A High-Avidity T-cell Receptor Redirects Natural Killer T-cell Specificity and Outcompetes the Endogenous Invariant T-cell Receptor

The role played by cytotoxic T cells in tumor control has been recognized for years (1), but the clinical utility of their adoptive transfer has remained elusive due to the low frequency of high-avidity, tumor-specific T cells naturally circulating in peripheral blood (2). Advances in gene engineering have facilitated the introduction of tumor-antigen specificity in polyclonal T cells via expression of ectopic T-cell receptors (TCR; ref. 3). Using this strategy, clinically relevant numbers of TCR-redirected T cells for adoptive transfer can be generated and have produced encouraging clinical results (4, 5).

Despite the clinical success, a major issue of the TCR technology remains the expression of both the ectopic and endogenous TCR chains within the same T cell. This phenomenon can cause competition for binding to CD3 molecules, resulting in a reduced expression of the transgenic TCR (6) and TCR mispairing between the endogenous and transgenic TCR chains, causing potential autoreactivity (7). The retention of functional endogenous TCRs may exacerbate graft-versus-host disease (GvHD) if these cells are used as an allogeneic product (8).

Several strategies have been developed to enhance TCR transgenic expression, such as the inclusion of a codon-optimized sequence (9), the addition of extra disulfide bonds to the transgenic TCR chains (10), the introduction of murine residues in the TCR constant region domain (11), and the replacement of key amino acids in the framework of the variable region (3, 12). siRNA gene silencing (13) and TCR gene editing (14) have been explored to knockdown the endogenous receptor of T cells for replacement with a tumor-specific TCR. However, siRNA-mediated TCR knockdown is generally suboptimal, and the scale-up of gene editing for clinical use remains challenging and of unknown safety profile.

We thus have explored the use of natural killer T cells (NKT) as a TCR-based cell platform to overcome the issues associated with TCR engineering in T cells. Type 1 NKTs are an evolutionarily conserved subset of innate lymphocytes characterized by expression of the invariant TCR α-chain Vα24-Jα18 and reactivity to glycolipids presented by the monomorphic MHC class I–like molecule CD1d (15). In humans, the invariant TCR (iTCR) Vα24-Jα18 chain is paired almost exclusively with TCR Vβ11 (TRBV25-1) chains (16). We reasoned that the limited repertoire of the iTCR and its CD1d restriction should minimize, if not eliminate, both the TCR mispairing phenomena and the risk of GvHD when NKTs express a tumor-specific TCR. Here, we showed the successful TCR engineering, expansion, and HLA-restricted antitumor activity of TCR-redirected NKTs, and we dissected the mechanism by which the transgenic TCR displaces the endogenous iTCR from the cell surface of engineered NKTs.

The tumor cell lines SK-MEL-5 and M14 (melanoma), T2 (TAP transporter deficient), SupT1 (CD3⁺TCR⁻), and 293T cells were obtained from the ATCC between 2015 and 2018. The tumor cell line C8161 (melanoma) was obtained in 2016 from the tissue culture facility at University of North Carolina, Chapel Hill (TCF, UNC-CH). K562 cells were previously established as described by Quintarelli and colleagues (17). All cells were maintained in culture with the appropriate media, either RPMI-1640 (Gibco), DMEM (Gibco), or IMDM (Gibco) supplemented with 10% FBS (Sigma), 1% L-glutamine (Gibco), and 1% penicillin–streptomycin (Gibco) in a humidified atmosphere containing 5% CO₂ at 37°C. The M14 cell line was transduced with an SFG gamma retroviral vector encoding the HLA-A2*0201 molecule (17) to obtain the M14-A2 cell line upon cell sorting. SK-MEL-5, M14-wild-type (wt), and M14-A2 cells were transduced with an SFG gamma retroviral vector encoding the firefly luciferase gene and the fusion protein enhanced GFP (eGFP-FFluc; ref. 18). Cells were kept in culture for less than 6 consecutive months, after which aliquots from the original expanded vial were used. All tumor cell lines were routinely tested to exclude contamination with Mycoplasma and assessed for the expression of transgenes (such as HLA-A2, CD40L, CD80, OX40L, and eGFP) and tumor markers by flow cytometry to confirm identity.

Retroviral supernatants were prepared by transient transfection of 293T cells (17) and used to transduce T cells and NKTs (isolated as stated below). The HLA-A2–restricted TCRs specific for Tyrosinase_368-376 (Tyr-TCR recognizing the YMDGTMSQV peptide) and MART-1_26-35 (MART-1-TCR recognizing the ELAGIGILTV peptide) were obtained from Frankel and colleagues (19) and were cloned into the SFG retroviral vector (20). The HLA-A2–restricted survivin-specific TCR (Sur-TCR recognizing the LMLGEFLKL peptides 96–104) and the HLA-A2–restricted PRAME-specific TCR (PRAME-TCR recognizing the ALYVDSLFFL peptide p300) are previously described (21). The human CD8α cDNA (NM_001768) was cloned into the SFG retroviral vector. The human CD3 complex (CD3δ-2A-CD3γ-2A-CD3ϵ-2A-CD3ζ) was designed as previously reported (6) and synthetized from Geneart and cloned in the SFG retroviral vector.

Buffy coats from healthy volunteer blood donors were purchased from the Gulf Coast Regional Blood Center (Houston, TX). Peripheral blood mononuclear cells (PBMC) were isolated by Lymphoprep (Accurate Chemical and Scientific Corporation) density gradient centrifugation, according to the manufacturer's protocol. NKTs were purified from PBMCs using anti-iNKT microbeads (Miltenyi Biotech). NKTs were cultured in complete medium, consisting of 45% Click's medium (Irvine Scientific), 45% RPMI-1640 (Hyclone), 10% FBS (Hyclone), 1% L-glutammine (Gibco), and 1% penicillin–streptomycin (Gibco). Upon NKT-positive selection (LS column, Miltenyi Biotech), the negative fraction was used as feeder cells after irradiation (40 Gy, RS-2000 Biological System). For the first stimulation (S1), feeder cells were used once at day 0 at the PBMC:NKT ratio of 10:1. α-Galactosylceramide (αGalCer, 100 ng/mL, Diagnocine LLC) was added at day 0 and IL2 (200 IU/mL, STEMCELL Technologies) was added at days 0, 2, and 4. NKTs were transduced in retronectin (Takara, 7 μg/mL) 24-well coated plates (22) at day 6 when the NKT purity was more than 85%, and cells were restimulated with irradiated feeder cells, at the PBMC:NKT ratio of 5:1 (S2). The feeder cells were loaded with the peptide specific for each TCR (20 μmol/L, YMDGTMSQV for the Tyr-TCR, ELAGIGILTV for the MART-1-TCR, or ALYVDSLFFL for the PRAME-TCR from Genemed Synthesis) and αGalCer (100 ng/mL) in the presence of IL2 (200 IU/mL). NKTs were then further expanded for 10 days in the presence of IL2 (200 IU/mL). In selected experiments, NKTs, CD8⁺, and unselected T cells were isolated from the same buffy coat. T cells were positively (LS column, Miltenyi Biotech) selected using CD8-specific microbeads (Miltenyi Biotech). Unselected T cells and CD8⁺ T cells were activated, transduced, and expanded in complete medium with IL7 (10 ng/mL, PeproTech) and IL15 (5 ng/mL, PeproTech) as previously reported (23). Briefly, T cells were activated on anti-CD3 (1 μg/mL, Miltenyi Biotech) and anti-CD28 (1 μg/mL, BD Biosciences) 24-well coated plates. At day 2, T cells were transduced in retronectin (7 μg/mL) 24-well coated plates. On day 10, T cells were stimulated with peptide-loaded (20 μmol/L) and irradiated (80 Gy) artificial antigen-presenting cells (aAPC) consisting of K562 cells expressing CD40L, CD80, OX40L, and the HLA-A*02 molecules (17) at an aAPC:T-cell ratio of 1:4, and further expanded using IL7 (10 ng/mL) and IL15 (5 ng/mL) for 10 days. In selected experiments, NKTs were cotransduced with retroviral vectors encoding the CD8α or the CD3 complex.

T and NKT cells were stained with antibodies (Ab) against CD3 (APC-H7, clone SK7), CD62L (BV605, clone DREG-56), PD-1 (PE-Cy7, clone EH12.1), LAG3 (PE, clone T47-530), TIM3 (BV711, clone 7D3), CD45 (APC, clone 2D1), CD4 (PE-Cy7, clone SK3), and CD8 (Alexa Fluor 700, clone RPA-T8) from BD Biosciences. Tumor cells and aAPCs were stained with Abs against HLA-A2 (PE, clone BB7.2) and CD276 (BV421, clone 7-517) from BD Biosciences. The purity of NKTs was assessed by staining the cells with a tetramer specific for the iTCR (PE-conjugated human CD1d tetramer loaded with αGalCer; ProImmune) and the PE-conjugated Ab specific for TCR Vα24 chain (anti-iNKT, clone 6B11, from BD Biosciences). The expression of the Tyr-TCR in NKTs was assessed using a Tyr-TCR-specific pentamer (Pro5 MHC Pentamer A*02:01 YMDGTMSQV; ProImmune) followed by the staining with the Pro5 Fluorotag APC (ProImmune). NKT transduction efficiency with ectopic TCRs was assessed with an FITC-conjugated Ab specific for the TCR β12 chain (for the Tyr-TCR and MART-1-TCR; Beckman Coulter) or the TCR β14 chain (for the PRAME-TCR; Beckman Coulter), or stained with an APC-conjugated Ab specific for the murine TCR constant β chain (for the Sur-TCR; BD Biosciences). Data acquisition was performed on a BD FACSCanto II or BD LSRFortessa using the BD FACS-Diva software. Data analyses were performed with the FlowJo software. To count the number of CD3 and iTCR molecules expressed from NKTs, we used the PE Fluorescence Quantitation Kit (BD Quantibrite). In selected experiments, NKTs were costained with an anti-iTCR tetramer and anti-Tyr-TCR pentamer, and the iTCR⁺Tyr-TCR⁻, iTCR⁻Tyr-TCR⁺, and iTCR⁺Tyr-TCR⁺ NKTs were sorted using a FACSAria II at the Flow Cytometry Core Facility (UNC, Chapel Hill). After sorting, cells (purity >96%) were kept in culture overnight in complete media and IL2 (200 IU/mL) before performing the functional assays.

The cytotoxic activity of NKTs and CD8⁺ T cells was evaluated using a standard 5-hour ⁵¹Cr release assay (17). In brief, 5 × 10^{3 51}Cr (PerkinElmer)-labeled target cells [HLA-A2⁺ mismatched phytohemagglutinin (PHA) blasts; ref. 24; M14-wt, M14-A2, C8161, and SK-MEL-5) per well were plated in triplicate in a 96-well plate (v bottom) with different ratios (40:1, 20:1, 10:1, and 5:1) of effector cells and incubated for 5 hours at 37°C. The supernatant was collected and analyzed with a gamma-counter (PerkinElmer). HLA-A2⁺ (PHA) blasts were incubated for 2 hours at 37°C with specific peptides (100 nmol/L, YMDGTMSQV for the Tyr-TCR or ELAGIGILTV for the MART-1-TCR) and washed twice with complete medium. Target cells were incubated with medium alone or in 1% Triton X-100 (Sigma-Aldrich) to determine the spontaneous and maximum ⁵¹Cr release, respectively. The mean percentage of specific lysis of triplicate wells was calculated as follows: [(test counts − spontaneous counts)/(maximum counts − spontaneous counts)] × 100%.

SupT1 cells and NKTs were prepared according to the manufacturer's instructions (Abcam immunofluorescence protocol). Briefly, cells were fixed with cytofix buffer (BD Biosciences) and permeabilized with 0.1% Triron X in PBS. Cells were then incubated with anti-Vα24Jα18 (clone 6B11, Novus Biologicals) diluted at 1:800 in PBS/1% BSA. The Alexa Fluor 546 goat anti-mouse IgG1 (Invitrogen) diluted at 1:500 in PBS/1% BSA was used as a secondary Ab. Cells were mounted cover slips with one drop of ProLong Diamond Antifade Mountant with DAPI (Invitrogen). Data acquisition was performed on LSM700 Zeiss confocal microscopy using ZEN software (ZEISS Microscopy). Data analysis was performed with Fiji software.

DNA or RNA were extracted from sorted NKTs using the Qiagen DNeasy or RNeasy micro kits. NKT receptor β chain libraries were prepared using the Clontech SMARTer Human TCR α/β profiling kit, according to the manufacturer's protocol. Libraries were sequenced via 2 × 150 paired end on an Illumina MiSeq at the UNC High-Throughput Sequencing Facility. TCR variable chain calling was performed using IMGT/HighV-QUEST, with subsequent analyses performed using the tcR R package (25).

iTCR and Tyr-TCR expression and integration were measured in NKTs by qPCR with specific primers and probes and normalized to the 18S gene expression. Briefly, RNA was extracted from NKTs (RNeasy Plus Kit, Qiagen), and 1 μg of RNA was used for reverse transcription (5X VILO Reaction Mix and 10X SuperScript Enzyme Mix from Invitrogen). For the qPCR reaction (TaqMan 2X Universal PCR Master Mix from Life Technologies), 20 ng of cDNA were ran in duplicates. The relative expression was calculated as follows: 2^{− [(CTgene – CT18S) – CTVα24 in NT-NKTs]}. DNA was extracted from NKTs (DNeasy Blood and Tissue Kit, Qiagen) and 500 ng of DNA was used for the qPCR reaction. To calculate the number of copies of Tyr-TCR construct in NKTs post sorting, we created a standard curve with the SFG Tyr-TCR plasmid and the QuantStudio Real-Time PCR software automatically calculated the number of copies/100 ng DNA. Data acquisition was performed on a QuantStudio 6 Flex from Life Technologies using the QuantStudio Real-Time PCR software. The primers and probe for the 18S (Hs03003631_g1) were purchased from Thermo Fisher Scientific. Va24 primers: Forward 5′-TGTCAAGCTGGTCGAGAAAAG-3′ Reverse: 5′-TCATGAGCAGATTAAACCCGG-3′ Probe 5′-CCACTTTCAGGAGGAGGATTCGGAAC-3′. Tyr-TRAC primers: Forward 5′-ACTTCGACAGCCAGATCAAC-3′ Reverse: 5′-TGTGAAGCTGGTCTGATTG -3′ Probe 5′-TGCTGGACATGAAGGCGATGGA-3′. Tyr-TRBC primers: Forward 5′-TGGGTCAACGGCAAAGAG-3′ Reverse: 5′-TGGCATCTGAAGTGGTTCC -3′ Probe 5′-CCGATCCCCAGGCCTACAAAGAG-3′.

Three million tumor cells (M14-wt, M14-A2, C8161, or SK-MEL-5) were lysed with 81 μL of 2X Laemmil Sample Buffer (Bio Rad) and 9 μL of 2-mercaptoethanol (Fisher Scientific). Ten microliters of total protein lysates were separated by Mini-PROTEAN TGX Stain-Free Gels (Bio-Rad) electrophoresis. After protein transfer, membranes were incubated with 3 μL of anti-tyrosinase (Thermo Fisher, clone T311) or 15 μL of anti-MART-1 (Invitrogen, clone M2-7C10). After washes, membranes were incubated with 3 μL of HRP-conjugated goat anti-mouse IgG (H+L) secondary Ab (Invitrogen). Staining for β-actin (Santa Cruz Biotechnology) was used as loading control. Membranes were developed with SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Scientific).

NKTs and CD8⁺ T cells (10⁵ cells/well) were cocultured with tumor cell lines (M14-wt, M14-A2, C8161, or SK-MEL-5; 10⁵ cells/well) at an effector-to-target (E:T) ratio of 1:1 in 24-well plates, in complete medium, in the absence of cytokines. After 3 days of culture, cells were harvested and stained for CD3 (APC-H7, clone SK7 from BD Biosciences) and CD276 (BV421, clone 7-517 from BD Biosciences) monoclonal Abs (mAb) to detect T cells and tumor cells, respectively. Percentage of residual tumor cells in culture were enumerated by flow cytometry. NKTs (10⁵ cells/well) were also cocultured with HLA-A2⁺ T-cell blasts loaded with the YMDGTMSQV peptide (100 nmol/L) at an E:T ratio of 1:1 or activated with 10 μL of anti-Vα24 (clone 6B11, from BD Biosciences) in 1 mL of complete medium in the absence of cytokines. Culture supernatants were harvested after 24 hours of culture and IFNγ measured in 100 μL of supernatant with the DuoSet Human IFNγ ELISA kit (R&D Systems). Data acquisition was performed on a Synegrgy2 microplate reader (BioTek) using the Gen5 software.

The IFNγ ELISpot assay was performed as previously described (17). Briefly, 2 × 10⁴ NKTs or CD8⁺ T cells per well were plated in triplicate and then stimulated with 2 × 10⁴ T2 cells loaded with specific peptides (YMDGTMSQV for the Tyr-TCR, ELAGIGILTV for the MART-1-TCR, or ALYVDSLFFL for the PRAME-TCR). NKTs and CD8⁺ T cells were stimulated with PMA (25 ng/mL), ionomycin (1 μg/mL; both from Sigma-Aldrich), and IL2 (100 IU/mL) as a positive control. After 18 hours, IFNγ single-forming cells (SFC) were counted using the ELISpot reader (AID Classic ELR07) and analyzed using the Elispot7.0 software.

All mouse experiments were performed in accordance with UNC Animal Husbandry and Institutional Animal Care and Use Committee (IACUC) guidelines and were approved by UNC IACUC (IACUC ID: 49268).

Female and male NSG mice (7–9 weeks of age, obtained from the UNC Animal Core) were injected either subcutaneously (s.c.) with 0.5 × 10⁶ SK-MEL-5 tumor cells or intravenously (i.v.) via tail injection with 0.5–1 × 10⁶ FFluc-labeled M14-wt or M14-A2 tumor cells. In specific experiments, mice received low-dose irradiation (100 cGy) 24 hours before receiving i.v. 1 × 10⁶ FFluc-labeled M14-A2 cells. Seven days after tumor cell injection (day 0) and at days +5 and +12, mice were infused i.v. with 1 × 10⁷ or 5 × 10⁶ nontransduced (NT)-NKTs or Tyr-TCR NKTs (expressing or not eGFP-FFluc, and nonsorted for TCR expression). In selected experiments, nontransduced (NT)-T cells and T cells expressing the Tyr-TCR (Tyr-TCR-Ts and nonsorted for TCR expression) were also used. In selected experiments, recombinant human IL2 (1000 U/mouse) was administered intraperitoneally (i.p.) every 2 days for a total of 8 doses after the adoptive transfer of either NKTs or T cells. Melanoma tumor cell growth was monitored weekly either with caliper measurement for s.c. tumors, or by bioluminescence (BLI; total flux, photons/second) using the IVIS kinetic in vivo imaging system (PerkinElmer) for the metastatic models. Mice were sacrificed according to UNC guidelines for either tumor growth or occurrence of sign of discomfort, including graft-versus-mouse disease (weight loss, fur loss, and lethargy). When mice were sacrificed, peripheral blood was collected from the heart and spleen, liver, and tumor were smashed on cell strainers and washed with 2 mL of PBS. Peripheral blood, spleen, liver, and tumors were analyzed to detect the presence of NKTs or T cells [stained with Abs against CD3 (APC-H7, clone SK7) and CD45 (APC, clone 2D1)] by flow cytometry using CountBright absolute counting beads (Invitrogen). In selected experiments, tumors were isolated, homogenized in a total of 1 mL of PBS, and 100 μL of the supernatant were used to measure the human IFNγ using the DuoSet Human IFNγ ELISA kit (R&D Systems). Infiltrating NKTs were counted in the tumors by staining with Abs against CD3 (APC-H7, clone SK7) and CD45 (APC, clone 2D1) and counting beads (Invitrogen) and analyzing via flow cytometry.

Data were summarized as the mean ± SD. Student t test or two-way ANOVA were used to determine statistically significant differences between treatment groups, with Bonferroni's correction for multiple comparisons when appropriate (Prism 6: GraphPad Software). Survival analysis was performed using the Kaplan–Meier method (Prism 6: GraphPad Software). All P values less than 0.05 were considered statistically significant.

Isolated and cultured NKTs (Fig. 1A) robustly and consistently expanded ex vivo (fold increase: 17 ± 4 after the first stimulation, S1; Fig. 1B). By days 5 to 7 of culture (before retroviral transduction), 93% ± 4% of the isolated cells expressed the iTCR, as assessed by flow cytometry using a specific tetramer (Fig. 1B). As a proof of concept, NKTs were transduced with a gamma retroviral vector to express an HLA-A2–restricted, tyrosinase-specific TCR (Tyr-TCR; ref. 19). Tyr-TCR was detectable in 76% ± 8% NKTs (Tyr-TCR NKTs), and the expression remained stable over the course of the entire culture period (Fig. 1C). The growth kinetics of NKTs was not hampered by the expression of the Tyr-TCR, as 1.2 × 10⁸ ± 8.5 × 10⁷ cells were obtained at the end of the culture period (S2), a number that is sufficient for clinical adoptive transfer (Fig. 1D). Tyr-TCR NKTs retained the phenotypic characteristics of nontransduced NKTs, including preservation of CD4 and CD8 composition and expression of CD62L and exhaustion markers (Fig. 1E). Tyr-TCR NKTs showed HLA class I restriction, as they lysed HLA-A2⁺ PHA blasts loaded with the HLA-A*0201 tyrosinase YMD peptide (77% ± 28% lysis at the E:T ratio of 40:1), whereas control NKTs (NT-NKTs) showed negligible cytotoxic activity (Fig. 1F). Overall, NKTs could be successfully manipulated to express a functional MHC class I–restricted TCR.

As early as 3 to 4 days after retroviral transduction with the Tyr-TCR–expressing vector, Tyr-TCR NKTs consisted of two distinct subsets, namely, CD3⁺iTCR⁺ (59% ± 18%) and CD3⁺iTCR⁻ (41% ± 18%) cells (Fig. 2A). A similar downregulation of the iTCR was observed when NKTs were transduced with gamma retroviral vectors encoding other HLA-A2–restricted TCRs specific for survivin (Sur-TCR), PRAME (PRAME-TCR), and MART-1 (MART-1-TCR) antigens (Supplementary Fig. S1A), thus excluding that the observed emergence of CD3⁺iTCR⁻ NKTs was an exclusive phenomenon caused by the Tyr-TCR. Staining with iTCR and Tyr-TCR multimers revealed three distinct cell subsets: iTCR⁻Tyr-TCR⁺ (44% ± 15%), iTCR⁺Tyr-TCR⁺ (34% ± 11%), and iTCR⁺Tyr-TCR⁻ (19% ± 11%) cells. These three subsets were also identified when using other combinations of Abs (Vα24, Vβ11, and Vβ12 Abs) recognizing Tyr-TCR and iTCR (Fig. 2B). Using SupT1 cells (CD3⁺TCR⁻; ref. 26), we observed that ectopic TCRs outcompeted iTCRs for cell-surface expression (Supplementary Fig. S1B) and that iTCRs were retained within the cytoplasm (Supplementary Fig. S1C). When iTCR⁺Tyr-TCR⁺ and iTCR⁻Tyr-TCR⁺ NKTs were sorted, the Vα24 chain of the iTCR was detectable in both subsets by confocal microscopy but was confined to the cytoplasm in iTCR⁻Tyr-TCR⁺ NKTs, whereas it was expressed on the cell surface in iTCR⁺Tyr-TCR⁺ NKTs (Fig. 2C). Based on these results, we hypothesized that in iTCR⁻Tyr-TCR⁺ NKTs, the Tyr-TCR displaced the iTCR for CD3 usage, and as a consequence, it was preferentially transferred to the cell surface. To corroborate this premise, we transduced NKTs with a retroviral vector encoding the CD3γδϵζ genes (6) to provide an excess of CD3 to NKTs, and confirmed an increase of CD3 molecules on the cell surface (Fig. 2D). When Tyr-TCR⁻ NKTs were transduced with the CD3 encoding vector, we observed an increase of the iTCR⁺Tyr-TCR⁺ subset (from 34% ± 14% to 72% ± 3%, gated on Tyr-TCR⁺ cells, P < 0.0001; Fig. 2E). Overall, these data indicated that the ectopic TCR can displace the iTCR from CD3 binding in NKTs.

We then studied how the transgenic TCR displaces the endogenous iTCR in NKTs. We first analyzed if iTCR⁺Tyr-TCR⁺ and iTCR⁻Tyr-TCR⁺ subsets showed differences in coreceptor expression, but both CD4 and CD8 were equally distributed within the two subsets (Fig. 3A). Although the Vα24-Jα18 chain is known to preferentially pair with Vβ11 (TRBV25-1) chains in human NKTs (16), we explored if combination with distinct Vβ chains correlated with the specific NKTs subsets. iTCR⁺Tyr-TCR⁺ and iTCR⁻Tyr-TCR⁺ were sorted using specific TCR multimers (Supplementary Fig. S2A) and subsequently analyzed for TCRβ using RNA amplicon sequencing. Although TRBV25-1 was the dominant detected TCR Vβ chain (range, 90.54%–96.49%), a proportion of the repertoire was distributed across the other 47 Vβ chains (Fig. 3B). However, TCR Vβ chain distribution was conserved between iTCR⁺Tyr-TCR⁺ and iTCR⁻Tyr-TCR⁺ subsets, suggesting that the TCR Vβ repertoire differences did not segregate for iTCR expression. The majority of shared TCR clones (identified by the CDR3 sequence of the Vβ chain) was observed within each donor rather than within the iTCR⁺Tyr-TCR⁺ or iTCR⁻Tyr-TCR⁺ subsets (Supplementary Fig. S2B), indicating that the type of TCR Vβ chain associated with the Vα24-Jα18 was not affecting the binding with CD3, and thus the iTCR displacement in iTCR⁻Tyr-TCR⁺ NKTs.

We next reasoned that the iTCR expression may be associated with transgene integrant copy-number variations between the iTCR⁺Tyr-TCR⁺ and iTCR⁻Tyr-TCR⁺ subsets. We assessed the mRNA expression of both Tyr-TRAC and Tyr-TRBC chains (transgenic TCR) and of Vα24 (iTCR) in both iTCR⁺Tyr-TCR⁺ and iTCR⁻Tyr-TCR⁺ subsets and found the mRNA of both Tyr-TCR chains to be significantly higher than the mRNA expression of the iTCR chain (Fig. 3C). Both Tyr-TCR chains were more significantly expressed in iTCR⁻Tyr-TCR⁺ cells than in the iTCR⁺Tyr-TCR⁺ subset (P = 0.0004; Fig. 3C). The higher mRNA expression of Tyr-TRAC and Tyr-TRBC chains in iTCR⁻Tyr-TCR⁺ NKTs reflected the higher copy numbers of the transgene integrants (Fig. 3D). We then analyzed the functionality of sorted iTCR⁺Tyr-TCR⁺ and iTCR⁻Tyr-TCR⁺ NKTs and found that it was independent from the expression of the iTCR because only the iTCR⁺Tyr-TCR⁺ subset released IFNγ in response to the agonistic Vα24 mAb (clone 6B11; Fig. 3E), but both iTCR⁺Tyr-TCR⁺ and iTCR⁻Tyr-TCR⁺ subsets lysed and produced IFNγ in response to HLA-A2⁺ PHA blasts loaded with the HLA-A*0201–restricted YMD peptide (Fig. 3F). Overall, these data suggest that the ectopic TCR outcompeted the iTCR for CD3 binding when the α- and β-TCR chains were expressed at higher levels than iTCR chains, so that the HLA-restricted cytotoxic activity of TCR NKTs could be exploited regardless of the presence of the iTCR on the cell surface.

To measure the antitumor activity of Tyr-TCR NKTs against tumor cells expressing tyrosinase, we used the HLA-A2⁺Tyr⁺ melanoma cell line SK-MEL-5. The HLA-A2⁺Tyr⁻ melanoma cell line C8161 was used as a negative control. As additional controls, we tested the melanoma cell line M14 that is HLA-A2⁻Tyr⁺ (M14-wt) and the same cell line genetically modified to express the HLA-A2 molecule (M14-A2; Fig. 4A). Tyr-TCR NKTs consistently lysed the HLA-A2⁺Tyr⁺ M14-A2 and SK-MEL-5 cells but not the HLA-A2⁻Tyr⁺ (M14-wt) or HLA-A2⁺Tyr⁻ (C8161) tumor cells (Fig. 4B). Both iTCR⁺Tyr-TCR⁺ and iTCR⁻Tyr-TCR⁺ NKT subsets lysed the M14-A2 and SK-MEL-5 cells (Supplementary Fig. S3A). We then cocultured Tyr-TCR NKTs and control NT-NKTs with these melanoma cell lines for 3 days. Both M14-A2 and SK-MEL-5 cells were efficiently eliminated (<3% residual tumor cells) by Tyr-TCR NKTs, whereas M14-wt and C8161 cells overgrew and accounted for more than 85% of the cells at the end of the 3-day culture (Fig. 4C). NT-NKTs did not eliminate any of these tumor cells. The antitumor activity of Tyr-TCR NKTs was paralleled by IFNγ production, which was detected only when Tyr-TCR NKTs were cocultured with M14-A2 and SK-MEL-5 cells (Fig. 4D). The IFNγ production in response to melanoma cell lines was comparable in both iTCR⁺Tyr-TCR⁺ and iTCR⁻Tyr-TCR⁺ NKT subsets (Supplementary Fig. S3B). Overall, these data demonstrated the antitumor activity of NKTs genetically modified to express an HLA-A2–restricted TCR.

To assess if TCR avidity played a role in determining the antitumor activity of TCR-redirected NKTs, we used NKTs expressing an HLA-A2–restricted MART-1–specific TCR (MART-1-TCR), which is characterized by lower functional avidity compared with the Tyr-TCR (19). Transduction of NKTs with MART-1–TCR was comparable with that of Tyr-TCR (75% ± 13% vs. 78% ± 10%, respectively; Fig. 5A) and also in terms of expression [MART-1–TCR mean fluorescence intensity (MFI) = 22,377 ± 1,753 vs. Tyr-TCR MFI = 19,461 ± 2,336; Supplementary Fig. S4A]. However, MART-1–TCR NKTs had less IFNγ spot-forming cells against T2 cells loaded with decreasing concentrations of the specific peptide compared with Tyr-TCR NKTs (Fig. 5B), despite TCRs being expressed at comparable levels. Similarly, MART-1–TCR NKTs showed lower cytotoxic activity against HLA-A2⁺ PHA blasts loaded with the specific peptide than Tyr-TCR NKTs (Fig. 5C). In contrast, when expressed in CD8⁺ T cells, both Tyr-TCR and MART-1–TCR (Fig. 5D) showed similar functionality to one another, assessed by IFNγ SFCs in response to peptide-loaded T2 cells (Fig. 5E) and cytotoxic activity against peptide-loaded HLA-A2⁺ PHA blasts (Fig. 5F). We also compared the antitumor activity of Tyr-TCR– and MART-1-TCR–expressing T cells and NKTs against the SK-MEL5 cell line, which expresses both tyrosinase and MART-1 (Fig. 5G). Although CD8⁺ T cells expressing either the Tyr-TCR or MART-1–TCR eliminated HLA-A2⁺Tyr⁺MART-1⁺ melanoma cells in coculture experiments (Fig. 5H), only Tyr-TCR NKTs successfully eliminated these tumor cells (Fig. 5I). These data were confirmed using another low-affinity HLA-A2-restricted TCR specific for PRAME (PRAME-TCR). As observed for the MART-1-TCR, despite comparable transduction efficiency and MFI (Supplementary Fig. S4B), PRAME-TCR-NKT functions were inferior to those of PRAME-TCR-Ts, as the former had less IFNγ SFCs against T2 cells loaded with decreasing concentrations of the specific peptide (Supplementary Fig. S4C) and showed an inferior cytotoxic activity against peptide-loaded HLA-A2⁺ PHA blasts (Supplementary Fig. S4D).

Because CD8⁺ T cells expressing either MART-1-TCR or Tyr-TCR were equally effective, we explored if the functionality of MART-1-TCR NKTs, that were mostly CD4⁺ or CD4⁻CD8⁻ cells, could be rescued by overexpressing the CD8α subunit, as previously reported for CD4⁺ T cells (27). Coexpression of both a MART-1-TCR and human CD8α (>60% CD8α⁺), this modification failed to enhance the functional avidity of MART-1-TCR NKTs (Supplementary Fig. S5A–S5C). Thus, our data suggest that the intrinsic functional avidity of the TCR was critical to functionally redirect NKT activity via HLA class I–restricted TCRs.

Having shown that functional avid Tyr-TCR grants the antigen specificity of NKTs in vitro, we used a xenogeneic NSG melanoma mouse model to assess if Tyr-TCR NKTs controlled tumor growth in vivo. NKTs were administered i.v. in mice engrafted s.c. with the melanoma cell line SK-MEL-5 (Fig. 6A). Tyr-TCR NKTs exhibited superior control of tumor growth than NT-NKTs (317 mm³ ± 124 mm³ vs. 1121 mm³ ± 216 mm³, respectively, at day 64, P < 0.0001; Fig. 6B; Supplementary Fig. S6A). None of the NKT-treated mice developed signs of graft-versus-mouse toxicity. In harvested tumors, we observed a higher number of infiltrating NKTs in mice treated with Tyr-TCR NKTs compared with NT-NKTs (997 ± 650 NKTs vs. 351 ± 401 NKTs, respectively) and higher detection of IFNγ release (183.5 pg/mL ± 179.1 pg/mL vs. 12.1 pg/mL ± 13.7 pg/mL, respectively; Fig. 6C), suggesting that Tyr-NKTs were functional at the tumor site. In contrast, no difference in the numbers of Tyr-TCR NKTs and NT-NKTs was detected in peripheral blood, liver, and spleen, and the expression of the Tyr-TCR was detectable in all the Tyr-TCR NKTs isolated (Supplementary Fig. S6B and S6C).

To further assess in vivo the HLA restriction of the antitumor responses mediated by Tyr-TCR NKTs, we also generated a metastatic murine model in which NSG mice were injected i.v. with either M14-wt or M14-A2 cells modified to express the enhanced green fluorescent protein (eGFP) and the firefly luciferase gene (FFLuc or eGFP-FFLuc). Tumor bioluminescence (BLI) was reduced only in mice engrafted with M14-A2 cells and treated with Tyr-TCR NKTs (P < 0.0001; Fig. 6D). When euthanized between days 41 and 63, the number of liver tumor metastases in M14-A2 tumor–bearing mice was significantly reduced compared with M14-wt tumor–bearing mice (1 ± 2 vs. 30 ± 13, P < 0.0001; Fig. 6E). In the same mouse model, we also assessed in vivo the persistence of Tyr-TCR NKTs using NKTs labeled with eGFP-FFLuc and unlabeled tumor cells. Tyr-TCR NKT BLI was detectable for 7 days after infusion (Fig. 6F). NKT persistence was also corroborated by phenotypic analyses in the peripheral blood and spleen (1.6% ± 0.2% and 0.9 ± 1.2% hCD3⁺hCD45⁺ cells, respectively) of treated mice (Fig. 6G). To assess if Tyr-TCR NKTs produced superior antitumor control when their persistence was prolonged, we infused multiple doses of Tyr-TCR NKTs and compared the efficacy of Tyr-TCR NKTs and Tyr-TCR-Ts generated from the same donors (Fig. 7A). Both Tyr-TCR NKTs and Tyr-TCR-Ts, which were detectable in the circulation up to days 40 to 60 (Fig. 7B), successfully controlled tumor growth (Fig. 7C and D) and produced an improved overall survival compared with NT-NKTs and NT-Ts (P < 0.0001; Fig. 7E). The number of liver tumor metastases was significantly reduced in mice treated with either the Tyr-TCR NKTs or the Tyr-TCR-Ts compared with NT-NKTs and NT-Ts (P < 0.0001; Supplementary Fig. S6D). Tyr-TCR NKTs and Tyr-TCR-Ts were detectable in both the liver and spleen of mice sacrificed between days 40 and 60 (Supplementary Fig. S6E). However, 11 of the 18 mice (61%) receiving Tyr-TCR-Ts developed lethal graft-versus-mouse disease characterized by weight and fur loss and lethargy (Fig. 7F). This graft-versus-mouse disease also confounded the observed long-term tumor control mediated from the Tyr-TCR-Ts. Finally, we used the same in vivo models, but decreased the number of T cells and NKTs injected (5 × 10⁶ cells/injection; Supplementary Fig. S7A). The antitumor activity of Tyr-TCR NKTs and Tyr-TCR-Ts treated mice was comparable (Supplementary Fig. S7B–S7D), and the Tyr-TCR-Ts continued to cause a lethal graft-versus-mouse effect (Supplementary Fig. S7E). Thus, our data indicated that TCR NKTs could control tumor growth in vivo without causing reactivity against murine tissues.

We showed that the antigen specificity of human NKTs could be redirected toward tumor antigens by inserting an ectopic TCR. Our study critically outlined the advantages of using this platform for the expression of ectopic TCRs, namely, how the TCR could outcompete the iTCR for the cell-surface expression without the need for additional manipulation to knockdown endogenous TCR chains as required in T cells.

Adoptive transfer of polyclonal T cells expressing high-affinity TCRs have shown promising clinical responses (28). However, current issues related to ectopic TCR expression in T cells suggest that exploring alternative platforms for TCR expression may be beneficial to further enhance clinical activity and safety of TCR-engineered cells. We previously demonstrated that NKTs can be engineered to express a chimeric antigen receptor for redirected tumor target capabilities (29). Jiang and colleagues previously showed that an HLA class I–restricted TCR targeting an epitope of the 38-kDa protein of the Mycobacterium tuberculosis can be expressed by NKTs and react against DCs pulsed with the 38-kDa protein (30). However, redirecting NKTs for the targeting of tumor cells via HLA class I restriction has so far remained unexplored.

Here, we have shown that NKTs expressing an ectopic HLA class I–restricted TCR can be generated in less than 3 weeks of culture, and in sufficient numbers for the clinical translation. We showed that the ectopic TCR can displace the iTCR on a proportion of NKTs by outcompeting for binding to the CD3 complex. This phenomenon occurred when the mRNA of the ectopic TCR chains was overexpressed compared with the endogenous iTCR, with higher transcript numbers dictating which receptor was expressed on the cell surface of NKTs. Interestingly, Sommermeyer and colleagues previously showed that a transgenic TCR can outcompete the endogenous TCR in CMV-specific T-cell clones (31). Our study, thus, underlines that the restricted TCR repertoire of polyclonal NTKs, compared with the broad TCR repertoire of polyclonal T lymphocytes, overcomes the hurdle of gene editing required in polyclonal T cells to knockdown the endogenous TCRs by granting the displacement of the iTCR.

Independent from the expression of the iTCR, TCR NKTs proved to specifically recognize the peptide/HLA-A2 complex and exhibited a potent antitumor effect against HLA-A2⁺ tumor cells expressing the targeted antigen. TCR NKTs showed fast killing of the target cells and activation as demonstrated by the production of IFNγ. TCR NKTs maintained their specificity in vivo and mediated tumor regression leading to improved overall survival. In particular, TCR NKTs could control tumor metastasis within the liver. It has been reported that murine NKTs can accumulate in the liver of mice bearing hepatocellular carcinomas and inhibit the tumor growth in both primary and metastatic liver models (32). It remains to be demonstrated how human NKTs in NSG mice cause a robust antitumor effect in liver metastases. Our antitumor effects were significant, considering that, because NKTs are well known for orchestrating immune responses by inducing maturation of DCs, activation of T and B cells and transactivation of NK cells (33–35), our model likely underestimated the real potentials of TCR NKTs.

Finally, one further advantage of using NKTs rather than polyclonal T cells for TCR expression lies in the reduced risk of GvHD when both ectopic and native TCRs are retained by transduced cells. Preclinical and clinical evidence suggests that NKTs may act as negative regulators of GvHD, making them relevant in the allogeneic context (36–39). Corroborating these clinical suggestions are previous observations that CAR-engineered human NKTs do not cause graft-versus-mouse—a surrogate of GvHD—in NSG mice (29, 37–40). Here, we showed that TCR NKTs could not only be equally effective as TCR-expressing T cells in controlling tumor growth, but that, unlike T cells, they did not cause graft-versus-mouse in NSG mice. Thus, the manipulations needed for appropriate expression of a transgenic TCR and abrogation of the endogenous TCR in T cells appear dispensable in NKTs. Overall, we showed that using NKTs as a cell platform for TCR expression in cancer immunotherapy is feasible and provide significant advantages compared with T cells. Based on our current analysis, the only requirement for effective antitumor activity via HLA-restricted TCR by NKTs is the need for expressing TCRs characterized by high functional avidity.

L.S. Metelitsa reports receiving a commercial research grant from and has ownership interest (including patents) in Cell Medica. G. Dotti reports receiving a commercial research grant from Cell Medica and is a consultant/advisory board member for MolMed s.p.a. and Bellicum Pharmaceutical. B. Savoldo reports receiving a commercial research grant from Cell Medica. No potential conflicts of interest were disclosed by the other authors.

Conception and design: E. Landoni, G. Dotti, B. Savoldo

Development of methodology: E. Landoni, L.S. Metelitsa, G. Dotti, B. Savoldo

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E. Landoni, C.C. Smith, G. Fucá, Y. Chen, C. Sun, B. Savoldo

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E. Landoni, C.C. Smith, G. Fucá, Y. Chen, C. Sun, B.G. Vincent, L.S. Metelitsa, G. Dotti, B. Savoldo

Writing, review, and/or revision of the manuscript: E. Landoni, C.C. Smith, B.G. Vincent, L.S. Metelitsa, G. Dotti

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E. Landoni

Study supervision: G. Dotti, B. Savoldo

The imaging core and UNC Flow Cytometry core facilities are supported in part by an NCI cancer core grant, P30-CA016086-40. This work was supported in part by Cell Medica. B. Savoldo is supported in part by R01HL114564, and G. Dotti is supported in part by R01-CA193140.

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