mice (Fig. 2A). Flow cyto-metric analysis confirmed that all the effector OT-I cells expressedlow levels of inhibitory receptors and high levels of CD44,indicating sufficient activation of the donor cells before transfer(Supplementary Fig. S3A). aGC-loaded APCs were intravenouslyinjected 8 days after OT-I adoptive transfer when the transferredOT-I cells became exhausted, considering the high expression ofPD-1, Tim3, and LAG3 (Supplementary Fig. S3B). We confirmedthat the transferred OT- I cells had defects in producing effectormolecules compared to effector OT- I cells upon restimulationwith peptides (Supplementary Fig. S3C). Treatment of aGC-loaded APCs resulted in delayed tumor growth, which wasdependent on CD8 T cells (Fig. 2B). Two days after the admin-istration of aGC-loaded APCs, tumor-infiltrating exhausted OT-Icells expanded and produced more effector molecules, such asIFNg , TNFa, CD107a, and granzyme B, than those in control

Bae et al.





APC–treated mice (Fig. 2C and D). In addition, the expressionlevels of T-bet, ki-67, andCD28 on tumor-infiltratingOT-I cells inaGC-loaded APC-treated mice were increased compared withthose levels in the control APC–treated mice. We also observedthat PD-1 expression in tumor-infiltrating OT-I cells was signif-icantly downregulated by aGC-loaded APC administration (Fig.2E). Consistent with the phenotypic changes, the cytotoxicity oftumor-infiltrating OT-I cells against B16F10-OVA tumor cells wasaugmented by the treatment ofaGC-loaded APCs (Fig. 2F). Takentogether, these data indicate that treatment of aGC-loaded APCsenhanced antitumor immunity by restoring the polyfunctional-ity, proliferative capacity, and cytotoxicity of the tumor antigen–specific CD8 T cells.

Exhausted CD8 T cells in the tumor can be reinvigorated byaGC-induced cytokines

Given that iNKT cells stimulated by aGC produce variouscytokines, we hypothesized that cytokines induced by iNKT cellactivation via aGC-loaded APC transfer might have affectedexhausted CD8 T cells in the tumor environment. To identifythe NKT cell activation–induced cytokine(s) responsible forthis effect, we obtained cytokine mixtures (aGC Supernatant;aGC Sup) by culturing total splenocytes with aGC for 24 hours.Exhausted OT-I cells, isolated as described in Fig. 3A, werecocultured with peptide and APCs in the presence of aGC Supor control Sup for 3 days (Fig. 3A). In result, aGC Sup enhancedthe Ag-specific expansion of exhausted OT-I cells compared

Figure 1.

Administration of aGC-loaded APCs inhibits tumor growth in an anti-PD-1–resistant tumor model. A–H, C57BL/6 mice were subcutaneously injected withB16F10-OVA. A and B, CD8 T cells were obtained from tumors and tumor-draining LNs (TdLN) on the indicated days after tumor inoculation. Representative flowcytometry plots and graphs show the intensity of PD-1 expression. Data are representative of two ormore independent experiments. C–E, Tumor-bearingmice weretreated with 300 mg of control IgG or anti-PD-1 on days 6, 9, 12, 15, and 18 (anti-PD-1 working model) or on days 12, 15, and 18 (anti-PD-1-resistant model).F–H, Tumor-bearing mice were treated with control IgG or anti-PD-1 on days 12, 15, and 18 (anti-PD-1-resistant model) or injected with B/Mo/aGC or their controlB/Mo on days 12 and 18. To deplete CD8 T cells, tumor-bearingmice were intraperitoneally injected with anti-CD8 (2.43) every three days, starting on the day beforeinitiation of the treatments. D and G, Tumor growth was measured three times weekly, and tumor sizes are presented as the means� SEM. Statistical differences intumor sizes were determined by two-way ANOVA with Bonferroni multiple comparison tests. E and H, Survival curves from data in D and G. The data arerepresentative of two independent experiments that included 4 to 6 mice per group. Comparisons were performed using the log-rank (Mantel–Cox) test(� , P < 0.05; �� , P < 0.01; �� , P < 0.001; �� , P < 0.0001).






with control Sup (Fig. 3B). aGC Sup–stimulated exhausted OT-I cells expressed higher level of effector molecules, includingIFNg , TNFa, CD107a, and granzyme B than those of controlSup–treated cells (Fig. 3C). We observed a similar enhancementof effector functions of exhausted OT-I cells by treatment withthe culture supernatant obtained from splenocytes stimulatedwith aGC-loaded APCs (Supplementary Fig. S4A and S4B).Moreover, expression of T-bet and ki67 was elevated in aGCSup–treated OT-I cells (Fig. 3D). CD28 expression was mar-ginally increased by aGC Sup treatment compared with thecontrol, while the PD-1 level was not significantly changed. A

recent study has demonstrated that CD8 T cells in tumors showtwo distinct chromatin states: CD38lowCD101low CD8 T cellsare in a reversible dysfunctional state, whereas CD38hiCD101hi

CD8 T cells are in a fixed dysfunctional state that is resistant toreprogramming. Tumor-infiltrating CD38hiCD101hi CD8 Tcells exhibit a defect in regaining the ability to secrete cytokineseven after exposure to IL15, which can restore the function ofCD38lowCD101low CD8 T cells (42). To delineate that thisdichotomy could also be applied to the reinvigorating effectof aGC Sup, we sorted each CD38hiCD101hi and CD38lowC-D101low cells from the tumor and treated with aGC Sup in our

Figure 2.

aGC-loaded APCs augment the antitumor immunity of Ag-specific exhausted CD8 T cells in vivo. A, Experimental scheme. In vitro–generated effector congenicOT-I CD8 T cells were transferred into B16F10-OVA–bearingmice. B/Mo/aGC or their control B/Mowere injected 8 days after in vivoOT-I adoptive transfer. To depleteCD8 T cells, tumor-bearing mice were intraperitoneally injected with anti-CD8 (2.43) every three days, starting one day before the treatment of aGC-loadedAPCs. B, Tumor growth was measured three times weekly, and tumor sizes are presented as the means� SEM. Data show one representative experiment (6–9 miceper group) out of two independent experiments. Statistical differences in tumor sizes were determined by two-way ANOVA with Bonferroni multiple comparisontests. C–F, TIL CD8 T cells were analyzed 2 days after the aGC-loaded APC transfer. C, Representative flow cytometry plots are shown for Ag-specific CD8 T cells intumors. The graphs show the frequency and quantification of Ag-specific CD45.1þOT-I cells in TILs. D, Intracellular staining for the coproduction of IFNg , TNFa,CD107a, and granzyme B upon the stimulation of TIL CD8 T cells with OT-I peptide. E, T-bet, Ki67, PD-1, and CD28 in CD45.1þ Ag-specific CD8 T cells were analyzedby flow cytometry. Overlaid histogram plots are representative of three independent experiments (shaded histograms, control B/Mo–treated group; filledhistograms, aGC-loaded B/Mo–treated group; open histograms, isotype control). The mean fluorescence intensities of T-bet, Ki67, PD-1, and CD28 on the tumor-infiltrating congenic OT-I cells between vehicle and aGC-loaded APC-treated mice are shown. The data are presented as the means � SEM and are representativeof two or more independent experiments. F, Analysis of the cytotoxicity of Ag-specific CD8 T cells in tumors 2 days after the aGC-loaded APC treatment.51Cr-labeled B16F10-OVA tumor cells were used as target cells. The data are presented as the means � SEM and representative of two independent experiments.The data in C–F were analyzed using an unpaired two-tailed Student t test (n.s., nonsignificant; � , P < 0.05; �� , P < 0.01; �� , P < 0.001; �� , P < 0.0001).

Bae et al.





Figure 3.

Exhausted CD8 T cells in tumors can be reinvigorated by aGC-induced cytokines. A, Experimental scheme. In vitro–generated effector congenicOT-I cells were transferred into B16F10-OVA bearing mice. Eight days after in vivo OT-I adoptive transfer, TIL CD8 T cells were isolated and restimulatedwith control or aGC Sup (aGC S; 50%), OT-I peptide, and CD3-depleted APCs for 3 days. B and C, Representative flow cytometry plots are shownfor Ag-specific CD8 T cells from tumors and intracellular cytokine staining for the coproduction of IFNg , TNFa, CD107a, and granzyme B. Graphs show asummary of the frequency of Ag-specific CD8 T cells and cytokine coproduction. D, The mean fluorescence intensities of T-bet, Ki67, PD-1, and CD28 on thetumor-infiltrating congenic OT-I cells between the control and aGC Sup–treated group are shown. B–D, Statistical differences between the controland aGC Sup–treated group were determined using an unpaired two-tailed Student t test. E, In the same system described in A, to neutralize cytokines in vitro,the indicated antibodies (10 mg/mL) were added in the aGC Sup–treated group during restimulation for 3 days. The frequency of Ag-specific CD8 T cellsand cytokine coproduction were analyzed by flow cytometry. Statistical differences between the aGC Sup–treated group with or without cytokineneutralization were determined using two-way ANOVA with Bonferroni multiple comparison tests. Data in B–E are presented as the means � SEM and arerepresentative of three independent experiments (n.s., nonsignificant; � , P < 0.05; �� , P < 0.01; �� , P < 0.001; �� , P < 0.0001).






in vitro restimulation experiment. Intriguingly, CD38hiCD101hi

as well as CD38lowCD101low exhausted OT-I cells, were sus-ceptible to aGC Sup–mediated reinvigoration of the effectorfunction (Supplementary Fig. S5A and S5B). Next, we examinedwhich cytokines induced by aGC stimulation were involved inrestoring CD8 T-cell exhaustion. Using an in vitro cytokine-neutralizing system, we found that blockade of either IL2 orIL12 significantly abrogated the recovery of the expansion andeffector molecule expression of exhausted OT-I cells induced byaGC Sup (Fig. 3E). These results indicate that iNKT stimulationby aGC restores the effector function of exhausted CD8 T cellsvia IL2 and IL12 induced by NKT cell activation.

IL2 and IL12 induced by aGC-loaded APCs promote antitumorimmunity in vivo

Having confirmed that IL2 and IL12 induced by aGC stimu-lation were responsible for the restoration of exhausted CD8 Tcells in vitro,we analyzed the cellular sources of IL2 and IL12 afterthe administration of aGC-loaded APCs by flow cytometry. Wefound that splenic iNKT cells mainly produced IL2, while bothCD11cþ DCs and CD11bþ myeloid cells contributed to IL12production by splenocytes (Fig. 4A). We sought to evaluatewhether these cytokines were also attributed to enhanced anti-

tumor activity by aGC-loaded APCs. To achieve this goal, tumor-bearing mice were treated with anti-IL2 and/or anti-IL12p40neutralizing mAbs concomitant with the aGC-loaded APCtreatment. Administration of either anti-IL2 or anti-IL12p40completely abrogated aGC-loaded APC-mediated regression oftumor growth, suggesting that both IL2 and IL12 contributed tothe antitumor effect of the aGC-loaded APC treatment (Fig. 4Band C).

IL2 and IL12 also increase the effector function of tumor-infiltrating CD8 T cells in patients with cancer

We next investigated whether the effector function of humantumor-infiltrating lymphocyte (TIL) CD8 T cells could beenhanced by the cytokines induced by iNKT cell stimulation. Tothis end, culture supernatants were prepared by stimulating iNKTcells among PBMCs isolated from healthy subjects with aGC-loaded APCs for 24 hours. When TILs isolated from patients withcolorectal cancer were stimulated with the aGC-treated culturesupernatant, the frequency of IFNgþTNFaþ effector CD8 T cellswas significantly increased compared with that in the controls(Fig. 5A and B). To directly test whether the two cytokinesidentified from the mouse models were also involved in thereinvigorating effect of the aGC-treated culture supernatant, the

Figure 4.

The antitumor effect of aGC-loaded APCs is dependent on IL2 and IL12. A, IL2 production by iNKT cells, T cells and CD3� non-T cells, and IL12 productionby CD11cþ, CD11bþ, and B cells within the splenocytes isolated from B16F10-OVA tumor-bearingmice were analyzed 6 hours after aGC-loaded APC treatment. B andC, In vitro–generated effector congenic OT-I cells were transferred into B16F10-OVA–bearing mice. Eight days after in vivo OT-I adoptive transfer, aGC-loadedAPCs (B/Mo/aGC) or their control B/Mo were injected. To neutralize IL2 and IL12, anti-IL2 (JES6-5H4), anti-IL12p40 (C17.8), or both were administeredintraperitoneally (1 mg/mouse) 2 hours before the B/Mo/aGC transfer and then 24 and 48 hours after the first injection (500 mg/mouse). C, Tumor growthwas measured three times weekly, and tumor sizes are presented as the means � SEM. Statistical differences in A and C were determined by two-wayANOVA with Bonferroni multiple comparison tests (� , P < 0.05; ��, P < 0.01; �� , P < 0.0001).

Bae et al.





effector function of TIL CD8 T cells was examined after treatmentwith rhIL2, rhIL12, or both. Consistent with the conclusion inmouse studies, either rhIL2 or rhIL12 significantly increased thefrequency of IFNgþTFNaþ cells among TIL CD8 T cells comparedwith that in the PBS controls (Fig. 5C and D). While the additionof rhIL12 partially increased TNFa-producing cells among TIL

CD8 T cells, rhIL2 significantly induced both IFNg- and TNFa-producing TIL CD8 T cells (Fig. 5E). We observed little if anyadditive effect induced by the combination of rhIL2 and rhIL12.These results suggest that iNKT cell activation can contribute to thereinvigoration of human tumor-infiltrating exhausted CD8 T cellsvia the actions of IL2 and IL12.

Figure 5.

IL2 and IL12 increase the effector function of tumor-infiltrating CD8 T cells in patients with cancer. TILs isolated from patient tumors were treated withcontrol or human aGC Sup (aGC S) in A and B and control PBS, IL2 (10 ng/mL), IL12 (10 ng/mL), and IL2 plus IL12 overnight in C–E. A–E, Production of IFNg and/orTNFa by CD8 TILs upon anti-CD3 stimulation (1 mg/mL) for 5 hours, in the presence of the protein transport inhibitor brefeldin A and monensin during thelast 4 hours. A and C, Representative flow cytometry plots showing the coproduction of IFNg and/or TNFa by human aGC Sup and the indicated cytokinesrespectively.B,Arbitrary units of the percentages of IFNgþTNFaþ cells in TIL CD8 T cells are depicted using fold change normalized by control Sup–treated samples.D, Data showing the percentages of IFNgþTNFaþ cells in TIL CD8 T cells are depicted. E, The plots depict the mean fluorescence intensities of IFNg or TNFafromCD8 T cells in tumors, by the addition of the indicated cytokines. The values determined for individual patients are depicted, and the lines inD denote themeans.Data in B, D, and E were analyzed using two-tailed paired t test (n.s., nonsignificant; � , P < 0.05; �� , P < 0.01).






Combined therapy with anti-PD-1 blockade and aGC-loadedAPC treatment

Wenext investigated the therapeutic efficacyofanti-PD-1 therapywas enhanced when combined withaGC-loaded APCs in the anti-PD-1–resistant tumormodel (Fig. 6A). Aswe demonstrated earlier,although tumor-bearing mice were refractory to four times of anti-PD-1 treatment from day 13 after tumor inoculation, aGC-loadedAPCs substantially suppressed tumor growth if administered twiceat day 13 and six days later. Importantly, the combination ofaGC-loaded APCs with anti-PD-1 completely eradicated the tumors andprolonged overall survival of tumor-bearing mice. The antitumoreffect of combination therapy was dependent on CD8 T cells, asshown that this effect was almost completely abrogated by CD8 T-cell depletion (Fig. 6B and C). In the MC38 tumor model, thecombination of aGC-loaded APCs with anti-PD-1 also inhibitedtumor growth more effectively than treatment with aGC-loadedAPCs alone (Supplementary Fig. S6). Altogether, these resultsindicate that the combination of aGC-loaded APCs and PD-1blockade elicits strongCD8T-cell–dependent antitumor responses.

DiscussionDespite the recent success of using an anti-PD-1 antibody as a

first-line therapy against advanced cancers, more than half ofpatients with cancer tested in clinical trials remained refractory tothe therapy (6). To improve the clinical outcomes of anti-PD-1

therapy,many combination strategies are under development (9).Oneof themain goals of combination therapies is overcoming theunresponsiveness of exhausted CD8 T cells in the tumor to anti-PD-1 therapy. Here, we found that iNKT cell activation by cell-associated aGC could inhibit the progression of anti-PD-1–resis-tant tumors by restoring the effector function of exhausted CD8 Tcells in tumors. We also revealed that IL2 and IL12 produced afteriNKT cell activationwere themain components that reinvigoratedexhausted CD8 T cells. While activated iNKT cells were a mainproducer of IL2, IL12 was primarily produced by dendritic cellsand myeloid cells. It seems that IL12 induction in APCs in thisexperimental setting could have been attributed to IFNg producedby either activated iNKT cells or transactivated NK cells (20, 43).

The functional relevance of IL2 and IL12 in reinvigoratingexhausted CD8 T cells in tumor models corresponds well withearlier studies that demonstrated similar effects of these cytokineson virus infection–induced exhausted T cells. A previous reportshowed that IL2 enhanced proliferative potential and granzyme Bproduction by virus-specific CD8 T cells in LCMV chronic infec-tionmodels (12). In addition, IL12 but not type 1 IFN rescued theexhausted HBV-specific CD8 T cells by inducing downregulationof PD-1 and upregulation of T-bet (13). In agreement with thesestudies, we confirmed that the increased effector function andantitumor immunity of tumor-infiltrating exhausted CD8 T cellsbyaGC-loaded APC treatmentwas dependent on IL2 and IL12. Inaddition to these cytokines, IL21 can also play a crucial role in the

Figure 6.

Combined therapy with anti-PD-1 blockade and aGC-loaded APCs. A–C, B16F10-OVA-bearing mice were treated with control IgG or anti-PD-1 on days 13, 16, 19,and 22 (starting when the tumor size exceeded over 200 mm3). They were coadministered with B/Mo/aGC or control B/Mo on days 13 and 19. To depleteCD8 T cells, tumor-bearingmicewere intraperitoneally injectedwith anti-CD8 (2.43) every three days, starting on day 12.B, Tumor growthwasmeasured three timesweekly, and tumor sizes are presented as the means � SEM. Statistical differences in tumor sizes were determined by two-way ANOVA with Bonferronimultiple comparison tests. C, Survival curves from the data in B. The data included 5 to 6 mice per group. Comparisons were performed using the log-rank(Mantel–Cox) test (� , P < 0.05; ��, P < 0.01; ��, P < 0.001).

Bae et al.





maintenance and polyfunctionality of exhausted CD8 T and NKcells (14, 15, 31). However, we could not find any role for IL21 inthe restoration of exhausted CD8 T cells followingaGC treatmentin our in vitro system, although further analyses in more physi-ologically relevant in vivo settings are warranted.

A recent study demonstrated that exhausted CD8 T cells grad-ually upregulated the expression of CD38 and CD101 with fixedchromatin states, which was associated with their rigid dysfunc-tional status (42). In that study, the effector function ofCD38hiCD101hi CD8 T cells was not reversed by IL15, a cytokineknown to enhance effector functions of tolerant CD8 T cells intumor models (16, 42). Although our data showed that aGC-induced cytokines could partially enhance IFNg and TNFa pro-duction by CD38hiCD101hi CD8 T cells, it remains unclearwhether they influence chromatin states. Blocking PD-1 signalingreinvigorates exhausted CD8 T cells in LCMV infection models,while the epigenetic fate of T cells is not significantly affected (44).In addition, patterns of chromatin accessibility of exhausted CD8T cells are only slightly changed by anti-PD-L1 treatment intumor-bearing mice (39). Therefore, further studies are neededto examine whether cytokines induced by aGC can change thechromatin states of exhausted CD8 T cells.

Previous studies have identified PD-1hi and PD-1int subsets ofexhausted CD8 T cells in chronic viral infection models (45). PD-L1 blockade reduces spontaneous apoptosis and enhances theexpansion and protective immunity of PD-1int CD8 T cells only(45). CD8 T cells marked with a transcriptional profile of T-bethi

(PD-1int) represent the progenitor T-cell subset and are ultimatelyconverted to Eomeshi (PD-1hi) CD8 T cells, which represent theterminally differentiated effector progeny (46). Our data con-firmed the downregulation of PD-1 and increase in T-bet ontumor-infiltrating, Ag-specific exhausted CD8 T cells in aGC-loaded APC-treatedmice, whichmight be one of themechanismsby which aGC-loaded APCs overcome the exhausted phenotypeof CD8 T cells. In addition, recent studies have shown that CD40stimulation by agonistic Ab can overcome resistance to anti-PD-1therapy by converting PD-1hi T cells into PD-1low T cells (47).Furthermore, CD28 costimulation is also required for the effec-tiveness of anti-PD-1 therapy (11). Our results showed that theadministration ofaGC-loaded APCs induced the downregulationof PD-1 and upregulation of CD28 on Ag-specific exhausted CD8T cells in tumors, and these phenotypic changesmight provide therationale for the ability of PD-1 blockade to enhance the antitu-mor effect of aGC-loaded APC treatment.

Administration of aGC is known to induce iNKT cell anergy bytriggering the upregulation of PD-1 on iNKT cells (48). We andothers have previously demonstrated that blocking the PD-1/PD-L1 interaction can prevent anergy induction in iNKT cells andenhance the antitumor effects of iNKT cells (48–50). In advancedtumors, NK cells also express Tim-3 and PD-1 and become

functionally exhausted (31). In this regard, aGC-loaded APCs incombination with PD-1 blockade may have effects on exhaustedNK cells and anergic iNKT cells, as well as exhausted CD8 T cells(32). Although the administration of aGC-loaded APCs couldreverse the exhausted phenotype of CD8 T cells, it was notsufficient to completely reject the tumor mass. Thus, we suggestthat combined therapy between aGC-loaded APCs and PD-1blockades not only reinvigorates exhausted CD8 T cells, but alsoboosts diverse effector arms of antitumor immunity.

This is thefirst report to demonstrate that the cytokines inducedby aGC can restore the antitumor effector function of exhaustedCD8 T cells in anti-PD-1–resistant mouse tumor models. Amongthe cytokines, IL2 and IL12 are crucial for enhancing cytokineproduction by exhausted CD8 T cells in tumor-bearing mice andin patients with cancer. Furthermore, we have clearly shown thesynergism of the antitumor effect between aGC-loaded APCtreatment and PD-1 blockade. Therefore, our study providesevidence for the application of cell-associatedaGCas an adjuvantin anti-PD-1 therapy for patients with cancer.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Authors' ContributionsConception and design: E.-A. Bae, H. Seo, C.-Y. KangDevelopment of methodology: E.-A. Bae, H. Seo, I. Jeon, S.J. Shin, C.-Y. KangAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): E.-A. Bae, I. Jeon, K.-S. Shin, B.S. Min, Y.D. Han,S.J. ShinAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): E.-A. Bae, H. Seo, B.-S. Kim, B. Song, C.-Y. KangWriting, review, and/or revisionof themanuscript: E.-A. Bae,H. Seo, B.-S. Kim,C.-H. Koh, B. Song, I.-K. Kim, C.-Y. KangAdministrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): E.-A. Bae, J. Choi, K.-S. Shin, C.-H. Koh, S.J. ShinStudy supervision: H. Seo, C.-Y. Kang

AcknowledgmentsWe appreciate the members of Kang's laboratory for technical support and

the staff of theNational Center for Interuniversity Research Facilities (NCIRF) atSeoul National University for assistance with the cell sorting by flow cytometry(FACSARIA II, BDBiosciences).We acknowledge theNIHTetramer Core Facilityfor providing the Mouse CD1d-PBS-57–loaded tetramers. C.-Y. Kang wassupported by grants from the Basic Science Research Program (NRF-2015R1A2A1A10055844) and the Bio & Medical Technology DevelopmentProgram (NRF-2016M3A9B5941426).

The costs of publication of this articlewere defrayed inpart by the payment ofpage charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received March 8, 2018; revised June 4, 2018; accepted July 11, 2018;published first July 16, 2018.

References1. Wherry EJ. T cell exhaustion. Nat Immunol 2011;12:492.2. Schietinger A, Greenberg PD. Tolerance and exhaustion: defining

mechanisms of T cell dysfunction. Trends Immunol 2014;35:51–60.

3. Sakuishi K, Apetoh L, Sullivan JM, Blazar BR, Kuchroo VK, Anderson AC.Targeting Tim-3 and PD-1 pathways to reverse T cell exhaustion and restoreanti-tumor immunity. J Exp Med 2010;207:2187–94.

4. Pardoll DM. The blockade of immune checkpoints in cancer immuno-therapy. Nat Rev Cancer 2012;12:252.

5. Apetoh L, Smyth MJ, Drake CG, Abastado J-P, Apte RN, Ayyoub M, et al.Consensus nomenclature for CD8þ T cell phenotypes in cancer. Oncoim-munology 2015;4:e998538.

6. Topalian SL,Hodi FS, Brahmer JR,Gettinger SN, SmithDC,McDermottDF,et al. Safety, activity, and immune correlates of anti–PD-1 antibody incancer. N Engl J Med 2012;366:2443–54.

7. Rizvi NA, Mazi�eres J, Planchard D, Stinchcombe TE, Dy GK, Antonia SJ,et al. Activity and safety of nivolumab, an anti-PD-1 immune checkpointinhibitor, for patients with advanced, refractory squamous non-small-cell






lung cancer (CheckMate 063): a phase 2, single-arm trial. Lancet Oncol2015;16:257–65.

8. Wolchok JD, Kluger H, Callahan MK, Postow MA, Rizvi NA, Lesokhin AM,et al. Nivolumab plus ipilimumab in advanced melanoma. N Engl J Med2013;369:122–33.

9. Mahoney KM, Rennert PD, Freeman GJ. Combination cancer immuno-therapy and new immunomodulatory targets. Nat Rev Drug Discov2015;14:561.

10. Mescher MF, Curtsinger JM, Agarwal P, Casey KA, Gerner M,Hammerbeck CD, et al. Signals required for programming effectorand memory development by CD8þ T cells. Immunol Rev 2006;211:81–92.

11. Kamphorst AO, Wieland A, Nasti T, Yang S, Zhang R, Barber DL, et al.Rescue of exhausted CD8 T cells by PD-1–targeted therapies is CD28-dependent. Science 2017;355:1423–7.

12. West EE, Jin H-T, Rasheed A-U, Penaloza-MacMaster P, Ha S-J, Tan WG,et al. PD-L1 blockade synergizes with IL-2 therapy in reinvigoratingexhausted T cells. J Clin Invest 2013;123:2604–15.

13. Schurich A, Pallett LJ, Lubowiecki M, Singh HD, Gill US, Kennedy PT, et al.The third signal cytokine IL-12 rescues the anti-viral function of exhaustedHBV-specific CD8 T cells. PLoS Pathog 2013;9:e1003208.

14. Elsaesser H, Sauer K, Brooks DG. IL-21 is required to control chronic viralinfection. Science 2009;324:1569–72.

15. John SY, Du M, Zajac AJ. A vital role for interleukin-21 in the control of achronic viral infection. Science 2009;324:1572–6.

16. Teague RM, Sather BD, Sacks JA, HuangMZ, Dossett ML, Morimoto J, et al.Interleukin-15 rescues tolerant CD8þ T cells for use in adoptive immu-notherapy of established tumors. Nat Med 2006;12:335.

17. Bendelac A, Savage PB, Teyton L. The biology of NKT cells. Annu RevImmunol 2007;25:297–336.

18. KronenbergM. Toward an understanding ofNKT cell biology: progress andparadoxes. Annu Rev Immunol 2005;26:877–900.

19. Matsuda JL, Naidenko OV, Gapin L, Nakayama T, Taniguchi M,Wang C-R,et al. Tracking the response of natural killer T cells to a glycolipid antigenusing CD1d tetramers. J Exp Med 2000;192:741–54.

20. Metelitsa LS, Naidenko OV, Kant A, Wu H-W, Loza MJ, Perussia B, et al.Human NKT cells mediate antitumor cytotoxicity directly by recognizingtarget cell CD1d with bound ligand or indirectly by producing IL-2 toactivate NK cells. J Immunol 2001;167:3114–22.

21. Sag D, Krause P, Hedrick CC, Kronenberg M, Wingender G. IL-10–pro-ducingNKT10 cells are a distinct regulatory invariantNKT cell subset. J ClinInvest 2014;124:3725–40.

22. Coquet JM, Kyparissoudis K, Pellicci DG, Besra G, Berzins SP, Smyth MJ,et al. IL-21 is produced byNKT cells andmodulates NKT cell activation andcytokine production. J Immunol 2007;178:2827–34.

23. KitamuraH, Iwakabe K, Yahata T,Nishimura S-i, Ohta A,Ohmi Y, et al. Thenatural killer T (NKT) cell ligand a-galactosylceramide demonstrates itsimmunopotentiating effect by inducing interleukin (IL)-12 production bydendritic cells and IL-12 receptor expression on NKT cells. J Exp Med1999;189:1121–8.

24. Tomura M, YuW-G, Ahn H-J, Yamashita M, Yang Y-F, Ono S, et al. A novelfunction of Va14þ CD4þ NKT cells: stimulation of IL-12 production byantigen-presenting cells in the innate immune system. J Immunol1999;163:93–101.

25. Nair S, Dhodapkar MV. Natural killer T cells in cancer immunotherapy.Front Immunol 2017;8:1178.

26. Cerundolo V, Silk JD, Masri SH, Salio M. Harnessing invariant NKT cells invaccination strategies. Nat Rev Immunol 2009;9:28.

27. Fujii S-i, Shimizu K, Kronenberg M, Steinman RM. Prolonged IFN-g–producing NKT response induced with a-galactosylceramide–loadedDCs. Nat Immunol 2002;3:867.

28. Chung Y, Kim B-S, Kim Y-J, KoH-J, Ko S-Y, KimD-H, et al. CD1d-restrictedT cells license B cells to generate long-lasting cytotoxic antitumor immunityin vivo. Cancer Res 2006;66:6843–50.

29. Kim YJ, Ko HJ, Kim YS, Kim DH, Kang S, Kim JM, et al. a-Galacto-sylceramide-loaded, antigen-expressing B cells prime a wide spectrumof antitumor immunity. Int J Cancer 2008;122:2774–83.

30. Ko H-J, Lee J-M, Kim Y-J, Kim Y-S, Lee K-A, Kang C-Y. Immunosuppressivemyeloid-derived suppressor cells can be converted into immunogenic

APCs with the help of activated NKT cells: an alternative cell-basedantitumor vaccine. J Immunol 2009;182:1818–28.

31. Seo H, Jeon I, Kim B-S, Park M, Bae E-A, Song B, et al. IL-21-mediatedreversal of NK cell exhaustion facilitates anti-tumour immunity in MHCclass I-deficient tumours. Nat Commun 2017;8:15776.

32. Seo H, Kim BS, Bae EA, Min BS, Han YD, Shin SJ, et al. IL21 therapycombined with PD-1 and Tim-3 blockade provides enhanced NK cellantitumor activity against MHC class I-deficient tumors. Cancer ImmunolRes 2018;6:685–95.

33. Kim I-K, Kim B-S, Koh C-H, Seok J-W, Park J-S, Shin K-S, et al. Glucocor-ticoid-induced tumor necrosis factor receptor–related protein co-stimula-tion facilitates tumor regression by inducing IL-9–producing helper T cells.Nat Med 2015;21:1010.

34. Lee KA, Bae EA, Song YC, Kim EK, Lee YS, Kim TG, et al. A multimericcarcinoembryonic antigen signal inhibits the activation of human T cells bya SHP-independent mechanism: a potential mechanism for tumor-mediated suppression of T-cell immunity. Int J Cancer 2015;136:2579–87.

35. Kamata T, Suzuki A,Mise N, Ihara F, TakamiM,Makita Y, et al. Blockade ofprogrammed death-1/programmed death ligand pathway enhances theantitumor immunity of human invariant natural killer T cells. CancerImmunol Immunother 2016;65:1477–89.

36. Kim Y-J, Han S-H, Kang H-W, Lee J-M, Kim Y-S, Seo J-H, et al. NKT ligand-loaded, antigen-expressing B cells function as long-lasting antigen present-ing cells in vivo. Cell Immunol 2011;270:135–44.

37. Kim E, Seo H, Chae M, Jeon I, Song B, Park Y, et al. Enhanced antitumorimmunotherapeutic effect of B-cell-based vaccine transduced with mod-ified adenoviral vector containing type 35 fiber structures. Gene Ther2014;21:106.

38. Kim E-K, Jeon I, Seo H, Park Y-J, Song B, Lee K-A, et al. Tumor-derivedosteopontin suppresses antitumor immunity by promoting extramedul-lary myelopoiesis. Cancer Res 2014;74:6705–16.

39. Mognol GP, Spreafico R, Wong V, Scott-Browne JP, Togher S, Hoffmann A,et al. Exhaustion-associated regulatory regions in CD8þ tumor-infiltratingT cells. Proc Natl Acad Sci 2017;114:E2776–E85.

40. LeeKA, ShinKS, KimGY, SongYC, Bae EA, Kim IK, et al. Characterization ofage-associated exhausted CD8þ T cells defined by increased expression ofTim-3 and PD-1. Aging Cell 2016;15:291–300.

41. Juneja VR,McGuire KA,Manguso RT, LaFleurMW,CollinsN,HainingWN,et al. PD-L1 on tumor cells is sufficient for immune evasion in immuno-genic tumors and inhibits CD8 T cell cytotoxicity. J Exp Med 2017;214:895–904.

42. Philip M, Fairchild L, Sun L, Horste EL, Camara S, Shakiba M, et al.Chromatin states define tumour-specific T cell dysfunction and reprogram-ming. Nature 2017;545:452.

43. Yang Y-F, TomuraM,OnoS,Hamaoka T, FujiwaraH. Requirement for IFN-g in IL-12 production induced by collaboration between Va14þNKT cellsand antigen-presenting cells. Int Immunol 2000;12:1669–75.

44. Pauken KE, Sammons MA, Odorizzi PM, Manne S, Godec J, Khan O, et al.Epigenetic stability of exhausted T cells limits durability of reinvigorationby PD-1 blockade. Science 2016;354:1160–5.

45. Blackburn SD, Shin H, Freeman GJ, Wherry EJ. Selective expansion of asubset of exhaustedCD8T cells byaPD-L1blockade. ProcNatl Acad SciU SA 2008;105:15016–21.

46. Paley MA, Kroy DC, Odorizzi PM, Johnnidis JB, DolfiDV, Barnett BE, et al.Progenitor and terminal subsets of CD8þ T cells cooperate to containchronic viral infection. Science 2012;338:1220–5.

47. Ngiow SF, Young A, Blake SJ, Hill GR, Yagita H, Teng MW, et al. AgonisticCD40 mAb-driven IL12 reverses resistance to anti-PD1 in a T-cell–richtumor. Cancer Res 2016;76:6266–77.

48. Chang W-S, Kim J-Y, Kim Y-J, Kim Y-S, Lee J-M, Azuma M, et al. Cuttingedge: Programmed death-1/programmed death ligand 1 interaction reg-ulates the induction and maintenance of invariant NKT cell anergy. JImmunol 2008;181:6707–10.

49. Parekh VV, Lalani S, Kim S, Halder R, AzumaM, Yagita H, et al. PD-1/PD-Lblockade prevents anergy induction and enhances the anti-tumor activitiesof glycolipid-activated invariant NKT cells. J Immunol 2009;182:2816–26.

50. Durgan K, Ali M, Warner P, Latchman YE. Targeting NKT cells and PD-L1pathway results in augmented anti-tumor responses in amelanomamodel.Cancer Immunol Immunother 2011;60:547–58.


Bae et al.




2018;78:5315-5326. Published OnlineFirst July 16, 2018.Cancer Res Eun-Ah Bae, Hyungseok Seo, Byung-Seok Kim, et al. CellsEnhances Antitumor Immunity by Reinvigorating Exhausted CD8 T

Resistant Tumor Model−Activation of NKT Cells in an Anti-PD-1

Updated version

10.1158/0008-5472.CAN-18-0734doi:

Access the most recent version of this article at:

Cited articles

http://cancerres.aacrjournals.org/content/78/18/5315.full#ref-list-1

This article cites 50 articles, 21 of which you can access for free at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/78/18/5315To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-18-0734

http://cancerres.aacrjournals.org/content/78/18/5315.full#ref-list-1

http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://cancerres.aacrjournals.org/content/78/18/5315


Documents

Activation of NKT Cells in an Anti-PD-1 Resistant Tumor Model … · Tumor Biology and Immunology Activation of NKT Cells in an Anti-PD-1–Resistant Tumor Model Enhances Antitumor