Preclinical assessment of MAGE-A4-specific TCR-NK cells against solid tumors

Abstract

T cell (Tc) receptor (TCR)-based cell therapies have shown clinical efficacy across many cancer types and represent an attractive strategy for targeting solid tumors. However, the immunosuppressive tumor microenvironment, downregulation of target antigen and HLA, and the need for an autologous source limit the efficacy and the accessibility of TCR-Tc therapies. Early clinical trials have shown the potential of natural killer cells (NKs) as a therapy to treat hematological and solid cancers. Allogeneic NKs, engineered to express a TCR, represent a novel and promising strategy overcoming the limitations of T and NKs therapies.

Here we describe the development of a product consisting of NKs engineered to express an affinity-enhanced TCR recognizing MAGE-A4, a clinically validated tumor antigen expressed across several solid tumors.

The introduction of the TCR does not disrupt the innate functionality of NKs and adds TCR-mediated specific killing of antigen-positive targets. In fact, the innate potential of the NKs appears to be enhanced by the presence of the CD3-TCR complex, creating NKs with increased potency. TCR-NKs are faster, more potent than TCR-Tc and retain killing activity in the absence of TCR target antigen thus potentially overcoming tumor heterogeneity and/or antigen loss. Lastly, TCR-NKs are not activated when co-cultured with normal cells, displaying a safe profile.

Combining the innate cytotoxicity of NKs with MAGE-A4-specific targeting of an affinity-enhanced TCR, results in a potent and safe cellular product representing a promising and novel therapeutic off-the-shelf paradigm for the treatment of many solid cancers.

Graphical Abstract

Graphical Abstract.

Introduction

Over the past decade, major advances have been made for cancer patients with novel immunotherapeutic approaches, including cell therapies [1]. Chimeric antigen receptor-T (CAR-T) cell therapies have shown remarkable clinical responses in hematological cancers and led to several approvals and, in some cases, complete tumor eradication and remission. Nonetheless, they have so far failed to realize equally significant responses in solid tumors [2], which are particularly challenging to treat, given their unfavorable microenvironments and heterogeneity [3, 4].

While CAR-T cells are limited to targeting cell surface antigens, T cells engineered with tumor antigen-specific T cell receptors (TCRs) can target virtually any intracellular peptide, resulting in some of the most advanced clinical responses observed in solid cancers today [5–8]. This has led to an approval of an engineered TCR-T-based therapy for the treatment of synovial sarcoma [7]. However, virtually all late-stage solid cancer patients undergoing treatment, including those treated with TCR-T therapies, will relapse. One of the dominant reasons for this is the heterogeneity of solid tumors. Since not all tumor cells express the target antigen and can evolve to lose the antigen and/or human leukocyte antigen (HLA) on treatment, resistance to treatment and ultimately relapse will occur [6, 7, 9]. Further limitations include the scalability and accessibility of these treatments. Most TCR-T cell therapies are autologous, where each treatment is manufactured specifically for each patient, making them costly and difficult to scale for widespread use. Furthermore, the manufacturing process can delay treatment, potentially allowing disease progression. TCR-T cell therapies are also associated with toxicities such as cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS) [10, 11] requiring specialist centers for treatment and further limiting broad accessibility.

Natural killer (NK) cells offer an alternative to T cells as potent cytotoxic cells, capable of overcoming some of the limitations hindering T cells. Unlike T cells, NK cells are not antigen-specific and can recognize a variety of ligands on cancer cells including lost or downregulated target antigens [12, 13]. Many cancer cells downregulate classical HLA-I molecules as an escape mechanism from T cell recognition. In this situation, similarly to T cells, TCR-NK cells would lose the activation through the TCR-HLA/peptide interaction but the innate recognition should be maintained if not enhanced due to the reduction in the NK inhibitory signal mediated by the interaction of killer-cell immunoglobulin-like receptors (KIR) and Leukocyte immunoglobulin-like receptors (LILR) with classical HLA-I molecules [14, 15].

The development of NK cell therapies over the past 30 years has consistently shown a favorable safety profile, even at high doses, with varying outcomes for efficacy. In the autologous setting, NK cell activity is inhibited in patients with cancer largely due to the matching of KIRs to the patients’ HLA. However, in contrast to T cells, NK cells do not cause graft-versus-host disease (GvHD) and can be used in an allogeneic format, where the KIR/HLA mismatch has also been shown to mediate a more powerful anti-tumor response [16]. As opposed to allogeneic T cells, there is no requirement to remove endogenous TCRα and TCRβ chains to overcome the potential of GvHD. Several clinical studies have demonstrated the safety of allogeneic NK cells [17–20] and showed promise in hematological and solid tumor malignancies [21, 22].

Despite several advantages, NK cells do not infiltrate solid tumors as efficiently as T cells, and their lack of antigen specificity prevents them from being specifically guided into the tumor. Introducing a TCR on NK cells represents a potential solution to these limitations, as the TCR specifically directs the product to a given target peptide presented by HLA complexes on tumor cells.

The TCR-NK concept emerged from the desire to combine the clinically proven solid cancer targeting capability of TCRs with the broad cancer detection, high potency, and safety profile inherent in NK cells [23–25]. This novel approach has the potential to target solid cancers, and detect many, rather than one or very few features, on heterogenous tumors, addressing a major challenge limiting the durability of current solid cancer treatments. To our knowledge, we show for the first time the expression of a functionally enhanced-engineered TCR, together with CD3, and the CD8 co-receptor, to form a functional receptor complex on NK cells. TCR-NK cells can bind to tumor cells and become activated through both innate NK receptors and the TCR.

This novel TCR-NK strategy offers multiple advantages: a potentially favorable safety profile, made readily available off-the-shelf, acts as a rapid serial killer, and can eliminate tumors that escape recognition by T cells.

The first product under development, named ZI-MA4-1a, targets a clinically validated tumor-associated protein, MAGE-A4. MAGE-A4 is part of the cancer testis antigen family, which comprises antigens not expressed in normal tissues (apart from the testis) though widely expressed across solid cancers. Based on The Cancer Genome Atlas Program (TCGA), MAGE-A4 is expressed in a significant proportion of multiple solid tumors including: synovial sarcoma (67%), non-small cell lung cancer (NSCLC) (55.9%), head and neck cancers (37.6%), myxoid/round cell liposarcoma (34%), esophageal cancer (33%), ovarian cancer (31.2%), urothelial/bladder cancer (30.9%), melanoma (21.6%), and stomach cancer (16.8%).

Treatments targeting MAGE-A4-derived peptides including the peptide target for the current study have been investigated in numerous clinical trials with positive clinical benefit across a variety of solid cancers. The first ever engineered TCR T cell therapy to be approved targets MAGE-A4 and is used for the treatment of synovial sarcoma. Objective clinical responses with this therapy and others targeting MAGE-A4 have also been observed in ovarian, head and neck, non-small cell lung cancers as well as others [5, 7].

In this article, we describe the preclinical characterization of a novel allogeneic cell therapy product: an NK cell expressing an engineered TCR (KVL-a) recognizing the HLA-A2 restricted MAGEA4_286–294 peptide KVLEHVVRV. KVL-a TCR has been engineered to increase the potency to tumor antigens. By combining a functionally enhanced TCR with the innate cancer detection and cytotoxicity of NK cells, we enhance the overall potency of the cell product, enabling efficient killing of heterogenous and diverse tumors, which often limit the effectiveness of current treatments.

Materials and methods

Cell lines and normal cells

Cancer cell lines used in this paper were purchased from ATCC or DMSZ and were cultured and maintained for a maximum of 10 passages or up to 10 weeks in their respective media following the manufacturer’s recommendations. A detailed list of the cancer cell lines used in this paper is provided in Supplementary Table S2. Normal cells and iCells from HLA-A*02:01 donors were purchased from ScienCell and PromoCell, and FujiFilm, respectively. Cell lines authentication and mycoplasma testing were performed at Eurofins using qPCR and DNA STR profiling, respectively. The HLA-A*02:01 status of cancer cell lines was confirmed by flow cytometry staining with anti-HLA-A2 BV510 antibody (BioLegend). Selected cell lines were engineered to express HLA-A*02:01, Nuclight Red, and Nuclight Green using lentiviral vectors (Eurofins and Sartorius Cat# 4625 and 4624, respectively). MAGE-A4 knockout clones were generated from NCI-H1703 cells by transfection with Human MAGE-A4 Multi-guide mod sgRNA (Synthego SO#10041284) and Cas9 Nuclease 2NLS (Synthego, M22969-01) following the Lonza SE Cell Line 4D-Nucleofector™ X Kit nucleofection protocol for nucleofection-program EN-130. After transfection, single cells were sorted into a 96-well plate using a FACSAria cell sorter (BD Biosciences) and subsequently expanded. Clones were verified for MAGE-A4 knockout by qPCR using Mage-A4 gene-specific primers as described below.

Blood products

Whole blood was collected through venipuncture from healthy volunteers who provided written informed consent at Zelluna Immunotherapy, after approval from the Norwegian ethical committee (REK #231235). After collection, peripheral blood mononuclear cells (PBMCs) were isolated through density gradient centrifugation using SepMate tubes (StemCell Technologies) and Lymphoprep (47421, ProteoGenix), cryopreserved in Cryostor CS10 (C2874-100ML, Merck Sigma Aldrich) and stored in liquid N₂ until use. Frozen leukapheresis were purchased from StemCell Technologies (200-0132) and AllCells (LP CR Solo 2.5-3.0B) and were stored at −150°C until use. All human material and associated information were handled and stored in accordance with approvals obtained from the relevant regional ethics committee (REC).

Generation of lentiviral vectors

HEK293T (CRL-3216, ATCC) cells were plated in DMEM (41965120, Thermo Fisher Scientific) supplemented with 5% FBS (10500064, Thermo Fisher Scientific) and 50 μg/ml gentamycin (L0012, Biowest) in 6-well plates previously coated with poly-L-lysine (P6282-5MG, Merck Sigma Aldrich). When completely adherent, HEK293T cells were transfected by adding a mix of X-tremeGene 9 (6365787001, Merck Sigma Aldrich), helper plasmids pALD-Rev-K, pALD-GagPol-K, and pALD-VSV-G-K (5032, 5034, 5036, Aldevron), and 1 μg of plasmid of interest [KVL-a, KVL-01-WT, or irrelevant TCR (IrrTCR), Genscript], or nuclease free H₂O as mock control. Cells were first incubated for 16–18 h at 37°C 5% CO₂ before the medium was changed to DMEM supplemented with 5% FBS and 50 μg/ml gentamycin. After another 48 h of incubation at 37°C 5% CO₂, the supernatant was filtered through a 0.45-µm syringe filter and stored at +4°C until use.

T cell production

Frozen PBMCs were thawed, and the T cell population within the PBMCs was expanded and stimulated by plating in anti-CD3 and anti-CD28 antibody (Thermo Fisher Scientific)-coated plates. After expansion, the cells were transduced with in-house-produced lentiviral vectors containing the KVL-01-WT TCR, an IrrTCR, a clinically approved MAGE-A4 TCR (ClinTCR) or the KVL-a TCR. Alternatively, T cells were transduced with commercially produced and purified lentiviral vectors (VectorBuilder) containing the KVL-a TCR. TCR-T cells were cultured in X-Vivo15 (02-060F, Lonza) supplemented with 10% serum replacement (A2596101, Thermo Fisher Scientific) and 50 IU/ml interleukin (IL)-2, 2 ng/ml IL-7, and 2 ng/ml IL-15 (AF-200-02, AF-200-07, AF-200-15, Peprotech) and used fresh for peptide titration and assessment of potency in ELISpot. Alternatively, transduced TCR T cells were cryopreserved in Cryostor CS10 (C2874-100ML, Merck Sigma Aldrich) and stored in liquid N₂ until use. T cells were thawed and cultured for 24 h in X-Vivo15 supplemented with 10% serum replacement, IL-2, IL-7, and IL-15 before use in impedance killing assay.

TCR-NK production

Frozen leukapheresis were thawed using a Plasmatherm (Barkey, Azenta Life Sciences) and NK cells were enriched using a triple depletion strategy with MACS beads (Miltenyi Biotec). The cells were expanded in CTS medium with supplement (A5019001, GIBCO Thermo Fisher Scientific), containing 5% human AB serum (535-HI-GI, Grifols), IL-2, and IL-15 (170-076-146, 170-076-114, Miltenyi Biotec) and transduced with two separate third-generation lentiviral vectors (VectorBuilder) containing: (i) all the CD3 subunits and (ii) KVL-a, KVL-01-WT, or an IrrTCR, together with the CD8α chain (as co-receptor). The cells were expanded, and the transduced population was positively selected using CD3 Microbeads (130-050-101, Miltenyi Biotec). The selected cells were pulsed with feeder cells (produced in house), harvested on Day 21–28, cryopreserved in Cryostor CS10 (C2874-100ML, Merck Sigma Aldrich) with 2.5% HSA (05-730-1E, Sartorius), or in Plasmalyte (AFE0324D, Baxter)/HSA/Cryostor CS10 (3:1:4 ratio) and stored in liquid N₂ until use. The TCR-NK cells were thawed and cultured in CTS medium with supplement (A5019001, GIBCO Thermo Fisher Scientific), containing 5% human AB serum (535-HI-GI, Grifols), IL-2, and IL-15 (170-076-146, 170-076-114, Miltenyi Biotec) for 72 h before use in the functional assays. The culturing media was supplemented with varying concentrations of IL-2 and IL-15 depending on the stage of the process.

Peptide titration and potency assessment of TCR-T cells

The sensitivity of the TCR to peptides and the reactivity of the TCR-T cells to target cell lines were determined using an interferon gamma (IFN-γ) ELIspot assay. For sensitivity testing, T2 cells were pulsed with titrated amounts of the KVL or DYHC1 peptides (Genscript), using 12 different concentrations ranging from 1 pM to 1 µM, in addition to an irrelevant peptide at 1 µM. The assays were performed in 96-well plates (MSIPS4510, Millipore) coated with anti-IFN-γ (clone 1 D1K, Mabtech). Pulsed T2 cells, or target cell lines, were co-cultured with effector T cells at an effector to target (E:T) ratio ranging from 1:5 to 1:10 (predetermined for each donor based on previously observed reactivity) for 24 h before the assay was developed using a Human IFN-γ kit and substrate (3420-2A and 3650-10, Mabtech). The IFNγ release was analyzed on the ELISpot plate reader ImmunoSpot S6 Versa (CTL), and the Non fit linear regression was calculated using GraphPad Prism.

Impedance-based killing assay

Impedance-based killing was assessed using the Maestro TrayZ platform from Axion biosystems. Target cancer cells were grown for 24 h in 96-well CytoView-Z plates (AXI Z96-IMP-96B-5, Axion) precoated with 0.6 µg/cm² fibronectin (5050-1MG, Cell Systems) per well. The impedance (ohms, Ω) was measured for 24 h before effector TCR-T or TCR-NK cells were added at E:T ratios of 1:1. The decrease in resistance was used to quantify the dying target cells. Data analysis was performed using the AxIS Z software according to the manufacturer’s instructions.

Flow-based killing assay

NCI-H1703 and A-431 target cells labeled with CellTrace Violet reagent (C34577, Thermo Fisher Scientific) at 1 µM working concentration for 20 min at 37°C and cultured overnight before addition of ZI-MA4-1a, KVL-a T cells, or ClinTCR-T cells with an E:T ratio of 4:1. Effector and targets were co-cultured for 4 or 24 h before all cells were harvested using StemPro™ Accutase™ Cell Dissociation Reagent (A1110501, Thermo Fisher Scientific) and stained with 10 ng/sample of 7-aminoactinomycin D (A1310, Thermo Fisher Scientific). Target cells cultured alone were used as control to determine the spontaneous cell death. Samples were acquired in triplicates using the CytoFLEX flow cytometer and CytExpert software (Beckman Coulter, USA). Data were analyzed with FlowJo software 10.10.0 (Tree Star).

Spheroid assay

Spheroids were produced by culturing a 1:1 mix of A-431_NLG and A-431_A2_NLR in ultra-low attachment U-bottom 96-well plates (444-1020, Corning) in RPMI (21875091, Thermo Fisher Scientific) supplemented with 10% FBS (10500064, Thermo Fisher Scientific), 50 μg/ml gentamycin (L0012, Biowest), and 2.5% Corning Matrigel Basement Membrane Matrix (CLS354234-1EA, Merck Sigma Aldrich). The spheroids were cultured for 72 h before TCR-NK were added at a 10:1 E:T ratio. The growth of the spheroids was monitored by imaging the plates in the IncuCyte instrument (Sartorius) using the spheroid module, 4× objective, 300 ms green fluorescence exposure, and 400 ms red fluorescence exposure. Images were taken every 6 h for 96 h. Images were analyzed using Incucyte software (Sartorius).

Real-time quantitative PCR MAGE-A4

Total RNA was isolated from cell pellets using Homogenizer tubes (Invitrogen, Thermo Fisher Scientific) and the PureLink RNA Mini kit (Life Technologies), including the On-column PureLink DNase treatment protocol using the PureLink DNase set (Life Technologies) according to the manufacturer’s protocol. The concentration and the quality of RNA were measured on Nanodrop One (Thermo Fisher Scientific). RNA was stored at −80°C if not used directly for the generation of cDNA. 50–500 ng total RNA was used for the cDNA synthesis using SuperScript IV VILO Master Mix (11756050, Thermo Fisher Scientific) according to the manufacturer’s protocol. The thermal protocol was performed using the SimpliAmp Thermal Cycler (Thermo Fisher Scientific). cDNA products were kept on ice until qPCR analysis or frozen at −20°C for longer storage. For the qPCR, MAGE-A4-specific primers (5′ CAGGAAACTGCTCACCCAAG-3′ and 5′ GGTAGGCAATGCGAACTCTT-3′) and a MAGE-A4-specific VIC-labeled TaqMan probe (5′ ACCGGCAGGTACCCGGCAGT-3′) were used in multiplex reactions with a FAM-labeled housekeeping gene control, the TaqMan gene expression assay, GAPDH, Hs99999905 m1 (all primers and probes by Applied Biosystems). The assays were carried out using 1× TaqMan Fast Advanced Master Mix for qPCR (4444557, Applied Biosystems), 500 nM gene-specific primers, 250 nM TaqMan probe, and 1× GAPDH TaqMan gene expression assay. All reactions were carried out in triplicates, including samples, no template control, and a 7-point standard curve with five-fold dilutions of a fragmented MAGE-A4 gene. No reverse transcriptase controls were included as single reactions. The qPCR was performed using a QuantStudio3 Real-Time PCR System (Applied Biosystems, Thermo Fisher Scientific). Data were analyzed using the QuantStudio Design & Analysis software v1.5.3 (Applied Biosystems). A relative standard curve method, using GAPDH for normalization, was used to quantify the expression of MAGE-A4.

Degranulation and IFNg production assay

The degranulation capacity of TCR-NK cells upon recognition of cancer target cells was assessed by 5 h co-culture of TCR-NK cells with cancer cell lines at an effector to target (E:T) ratio 1:1 in the presence of anti-human CD107a-PE-Cy7 antibody. GolgiStop™ Protein Transport Inhibitor (554724, BD Biosciences) was added after the first hour of co-culture at final concentration of 0.2 µl per sample. TCR-NK cells alone were used as negative control, and TCR-NK cells stimulated with eBioscience™ Cell Stimulation Cocktail (00-4970-93, Thermo Fisher Scientific) were used as positive control. Each co-culture condition was set up in duplicate. After the co-culture, effector cells were stained with Fixable Viability Stain 780 (565388, BD Biosciences) followed by labeling with anti-human CD56-APC, and anti-human CD3-BV421 or anti-human CD3-BV785 antibody. Cells were washed with DPBS containing 2% FBS, fixed by incubation with 4% Paraformaldehyde (22023, Biotium), and stored at 4°C. The next day the samples were analyzed by flow cytometry using the CytoFLEX (Beckman Coulter) or the Northern Lights (Cytek) flow cytometers. Data were analyzed using FlowJo software (BD). The reactivity of ZI-MA4-1a toward normal cells and iCells was assessed by degranulation assay as described above for cancer cell lines. Co-culture of effector cells with NCI-H1703 cancer cell line was included as additional control for potency of effector cells. The reactivity of ZI-MA4-1a toward PBMCs was assessed by degranulation assay as described above. Before co-culture, PBMCs were labeled with CellTrace Violet reagent (Thermo Fisher Scientific, C34577) at 1 µM working concentration for 20 min at 37°C, then the cells were washed with complete RPMI medium. PBMCs pulsed with 1 µM KVL peptide were used as control.

Production of soluble effector molecules by TCR-NK cells

To assess TCR-NK cells’ capacity of secreting selected soluble effector molecules, TCR-NK cells were co-cultured with cancer cell lines for 6 h at E:T ratio 2:1. The co-culture supernatant was then collected in new 96-U-bottom plates and frozen at −80°C. The concentration of effector molecules was measured in thawed supernatants by LEGENDplex™ assay (BioLegend) using: LEGENDplex HU Proinflammatory Chemokine Panel 1, LEGENDplex Hu CD8/NK Panel V02 (740985 and 741187, BioLegend) and a custom kit (BioLegend) containing the following capture beads: IFNγ, TNFα, soluble Fas-Ligand, GM-CSF, granzyme A, granzyme B, CCL1, CCL3, and CCL4. The assay was performed according to the vendor’s instructions. Each supernatant was assayed in duplicate. Samples were analyzed using CytoFLEX (Beckman Coulter) flow cytometer. Data were analyzed using LEGENDplex™ Data Analysis Software Suite.

Cytotoxicity toward heterogenous cancer cells

To assess TCR-NK cell’s capacity of killing the cancer target cells in a heterogenous culture, A-431_A2_NLR and A-431_NLG target cells were mixed at 1:1 ratio and seeded at a density of 10 000 cells per well in a 96-well Incucyte^®ImageLock plates (Sartorius, Cat. No. BA-04856). As controls for data analysis, 10 000 A-431_A2_Nuclight Red (NLR) cells or A-431_Nuclight Green (NLG) cells per well alone were seeded. Target cells were incubated in 5% CO₂ 37°C incubator overnight. KVL-a-NK cells, IrrTCR-NK cells, or ClinTCR-T cells were added to the target cells at an E:T ratio of 1:1. Samples were imaged in the IncuCyte instrument (Sartorius) using Image Lock scan type, 10× objective, 300 ms green fluorescence exposure, and 400 ms red fluorescence exposure. Three images per well were taken every 10 min for 24 h. Images were analyzed using Incucyte software (Sartorius).

Serial killing assay

A-431_NLG and A-431_A2_NLR were cultured for at least two passages prior to the assay setup. The effectors were thawed 72 h prior to the co-culture. Twenty-four hours prior to the addition of the effectors, the targets were harvested, plated in completed RPMI (GIBCO, 10% FBS + RPMI), and placed in the incubator overnight. On the day of the assay, the effectors were counted, resuspended in complete RPMI, and added to the target cells at a 4:1 E:T ratio. On the same day, a new plate of targets was set up. The plates were placed in the incubator containing the Incucyte (Sartorius AG), and phase contrast, green, and red fluorescent images were captured every 2 h continuously. Twenty-four hours later, excess media was removed from the co-culture plate and then the effectors were transferred onto the new plate of targets plated the day before and a new plate of targets was set up. This was repeated every 24 h until the assay was terminated. Analysis was carried out by measuring the red and green object count on the Incucyte software and normalizing to the 0-h time point of each round of killing and plotted on GraphPad.

Flow cytometry

A detailed list of the antibodies used in this paper is available in Supplementary Table S1. To check the expression of CD3 and TCR in ZI-MA4-1a and untransduced NK cells, samples were stained 72 h after thawing with the Fixable Viability Dye eFluor™ 506 (FVD506) (65-0866-14, Thermo Fisher Scientific, USA) for 10 min at RT, followed by the staining with the following monoclonal antibodies: BV785 Anti-CD3, APC-Anti-CD56, PE-anti-TCR vβ13.2 for 20 min at 4°C in the dark; cells were washed twice with DPBS + 2% FBS (105000064, Gibco/Thermo Fisher Scientific, USA) and resuspended in the same buffer for analysis on the same day. The cells were acquired at CytoFLEX flow cytometer using CytExpert software (Beckman Coulter, USA). For the extended NK phenotype, the cells were stained 48 h after thawing with the ViaDye™ Red Fixable Viability Dye (SKU R7-60008, Cytek Biosciences, USA) for 15 min at RT; cells were washed with DPBS + 2% FBS (Gibco/Thermo Fisher Scientific, USA) followed by incubation with Human BD Fc Block™ (Clone Fc1, 564220, BD Biosciences, USA) for 5 min at RT. Cell surface staining was performed with the following monoclonal antibodies: PE-Vio-615-Anti-NKG2A, BV786-Anti-NKG2C, Alexa Fluor-647-Anti-NKG2D, BV711-Anti-NKp30, BV480-Anti-NKp44, PE-Vio-770-Anti-NKp80, cFluorV450-Anti-CD16, cFluor R720-Anti-CD56, APC-Anti-CD57, BV605-Anti-TIGIT (CD155), BB515-DNAM-1 (CD226), PE-Anti-TRAIL (CD253) for 20 min at 4°C in the dark; cells were washed twice with DPBS + 2% FBS (Gibco/Thermo Fisher Scientific, USA) and resuspended in the same buffer for analysis on the same day. The cells were acquired at Northern Lights™ flow cytometer using SpectroFlo^® software (Cytek Biosciences, USA). Flow cytometric data were analyzed with FlowJo software 10.10.0 (Tree Star).

CFSE proliferation assay

TCR-NK cells were thawed and cultured for 72 h before labeling with 0.5 µM carboxyfluorescein succinimidyl ester-SE (CellTrace™ CFSE Cell Proliferation Kit for flow cytometry, Thermo Fisher Scientific) for 20 min at 37°C. The staining reaction was quenched by adding five times the volume of complete media and subsequent washing of the cells twice in media. CFSE-labeled TCR-NK were co-cultured with A-431 cells or A-431_A2 at an E:T ratio of 1:1 for 4 days. At Days 1 and 4, proliferation of TCR-NK cells was assessed by determining the CFSE dilution using a CytoFLEX flow cytometer and CytExpert software (Beckman Coulter, USA). Data were analyzed using the FlowJo software 10.10.0 (Tree Star).

Results

Engineered MAGE-A4 TCR delivers higher potency than wild-type TCR

The wild-type (WT) TCR (KVL-01-WT) was derived from antigen-specific T cells of an HLA-A*02:01-positive healthy donor that specifically recognized the MAGE-A4_286–294 peptide KVLEHVVRV (hereafter named KVL peptide). To increase the functionality of the WT TCR, it was engineered, using a variety of methods [26] by introducing mutations into the complementarity-determining regions (CDRs) of the TCR variable chains (data not shown). Point mutations in the HLA and peptide-binding regions of the TCR chains have been previously shown to have the ability to increase affinity of a TCR [27].

Screening of the mutated TCRs was conducted using in vitro assays to identify candidates with enhanced sensitivity and potency compared to the WT TCR, while ensuring no significant cross-reactivity had been introduced. A pool of 150 TCRs, each containing single or multiple point mutations in the CDR regions, were cloned into lentiviral vectors and transduced into primary human T cells. The sensitivity of the TCRs was evaluated by co-culturing TCR-transduced T cells with T2 target cells pulsed with increasing concentrations of the KVL peptide. Reactivity to naturally expressed antigen was evaluated using a panel of HLA-A*02:01+ MAGE-A4+ and MAGE-A4-cell lines (data not shown).

The screening process led to the selection of an engineered TCR, named KVL-a. At comparable expression levels and with comparable expression density (Fig. 1a), KVL-a showed increased sensitivity and potency compared to KVL-01-WT. Activation, measured by IFNγ release, was detected for KVL-a at target peptide concentrations ranging from 1 to 10 pM. In contrast, KVL-01-WT activation was observed at concentrations between 100 and 500 pM (Fig. 1b), confirming the superior sensitivity of KVL-a. Additionally, KVL-a produced more IFNγ than KVL-01-WT when co-cultured with the NCI-H1703 and C-33A cell lines, both of which naturally express MAGE-A4 and the relevant HLA-A allele (Fig. 1c, Supplementary Figure S1a and S1b). As expected, no significant IFNγ was detected in response to the MAGE-A4-negative cell line SW620. Interestingly, the shift in sensitivity of KVL-a compared to KVL-01-WT was more significant against the C-33A cell line compared to NCI-H1703. C-33A expresses low levels of HLA-A*02:01 (Supplementary Figure S1b), suggesting an enhanced benefit of KVL-a in the context of lower physiological levels of the peptide-HLA ligand.

Figure 1.

Comparison between KVL-01-WT and KVL-a in T cells. (a) Median fluorescence intensity (MFI) of TCR expression on KVL-a and KVL-01-WT transduced T cells. Each dot represents one batch of TCR-T cells. Statistical significance is calculated using the Wilcoxon matched-pairs signed rank test. (b) IFNγ release of KVL-01-WT, KVL-a, and IrrTCR T cells measured in ELISpot, in response to co-culture with T2 cells pulsed with increasing concentration of target peptide. Datapoints represent the average of three technical replicates ± SD from one representative experiment. Connecting lines show the non-linear fit curve. (c) IFNγ release of KVL-01-WT, KVL-a, and IrrTCR T cells measured in ELISpot, in response to co-culture with cancer cell lines at an E:T ratio of 1:5. Each box represents the average of 4–5 independent experiments ± SD. (d) Percentage of cytolysis after 24 h of co-culture of KVL-01-WT, KVL-a, and IrrTCR T cells with cancer cell lines at an E:T ratio of 1:1. Percentage of cytolysis is measured in an impedance assay. Each datapoint represents the average of three technical replicates from one experiment; bars represent the average of three independent experiments ± SD.

To expand the pool of cancer cell lines used to test KVL-a, two cell lines expressing high levels of MAGE-A4 protein were added to the panel of targets: A431, derived from an epidermoid carcinoma, and KYSE-150, derived from esophageal carcinoma. Both cell lines were engineered to express HLA-A*02:01 and were used alongside their WT counterpart to analyze and dissect the TCR specificity. An overview of the cell lines used in this paper is provided in Supplementary Table S2, and the MAGE-A4 and HLA-A*02:01 status of the cell lines are illustrated in Supplementary Figure S1a and S1b.

The superior potency of KVL-a was confirmed in cytotoxicity assays against this panel of cell lines, all expressing the target protein MAGE-A4 and the HLA-A*02:01 molecule. In a 24-h impedance-based killing assay, KVL-a T cells showed a consistent advantage over KVL-01-WT T cells in killing three of the four cell lines tested. No significant difference between the two TCRs was observed against NCI-H1703, which were efficiently eradicated as they express high levels of both MAGE-A4 and HLA-A*02:01 and therefore, in the context of T cells, WT TCR seemed to perform similarly to the engineered TCR (Fig. 1d). These data confirm that the activation advantage of KVL-a over KVL-01-WT translates into improved cytotoxicity.

TCR introduction augments NK function without altering core phenotype

To combine the specificity of a MAGE-A4 targeting TCR with the innate anti-tumor features of NK cells, peripheral blood NK cells were isolated from healthy donor leukapheresis and transduced to express a functional TCR-CD3 complex with a CD8 coreceptor using lentiviral vectors. After transduction, the TCR-NK cells were expanded ex vivo using a feeder-based approach before characterization and use in functional assays. The resulting cellular product was called ZI-MA4-1a. At the end of the process, ZI-MA4-1a comprised a pure population of CD56+ cells with variable levels of TCR and CD3 expression (Fig. 2a and b).

Figure 2.

Engineered TCR enhances NK function and maintains cell phenotype a) representative flow cytometry dot plots showing the percentage of cells expressing CD56 (left dot plot) in ZI-MA4-1a, as well as CD3 and TCR Vb13.2 within the CD56+ population in ZI-MA4-1a (central dot plot) and in untransduced NK cells (right dot plot). Cells are gated on FSC/SSC > Singlets > Live cells, or on FSC/SSC > Singlets > Live cells > CD56+. (b) Frequency of ZI-MA4-1a cells expressing CD56, and CD56/CD3/TCR Vb13.2 in n = 15 ZI-MA4-1a batches. Each dot represents the percentage expression of one individual batch. (c) Heatmap showing the percentage of cells positive for different NK cell markers within the CD56+ population in ZI-MA4-1a from 12 batches. Cells are gated on FSC/SSC > Singlets > Live cells > CD56+. (d) Percentage of degranulating cells expressed as % of CD56+CD107a+ cells in ZI-MA4-1a and untransduced NK cells against K562 at an E:T ratio of 1:1. Bar graphs represent the average of two technical replicates ± SD. (e) Cytotoxicity of ZI-MA4-1a and KVL-01-WT-NK against the NCI-H1703 cell line evaluated using an impedance-based assay. Target cells were cultured for 24 h before the addition of effectors cells at E:T ratios of 1:2 and 1:5 for an additional 24 h. Normalized resistance indicates changes in impedance, indicative of target cell killing and detachment.

The NK cells were phenotypically characterized after expansion. Although mostly consistent, the phenotype of the CD56+ cells shows, as expected, some degree of variability attributed to the starting leukapheresis material (Fig. 2c). Overall, we observe high expression of the activating receptors NKG2D, DNAM-1, NKp30, and NKp44. Other activating receptors, such as NKp80 and NKG2C, were more variably expressed. Importantly, most of ZI-MA4-1a batches expressed intermediate to high levels of CD16 and TRAIL, suggesting that these cells can mediate antibody-dependent and death-receptor-mediated cytotoxicity in addition to activating receptor-mediated cytotoxicity. Furthermore, ZI-MA4-1a expresses high levels of NKG2A and low levels of CD57, suggesting that the cells are not terminally mature and retain their proliferation potential. In addition, we looked at the distribution of the activating and inhibitory receptors on the NKG2A-positive population compared to the NKG2A-negative. We observed that in 12 batches of ZI-MA4-1a, the NKG2A+ population expresses significantly higher levels of NKp30, NKp44, NKp80, and TRAIL compared to the NKG2A-cells (Supplementary Figure S1d).

The ability of the cells to degranulate in response to K562 was not affected by the introduction of the TCR as shown by similar levels of CD107a expression in ZI-MA4-1a compared to untransduced CD56+ cells (Fig. 2d). Overall, the process produces a pure CD56+ product with a functional and stable expression of CD3 and TCR. The resulting TCR-NK cells maintain the phenotypic characteristics of PB-NK cells and are functionally active. Differences in the transduction rates and phenotypes of ZI-MA4-1a were attributed to the variability of the research grade leukapheresis material and are a result of the functional and phenotypical heterogeneity of human NK cells, also observed by others [28].

Finally, to confirm that the increased potency of the KVL-a TCR over the KVL-01-WT observed in T cells also translate to an advantage in NK cytotoxicity, we compared ZI-MA4-1a to KVL-01-WT-NK cells in an impedance-based killing assay against the NCI-H1703 cell line. At an E:T ratio of 1:5, ZI-MA4-1a shows a significant advantage over the KVL-01-WT-NK. The difference between ZI-MA4-1a and KVL-01-WT-NK is reduced at E:T ratios of 1:2, but importantly all target cells are eradicated by ZI-MA4-1a, while KVL-01-WT-NK fail to eliminate the NCI-H1703 and the cells start growing again at later time points (Fig. 2e). These data show, for the first time, the potency benefit of an engineered TCR in NK cells compared to a WT TCR.

KVL-a TCR augments NK cell effector functions and expansion

The introduction of KVL-a TCR into NK cells enhances their innate activity by adding antigen specificity. To test the advantages brought by the TCR, ZI-MA4-1a was tested in a degranulation assay against a panel of cancer cell lines. In addition to the A-431/A-431_A2 and KYSE-150/KYSE-150_A2 pairs, a MAGE-A4 knock-out (KO) was generated to create an antigen-negative counterpart to the NCI-H1703 cell line. As a control to ZI-MA4-1a, NK cells were transduced with an irrelevant TCR with the same HLA restriction, but different peptide specificity (IrrTCR-NK). To distinguish the added effect of the TCR, the cells were analyzed for the percentage of CD107a+ cells within the CD56+CD3+ cells (CD56+TCR+), or the CD56+CD3− (CD56+TCR−) populations. CD3+ cells in ZI-MA4-1a degranulated more than CD3− cells in response to the MAGE-A4+/HLA-A*02:01+ targets A431_A2, KYSE-150_A2, and NCI-H1703 (Fig. 3a). In contrast, no differences in degranulation between CD3+ and CD3− populations were observed when ZI-MA4-1a was co-cultured with targets not expressing MAGE-A4 or the HLA-A*02:01 allele, or when using IrrTCR-NK as effectors. This demonstrates that the TCR+ cells have an advantage in degranulation and that this advantage is TCR-specific (Fig. 3a). Interestingly, a small trend for higher degranulation of the CD3+ cells compared to the CD3− was also observed for the IrrTCR-NK, especially against A431_A2 cells.

Figure 3.

KVL-a TCR strengthens NK activity and promotes proliferation. (a) Percentage of degranulating cells expressed as % of CD56+CD107a+ cells in CD3+ or CD3− populations of the ZI-MA4-1a and IrrTCR-NK cells against different cancer cell lines at an E:T ratio of 1:1. Each datapoint represents the average of two technical replicates in TCR-NK cells from one batch. Statistical significance is calculated with Mann–Whitney test. (b) Cytotoxicity of ZI-MA4-1a against the KYSE-150, KYSE-150_A2, A-431, and A-431_A2 cell lines evaluated using an impedance-based assay. Target cells were cultured for 24 h before the addition of effectors cells at E:T ratio of 1:1 for an additional 50 h. Normalized resistance indicates changes in impedance, indicative of target cell killing and detachment. (c) Serial killing ability of ZI-MA4-1a and IrrTCR-NK cells. Effector cells were co-cultured with A-431_A2_NLR cancer cell line at an E:T ratio of 4:1 during several rounds of 24 h. Every 24 h, the TCR-NK cells were transferred to a new plate containing new target cells. Target cell growth was measured with Incucyte by imaging and quantifying the red fluorescent signal. (d) Proliferation of ZI-MA4-1a and IrrTCR-NK cells measured with a CFSE assay. The bar graph represents the difference in CFSE median fluorescence intensity (MFI) between ZI-MA4-1a and IrrTCR-NK co-cultured with A-431 and A431-A2. Each datapoint represents the average of two technical replicates from one batch, bars represent the average of three batches from three independent experiments. Statistical significance is calculated with t-test. Histograms show the CFSE staining from one representative batch. (e) Cytokine and chemokine secretions of ZI-MA4-1a are measured after 6 h of co-culture with A-431 or A-431_A2 cells. The concentration of the analytes was measured in flow cytometry using the LEGENDplex assay. Each datapoint represents the average of three technical replicates from one ZI-MA4-1a batch. Statistical significance is calculated with Mann–Whitney test. ns: not significant. P value below .05 was considered significant.

To test that the antigen-specific degranulation translates to increased cytotoxicity, ZI-MA4-1a was tested in an impedance-based killing assay. Both ZI-MA4-1a and IrrTCR-NK were able to kill the KYSE-150 and KYSE-150_A2 targets through innate activation, but the added antigen specificity of ZI-MA4-1a led to a significantly higher degree of cytotoxicity as shown by more substantial elimination of the KYSE-150_A2 cells (Fig. 3b). The cytotoxic advantage brought by KVL-a TCR was particularly evident toward the NK-resistant A-431 target cells. In this setting, the contribution of KVL-a was essential in mediating the complete eradication of A-431_A2 cells (Fig. 3b).

NK cells are known for their ability to perform serial killing. We hypothesized that the addition of a TCR could expand the serial killing capacity of the NK cells by increasing the numbers of killing rounds. ZI-MA4-1a and IrrTCR-NK cells from two different donors were co-cultured with target cells that were renewed every 24 h for a maximum of 10 rounds. We observed more enduring killing capacity of ZI-MA4-1a compared to IrrTCR-NK in both donors tested (Fig. 3c and Supplementary Figure S2a). The degree and the number of rounds of serial killing was donor-dependent, with a potential for ZI-MA4-1a to reach seven rounds of killing, compared to four rounds for IrrTCR-NK (Fig. 3c). Interestingly, the innate ability of IrrTCR-NK to kill target cells was enhanced in the third round. This enhancement coincided with an observed increase in the number of TCR-NK cells in the culture for both ZI-MA4-1a and IrrTCR-NK (data not shown), and consequently a skewed effector-to-target ratio. This shift likely explains the higher cytotoxicity observed.

To measure the impact of the TCR on proliferation, an assay was designed to follow the expansion of ZI-MA4-1a and IrrTCR-NK during co-culture with antigen-positive and antigen-negative cells. Cell growth, measured by flow cytometry, revealed a significant increase in proliferation when ZI-MA4-1a encountered their specific target, A-431_A2 cells. This effect was not observed for IrrTCR-NK demonstrating that it is dependent on the expression of MAGE-A4 on the targets (Fig. 3d). In addition, neither ZI-MA4-1a nor IrrTCR-NK proliferated in absence of target cells (data not shown).

Next, cytokine production by ZI-MA4-1a upon encountering target cells was assessed. ZI-MA4-1a produced most of the effector molecules characteristic of NK cells. The levels of the IFNγ, TNFα, sFasL, GrzB, and the chemokines CCL3 and CCL4 were significantly higher after co-culture with A431_A2 cells compared to A431. Interestingly, IFNγ and TNFα were detected at very low levels or almost immeasurable in the absence of TCR engagement, suggesting that the TCR can turn on cytokine release in the context of almost insignificant levels otherwise. Slightly increased levels of GM-CSF, GrzA, and CCL1 were also detected, although those did not reach statistical significance (Fig. 3e).

Collectively, these data demonstrate that the addition of an engineered TCR/CD3 complex and CD8 co-receptor on NK cells enhances cytotoxicity, serial killing, proliferation, and cytokine production of TCR-NKs, all of which may serve to strengthen the potential of this modality in the treatment of solid cancers.

KVL-a TCR introduction maintains the safety of NK cells in the context of normal tissues

Allogenic NK cells have been widely used in the clinic and are considered to have a favorable safety profile as they are not known to induce GvHD or CRS [29]. However, it was important to assess whether the introduction of a TCR has the potential to alter the safety profile of NK cells by evaluating on-target and off-target cross-reactivity potential against normal tissues. Initially, the safety profile of the KVL-a TCR was extensively tested in T cells, both from a molecular and functional perspective (data not shown). To confirm that the safe profile of KVL-a translates to safe TCR-NK cells, ZI-MA4-1a was tested in degranulation assays against MAGE-A4-negative normal cells isolated from HLA-A*02:01-positive healthy donors. The panel of normal cells tested comprised different cell types derived from critical organ systems. The CD3+ cells of ZI-MA4-1a degranulated at the same levels as the CD3− population when co-cultured with all the normal cells tested (Fig. 4a and data not shown), showing no TCR-mediated reactivity toward the normal cells.

Figure 4.

ZI-MA4-1a does not react to normal tissues. Percentage of degranulating cells expressed as % of CD107a+CD56+ cells in CD3+ or CD3− populations of the ZI-MA4-1a after co-culture at an E:T ratio of 1:1 with (a) normal cells (b) iCells, and (c) PBMC. Each datapoint represents the average of two technical replicates of TCR-NK cells from one batch. Statistical significance is calculated with Mann–Whitney test. ns: not significant. P value below .05 was considered significant.

To further investigate the safety profile of ZI-MA4-1a, its reactivity toward induced pluripotent stem cells (iPSC)-derived astrocytes, endothelial cells, cardiomyocytes, and hepatocytes (iCells) was assessed. iPSC-derived normal cells express a more physiologically relevant proteome compared to fully differentiated normal cells, and they have been shown to uncover potential cross-reactivities that have not been identified using only cultured normal cells [30]. CD3+ ZI-MA4-1a did not degranulate above the levels of the CD3− population when co-cultured with iPSC-derived cells (Fig. 4b), further supporting the absence of cross-reactivity of ZI-MA4-1a. To ensure that both normal cells and iPSC-derived cells expressed sufficient levels of HLA-A*02:01 molecules to present processed peptides to KVL-a, the cells were pulsed with 1 µM of the KVL peptide. Once pulsed, all cells tested in the assay were able to elicit a response from ZI-MA4-1a, demonstrating their potential for effective peptide presentation on the HLA-A*02:01 molecules (data not shown).

Finally, to exclude reactivity toward hematopoietic cells, ZI-MA4-1a was tested against HLA-A*02:01-positive PBMCs from two different healthy donors. In this case as well, no reactivity of the CD3+ ZI-MA4-1a cells above the CD3− population was detected (Fig. 4c).

Taken together, these results demonstrate that the addition of KVL-a does not introduce unspecific reactivity of ZI-MA4-1a toward normal cells.

ZI-MA4-1a outperforms TCR-Tc

Tumor heterogeneity and tumor immune-escape mechanisms, including antigen and/or HLA loss, are one of the limiting factors for traditional TCR-T cell-based therapies. TCR-NK cells may overcome some of these limitations due to their dual mechanism of action represented by the TCR and the potential for multiple innate receptors to engage different cancer antigens, triggering both killing and cytokine secretion. To highlight the potential of ZI-MA4-1a to address tumor heterogeneity and to assess the ability of TCR-NK cells to kill 3D structures, we generated the NK cell killing resistant A-431 and A-431_A2 cell lines stably expressing a nuclear green (Nuclight Green, NLG), or red (Nuclight Red, NLR) dye. ZI-MA4-1a was cultured with spheroids comprising a 1:1 mix of A-431_NLG and A-431_A2_NLR, and its cytotoxicity was assessed in an Incucyte-based assay. ZI-MA4-1a killed all A-431_A2_NLR cells within the first 24 h of the assay and greatly reduced the number of A-431_NLG cells within the spheroid. IrrTCR-NK killed both targets at a lower rate compared to ZI-MA4-1a (Fig. 5a). Interestingly, within the heterogeneous tumor, ZI-MA4-1a displayed a better innate cytotoxicity against A-431_NLG compared to IrrTCR-NK cells further supporting the notion that TCR triggering enhances NK innate killing function. Tested in a serial killing assay, ZI-MA4-1a maintained the dual targeting potential for up to five rounds (Supplementary Figure S3a). Again, we observed that the innate killing of A-431_NLG by ZI-MA4-1a was enhanced in the heterogeneous tumor compared to the monoculture (Supplementary Figure S3a and S3b), continuing to suggest that the engagement of the TCR on ZI-MA4-1a not only increases the killing of the antigen-specific targets, but also enhances the innate killing of the TCR-NK cells in the context of resistant tumor cells. We compared the phenotype of ZI-MA4-1a cultured with A-431 or A-431_A2 at baseline and after one and three rounds of stimulation. During co-culture DNAM-1, NKp30, NKp44, TIGIT, and TRAIL were progressively downregulated in comparison to ZI-MA4-1a cells cultured alone. The downregulation of these receptors in the ZI-MA4-1a cells that were co-cultured with the A431_A2 was less pronounced or delayed (Supplementary Figure S3c), suggesting that the engagement of the TCR might reduce or delay the downregulation of the NK receptors, and potentially preserve the innate activity.

Figure 5.

ZI-MA4-1a outperforms TCR-T cells. (a) Kinetic of mixed spheroids killing by ZI-MA4-1a and IrrTCR-NK cells. The spheroids are composed of a 1:1 mix of A-431_A2_NLR and A-431_NLG cells. Cell killing is measured using Incucyte, based on changes in the fluorescence intensity of the total red and green objects. Images illustrate the growth of spheroids in one representative well, while quantification is displayed as graphs and represents the average of triplicate wells ± SD. (b) Cell death of single and mixed target cell cultures induced by ZI-MA4-1a, IrrTCR-NK, and ClinTCR-T cells monitored over 24 h using Incucyte. Images illustrate the cell killing in one representative well, while quantification is displayed as graphs and represents the average of triplicate wells ± SD. (c) Percentage of specific target cell death measured by 7-AAD staining in a flow-based killing assay. Effector and target cells were incubated at E:T ratio of 4:1 for 4 and 24 h.

The ability of ZI-MA4-1a to kill heterogeneous tumors was compared to TCR-T cells expressing a MAGE-A4-specific TCR currently used in the clinic (ClinTCR-T) in 2D mixed cultures (Fig. 5b). Both ZI-MA4-1a and ClinTCR-T killed A431_A2_NLR target cells, with ZI-MA4-1a being faster and more efficient. ClinTCR-T only killed the antigen-positive targets within the mixed population while ZI-MA4-1a also killed the A431_NLG target cells, leading to a significant reduction of the total mixed population after 24 h. This highlights the effectiveness of ZI-MA4-1a in a mixed (i.e. heterogenous) population of cancers which is limiting the effectiveness of ClinTCR-T. Finally, to ensure that the differences in killing capacities between ZI-MA4-1a and ClinTCR-T were not due to the different TCRs, the cytotoxicity of ZI-MA4-1a (KVL-a in NKs), KVL-a-Tc (KVL-a in T cells), and ClinTCR-T were compared. Consistent with previous results, ZI-MA4-1a demonstrated superior killing of two target cell lines compared to both KVL-a-T and ClinTCR-T after 4 and 24 h (Fig. 5c) showing that while the KVL-a and ClinTCR demonstrate comparable potency in T cells, the introduction of KVL-a into NK cells (ZI-MA4-1a) enhances the overall effectiveness of the product.

Discussion and conclusion

NK cells, with their innate ability to recognize and target transformed cells, are considered the first line of defense against tumors [31]. Several studies have shown that the presence of tumor-infiltrating NK cells correlates with overall survival in patients across various solid cancers [32]. NK cells can overcome some of the limitations of T cells. Firstly, NK cells have consistently shown a favorable safety profile in both hematological and solid tumor malignancies [19, 21, 33]. Unlike T cells, NK cells are not known to cause GvHD and can therefore be used in an allogeneic setting, overcoming the challenges of autologous cell manufacturing. Secondly, the cytolytic activity of NK cells can be triggered by a variety of stress ligands on cancers and by the down regulation or absence of human leukocyte antigen (HLA). This gives the NK cells the advantage of being activated by a diverse array of cancer ligands, rather than a single one. This makes them less susceptible to detection avoidance due to antigen loss or HLA downregulation. Upon activation, NK cells secrete cytokines and chemokines that can modulate the function of other innate and adaptive immune cells.

Despite their advantages, NK cells are known to have poor tumor infiltration, and limited persistence in vivo [34]. To address these limitations, we have introduced an engineered MAGE-A4 targeting TCR in NK cells. In this study, we present the preclinical characterization of a novel MAGE-A4-targeting TCR-NK cell product called ZI-MA4-1a.

MAGE-A4 is a member of the MAGE protein family of cancer/testis antigens [35]. MAGE-A4 is expressed in a number of solid tumors, including synovial sarcoma (SS), myxoid/round cell liposarcoma (MRCLS), non-small-cell lung cancer (NSCLC), head and neck squamous cell carcinoma (HNSCC), ovarian, urothelial, melanoma, and gastroesophageal cancers [36]. The expression of MAGE-A4 in normal tissues is highly restricted to the testis and placenta, which are considered immune-privileged sites [37, 38]. Therefore, based on the normal tissue distribution of these antigens, the target is deemed safe. Tumor antigens often derive from self-proteins, and consequently, high-affinity natural TCRs targeting these proteins are negatively selected during thymic maturation. This process results in a scarcity of mature circulating T cells capable of recognizing tumor antigens with high affinity.

The WT MAGE-A4 TCR (KVL-01-WT) was isolated from a healthy donor and initially showed moderate sensitivity to the target peptide. To increase its functionality, single-nucleotide mutations were introduced in the CDRs of the TCR.

After screening a pool of approximately 150 TCRs, the lead TCR candidate, KVL-a, was selected for its superior sensitivity and potency in T cells compared to KVL-01-WT, while retaining a favorable specificity profile.

The ultimate goal of engineering the KVL-01-WT TCR was to be able to increase functional activity without introducing off-target sensitivities. The quality of T cell responses is dictated by the combination of physical and biological parameters like TCR affinity, TCR and peptide-major histocompatibility complex (pMHC) density, co-stimulation, and environmental factors[26]. By introducing point mutations in the CDR regions, we achieved increased sensitivity and functionality without modifying the surface density of the TCR. As no other variables were introduced in the comparison between KVL-01-WT and KVL-a in T cells, the increased functionality is likely to be a consequence of increased affinity.

Once selected, KVL-a was introduced into NK cells along with the CD3 complex and the CD8 co-receptor. The resulting cellular product, ZI-MA4-1a, expressed a fully functional TCR as well as the canonical NK cell receptors.

Donor variability in cell therapies is well documented and can influence the functional and phenotypic characteristics of PB-NKs. We also observed such variability, however, as development has continued, the upscaled manufacturing process has been optimized to specifically address donor variability ensuring the selection of cells with the most optimal attributes.

ZI-MA4-1a was activated in a TCR-specific manner as demonstrated by higher degranulation in the CD3+/TCR+ compared to the CD3−/TCR− populations in response to target cells expressing MAGE-A4. It also exhibited higher cytotoxicity toward relevant target cells compared to TCR-NK cells expressing an irrelevant TCR. In addition to the benefits provided by the TCR, our product retained the innate activation potential of NK cells, as demonstrated by degranulation against K562 cells at levels comparable to untransduced NK cells.

In addition to the antigen specificity brought by the TCR, we also observed a consistent advantage in degranulation against antigen-negative targets for the CD3+ population compared to the CD3− of both ZI-MA4-1a and IrrTCR-NK, i.e. innate NK cell function is boosted when the TCR is present. This difference was notably evident in the NK-resistant A-431 cell line. The CD3ζ zeta molecule is important for NK cell activation, particularly for the antibody-dependent cellular cytotoxicity (ADCC) function, as it associates with the Fc receptor CD16 and provides downstream signaling [39–41]. CD3ζ also associates with the NK receptors NKp30 and NKp46, transmitting activation signal via phosphorylation [39]. We hypothesized that the transduction of CD3ζ alongside the TCR strengthens the function of these innate receptors as well as the TCR. A beneficial role for additional CD3ζ in TCR-NK cells is supported by our own observations and its exact function in the context of TCR-NK cells innate potential is currently under further investigation. In addition, a more detailed analysis of the phenotypical characteristics of the CD3+TCR+ compared to the CD3−TCR− cells within ZI-MA4-1a could suggest whether the introduction of the CD3 and TCR can alter the expression patterns of activating and inhibitory receptors and could therefore explain the activation advantage observed in the CD3+ population.

TCR-NK cell therapy offers several advantages compared to TCR-T cell therapies. Tumor-escape mechanisms, and in particular antigen and HLA loss, render TCR-T cell therapy susceptible to failure and potential disease relapse. TCR-NK dual targeting allows the simultaneous elimination of both antigen-positive and -negative targets through the TCR, and the innate mechanisms, respectively. This approach is beneficial in case of antigen loss and in the presence of a heterogeneous tumor. By combining innate cytotoxicity with TCR targeting, we demonstrate here that ZI-MA4-1a has an advantage over T cells expressing either KVL-a or a clinically proven TCR. Activation of NK cells is faster than T cells, and in heterogeneous cultures, ZI-MA4-1a can kill both antigen-positive and antigen-negative cells, whereas T cells only kill antigen-positive targets. We also anticipate a role for the TCR in guiding the NK cells to the tumor, similar to the homing role of a TCR in T cells [42], a feature that puts TCR-NK cell in an advantageous position compared to other NK cell-based therapies, including bispecifics like BiKE and TriKE.

Peripheral blood NK cells are characterized by a high functional heterogeneity [43] and several subsets have been described in the literature [44, 45]. The ability to perform serial killing is a characteristic of specific subclasses of NK cells [46]. We tested the serial killing capacity of ZI-MA4-1a and we found that the addition of KVL-a prolongs the functional life of the NK cells by adding a further 2–3 rounds of target killing compared to IrrTCR-NK. In a heterogeneous culture, ZI-MA4-1a cytotoxicity against antigen-positive targets is sustained for longer compared to the cytotoxicity against antigen-negative targets, highlighting again the advantage brought by the TCR. The sustained and prolonged cytotoxicity suggests that the functional exhaustion of ZI-MA4-1a is delayed compared to IrrTCR-NK. In particular, we observed that the progressive downregulation of NK cell markers on ZI-MA4-1a after co-culture with target cells is reduced or delayed when the TCR-NK cells are cultured with relevant target cells, suggesting a protective mechanism of the TCR engagement on the innate features of the NK cells. This characteristic of ZI-MA4-1a, together with the increased proliferation observed after encounter with antigen-positive targets, has an important significance for the functional longevity of NK cells. A longer persistence in the tumor microenvironment may also lead to prolonged recruitment and interactions with other immune cells and better overall tumor clearance. The potential for increased immune cell recruitment is supported by our observations. When activated by antigen-positive target cells, ZI-MA4-1a produces cytokines and chemokines and could therefore contribute to recruitment of additional immune cells and initiation of an immune response in the tumor microenvironment. CCL3 and CCL4, produced by NK cells, can interact with immature dendritic cells which in turn promote the recruitment of effector CD8+ T cells to the tumor microenvironment [47]. It remains to be assessed if the longer functional lifespan of ZI-MA4-1a observed in vitro translates to better persistence in vivo. However, as an off-the-shelf NK cell therapy with low toxicity, ZI-MA4-1a allows patients to be re-dosed, further prolonging presence of functional NK cells in vivo.

The wide array of receptors expressed by NK cells offers a pool of potential targets for combinational therapy. Activating receptors can be engaged and several checkpoints can be inhibited [48]. In addition, NK cells express CD16, which by binding to the Fc portion of antibodies can trigger ADCC. Phenotypic analysis of ZI-MA4-1a shows that the cells consistently express CD16, and high levels of classical activating and inhibitory NK receptors, opening the possibility of introducing combinational therapies to further tailor and strengthen its anti-tumor efficacy [49]. The high heterogeneity of human peripheral blood NK cells can influence the clinical outcomes in NK cell-based cell therapies [50]. In particular, the presence of the inhibitory receptor NKG2A might limit the efficacy of the cell product due to the widespread expression of HLA-E molecules among tumor types [51]. The in vitro activation and killing ability of ZI-MA4-1a do not seem to be influenced by the levels of NKG2A expression (data not shown). In our analysis, we show that the cells expressing NK activating receptors are more abundant in the NKG2A-positive population compared to the NKG2A-negative, suggesting that despite the potential for inhibition, the NKG2A-positive population has, in addition to the TCR, many innate activating receptors to counterbalance it. Monitoring the changes in phenotype and subset distribution of ZI-MA4-1a after infusion and possibly correlating it with clinical outcome could give more information on the specific role of different subsets in TCR-NK cell therapy. This information may enable further development of the platform. Decades of NK cell therapy have shown the safe profile of allogeneic NK cell [29]. Nevertheless, it is important to ensure that introducing a TCR does not lead to unwanted toxicity by thorough safety testing of the final product. To maximize genomic coverage, the reactivity of ZI-MA4-1a was tested against a panel of normal cells representing different critical organ systems and cells from the hematopoietic system. The results showed no significant TCR-mediated reactivity against normal cells, as well as against induced-pluripotent stem cell derived (iPSC) cells, supporting the lack of off-target reactivity and the safety of ZI-MA4-1a.

In conclusion, we describe here the preclinical characterization of ZI-MA4-1, a novel MAGE-A4 targeting TCR-NK cell product. ZI-MA4-1 combines the innate potential of NK cells with the specificity of an engineered TCR, which results in NK cells with antigen-specific targeting and increased innate function within heterogeneous tumors. These characteristics give our product the dual functionality required to tackle heterogeneous tumors. Based on the potency and safety profile, ZI-MA4-1 is advancing toward a clinical trial application for the treatment of patients with advanced solid tumors.

Author contributions

Margherita Boieri (Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Visualization, Writing—original draft, Writing—review & editing), Justyna Kmiecik (Conceptualization, Formal analysis, Investigation, Methodology, Validation, Writing—review & editing), Maja Sandve (Formal analysis, Investigation, Methodology, Validation), Zara Hannoun (Conceptualization, Formal analysis, Investigation, Methodology, Validation), Martha Haugstøyl (Formal analysis, Investigation, Methodology, Validation), Ines Cardoso (Conceptualization, Formal analysis, Investigation, Methodology, Supervision, Validation), Sarah Vollmers (Formal analysis, Investigation, Methodology, Validation), Anja Oldenburg (Formal analysis, Investigation, Methodology, Validation), Luz Mora-Velandia (Conceptualization, Formal analysis, Investigation, Methodology, Validation), Camilla Sletten (Formal analysis, Investigation, Methodology, Validation), Giulia Malachin (Formal analysis, Investigation, Methodology, Validation), Artur Cieslar-Pobuda (Formal analysis, Investigation, Methodology, Validation), Liliane Christ (Formal analysis, Investigation, Methodology, Validation), Pimthanya Wanichawan (Formal analysis, Investigation), Dennis Clement (Formal analysis, Investigation), Michelle Saetersmoen (Formal analysis, Investigation), Frida Haugen (Formal analysis, Investigation), Amanda Ruud (Formal analysis, Investigation), Julia Ino (Project administration), Pranav Oberoi (Supervision), Anders Holm (Funding acquisition, Resources, Writing—review & editing), Emilie Gauthy (Conceptualization, Project administration, Supervision, Writing—review & editing), Namir Hassan (Conceptualization, Funding acquisition, Supervision, Writing—review & editing), Sylvie Pollmann (Conceptualization, Project administration, Supervision, Writing—original draft, Writing—review & editing), and Luise Weigand (Conceptualization, Funding acquisition, Project administration, Supervision, Writing—original draft, Writing—review & editing)

Supplementary material

Supplementary material is available at Immunotherapy Advances online.

Funding

This work was financially supported by the Norwegian Research Council under the “Innovation Projects for the Industrial Sector” program (project number 309801).

Ethical approval

Whole blood was collected from healthy volunteers who provided written informed consent at Zelluna Immunotherapy, after approval from the Norwegian ethical committee (REK #231235)

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author. The MAGE-A4 TCR sequences used in this study are described in the patent application titled: “ANTI-MAGE-A4 T CELL RECEPTORS,” with reference number PCT/EP2024/052478.

Permission to reproduce

Graphical abstracted created in BioRender. Holm, A. (2025) https://BioRender.com/qh3tgs1