CD8⁺ T cell antitumor immunity via human iNKT-DC conjugates

Abstract

Invariant Natural Killer T (iNKT) cells are a conserved T lymphocyte population capable of acting on dendritic cells (DCs) to potently amplify downstream immune responses. However, the processes underlying such iNKT adjuvancy remain poorly understood. Here, we showed that allogeneic human CD4⁺ iNKT cells form stably adhered bi-cellular complexes with monocyte-derived DCs that migrated together as pairs and showed extended DC calcium signaling. Compared to DCs treated with the synthetic adjuvant monophosphoryl lipid A (MPLA), DCs complexed with iNKT cells had elevated expression of MHC class I and multiple costimulatory molecules including 4-1BBL, OX40L, and IL-15Rα, while the iNKT cells expressed CD70. Consistent with this distinctive co-stimulatory profile, iNKT-DC complexes were efficient activators of CD8⁺ T cells. Administering iNKT-DC complexes as a cellular immunotherapy in a xenograft model of aggressive human B cell lymphoma resulted in rapid reduction in tumor mass, antigen-specific B cell clearance, and transcriptional activation indicative of enhanced T cell proliferation and effector responses. iNKT-DC immunotherapy was effective at late stages of tumor progression that were refractory to immune checkpoint blockade immunotherapy, suggesting that the consortium of activating signals provided by iNKT-DC complexes rejuvenates exhausted antitumor immunity. Finally, allogeneic CD4⁺ iNKT cells formed similar complexes with monocyte-derived DCs from Head and Neck Cancer patients and promoted tumor antigen-dependent CD8⁺ T cell activation. These results show that monocyte-derived DCs paired with allogeneic CD4⁺ iNKT cells act as a potent antitumor cellular immunotherapy that activates antigen-specific CD8⁺ T cell immunity.

Introduction

Invariant Natural Killer T (iNKT) cells are a subset of innate-like T cells, present in all individuals, that have antitumor activity through both cytotoxic and adjuvant-like effects (1–3). The CD4⁻ subset of human iNKT cells is highly cytolytic, whereas CD4⁺ iNKT cells are characterized by a poly-functional profile that includes both regulatory and adjuvant-like activities (4,5). Due to their invariant TCRs and recognition of highly conserved lipid antigens presented by non-polymorphic CD1d molecules, human iNKT cells lack alloreactivity and are therefore ‘donor unrestricted’ (6). Consistent with this, recent clinical trials have shown that allogeneic iNKT cell immunotherapy is well-tolerated at doses up to one billion cells (7,8). Moreover, iNKT cells engineered to express a chimeric antigen receptor (CAR) mediate potent antitumor effects, and these occur not only through direct killing of cancer cells but also through adjuvant-like effects that enhance the antitumor activity of CTLs and NK cells (9,10). Thus, due to their lack of alloreactivity and their multi-faceted antitumor activity, iNKT cells are a highly promising platform for developing off-the-shelf cellular immunotherapies.

Because of their adjuvant-like activity iNKT cells may also be particularly useful in combination with dendritic cell (DC)-based immunotherapies. DC-based immunotherapies offer the potential to selectively activate tumor-specific patient T cells for highly specific antitumor responses (11). A specialized subtype of DCs, called cDC1, is critical for activating CD8⁺ cytotoxic T lymphocytes (CTLs) against cancers in vivo (12). The cDC1 subset cross-presents tumor antigens on MHC class I molecules and effectively primes CTLs (13). However, cDC1s are a rare cell type, and it is difficult to isolate a sufficient number to generate a clinical immunotherapy. Instead, monocyte-derived DCs are the main population that has been used clinically, as monocytes are comparatively abundant and can be converted to DCs through cytokine exposure ex vivo (14). Immunotherapies of this type have been well-tolerated clinically and have shown therapeutic potential (15–18). Yet, while monocyte-derived DCs are potent activators of CD4⁺ T cells they are not highly efficient activators of CD8⁺ T cells, which are key for effective tumor control (13). Thus, enhancing CD8⁺ T cell activation by monocyte-derived DCs is a critical step to improve their immunotherapeutic efficacy.

Studies in murine models have shown that the antitumor adjuvancy activity of iNKT cells is largely due to their effects on DCs. When a highly potent synthetic glycolipid antigen (α-galactosylceramide, α-GalCer) is administered in vivo, it is rapidly taken up and presented by CD1d⁺ DCs in draining lymph nodes. iNKT cell recognition of α-GalCer leads them to secrete IFN-γ and upregulate CD40L, which in turn induce the DCs to upregulate co-stimulatory ligands and produce IL-12p70, enabling enhanced activation of antitumor effectors (2,3). Human iNKT cells can be similarly activated by monocyte-derived DCs that have taken up α-GalCer, however, this typically leads to iNKT killing of the DCs (19). In contrast, in the absence of α-GalCer human iNKT cells are activated by cellular lipids that are less potent agonists (20). In this context, we have found that there is little DC killing by human CD4⁺ iNKT cells, and instead the iNKT cells induce sustained DC calcium signaling that is associated with a protective sterile inflammatory response (21). Therefore, iNKT-DC interactions in the absence of α-GalCer may be sufficient for their adjuvancy effects, without leading to loss of DCs through iNKT cytotoxicity.

We recently reported that administering human CD4⁺ iNKT cells as a cellular immunotherapy in the absence of α-GalCer led to marked antitumor effects in a murine xenograft model of aggressive human B cell lymphoma (22). The CD4⁺ iNKT cells did not directly kill the lymphoma cells and instead were associated with increased activity of antigen-specific T cells, suggesting an adjuvancy mechanism. Based on these findings, we hypothesized that pre-incubating human CD4⁺ iNKT cells with monocyte-derived DCs might lead to an adjuvancy effect that could be leveraged to improve DC-based immunotherapies. We show here that human CD4⁺ iNKT cells and monocyte-derived DCs form stable bicellular conjugates, creating a highly stimulatory cellular nexus that converts the monocyte-derived DCs into powerful activators of CD8⁺ T cells and potently promotes antitumor immunity.

Materials and Methods

Study Design.

The goals of this study were to investigate how human CD4⁺ iNKT cells interact with monocyte-derived DCs to promote T cell activation and to assess whether this pathway can be leveraged as a cellular immunotherapy for cancer. The study involved controlled laboratory experiments using primary and short-term cultures of human cells from healthy donors and cancer patients and included three main components: i) live cell microscopic imaging and flow cytometric analysis of interactions between iNKT cells and DCs in vitro; ii) functional assays in vitro to test iNKT-DC activation of primary human T cells from healthy donors and cancer patients; iii) in vivo analysis of human antitumor responses in a xenotransplant model. Our pre-specified hypothesis was that exposing monocyte-derived DCs to CD4⁺ iNKT cells would lead to improved DC-mediated activation of MHC-restricted T cells. After initiation of data analysis we modified this to postulate that iNKT-DC conjugates would particularly promote CD8⁺ T cell responses. Sample sizes for in vitro experiments were determined by human tissue availability and were based on prior experience with similar analyses in our laboratory; sizes of experimental groups for in vivo experiments were determined by the availability of human tissue to set up the experimental model, with statistical power achieved by aggregating results from multiple independent experiments. Timing of data collection was carried out according to pre-specified timepoints for each type of assay or analysis. None of the data from in vitro experiments were excluded; for in vivo experiments we set a prospective criterion of excluding mice that had no visible tumor tissue at day 29-31 post-transplant and that also showed less than 1% human cells in spleen, as these mice were considered to have insufficient human cell engraftment to support tumorigenesis. This resulted in exclusion of 7 out of a total of 147 mice. Outliers were not excluded from any analysis. Experimental replicates are shown in figures or indicated in legends; all experiments were performed using tissue samples and iNKT cells from multiple unrelated human donors. For in vivo experiments, transplanted mice were randomly assigned to experimental groups, and investigators were blinded to experimental groups for assessment of outcomes.

Human tissue samples.

All use of human tissue samples was performed in accordance with the U.S. Common Rule and with Institutional Review Board (IRB) approval. Peripheral blood samples were collected under written informed consent from 15 healthy, non-pregnant adult donors (age 18-65) in accordance with University of Wisconsin IRB protocol #2018-0304. Twenty-eight de-identified human umbilical cord blood samples and 2 de-identified adult human bone marrow samples obtained from accredited suppliers were used in accordance with University of Wisconsin IRB exemption #2023-0392. Cord blood samples of less than 70 ml volume or containing more than 30% red blood cells after density gradient centrifugation were excluded from the study. De-identified blood samples from 13 patients diagnosed with HPV16⁺ Head and Neck squamous cell carcinoma were obtained through the University of Wisconsin Translational Science BioCore BioBank and used in accordance with University of Wisconsin IRB protocol #2021-0244. Patient samples were neither included nor excluded on the basis of disease stage, treatment history, age, or other donor demographic factors. All samples were stored at room temperature in the presence of anti-coagulant and used within 48h after collection.

Generation of monocyte-derived DCs.

Mononuclear cells were isolated from human tissue samples by density gradient centrifugation using Ficoll-Paque Plus (GE Healthcare). Monocytes were purified using CD14 Microbeads (Miltenyi Biotec Cat# 130-050-201, RRID:AB_2665482) on an autoMACSPro magnetic sorter (Miltenyi Biotec RRID:SCR_018596) and were cultured for three days in “DC medium” [RPMI 1640 (Corning Cat# 10-040-CV) containing 2mM L-glutamine (Corning Cat# 25-005-CI), 10% heat-inactivated fetal bovine serum (GeminiBio Cat# 100-106)] supplemented with 200U/ml IL-4 (Peprotech Cat# 200-04) and 300U/ml GM-CSF (Peprotech Cat# 300-03). Where indicated, DCs were co-cultured with iNKT cells at a 1:1 ratio, or with 500ng/ml Monophosphoryl Lipid A (Avanti Polar Lipids Cat# 699800P), or alone, in “assay medium” [RPMI 1640, containing 2mM L-glutamine, 10% heat-inactivated Bovine Calf Serum (HyClone Cat# SH30072.03)].

iNKT cells.

Polyclonal and clonal CD4⁺ iNKT cell lines were generated from PBMCs of healthy adult donors as previously described (22). Briefly, CD4⁺ iNKT cells were sorted using human CD1d tetramers loaded with the PBS-57 lipid analogue of α-GalCer (NIH tetramer facility, Emory University), and expanded in the presence of irradiated PBMCs and 1μg/ml phytohemagglutinin (Sigma-Aldrich Cat# L1668), in “T cell culture medium” [RPMI1640 containing 2mM L-glutamine, 10% heat-inactivated fetal bovine serum, 5% heat-inactivated bovine calf serum, 3% pooled human AB serum (GeminiBio Cat# 100-812)] supplemented with 200U/ml recombinant human IL-2 (Peprotech Cat# 200-02). After expansion, iNKT cells were maintained in T cell culture medium supplemented with IL-2 without additional stimulation. Expanded iNKT cells were regularly tested using PBS-57 loaded CD1d tetramer (NIH tetramer facility, Emory University) or 6B11 mAb (BioLegend Cat# 342916 RRID:AB_2721323) to confirm that purity was ≥97%. Eight different polyclonal CD4⁺ iNKT lines generated from 6 donors, and 5 clonal iNKT cell lines from 2 donors were used.

EBV.

Experiments were performed using the lytic M81 strain of EBV. An M81 bacmid containing genes for green fluorescent protein (GFP) and hygromycin B resistance was constructed using bacterial artificial chromosome technology as described (23), and was validated through next generation sequencing. Viral particles were produced from stably infected 293 cells (ATCC Cat# CRL-1573 RRID:CVCL_0045), and infectious titer was determined by assessing fluorescence after titrated infection of Raji cells (ATCC Cat# CCL-86 RRID:CVCL_0511), as previously described (24). EBV antigen was prepared by inactivating M81-infected 293 culture supernatant for 30min under UV light, followed by sonication for 30min, and filtration through 0.450 μm cellulose acetate membrane. Lymphoblastoid cell lines were generated as previously described, by M81 EBV infection of primary human splenic B cells from EBV-negative donors (22).

Live cell imaging.

Imaging was performed in assay medium in a humidified incubator at 37°C and 5% CO₂, using an IncuCyte S3 Live-cell Imaging and Analysis System (Sartorius). To assess iNKT-DC contact, DCs were labeled with CFSE (Invitrogen Cat# C34570) or CFDA-SE (ThermoFisher Cat# C1157) and mixed at a 1:1 ratio with iNKT cells labeled with CMTPX (Invitrogen Cat# C34552) in flat bottom 96-well plates (Corning Cat# 3799) at a density of 8,000 cells/well. The wells were imaged with a 4x objective at 30min intervals for up to 4 days, and the area of red-green fluorescence overlap was calculated using the IncuCyte platform whole well Standard Analysis software. For analysis of cell motility, ImageLock 96-well plates (Sartorius) were coated at 5μg/cm² with purified human plasma Fibronectin (EMD Millipore Cat# FC010) dissolved in sterile PBS then washed with sterile PBS followed by assay medium, and a total of 2x10⁴ cells/well of iNKT cells (CMTPX-labeled), DCs (CFDA-SE-labeled), or 1:1 iNKTs + DCs were added in a volume of 100μl assay medium. The plates were briefly centrifuged to pellet the cells, incubated at 37°C for 30 min to allow cell adherence to the fibronectin, then the supernatant was removed and replaced with assay medium containing 0.3mg/mL Cultrex Rat Collagen I (R&D Systems Cat# 3440-005-01). The wells were then imaged with a 10x objective at 4 min intervals for up to 6 hours. Approximately 75-80 randomly chosen cells in each condition were manually tracked in serial images for a 1h period from ~2-3h after initiation of imaging, using the NIH ImageJ software Manual Tracking plug-in (developed by Fabrice Cordelieres, available on https://imagej.net). Migration plots and average cell displacement values (Euclidian distance) were generated using the Ibidi Chemotaxis and Migration Tool ImageJ software plug-in (https://ibidi.com). To investigate DC calcium signaling, the DCs were intracellularly loaded with 2μM Fluo-4 (Invitrogen Cat# F14201) in calcium-free PBS according to the manufacturer’s instructions. Cells were washed and plated in flat-bottom 96-well plates in assay medium alone, or at a 1:1 ratio with CMTPX-labeled iNKT cells, and imaged every 3min for 8h using a 4x objective. Individual DCs or iNKT-DC pairs were selected for analysis based on their consistent presence in serial image frames over a 6h period, and integrated Fluo-4 fluorescence intensity of individual cells was calculated using NIH ImageJ software.

Flow cytometry.

To assess the fraction of tightly adhered iNKT cells and DCs, DCs labeled with 5μM CFSE and iNKT cells labeled with 5μM CTV (Invitrogen Cat# 34557) were mixed at a 1:1 ratio and co-incubated in assay medium for the indicated times. Cells were resuspended with vigorous pipetting, then washed at 4 °C using “FACS buffer” [phosphate buffered saline (PBS) containing 2% bovine calf serum, 2 mM EDTA (Fisher Scientific Cat# S657) and 0.05% NaN3 (Sigma-Aldrich Cat# 26628-22-8)], and analyzed on a LSRII (Becton Dickinson) flow cytometer. For analysis of cell surface markers, cells were resuspended in FACS buffer and blocked with 20% human AB serum for 15min, then stained for 30min at 4°C with fluorescently-labeled antibodies against the indicated markers. Antibodies used were as follows: human CD1d clone 51.1 (Cat#350321 RRID:AB_2814280); CD3 clone OKT3 (Cat # 317312 RRID:AB_571883); CD4 clone OKT4 (Cat# 317407 RRID:AB_571950); CD8a clone RPA-T8 (Cat# 301039 RRID:AB_11126985) or HIT8a (Cat# 344712 RRID:AB_2044008); CD14 M5E2 (Cat# 301839 RRID:AB_2561366); CD19 clone HIB19 (Cat# 302258 RRID:AB_2629691) or SJ25C1 (Cat# 363008 RRID:AB_2564171); CD20 clone 2H7 (Cat# 302304 RRID:AB_314252); CD25 clone BC96 (Cat# 302606, RRID:AB_314276); CD27 clone O323 (Cat# 302808, RRID:AB_314300); CD40 clone HB14 (Cat# 313014 RRID:AB_2563958); pan-CD45 clone HI30 (Cat# 304036 RRID:AB_2561940) or 2DI (Cat# 368504 RRID:AB_2566352); CD70 113-16 (Cat# 355112 RRID:AB_2687254); CD127 clone A019D5 (Cat# 351350, RRID:AB_2715894); CD137L (4-1BBL) clone 5F4 (Cat# 311505 RRID:AB_2561310); CD80 clone 2D10 (Cat# 305216 RRID:AB_528875); CD83 clone HB15E (Cat# 305326 RRID:AB_2561775); CD86 clone IT2.2 (Cat# 305416 RRID:AB_528883); CD95 clone DX2 (Cat# 305630 RRID:AB_2562893); CD107a clone H4A3 (Cat# 328612 RRID:AB_1227506); CD209 clone 9E9A8 (Cat# 330104, RRID:AB_1134048); CD215 clone JM7A4 (Cat# 330208 RRID:AB_2124601); CD273 (PD-L2) clone MIH18 (Cat# 345508 RRID:AB_2162176); CD274 (PD-L1) clone 29E.2A (Cat# 329714 RRID:AB_2563852); FoxP3 clone 259D (Cat# 320216, RRID:AB_2104902); HLA-ABC clone W6/32 (Cat# 311438 RRID:AB_2566306); HLA-A2 clone BB7.2 (Cat# 343320, RRID:AB_2566767); HLA-DR clone L243 (Cat# 307629 RRID:AB_893575); IL-2 clone MQ1-17H12 (Cat# 500324 RRID:AB_2125595); IFN-γ clone 45.B3 (Cat# 502505 RRID:AB_315230); iNKT TCR clone 6B11 (Cat# 342916 RRID:AB_2721323); murine CD45.1 clone A20 (Cat# 110748 RRID:AB_2564295); isotype controls mouse IgG1 clone MOPC-21 (Cat# 400126 RRID:AB_326448), IgG2a clone MOPC-173 (Cat# 400229 RRID:AB_326477), IgG2b clone MG2B-57 (Cat# 400334 RRID:AB_493780), all from BioLegend; as well as CD252 (OX40L) clone ik-1 (Cat# 558164 RRID:AB_647195) and CD86 clone FUN-1 (Cat# 561124 RRID:AB_10564087) from BD Biosciences; CD8β clone SIDI8BEE (eBioscience Cat# 25-5273-42 RRID:AB_11219680). For analysis of intracellular molecules, cells were stained for surface antigens then fixed and permeabilized using the Cyto-Fast buffer set (BioLegend Cat# 426803) prior to intracellular staining. For tetramer staining, T cells were treated with 50nM Dasatinib (Axon Medchem Cat# 1392) for 30min at room temperature, followed by incubation with a mix of Alexa647-labeled HLA-A*02 tetramers (NIH Tetramer Core) loaded with the following synthesized peptides (GenScript): BMLF1 259-267 (GLCTLVAML), BRLF1 109-117 (YVLDHLIVV), BMRF1 208-216 (TLDYKPLSV), EBNA3C 284-293 (LLDFVRFMGV), LMP1 125-133 (YLLEMLWRL), LMP1 159-167 (YLQQNWWTL), LMP2 329-337 (LLWTLVVLL), LMP2 243-251 (TVCGGIMFL), LMP2 356-364 (FLYALALLL), LMP2 426-434 (CLGGLLTMV), or with APC-labeled HLA-A*02 tetramers loaded with peptides from the HPV16 E7 protein: E7 11-19 (YMLDLQPET), E7 82-90 (LLMGTLGIV), E7 86-93 (TLGIVCPI). Flow cytometric analysis was performed using LSRII (Becton Dickinson) or AttuneNxT (Thermo Fisher Scientific) instruments and data analyzed using FlowJo v9 and v10 software (BD Biosciences). Imaging flow cytometry was performed using an Amnis ImageStream MarkII (Cytek) imaging cytometer, and data processed using Ideas v6.2 (EMD Millipore) software. For all analyses, light scatter gates (SSC-A vs. SSC-H and FSC-A vs. FSC-H, then SSC-A vs. FSC-A) were first applied to eliminate particles, aggregates, and dead cells, then the indicated specific populations were identified based on fluorescence staining. Absolute cell numbers were calculated based on Precision Count Beads (BioLegend Cat# 424902) or volumetric cell count of the Attune NxT instrument.

Confocal microscopy.

DCs labeled with 5μM CTV and unlabeled iNKT cells were mixed at a 1:1 ratio and co-incubated in DC medium for 24h, then seeded on fibronectin coated glass coverslips, centrifuged at 300xg for 10 min, fixed with fresh 4% formaldehyde (EMC Cat# 15710), and permeabilized with 0.5% saponin (Acros Organics Cat# 74499-23-3), prior to staining with CD70 mAb clone 113-16 (BioLegend Cat# 355101, RRID:AB_2561428) followed by Alexa488-labeled goat anti-mouse IgG (Invitrogen Cat# A-11029 RRID:AB_2534088). Cells were mounted under ProLong Gold (Invitrogen Cat# P36970) and images acquired using a 63x oil immersion objective on a Zeiss LSM 800 confocal microscope, using Zen 3.4 software.

T cell proliferation.

Untouched T cells were magnetically sorted from PBMC using Pan T Microbeads (Miltenyi Biotec Cat# 130-096-535 RRID:AB_3695577) and labeled with 5μM CFSE. Autologous monocyte-derived DCs were incubated for 24h in DC culture medium alone, or with a 1:1 ratio of iNKT cells (iNKT-DCs), or with 500 ng/ml MPLA. The DCs or iNKT-DCs were then incubated for 2h with antigen or mock-treated. Antigen preparations used were inactivated EBV (equivalent to 2.5U EBV/1000 DC), or 5ng/ml dialyzed recombinant HPV16 E7 protein (Sino Biological Cat# 40965-V07E). CFSE-labeled T cells were added at a T:DC ratio of 50:1 and cells were placed in round bottom 96-well plates (Corning) and kept in a humidified incubator at 37°C with 5% CO₂. Cells were harvested after 3, 7, or 10 days and proliferation of CD4⁺ and CD8⁺ T cells was assessed by flow cytometry, with iNKT cells gated out using 6B11 mAb staining. Where indicated, 5μg/ml of the following blocking antibodies were included: CD70 clone BU69 (Ancell Corporation Cat# 222-820 RRID:AB_3712806); CD1d clone CD1d51.1 (BioLegend Cat# 350321, RRID:AB_2814280), clone CD1d55 or CD1d59 (purified in house); ICAM-1 clone HCD54 (BioLegend Cat# 322721, RRID:AB_2832632); murine IgG1 clone MOPC-21 (BioLegend Cat# 400165, RRID:AB_11150399) and IgG2b clone MG2b-57 (BioLegend Cat# 401215, RRID:AB_3097073).

Murine xenograft model of human EBV-driven B cell lymphoma.

Animal husbandry and experimental use were approved by the University of Wisconsin IACUC, under protocol #M005199. Human umbilical cord blood mononuclear cells (CBMCs) were purified by density gradient centrifugation and monocytes were isolated using CD14 microbeads as described above. Remaining CBMCs were suspended in assay medium and incubated with 2000U M81 EBV for 2h at 37°C and 5% CO₂. CBMCs (10⁷ cells per mouse) were injected intraperitoneally into 6–10 week-old NOD.Cg-Prkdc^scidIl2rg^tm1Wjl/SzJ (NSG) mice (Jackson Labs RRID:IMSR_JAX:005557). Mice were maintained in a specific pathogen-free facility using microisolator cages, with sterilized bedding, food and water. Immunotherapeutic cells (iNKT cells, DCs, or iNKT-DC co-cultures) were administered intravenously at the indicated numbers through retro-orbital injection at day 25 after transplantation of CBMCs. Mice were euthanized after 26–31 days and macroscopically visible tumor tissue and spleens were collected for analysis. Both female and male mice were used for experiments, with no significant differences observed in tumor mass or effect of iNKT-DC immunotherapy (Fig S1A and S1B).

Cytotoxicity assays.

Mice were transplanted with CBMCs that were pre-treated with EBV or mock-treated (uninfected). Monocytes from the same CBMC sample were used to generate autologous monocyte-derived DCs and stored cryopreserved. Mice were sacrificed after 29 days, and human T cells were magnetically sorted by negative selection from spleens and expanded for 3-5 days in T cell medium supplemented with 200U/ml IL-2. Splenic target cells were labeled with CFSE and combined at a 2:1 ratio with autologous T cells. Assays were performed in triplicate wells of 96-well flat-bottom plates in assay medium containing 500nM propidium iodide (PI, Invitrogen Cat# P3566), with or without 2% added DCs, iNKT-DCs, or iNKT cells. Wells were imaged for 48h using an IncuCyte S3 instrument in a humidified incubator at 37°C with 5% CO₂. Data analysis was carried out using the IncuCyte platform’s Standard Analysis v2023A software. Cell masking was performed to identify CFSE singly stained (live) and CFSE/PI double-positive (dead) target cells. Specific lysis was calculated by determining the dead fraction of the total target cells and subtracting spontaneous lysis.

Histological analysis.

Tissues were fixed with fresh 4% formaldehyde in PBS at 4°C overnight, followed by 70% ethanol, then embedded in paraffin and sectioned at 5μm. Histopathology was evaluated after hematoxylin-eosin staining. To assess whether tumor-reducing immunotherapy was associated with a reduction of EBV antigens, fixed tissue slides from randomly selected control mice and iNKT-DC treated mice with tumor mass ≤ 80% of the mean of the control group were selected for immunofluorescence analysis. Spleen tissue mounted on slides was deparaffinized, antigen retrieved by boiling for 15min, followed by blocking and staining with specific mAbs, and detection by fluorescently labeled polyclonal antibodies. mAbs specific for the following antigens were used: EBV BZLF1 clone BZ1 (Santa Cruz Cat# sc-53904 RRID:AB_783257), EBV LMP1 clone CS1-4 (Abcam Cat# ab78113 RRID:AB_1566182), EBV LMP2A clone 15F9 (Invitrogen Cat# MA1-81921 RRID:AB_936576), granzyme B clone 11F1 (Biocare Medical Cat# ACI3202A RRID:AB_3712808), CD8 clone SP16 (Biocare Medical Cat# CRM 311 RRID:AB_2750579), CD20 clone L26 (Biocare Medical Cat# CM 004 RRID:AB_2750580), and CD3 clone EP41 (Biocare Medical Cat# CME 324 RRID:AB_2923282). Detection antibodies were goat anti-rat IgG (BioLegend Cat# 405422, RRID:AB_2563301), goat anti-mouse IgG (Thermo Fisher Scientific Cat# A-11029, RRID:AB_2534088) and goat anti-rabbit IgG (Molecular Probes Cat# A-21070, RRID:AB_2535731). Tissues were counterstained with DAPI (Sigma-Aldrich Cat# D9542) to identify cell nuclei and images acquired with a 10x objective were analyzed on a Zeiss LSM 800 confocal microscope to quantitate the number of positive cells in fields of defined unit size (0.8mm² ).

Transcriptomic analysis.

Samples of spleen and tumor tissue were collected from paraffin-embedded tissue blocks of paired iNKT-DC treated and control mice from 3 independent experiments. Total RNA was extracted and analyzed by a barcoded probe microarray system (NanoString Technologies Inc.) using the Human Immune Exhaustion panel (Cat. #115000475), which includes 773 cancer, viral and immune-associated genes, and 12 housekeeping genes. This microarray system uses unique fluorescent barcodes attached to capture probes that are highly specific for human gene transcripts. Each capture probe binds directly to RNA transcripts present in the sample without PCR amplification. Barcode-labeling is then imaged to obtain a count of gene transcripts present in the sample. Only genes for which the raw signal was greater than two standard deviations from the geometric mean of the negative control probes were included in the analysis (536 genes). Normalization to housekeeping and differential gene expression was calculated with R software version 4.4.0 (25), using the NanoStringDiff R software package (26). Genes were selected as differentially expressed if their absolute value of Log₂ fold-change was above 0.5 and their p value was less than 0.05, using gene-wise likelihood ratio tests performed using the glm.LRT() function. Differentially expressed genes that met these criteria were assigned to a cell type based on gene expression patterns from the Monaco data set shown in The Human Protein Atlas (27), then plotted using Enhanced Volcano (https://github.com/kevinblighe/EnhancedVolcano). Proportional Euler diagrams showing the amount of overlap between results from spleen and tumor were generated using the Venneuler R software package (https://www.rforge.net/venneuler/); biological functions categories were assigned using UniProt (https://www.uniprot.org) (28).

Statistical analysis.

Normalized tumor mass was calculated by dividing the raw tumor mass for each mouse by the mean tumor mass of the control group from the same CBMC sample. Where indicated, flow cytometric staining was normalized by that of corresponding controls run in parallel to show fold-upregulation. For statistical analyses involving comparisons of samples sets to a hypothetical mean we used a one sample t-test; for comparisons involving technical replicates with a normal distribution we used an unpaired parametric t-test with Welch’s correction; for comparisons between groups that included biologically independent samples we used an unpaired nonparametric t-test (Mann-Whitney); for comparisons of the same samples tested in parallel under two different conditions we used a paired non-parametric t-test (Wilcoxon matched-pairs signed rank); for analysis of probability of survival we used a log-rank (Mantel-Cox) test. All tests were two-sided, with α=0.05. Analyses were performed using GraphPad Prism10 software.

Data and materials availability:

All data are represented in the main text or the supplementary materials. Data files are deposited in the Open Source Framework Repository (OSF.io), and are available at the following URL: https://doi.org/10.17605/OSF.IO/2TFQP. Expanded iNKT cells will be shared after execution of a Material Transfer Agreement (Simple Letter Agreement or Uniform Biological Materials Transfer Agreement).

Results

Formation of sustained signaling complexes with monocyte-derived DCs.

We used live cell time-lapse fluorescent imaging to investigate interactions between human CD4⁺ iNKT cells and monocyte-derived DCs in the absence of added lipid antigens. iNKT cells and DCs came into close contact with each other over the first ~8h of co-culture, and the area of inter-cell contact was maintained for at least 96h (Fig 1A). Flow cytometric analysis revealed that a population of tightly adhered iNKT-DC conjugates formed within 2h that resisted vigorous pipetting and EDTA treatment and was maintained for at least 96h (Fig 1B and S2A). Stable iNKT-DC conjugates were formed using CD4⁺ iNKT cells and DCs derived from multiple different donors, with no difference in frequency when the iNKT cells and DCs were autologous or allogeneic to each other (Fig 1C). Conjugate formation was partially inhibited by addition of either CD1d or ICAM blocking antibodies, and substantially reduced in the presence of both antibodies, indicating roles for CD1d recognition and integrin-mediated adhesion (Fig 1D).

Fig. 1. CD4⁺ iNKT cells form complexes with monocyte-derived DCs.

A) DC and iNKT cell contact area over time assessed by live cell fluorescent imaging. Green (CFSE) labeled DCs were co-incubated with a 1:1 ratio of red (CMTPX) labeled iNKT cells and microscopically imaged at 30-minute intervals. Plot shows mean ± SD of calculated area of red-green overlap for 3 replicate wells; results representative of 2 independent experiments. B) Flow cytometric analysis of CFSE-labeled DCs and CTV-labeled iNKT cells cultured alone or at a 1:1 ratio for the indicated times. Numbers in gates are percent of total culture. Similar results were observed in 5 independent experiments. C) DCs were co-incubated with autologous or allogeneic iNKT cells. Graph shows percent of DCs in conjugates after 24h; symbols show independent biological replicates. D) DCs and iNKT cells were co-incubated in presence of CD1d, ICAM-1 or isotype control mAbs. Graph shows percent of DCs in iNKT-DC conjugates compared to no antibody control. Symbols show independent biological replicates; P values by one-sample t-test. E) Imaging flow cytometry showing a representative iNKT-DC complex with CD209 staining in green and CD3 in blue. Graph on right shows cellular composition of iNKT-DC complexes from 3 independent co-cultures. P values by unpaired t-test with Welch’s correction. F) Live cell fluorescent imaging was used to track DC motility over a 1h period for DCs cultured alone or with iNKT cells. Plots show individual cellular motility tracks, with origins set at the center. Left plot shows DCs cultured alone, middle and right plots shows DC+iNKT co-culture, with middle plot showing unpaired DCs within the co-culture and right plot showing adhered iNKT-DC pairs. Graph on right shows calculated euclidian distances travelled for each tracked cell; white symbols below dotted line indicate cells that showed no migration in the assessed period. P values by Mann-Whitney t-test; bars show means. G) Intracellular calcium levels over time in DCs paired with iNKT cells (green lines) or DCs cultured alone (pink lines). DCs were labeled with Fluo-4 calcium indicator dye and microscopically imaged at 30min intervals between 2h and 8h from the start of the co-culture. Graph on right shows the corresponding area under the curve (AUC) values; P value by Mann-Whitney t-test. Data from one experiment of two with similar results.

Imaging flow cytometry demonstrated that adhered iNKT cells and DCs were nearly all composed of 1:1 pairings (Fig 1E). The fraction of DCs in conjugates remained stable over a 96h period in 1:1 iNKT-DC co-cultures, and increasing the proportion of iNKT cells only transiently elevated conjugate frequency suggesting that 1:1 iNKT-DC pairings were favored (Fig S2B). Through time-lapse live cell fluorescent imaging, we observed individual iNKT-DC pairs that remained associated for up to 3 days in (Movie S1). Time-lapse analysis of cellular motility revealed that unpaired DCs showed little or no migration (Movies S2–S4), whereas DCs that were paired with iNKT cells showed significantly increased migration distances (Fig 1F), with associated iNKT-DCs seen migrating together (Movies S5–S7). Cytoplasmic Ca⁺⁺ levels were also elevated for at least 8h in DCs that were in contact with iNKT cells compared to DCs cultured alone (Fig 1G). Together, these results demonstrate that cultured human CD4⁺ iNKT cells spontaneously form bi-cellular motile complexes with monocyte-derived DCs that entail sustained calcium signaling within the DCs.

Upregulation of co-stimulatory ligands.

By flow cytometric analysis we observed elevated cell surface expression of a variety of maturation and stimulatory ligands on iNKT-DC conjugates compared to DCs cultured alone (Fig 2A). We also noted that iNKT-DCs expressed higher levels of several key co-stimulatory molecules than DCs that were treated with the synthetic adjuvant MPLA. Specifically, while iNKT-DC conjugates and MPLA-treated DCs showed similar upregulation of class I and II HLA, CD83, CD80, and CD86, iNKT-DC conjugates had significantly higher expression levels of CD70, IL-15Rα chain (CD215), 4-1BBL (CD137L), OX-40L (CD134L), and lower levels of PDL-1 (CD274) than MPLA-treated DCs (Fig. 2A). Cell surface expression levels of these co-stimulatory molecules were maintained on iNKT-DC conjugates for at least 4 days of culture (Fig S3A). Thus, iNKT-DC conjugation appeared to promote sustained upregulation of multiple stimulatory ligands, with little upregulation of inhibitory ligands.

Fig. 2. iNKT-DC conjugates upregulate expression of co-stimulatory molecules.

A) Stimulatory ligand upregulation on iNKT-DC conjugates compared to MPLA-treated DCs. Graph shows fold-upregulation over monocyte-derived DCs cultured alone, with dashed line indicating a value of 1 (no upregulation). Symbols show independent biological replicates; P values by Man-Whitney: **** P<0.0001; *** P<0.005; ** P<0.01; * P<0.05. B) iNKT-DCs were stained for cell surface markers and analyzed by imaging flow cytometry. Panels on left show phase contrast views; images on right show staining for CD209 in green (DCs) and CD3 in blue (iNKTs) with the indicated co-stimulatory molecules in red. C) Confocal microscopic analysis of CD70 expression showing representative images of 0.2 μm slices from permeabilized cells from cultures of DCs alone (left), iNKT cells alone (middle), or iNKT+DC co-culture (right). DCs were labeled with CTV (purple); CD70 staining is shown in green. Similar results observed in 2 independent analyses. D) Flow cytometric analysis was performed to assess CD70 expression by iNKT cells ex vivo, or after in vitro expansion, or after 24h co-culture with DCs. Graph shows CD70 staining normalized by isotype. Symbols show independent biological replicates, pairing lines show non-conjugated and conjugated iNKT cells from the same iNKT-DC co-culture. P values for iNKT cells conjugated to DCs vs. non-conjugated iNKT cells by Wilcoxon matched pairs test, Mann-Whitney test for other comparisons.

We noted that within iNKT-DC co-cultures, the iNKT-DC conjugates consistently showed significantly higher cell surface expression of co-stimulatory molecules than non-conjugated DCs in the same cultures (Fig S3B), suggesting that cell contact was important for their up-regulation. To investigate, we cultured DCs alone or in the presence of transwell inserts containing either unstimulated or anti-CD3 activated iNKT cells to allow exposure to soluble factors produced by iNKT cells. DCs cultured with transwell inserts containing unstimulated iNKT cells showed little detectable increase in any of the ligands tested, whereas inserts containing CD3-activated iNKT cells led to significant upregulation of HLA-DR, CD83, CD86, and CD40, suggesting upregulation of these ligands by soluble factors from activated iNKT cells (Fig S3C). In contrast, DCs in transwell cultures did not upregulate CD70, 4-1BBL, OX40L, or IL-15Rα, suggesting these ligands depend on iNKT-DC contact.

We used imaging flow cytometry to further investigate costimulatory molecule expression. iNKT cells and DCs were co-cultured for 24h and stained for DC-SIGN (CD209) and CD3 to allow for identification of DCs and iNKT cells, respectively (Fig 2B). We found that CD40, CD80, CD86, 4-1BBL, and IL-15Rα were expressed by the DCs (Fig 2B). In contrast, CD70 was expressed by the iNKT cells (Fig 2B). Confocal microscopy of permeabilized cells confirmed that DCs did not express detectable CD70, and CD70 was observed solely on iNKT cells (Fig 2C). Flow cytometric analysis of primary CD4⁺ iNKT cells from healthy adult donors showed little or no detectable cell surface CD70 expression, whereas CD70 was significantly elevated on CD4⁺ iNKT cells expanded in vitro (Fig 2D). iNKT cells co-cultured with DCs showed no increase in expression of CD70 on non-conjugated cells, but significantly increased expression on iNKT cells that were conjugated with DCs (Fig 2D). Thus, iNKT cell expansion in vitro led to increased CD70 expression, and levels were further increased after conjugation with DCs.

Enhanced CD8⁺ T cell activation.

To investigate whether this co-stimulatory molecule upregulation translates into better T cell activation, we isolated total T cells from peripheral blood mononuclear cells (PBMCs) of Epstein-Barr virus (EBV) exposed adults, and prepared autologous monocyte-derived DCs. The DCs were pre-incubated with MPLA or with CD4⁺ iNKT cells in the presence of EBV antigen or vehicle control. Fluorescently labeled T cells were co-cultured with iNKT cells alone, MPLA-treated DCs, or iNKT-DCs, or kept without stimulator cells, and T cell proliferation was assessed over a 10-day period by flow cytometry (Fig 3A). There was only minimal proliferation when the T cells were cultured alone or with iNKT cells. CD4⁺ T cells showed marked proliferation in response to MPLA-treated DCs and iNKT-DCs, but showed no increased proliferation in the presence of EBV antigen (Fig 3B). In contrast, CD8⁺ T cells showed EBV antigen-dependent responses to iNKT-DCs but not MPLA-treated DCs (Fig 3B), and there was 5-10 fold greater CD8⁺ T cell expansion in response to iNKT-DCs compared to MPLA-treated DCs (Fig 3C). Moreover, after activation by iNKT-DCs in the presence of EBV antigen we were able to detect a specific population of proliferated CD8⁺ T cells using EBV peptide-loaded HLA-A2 tetramers, whereas proliferated tetramer⁺ cells were not detectable above background after co-incubation with DCs and EBV antigen (Fig 3D). There was no significant difference in CD4⁺ or CD8⁺ T cell proliferation when autologous or allogeneic iNKT cells were used in the co-cultures, indicating that the responses we observed were not due to alloreactivity (Fig 3E). Thus, iNKT-DCs activated CD8⁺ T cells more efficiently than MPLA-treated DCs, and promoted an antigen-dependent CD8⁺ T cell response.

Fig. 3. T cell activation by iNKT-DC conjugates.

A) T cells isolated from PBMC of a healthy adult donor were tested for proliferation in response to EBV antigen alone (−) or in the presence of iNKT cells, MPLA-treated DCs, or iNKT-DCs. Graphs show the percent of CD4 or CD8 T cells that had divided 3 or more times at the indicated time points. Results shown are from one representative experiment out of four. B) Percent of CD4 or CD8 T cells that had divided after 10 days of culture in the indicated conditions with or without EBV antigen. Symbols show independent biological replicates, pairings show +/− antigen conditions from the same experiment. P values by Wilcoxon matched pairs test. C) T cell expansion induced in presence of EBV antigen by MPLA-treated DCs vs. iNKT-DCs. Graph shows number of divided T cells in the APC conditions normalized by the number of T cells that divided in absence of APCs. Symbols show independent biological replicates, pairings show iNKT-DCs vs. MPLA-treated DCs from the same experiment. P values calculated by Wilcoxon matched pairs test. D) CTV-labeled T cells from a healthy HLA-A2⁺ donor were cultured for 10 days with DCs or iNKT-DCs in the presence of EBV antigen. Plots show CTV and EBV peptide-loaded HLA-A2 tetramer staining after gating on live cells expressing CD3 and CD8β. E) T cells were co-incubated with EBV antigen and iNKT-DCs made with either autologous or allogeneic CD4⁺ iNKT cells. Graph shows percent of CD4 or CD8 T cells that had undergone division after 10 days. Symbols show independent biological replicates, lines show means. F) T cells were co-incubated with iNKT-DCs and EBV antigen in presence of CD70 blocking mAb or isotype control. Graph shows T cell proliferation as a percentage of the no-antibody control condition. Symbols show independent biological replicates, lines show means. P value by Mann-Whitney test.

Since iNKT cells specifically contribute CD70 to iNKT-DC complexes, we tested the role of this co-stimulatory pathway. We found that CD27 (the ligand for CD70) is expressed by a substantial fraction of both CD4⁺ and CD8⁺ T cells, but not by monocyte-derived DCs (Fig S4). When T cells were co-incubated with EBV-antigen pulsed iNKT-DCs in the presence of CD70 blocking mAb we observed a significant reduction in CD8⁺ T cell proliferation, while CD4⁺ T cell responses were not impeded (Fig 3F). CD70 expression by iNKT-DCs thus appears to selectively enhance CD8⁺ T cell activation. However, since proliferation was only reduced by about 50%, other signals (e.g. from other co-stimulatory ligands expressed by the DCs) likely also play an important role in CD8⁺ T cell activation by iNKT-DCs.

Antitumor activity in vivo.

We next used a pre-clinical human xenograft model of EBV-driven B cell lymphoma to test the ability of iNKT-DC mixtures to promote antitumor immunity. In this model, human umbilical cord blood cells (containing B cells and T cells that are naive to EBV) are briefly exposed to the virus in vitro, then injected intraperitoneally into immunodeficient NSG mice. EBV drives neoplastic transformation of human B cells over the next 2-3 weeks, resulting in the formation of aggressive lymphomas in the peritoneal cavity (22,29–31). The peritoneal tumors typically start to invade organ tissue within 3-4 weeks and mortality usually occurs in 5-6 weeks. Importantly, the tumors become highly infiltrated by autologous human T cells, but these fail to control the cancer. However, administering immune checkpoint blockade (ICB) antibodies leads to significantly enhanced antitumor T cell responses, reduced tumor burden, and improved survival (30). Thus, this pre-clinical model enables assessment of immunotherapies aimed at activating suppressed antitumor responses.

To test iNKT-DCs as a cellular immunotherapy, we set up this EBV-driven lymphoma model and generated autologous monocyte-derived DCs from the same samples. The DCs were pre-incubated for 24h with allogeneic CD4⁺ iNKT cells, and randomly chosen mice were injected intravenously with iNKT-DCs at 25 days post-transplant (a timepoint where tumors have already formed), while the remaining mice were not treated. Tumor burden of the whole cohort was assessed 6 days later. Mice injected with iNKT-DCs had significantly reduced tumor mass compared to those given no immunotherapy (Fig 4A). In contrast, mice given an equivalent dose of DCs or iNKT cells alone did not show reduction in tumor mass (Fig 4B). Dose-response experiments demonstrated that the antitumor activity of iNKT-DCs required ~10-fold lower cell doses than iNKT cells alone (Fig 4C). Histological examination revealed decellularized regions in iNKT-DC treated mice, suggesting clearance of areas previously occupied by tumor (Fig 4D). Analysis of tumor burden over time (day 26-33), showed that mice given iNKT-DCs underwent an abrupt reduction in tumor burden 3-4 days after administration of the immunotherapy (Fig 4E). To assess impact on survival we administered two iNKT-DC immunotherapy doses given one week apart (day 24 and 31). iNKT-DC treated mice showed significantly improved survival, with an 8.5 day median increase in survival time and a hazard ratio of 0.32 (Fig 4F). Moreover, approximately 33% of the iNKT-DC treated mice survived beyond 70 days, whereas without immunotherapy mortality was 100% within 51 days (Fig 4F). Together, these results demonstrated that iNKT-DC cellular immunotherapy induced a marked antitumor response in mice bearing established lymphomas.

Fig. 4. Antitumor activity of iNKT-DC cellular immunotherapy in vivo.

A) NSG mice were transplanted with EBV-infected human CBMCs to induce human B cell lymphoma formation, and after 25 days were intravenously injected with 0.5-1x10⁶ iNKT-DCs or not given immunotherapy (no I.T.), and total lymphoma mass was assessed 6 days later. Graph shows data aggregated data from 20 independent experiments, with each symbol showing results from an individual mouse and lines showing means. B) Tumor burden of mice that received no immunotherapy (no I.T.), 1x10⁶ iNKT-DC mixture, 1x10⁶ DCs alone, or 1x10⁶ iNKT cells alone. Each symbol shows results from an individual mouse; lines show means. C) Dose titration of iNKT-DCs compared to iNKT cells alone. Symbols show means ± SEM of n = 4-30 mice per dose. D) H&E stained tissue sections from lymphoma-invaded murine pancreas at 4x magnification, showing a mouse given no immunotherapy (left) and an iNKT-DC treated mouse (right). E) Tumor burden over time following iNKT-DC administration. Symbols show means ± SEM of n=3-31 mice per time point. F) Probability of survival for mice that received two-doses of iNKT-DC immunotherapy at days 25 and 31 (n=12)), or no immunotherapy (n=13). G) Tumor burden was evaluated at day 31 for mice given no immunotherapy, or treated with a single iNKT-DC dose at day 25, or with 3 doses of PD-L1 and CTLA4 immune checkpoint blockade antibodies (ICB 3x) at days 14, 19, and 24, or with a single ICB dose at day 25 (ICB 1x). Each symbol shows results from an individual mouse; lines show means. H) Probability of survival for mice that received no immunotherapy (n=6), 3 doses of ICB at days 14, 19, 24 (n=4), or two iNKT-DC doses at days 24 and 31 (n=5)). P values for (A), (B) and (G) by Mann-Whitney t-test; bars show means; P values for (F) and (H) by Mantel-Cox test.

To assess the relative potency of iNKT-DC immunotherapy, we performed head-to-head comparisons with ICB immunotherapy. Mice transplanted with EBV-treated CBMCs were given PD-1 and CTLA-4 antibody mixture (ICB), iNKT-DCs, or no immunotherapy. Whereas a single dose of iNKT-DCs had a statistically significant antitumor effect, one ICB dose administered at the same timepoint was not sufficient to reduce tumor burden (Fig 4G). Three ICB doses, given at 5 day intervals starting at day 14, produced a comparable antitumor effect as a single iNKT-DC treatment given at day 25 (Fig 4G). The three-dose ICB protocol also resulted in significantly improved survival (Fig 4H). However, whereas long-term survivors persisted in the iNKT-DC immunotherapy group, there was 100% mortality in the ICB group within 63 days (Fig 4H). These data underscore the efficacy of iNKT-DC immunotherapy, and suggest differences in mechanism of action compared to ICB, since iNKT-DC immunotherapy was effective when given after tumors have already formed (day 25) while ICB was only effective when initiated at an earlier timepoint (day 14).

Transcriptional changes.

To further investigate, we performed a micro-array based transcriptional analysis of 773 human cancer immunity-associated genes. Pairs of individual mice (one that received no immunotherapy, and one iNKT-DC-treated) were selected from three different experiments collected at day 29-31 (4-6 days post-immunotherapy), and samples of spleen and tumor tissue were tested from each mouse. Differential gene expression analysis revealed upregulation in iNKT-DC treated mice of genes encoding molecules that are selectively or predominantly expressed in T and/or NK cells, whereas downregulated genes were nearly all B cell-associated (Fig 5A). Genes upregulated in both spleen and tumor of iNKT-DC-treated mice included CD3, CD28, T cell Receptor Associated Transmembrane Adaptor 1 (TRAT1), CTLA-4, IL-7 receptor, and Vδ1⁺ γδ TCR genes (Fig 5A, Table S1).

Fig 5. Transcriptional changes associated with iNKT-DC immunotherapy.

A) Expression of 773 cancer-immune related transcripts was analyzed by microarray using tumor and spleen tissue from 3 mice treated with iNKT-DC immunotherapy compared to 3 mice given no immunotherapy (paired mice from 3 independent experiments). Volcano plots show differential expression of the 536 genes with detectable expression for iNKT-DC treated vs. no immunotherapy, with cut-offs set at log₂ fold-change of 0.5 and p value of 0.05. Color coding indicates predominant cell-type expression. B) Venn diagrams showing number of shared vs. distinct differentially expressed genes between tumor and spleen tissue. The top 5 biological categories (with number of genes in each) are shown for the non-shared transcripts. C) Total numbers of human B cells in spleens of mice that received no immunotherapy or iNKT-DCs on day 25. Data aggregated from 8 different experiments assessed at day 31; each symbol shows results from an individual mouse, lines show means. P value by Mann-Whitney t-test, bars at means. D) Human T cells from spleens of 3 replicate mice in each condition were tested directly ex vivo for expression of intracellular IL-2. Plot shows percent of CD4⁺ or CD8⁺ T cells staining positively for IL-2; P values by unpaired parametric t-test with Welch’s correction. E) Tumor tissue from mice given no immunotherapy or iNKT-DCs at day 25 was harvested at day 28 and stained for CD8 (red) and granzyme B (green), with cell nuclei visualized by DAPI (blue). Images show representative fields for each condition, with white arrows indicating extracellular deposition of granzyme B.

However, distinct gene signatures were also observed in tumor and spleen. In the spleen iNKT-DC immunotherapy was mainly associated with upregulation of genes involved in signaling, cell cycle, transcription, and differentiation, suggesting increased T cell priming (Fig 5B, Table S2). Tumor tissue of treated mice showed upregulation of genes for chemokines and cell migration, cytokines, and NK receptors, suggesting an increased effector response (Fig 5B, Table S3). Flow cytometric analysis of spleen samples revealed significantly reduced total human B cell numbers in iNKT-DC-treated mice (Fig 5C), consistent with the down-regulation of B cell-associated transcripts, and elevated frequencies of CD4⁺ T cells that were positive for IL-2 directly ex vivo, which is an indicator of T cells actively undergoing proliferation (Fig 5D). Consistent with increased effector activity in tumors, immunofluorescence analysis of tumor tissue revealed areas showing extracellular granzyme B deposition in iNKT-DC-treated mice, whereas granzyme B deposition appeared minimal in mice not given immunotherapy (Fig 5E). Together, these results suggest that iNKT-DC immunotherapy induces immune responses both in tumors and spleen, with T cell priming being more predominant in spleen and effector responses more apparent in tumor.

To further investigate, we carried out an analysis to assess tissue localization of iNKT cells and DCs after injection of iNKT-DCs into xenografted mice bearing EBV-driven lymphomas. iNKT cells were labeled with CFSE and DCs with CTV, then co-incubated to form iNKT-DCs. Mice were injected intravenously and euthanized 48h later for collection of blood, liver, lung, spleen, and tumor. Flow cytometric analysis revealed clearly detectable CFSE-labeled iNKT cells in each of the collected tissues (Fig S5). There was no evidence of CTV-labeled DCs in blood, and only minimal CTV staining in cells isolated from liver, but CTV staining was clearly detected in the lung, spleen, and tumor samples, with tumor showing the highest frequency (Fig S5). We observed a few CFSE-CTV dual staining events only in spleen and tumor samples, consistent with adhered iNKT-DC conjugates, but overall these were rare (Fig S5). It is not clear whether iNKT-DC complexes tend to dissociate after injection into mice, or whether it is simply more difficult to recover adhered complexes due to adhesive interactions with other cells. Nevertheless, together these results suggest that immunotherapeutic iNKT cells and DCs traffic to both spleen and tumor, but may elicit distinct responses in each tissue.

Antigen-specific response.

In a total of 19 independent experiments using unrelated human CBMC samples, we found that iNKT-DCs comprised of allogeneic iNKT cells and autologous DCs consistently induced marked tumor burden reductions in the EBV-driven lymphoma model (Fig 6A). However, when the DCs in the iNKT-DC mixture were allogeneic to the CBMCs there was no detectable antitumor effect (Fig 6B). This requirement for MHC-matching of the immunotherapeutic DCs with the engrafted T cells suggested a critical role for HLA-mediated antigen presentation in tumor control mechanisms. Supporting this, immunofluorescence analysis revealed a selective reduction in B cells expressing EBV antigens (BZLF1, LMP1, LMP2A) in iNKT-DC-treated mice (Fig 6C).

Fig. 6. Induction of targeted effector responses.

A) Summary of tumor burden reduction following iNKT-DC immunotherapy in EBV-lymphoma model generated using 19 unrelated human tissue samples. Symbols show average tumor burden for iNKT-DC treated mice as a fraction of the corresponding no I.T control. P value by one sample t-test. B) Mice were given no immunotherapy (no I.T.), or treated with iNKT-DCs where the DCs were autologous (autol) or allogeneic (allo) to the CBMCs used to establish the model. In both cases the iNKT cells were allogeneic. Graph shows data aggregated from 4 independent experiments; each symbol shows results from an individual mouse, lines show means. P values by Mann-Whitney t-test, bars at means. C) Spleen tissue of mice given no immunotherapy or iNKT-DCs at day 25 was analyzed by immunofluorescence at day 31 for cells expressing the indicated EBV antigens. Graph shows the number of positively stained cells detected in a field size of 0.8mm². Data aggregated from 3 independent experiments. P values by Mann-Whitney test; bars at means. D) T cells isolated from EBV-infected or control mice transplanted with CBMCs alone (uninfected) were tested for killing of corresponding splenocytes. Where indicated a 1:50 ratio of iNKT-DCs, DCs alone, or iNKTs alone were included in the cytotoxicity assay wells. Plots show frequency of dead target cells over time after subtraction of spontaneous target cell death; symbols represent means ± SD of 3 replicates; shown is one representative experiment of two.

To further investigate, we isolated T cells and splenocytes (target cells) from tumor-bearing or uninfected CBMC-engrafted mice and co-incubated these ex vivo in the presence or absence of autologous DCs, iNKT cells, or iNKT-DCs. Lysis of target cells was assessed over a 48h period using an IncuCyte live cell imaging platform. As we have previously observed (22,30), T cells from tumor-bearing mice failed to lyse autologous target cells (Fig 6D). Ex vivo exposure to iNKT-DCs led to increased specific lysis of splenocytes from EBV-infected but not uninfected mice, whereas no increased lysis was observed when either DCs or iNKT cells alone were added (Fig 6D). iNKT-DC mixtures did not show direct cytolytic activity against EBV-infected target cells in vitro (Fig S6A), but did induce increased degranulation by CD8⁺ T cells from adult PBMC in the presence of EBV antigen (Fig S6B). Together these results suggest iNKT-DCs activate antigen-specific clearance of EBV-infected B cells by autologous T cells from tumor-bearing mice.

Activation of T cells from cancer patients.

We next investigated the ability of iNKT-DCs to activate T cells ex vivo from patients with human papilloma virus (HPV) positive head and neck squamous cell carcinoma (HNSCC). Monocytes from HNSCC patient blood were cultured with GM-CSF and IL-4, and assessed for DC differentiation by flow cytometry. The cells acquired a phenotype of immature monocyte-derived DCs, as characterized by upregulation of CD209, down-regulation of CD14, and elevated expression of HLA-DR and CD86 (Fig 7A). Co-culture of these DCs with iNKT cells resulted in the formation of tightly adhered complexes (Fig 7B). Compared to DCs alone, the iNKT-DC conjugates showed significantly increased expression of class I and II HLA molecules, CD40, CD70, 4-1BBL (CD137L), and IL-15Rα (CD215) (Fig 7C). Thus, monocyte-derived DCs from HNSCC patients formed iNKT-DC complexes that had elevated expression of key CD8⁺ T cell activating molecules.

Fig. 7. Activation of CD8⁺ T cells from cancer patients.

A) Generation of monocyte-derived DCs from head and neck squamous cell carcinoma (HNSCC) patients. Plot on left shows representative flow cytometric staining of freshly isolated monocytes; right plot shows DC-like phenotype after 3 days of culture in medium containing GM-CSF and IL-4. Graph on right shows expression of key differentiation markers, symbols show results from different patient samples, bars show means. B) Co-incubation of HNSCC patient DCs with allogeneic iNKT cells from healthy donors results in formation of tightly-adhered iNKT-DC conjugates. Plot on left shows representative flow cytometric staining; graph on right shows percent of DCs in conjugates, symbols show results from different donors, bars show means. C) Conjugation of HNSCC patient DCs with iNKT cells leads to increased expression of key T cell activating ligands. Paired symbols show results for DCs alone vs. iNKT-DC conjugates from the same patient; points below dotted line did not have positive staining. P values by Wilcoxon matched pairs t-test. D) CFSE-labeled T cells from blood of HPV16⁺ HNSCC patients were cultured in the indicated conditions in the presence or absence of recombinant HPV E7 protein. Graphs show percent of CD4 or CD8 T cells that had undergone cell division after 10 days. Pairings show +/− antigen conditions from the same experiment; P values calculated by Wilcoxon match pairs signed rank test. E) Plots showing flow cytometric analysis of CD8⁺ T cells from one representative experiment out of seven. T cells cultured with E7 antigen are shown in filled histograms (top percentages); cultures without E7 antigen shown by dotted lines (bottom percentages). F) Tetramer staining of T cells from HLA-A2⁺ patients. Plots on left show staining for CTV and HPV E7 peptide-loaded HLA-A2 tetramer after gating on live cells expressing CD3 and CD8β. Graph on right shows data from 3 HLA-A2⁺ patients on the percent tetramer positive of proliferated CD8⁺ T cells after 10 days of culture with DCs alone or iNKT-DCs in the absence (light symbols) or presence (dark symbols) of recombinant E7 antigen. P values by paired t-test.

To test for patient T cell activation we used recombinant HPV E7 protein, since it has previously been shown that this viral protein is an important tumor-associated antigen for patients with HPV⁺ cancer (32–34), with HPV-specific T cells typically present at low frequencies in patient peripheral blood but enriched in tumors (33). T cells were isolated from blood of eight unrelated HPV16⁺ HNSCC patients, labeled with CFSE, and cultured in medium with or without E7 antigen, in the presence of iNKT cells, MPLA-treated monocyte-derived DCs, iNKT-DCs, or no added cells. Flow cytometric analysis revealed minimal proliferation of CD4⁺ and CD8⁺ T cells after 7 days culture alone, or with iNKT cells, regardless of whether E7 antigen was added or not (Fig 7D). Co-culture with MPLA-treated DCs led to somewhat elevated percentages of proliferated CD4⁺ T cells, but this was not different in the presence vs. absence of E7 antigen, and CD8⁺ T cell proliferation was not increased in the presence of MPLA-treated DCs (Fig 7D). In contrast, for both CD4⁺ and CD8⁺ T cells the percentage of proliferated cells appeared increased in the presence of iNKT-DCs, and significant enhancement was observed in the presence of recombinant E7 antigen (Fig 7D). Co-culture with iNKT-DCs and E7 antigen was also typically associated with a numerical expansion of CD8⁺ T cells (Fig 7E and S7). Finally, for HLA-A2 positive patients, we observed detectable staining of proliferated CD8⁺ T cells by HLA-A2 tetramers loaded with E7 peptide after co-culture with iNKT-DCs (Fig 7F). These results support the potential for using CD4⁺ allogeneic iNKT cells to boost the ability of monocyte-derived DCs from cancer patients to activate autologous tumor-specific CD8⁺ T cells.

Discussion

We show here that human CD4⁺ iNKT cells and monocyte-derived DC form sustained bi-cellular conjugates that express multiple costimulatory receptors, activate antigen-specific CD8⁺ T cells, and induce potent antitumor effects. These findings are significant because they address one of the most challenging tasks in developing clinical immunotherapies, which is achieving effective activation of antitumor CTLs (35,36). CD8⁺ T cell dysfunction is a central problem that limits effective antitumor immunity. Effector T cells face a number of obstacles during cancer, including upregulation of checkpoint ligands due to chronic antigen and cytokine exposure, inhibition via additional tumor-associated suppressive pathways, and metabolic exhaustion driven by conditions in the tumor microenvironment (TME) (37,38). Immunotherapies such as immune checkpoint blockade (ICB) and chimeric antigen receptor (CAR) T cells provide a boost in cytolytic effectors that can drive tumor regression, but these are often still limited by immunosuppressive and metabolically challenging conditions within tumors (39,40). Hence, an important key to the puzzle may be the development of immunotherapy approaches that deliver a consortium of co-stimulatory signals that enables antitumor effectors to overcome TME-associated challenges.

Our results suggest iNKT-DCs may provide such signals through expression of key co-stimulatory molecules. We found that in addition to expressing the classic B7 family co-stimulators CD80 and CD86, iNKT-DC complexes showed significantly elevated expression of tumor necrosis factor super family (TNFSF) molecules such as OX40L (CD134L), 4-1BBL (CD137L), and CD70, as well as the component of the IL-15 receptor (IL-15Rα) that is responsible for transpresentation. It is well established that 4-1BBL, CD70, and IL-15 transpresentation all promote CD8⁺ T cell survival and effector functioning (41–44). Stimulation by 4-1BBL fosters metabolic resilience in CD8⁺ T cells that facilitates their antitumor effector activity, and for this reason its cytoplasmic signaling domain is frequently included in the design of chimeric antigen receptors (45). Additionally, IL-15 transpresentation immunotherapy leads to tumor eradication by rejuvenating tumor-resident CD8⁺ T cells (46). Thus, we hypothesize that the expression of both 4-1BBL and IL-15Rα by iNKT-DCs contributes signals that help drive antitumor responses.

Our analysis also suggests a key role for CD70, which is expressed by iNKT cells in our immunotherapy. Co-stimulation through CD70 has been shown to promote T cell expansion by allowing daughter cells to survive successive divisions (42,43). However, availability of CD70 is highly restricted in vivo, with APCs showing transient cell surface expression only after immune activation (47). Transgenic expression of CD70 was found to be sufficient to convert immature DCs to a phenotype that primed CD8⁺ T cells to become tumor-eradicating effectors, whereas without CD70 the DCs generated abortive clonal expansion, dysfunctional antitumor responses, and failed to induce CD8⁺ T-cell memory (48). We found that CD70 played an important role in induction of CD8⁺ but not CD4⁺ T cell proliferation by iNKT-DCs. Since prior findings have suggested that coordination with TCR signaling is critical for the role of CD70 in promoting T cell survival (49), we speculate that iNKT conjugation with DCs may be necessary for them to co-stimulate CD8⁺ T cells via CD70, as this pairing provides spatial and temporal concurrence of CD70 with antigenic stimulation provided by the DCs.

Our transcriptional analysis suggests iNKT-DC immunotherapy activates T cells and innate effectors, including Vδ1⁺ γδ T cells, but the activation signatures differed between spleen and tumor. In spleen tissue, iNKT-DC immunotherapy was associated with upregulated expression of transcripts involved in T cell signaling, proliferation, and differentiation , suggesting T cell priming is a central activity in this site. The sphingosine-1-phosphate receptor (S1PR) was also upregulated, which plays a critical role in emigration of effector CD8⁺ T cells from lymphoid tissues (50). We also observed downregulation of YES1, a molecule that has been associated with increased regulatory T cells in cancer patients (51). However, flow cytometric staining did not reveal an altered frequency of FoxP3⁺CD25^hi T_regs in spleens of mice that received iNKT-DC immunotherapy. These data suggest iNKT-DC immunotherapy promotes T cell priming in spleen tissue.

In tumor tissue of iNKT-DC-treated mice we observed upregulation of transcripts associated with recruitment of antitumor effectors (CCL5, CXCL9, CCR1, CCR5, CXCR3, and CXCR6), as well as receptors characteristic of cytolytic effector populations (KIR2DL1/2, KLRB1, KLRC1, KLRC2/3, KLRD1, KLRK1), transcription factors associated with effective antitumor immunity by CD8⁺ T cells (EOMES, TCF-1), and antitumor cytokine pathways (IL-15, IL-16, IL-21, IL-11Rα, IL-12Rβ, IL-27Rα) (Tables S1,S3). These signatures suggest iNKT-DC immunotherapy markedly remodels immune responses in tumor tissue, promoting effector responses. It is not clear whether the enhanced effector responses are due to recruitment of newly primed cells from other sites (e.g. spleen), or to activation of effector cells that were already residing within tumors.

Our data furthermore suggest that iNKT-DCs activate antigen-specific antitumor responses. The requirement for iNKT-DC complexes to be generated using autologous DCs suggests a key role for HLA-mediated T cell activation. Additionally, iNKT-DC immunotherapy led to clearance of B cells expressing EBV-antigens. While we did not pre-treat the iNKT-DCs with EBV antigens for these experiments, we expect that EBV antigens were available for uptake in vivo, since the strain of EBV used here (M81) is highly lytic and leads to substantial death of infected B cells. Our analysis of blood samples from HPV16⁺ HNSCC patients provides further strong support for the ability of iNKT-DCs to activate antigen-specific CD8⁺ T cell responses. We thus postulate that in both the EBV and HPV systems, uptake of exogenous viral antigens by iNKT-DCs leads to cross-presentation of viral peptides on class I HLA molecules of the DCs, enabling antigen-specific T cell activation. Prior studies have established that monocyte-derived DCs have the capacity for cross-presentation of exogenous antigens, although they are less efficient than the cDC1 subset (52). While our studies demonstrate that complexing with iNKT cells greatly enhances the ability of monocyte-derived DCs to activate CD8⁺ T cells, we cannot distinguish whether this is mainly due to upregulation of co-stimulatory molecules, or whether improved antigen cross-presentation also plays a role. An interesting area for future analysis will thus be to determine the impact of iNKT cells on DC cross-presentation.

We have previously found that interactions between CD4⁺ iNKT cells and peripheral blood monocytes leads to the formation of APCs that suppress T cell responses (53,54). Thus, the approach we took here of pre-coupling iNKT cells with DCs may help to channel their activity towards adjuvancy. Since the antitumor activity of iNKT-DC complexes was observed using multiple unrelated human samples, this appears to be a highly conserved adjuvancy pathway capable of promoting antitumor responses in the context of diverse human genotypes. Since similar adjuvant-like effects were observed using multiple polyclonal or clonal CD4⁺ iNKT lines generated from different healthy donors, this function appears to be a common feature of human CD4⁺ iNKT cells. It will be of interest in future studies to evaluate whether modifications such as pre-loading the DCs with tumor antigens, or including more highly cytolytic subsets of iNKT cells (e.g. double-negative or CD8⁺ iNKT cells) provide further enhancements to antitumor activity. Ultimately, we envision that it may be possible to use stem cell engineering technologies to generate DCs bearing patient-matched HLA molecules, in order to produce a fully off-the-shelf iNKT-DC immunotherapy.

Synopsis.

Here, the authors show that iNKT cells and monocyte-derived DCs form stably adhered bi-cellular complexes that activate CD8+ T cells and promote rapid antitumor effects at timepoints when immune checkpoint blockade immunotherapy no longer works.