Article Text
PDF
Abstract
Background Inhibitory receptor T-cell Immunoreceptor with Ig and ITIM domains (TIGIT) expressed by Natural Killer (NK) and T cells regulates cancer immunity and has been touted as the next frontier in the development of cancer immunotherapeutics. Although early results of anti-TIGIT and its combinations with antiprogrammed death-ligand 1 were highly exciting, results from an interim analysis of phase III trials are disappointing. With mixed results, there is a need to understand the effects of therapeutic anti-TIGIT on the TIGIT+ immune cells to support its clinical use. Most of the TIGIT antibodies in development have an Fc-active domain, which binds to Fc receptors on effector cells. In mouse models, Fc-active anti-TIGIT induced superior immunity, while Fc receptor engagement was required for its efficacy. NK-cell depletion compromised the antitumor immunity of anti-TIGIT indicating the essential role of NK cells in the efficacy of anti-TIGIT. Since NK cells express TIGIT and Fc-receptor CD16, Fc-active anti-TIGIT may deplete NK cells via fratricide, which has not been studied.
Methods CRISPR-Cas9-based TIGIT knockout (KO) was performed in expanded NK cells. Phenotypic and transcriptomic properties of TIGIT KO and wild-type (WT) NK cells were compared with flow cytometry, CyTOF, and RNA sequencing. The effect of TIGIT KO on NK-cell cytotoxicity was determined by calcein-AM release and live cell imaging-based cytotoxicity assays. The metabolic properties of TIGIT KO and WT NK cells were compared with a Seahorse analyzer. The effect of the Fc-component of anti-TIGIT on NK-cell fratricide was determined by co-culturing WT and TIGIT KO NK cells with Fc-active and Fc-inactive anti-TIGIT.
Results TIGIT KO increased the cytotoxicity of NK cells against multiple cancer cell lines including spheroids. TIGIT KO NK cells upregulated mTOR complex 1 (mTORC1) signaling and had better metabolic fitness with an increased basal glycolytic rate when co-cultured with cancer cells compared with WT NK cells. Importantly, TIGIT KO prevented NK-cell fratricide when combined with Fc-active anti-TIGIT.
Conclusions TIGIT KO in ex vivo expanded NK cells increased their cytotoxicity and metabolic fitness and prevented NK-cell fratricide when combined with Fc-active anti-TIGIT antibodies. These fratricide-resistant TIGIT KO NK cells have therapeutic potential alone or in combination with Fc-active anti-TIGIT antibodies to enhance their efficacy.
- cell engineering
- immune checkpoint inhibitors
- immunotherapy, adoptive
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information. Raw RNA sequencing data are deposited in the NCBI SRA database under accession number PRJNA1049053
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Treatments targeting T-cell Immunoreceptor with Ig and ITIM domains (TIGIT) generated a lot of excitement based on preclinical and early clinical results, although interim results from phase III trials of Fc-active anti-TIGIT were so far disappointing.
Natural killer (NK) cells are effector immune cells with high expression of TIGIT, particularly when activated, and based on murine studies are critical for the efficacy of anti-TIGIT therapeutics.
WHAT THIS STUDY ADDS
Fc-active anti-TIGIT binds TIGIT on NK cells and results in NK cell fratricide via antibody-dependent cell cytotoxicity and NK cell depletion.
Knockout of TIGIT on NK cells renders them resistant to fratricide with Fc-active anti-TIGIT and improves their metabolic fitness and cytotoxicity.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Administration of Fc-active anti-TIGIT drug candidates under development may result in the depletion of activated NK cells in treated patients which can negatively impact treatment outcomes.
Patient NK cells should be monitored in ongoing trials to determine if this mechanism potentially contributes to the lack of response or its short duration.
Co-administration of TIGIT KO NK cells may offer a potential approach to improve treatment outcomes of Fc-active TIGIT therapeutics and combinations thereof.
Introduction
Therapeutic blockade of the inhibitory T-cell Immunoreceptor with Ig and ITIM domains (TIGIT) is a rapidly developing cancer treatment strategy, with several anti-TIGIT antibodies in clinical trials in combination with other checkpoint inhibitors, predominantly programmed death-ligand 1 (anti-PD-L1).1 2 TIGIT negatively regulates the cancer immunity cycle by preventing cytotoxicity and proinflammatory cytokine secretion by natural killer (NK) and cytotoxic T cells, while also enhancing suppressive activities of regulatory T (Treg) cells. TIGIT competes with the activating receptor DNAX Accessory Molecule-1 (DNAM-1) for ligands CD155 (Poliovirus receptor - PVR) and CD112 (nectin-2 or PVRL2) which are commonly expressed on cancer cells and antigen-presenting cells (APCs),3 with close to 100-fold higher binding affinity for PVR.4 In mouse models with TIGIT−/− mice or using anti-TIGIT, TIGIT blockade alone or in combination with other checkpoint inhibitors enhanced antitumor immunity, prevented tumor growth, and improved survival.1 5–7 Early clinical testing of anti-TIGIT therapies has shown success.1 8 9 In phase II, anti-TIGIT (tiragolumab, Genentech) in combination with anti-PD-L1 (atezolizumab) enhanced overall response rates and median progression-free survival of patients with non-small cell lung cancers (NSCLC) with PD-L1-positive tumors compared with anti-PD-L1 alone.10 11 However, based on interim results of phase III clinical trials, tiragolumab in combination with atezolizumab did not elicit similar benefits in patients with locally advanced unresectable or metastatic PD-L1-selected NSCLC or extensive-stage small cell lung cancer (ES-SCLC).12 13 This highlights the need to understand the mechanism of function/off-target effects of anti-TIGIT therapeutic candidates and to identify alternative therapeutic strategies including potential combinations to support its clinical use. While all therapeutic anti-TIGIT antibodies in development have high target affinity and prevent TIGIT from binding its ligands, their Fc region varies greatly to either engage (Fc-active) or not engage (Fc-inactive) the Fcγ receptors on effector cells.
Most of the therapeutic anti-TIGIT antibodies, including the above-mentioned tiragolumab, are Fc-active,14 with some further engineered to enhance Fc receptor-based antibody-dependent cellular cytotoxicity (ADCC).15 16 The selection of therapeutic candidates with the Fc-active design was based on multiple independent murine studies demonstrating antitumor efficacy to be superior to Fc-inactive versions of the same antibodies.17 18 While Fc-active anti-TIGIT induced superior antitumor immunity and preferential Treg cell depletion with FcγR engagement required for the efficacy of anti-TIGIT antibodies,19 antibody-driven depletion of other TIGIT-expressing cells could have potentially negative consequences. Among those, NK cells, critically important for the efficacy of immunotherapies,20 have one of the highest levels of TIGIT expression, particularly when activated and/or tumor infiltrating.21–24
NK cells play a major role in immune surveillance and selectively kill cancer cells in a major histocompatibility complex-independent manner.25–28 NK cells directly kill tumor cells and secrete cytokines, chemokines, and recruit dendritic cells and T cells to prime an adaptive immune response.29–31 ,32 NK cells are an important population for the efficacy of checkpoint blockade therapies and low numbers of NK cells correlate to low response in patients with myeloma.33 A study by Zhang et al highlights the NK-cell dependency of TIGIT blockade therapeutic efficacy. TIGIT treatment increased the frequency of tumor-infiltrating NK cells expressing tumor necrosis factor (TNF)α, interferon (IFN)γ, and CD107a in CT26 tumor-bearing mouse models.5 Treatment with anti-TIGIT still slowed tumor growth and reduced tumor mass even in mice lacking T cells, suggesting that the improved tumor response could be achieved in the absence of functional adaptive immune cells.5 Furthermore, in a B16 pulmonary metastasis model lacking NK cells, TIGIT blockade had no therapeutic effect, even in the presence of TIGIT+-CD8+ T cells, and led to more metastasis and CD8+ T-cell exhaustion.5 The therapeutic effect of anti-TIGIT, anti-PD-L1, or their combination also depended on NK cells where their absence resulted in fewer tumor-infiltrating CD8+ T cells expressing IFNγ or TNFα.5 However, NK cells express FcγRIII (CD16), which may bind with Fc-active anti-TIGIT bound to TIGIT+ NK cells and result in fratricide and NK-cell depletion in treated patients. Thus, adoptive transfer of expanded NK cells may enhance the antitumor activities of anti-TIGIT antibodies, particularly with the use of TIGIT-depleted NK cells. TIGIT depletion before adoptive transfer may increase NK-cell antitumor activities by preventing TIGIT inhibitory signaling and may prevent NK-cell fratricide when combined with Fc-active anti-TIGIT therapeutics.
This study compared the function of wild-type (WT) and TIGIT knockout (KO) NK cells and studied the impact of TIGIT expression on their viability and cytotoxicity with Fc-active versus inactive anti-TIGIT. This study demonstrated that TIGIT KO significantly increased the cytotoxicity of expanded NK cells against several cancer cell lines. TIGIT KO NK cells showed better metabolic fitness with higher basal glycolytic rate compared with control NK cells when challenged with cancer cells. More importantly, this study demonstrated that tiragolumab induced NK-cell fratricide, resulting in NK-cell depletion and decreased cytotoxicity while TIGIT KO prevented fratricide of NK cells when combined with tiragolumab. Altogether, our study demonstrated that TIGIT KO enhanced NK-cell antitumor immunity, and adoptive therapy with TIGIT KO NK cells may offer a potential solution to improve the efficacy of Fc-active anti-TIGIT therapeutics.
Methods
Cell culture
Peripheral blood mononuclear cells (PBMCs) were separated from buffy coats from deidentified healthy donors, purchased from blood banks (OneBlood, Orlando, Florida or American Red Cross, Columbus, Ohio, USA), by density gradient (Ficoll-Paque Plus solution; GE Healthcare, Chicago, Illinois, USA) and used as the NK-cell source. For PM21-NK cells, NK cells were expanded with PM21 particles after T-cell depletion from PBMCs as previously described34 with the addition of interleukin (IL)-12 (10 ng/mL)+IL-15 (100 ng/mL)+IL-18 (50 ng/mL) on day 0 of culture. For FC-NK cells, NK cells were isolated by negative selection using RosetteSep Human NK Cell Enrichment Cocktail (Stem Cell Technologies, 15065, Vancouver, British Columbia, Canada). Purified NK cells were stimulated with irradiated CSTX002 feeder cells as previously described35 and 50 IU of human recombinant IL-2 (Novartis). Ex vivo expanded NK (exNK)-cell count and viability >95% were confirmed before use in experimentation.
A549, NCI-H358, NCI-H1650, and NCI-H1975 cells (American Type Culture Collection (ATCC)) and NCI-H1299 were stably transduced with NucLight Red (NLR) Lentivirus (Sartorius) to generate NLR-positive cell lines. A549-NLR, NCI-H358-NLR, NCI-H1650-NLR, NCI-H1975-NLR, and NCI-H1299-NLR-positive cells were enriched with puromycin selection and sorted for uniform positive populations (BD FACS Aria II). CHLA90, CHLA10, RH30, and SJBM2 were obtained from the Children’s Oncology Group Cell Line and Xenograft Repository (COGCell). The CTX002 feeder cells (K562 genetically modified to express 4-1BBL and membrane-bound IL-21, referred to hereafter as FC) were generated as previously described.36 All cancer cells were maintained in Roswell Park Memorial Institute (RPMI) media with 10% fetal bovine serum, 2 mM GlutaMAX and 1% antibiotic/antimycotic in a tissue culture incubator in a humidified atmosphere at 37°C supplemented with 5% (vol/vol) CO2. Cell lines were tested for mycoplasma (E-Myco Plus Mycoplasma PCR Detection kit, Bulldog-Bio, Portsmouth, New Hampshire, USA) and authenticated via Human STR Profiling (serviced by ATCC).
CRISPR-mediated TIGIT gene deletion
TIGIT KO NK cells were generated by electroporation of Cas9 RNP into PM21-NK or FC-NK cells on day 7 of expansion, containing gRNA targeting exon 1 (5-ACCCTGATGGGACGTACACT) or exon 3 (5- GTTCACGGTCAGCGACTGGA) of the TIGIT gene, as previously described37 38 using a Lonza 4D-Nucleofector system with pulse EN-138 or MaxCyte ATx with program NK-5. Unless noted in the text, WT NK cells were electroporated with only Cas9 as a control.
Flow cytometry
NK cells or cancer cells were stained with preconjugated protein-specific or the corresponding isotype control antibodies. Stained samples were acquired by Cytoflex (Beckman Coulter, Brea, California, USA) or Northern Lights 2000 Full Spectrum (Cytek, Fremont, California, USA) flow cytometer. Acquired data were analyzed with FlowJo software (V.10.6.2). An example gating strategy for NK cells is shown in online supplemental figure 1. For flow cytometry analysis of target proteins, the antibodies that were used are listed in online supplemental table 1.
Supplemental material
Mass cytometry
Metal-conjugated antibodies for mass cytometry were obtained from Fluidigm, or unlabeled antibodies were obtained from BioLegend or R&D and conjugated with heavy metals using Maxpar-X8 labeling reagent kits (DVS Sciences) according to manufacturer’s instructions and titrated for determination of optimal concentration. The antibodies and their respective heavy metal labeling are shown in online supplemental table 2. ExNK cells were thawed 1 day before staining to allow for NK-cell recovery overnight. After overnight recovery, NK cells were co-cultured with or without SJGBM-2 cells as targets at a 1:1 ratio. After co-incubation for 4 h, 1.5×106 cells were stained with 2.5 μM cell ID cisplatin (Fluidigm, 201064) in serum-free RPMI for 1 min for identification of viable populations, followed by staining with metal-conjugated antibodies as described previously,39 with the following minor modification: staining was performed in 5% FBS/0.01% sodium azide buffer in phosphate-buffered saline. Data were acquired on the CyTOF instrument (DVS Sciences). Events were collected on a CyTOF 2 mass cytometer (Fluidigm) to generate Flow Cytometry Standard (FCS) files. Median metal intensity for each marker is reported.
Cytotoxicity assays
Calcein-AM release assay was used to evaluate FC-NK-cell cytotoxicity against CHLA90, CHLA10, RH30, or SJBM2 as previously described.40 Briefly, tumor cell targets were loaded with 2 μg/mL of calcein-AM for 30 min. Cells were washed and incubated with WT or TIGIT KO NK cells at multiple effector/target (E:T) ratios for 4 h. Fluorescence was measured using a BioTek Synergy 2 plate reader (excitation: 485 nm/emission: 530 nm) and cytotoxicity (%) was calculated using the following equation:
Kinetic live-cell imaging cytotoxicity assays were conducted on an IncuCyte Live-Cell Analysis System (Sartorius, Göttingen, Germany) as previously described.24 41 Briefly, cancer cells expressing NLR (A549-NLR, NCI-H358-NLR, NCI-H1650-NLR, NCI-H1975-NLR, NCI-H1299-NLR, or SK-N-AS-NLR were used as target cells and seeded in monolayers or in ultra-low attachment plates to generate three-dimensional (3D) spheroids. WT and TIGIT KO PM21-NK cells were co-cultured with cancer cell spheroids at multiple effector versus target ratios as indicated for 3 days for two-dimensional (2D) assays and 7 days for spheroid assays. Tumor cell growth was determined by measuring red object count per well (ROC) in 2D assays or total red object integrated intensity (ROII) (RCU×µm2/image) of 3D spheroids. Relative cell growth of the target cells alone or in the presence of NK cells was determined by normalizing ROC or ROII to the value at time 0 (ROCt/ROCt=0 or ROIIt/ROIIt=0) when NK cells were initially added to determine normalized ROC (nROC) or normalized ROII (nROII). Cytotoxicity (%) was then determined based on the following equations:
Concentration-dependent cytotoxicity was determined and non-linear regression (curve fit) used to calculate half-maximal effective concentration(EC50). Curve fits for cytotoxicity over time were used to determine half-killing time (t1/2). Data for one donor for cytoxicity against H1975 were excluded due to poor goodness-of-fit value.
Metabolic assay
WT and TIGIT KO PM21-NK cells were cultured alone or with IL-12 (10 ng/mL)/IL-15 (100 ng/mL)/IL-18 (50 ng/mL) or with K562-PVR+-GFPLuc cells (E:T=3:1) overnight. The next day, it was confirmed by flow cytometry that no viable K562-PVR+-GFPLuc cells remained (online supplemental figure 2) and 1×105 NK cells were plated on poly D-lysine-coated 96-well plates. Glycolytic rate was determined using the Seahorse XF Glycolysis Stress Test Kit that measures extracellular acidification rate (Cat# 103020-100, Agilent Technologies, Santa Clara, California, USA) using a Seahorse XF96 analyzer (Agilent Technologies) and analyzed with Seahorse Wave Desktop Software (V.2.6.0).
Fratricide assay
WT or TIGIT KO PM21-NK cells were incubated alone, with isotype control, Fc-inactive UltraLEAF mouse IgG2b anti-hTIGIT antibody (BioLegend, California, USA) or Fc-active tiragolumab (SelleckChem, USA) for 24 h. After incubation, NK cells were stained with DRAQ7 and DiOC6 dye. Data were acquired with Northern Light 2000 Full Spectrum (Cytek) flow cytometer and analyzed with FlowJo software, gating on DRAQ7−DiOC6+CD3−CD56+ viable NK cells. To determine NK-cell viability, data were normalized to WT or TIGIT KO PM21-NK cells only for each group, respectively.
RNA-sequencing
Donor-matched expanded WT and TIGIT KO NK cells were purified with an NK-cell selection kit (EasySep CD56+ selection kit; StemCell Technologies). Total RNA was extracted (Direct-zol Microprep kit, Zymo Research) and RNA quality was evaluated by RIN value (TapeStation). Next, polyA selection, library preparation, and RNA sequencing (Genewiz, South Plainfield, New Jersey, USA) were performed. For analysis, the quality of raw RNA-sequencing (RNA-seq) data was determined (FastQC42) and adaptor and low-quality reads were trimmed (Trimmomatic43). Reads were then mapped to the hg38 human genome (HISAT244) and genome assembly and quantification of read counts performed (Stringtie44). Batch effects among samples were corrected with Combat seq.45 EdgeR46 was used to normalize gene expression and determine differentially expressed genes. The sequencing data were then uploaded to the Galaxy web platform and the public server usegalaxy.org was used to analyze the data.47 Ranked gene lists were generated by multiplying −log10(p value) with the sign of log2 fold-change (positive or negative) of individual genes obtained from edgeR analysis and used for preranked Gene Set Enrichment Analysis48 to determine enriched hallmark-gene sets in TIGIT KO NK cells compared with WT NK cells; ggplot2 was used to generate heatmaps of differentially expressed genes involved in enriched pathways.
Statistical analysis
All statistical analyses were done with GraphPad Prism V.9.3.1 unless mentioned in specific figure legends. Student’s t-tests (paired or unpaired) or two-way analysis of variance was used to determine the statistical significance of sample groups unless mentioned in figure legends. P<0.05 was considered statistically significant.
Results
TIGIT can be knocked out in ex vivo expanded NK cells
NK cells obtained from healthy human donors were expanded using clinically relevant methods (exNK cells) with PM21 particles (PM21-NK cells)34 or CSTX002 feeder cells (FC-NK cells).49 TIGIT KO exNK cells were generated by CRISPR-Cas9-based gene editing using TIGIT-specific guide RNA. Representative histograms of TIGIT KO PM21-NK cells (figure 1A) or FC-NK cells (figure 1B) and summary data from multiple donors showed that TIGIT KO efficiency was on average 88% in PM21-NK cells and 94% in FC-NK cells (figure 1C) (n=6–15 donors, each average of duplicate). PM21-NK-cell expansion was not affected by TIGIT KO. PM21-NK cells electroporated with Cas9 only (WT) or TIGIT-specific Cas9 RNP (KO) had a decline in expansion immediately after electroporation on day 7 of culture compared with donor-matched non-electroporated PM21-NK cells but growth rates recovered by day 12 of culture (figure 1D). There was no significant difference in fold-expansion between TIGIT KO and WT PM21-NK cells on day 14 of culture, 1444±1983-fold and 1292±1862-fold, respectively for PM21-NK cells (n=12 donors) and 1300±700-fold and 1260±750-fold for FC-NK cells (n=3 donors) (figure 1E). Similarly, there was no difference in NK-cell number on day 14 between donor and postelectroporation seeding density-matched WT and TIGIT KO FC-NK cells (1.16±8.3)×106 vs (1.02±7.1)×106 (n=6 donors) (figure 1F). These results demonstrate that TIGIT can be knocked out in ex vivo NK-cells propogated with clinically relevant methods with no deleterious effect on proliferation.
TIGIT can be knocked out in ex vivo expanded NK cells. Representative histograms from one donor demonstrate TIGIT KO (red) PM21-NK cells (A) or FC-NK cells (B) have minimal TIGIT expression compared with WT (black, gray fill) exNK cells. Isotype or unstained control are shown in gray. (C) Summary data from multiple donors (n=6–15 donors, each average of duplicates) show TIGIT KO efficiency in PM21-NK cells (circles) and FC-NK cells (triangles). (D) An example graph of NK-cell fold-change over time for one donor comparing expansion-matched, donor-matched non-electroporated PM21-NK cells (gray squares), PM21-NK cells electroporated with Cas9 only (WT, black circles), or TIGIT-specific Cas9 RNP (KO, red triangles). (E) There was no significant difference in fold-expansion of TIGIT KO (red) and WT exNK cells (black) on day 14 of culture for PM21-based expansion (circles) (n=12 donors) or for FC-based expansion (triangles) (n=3 donors). (F) Similarly, there was no difference in NK-cell number on day 14 between donor and postelectroporation seeding density-matched WT (black triangles) and TIGIT KO (red triangles) FC-NK cells (n=6 donors). Data are presented as histograms or scatter dot plots with mean and error bars representing SD. Statistical significance was determined by multiple paired t-tests. ExNK, ex vivo expanded NK; KO, knockout; NK, natural killer; TIGIT, T-cell Immunoreceptor with Ig and ITIM domains; WT, wild-type.
TIGIT knockout does not affect the basal phenotype of exNK cells based on donor-matched comparisons
To determine if TIGIT KO affects NK cell phenotype, cell surface expression of NK-cell activating and inhibitory receptors and death ligands was determined by flow cytometry on TIGIT KO PM21-NK cells and compared with WT PM21-NK cells (figure 2A). There were no significant changes in the percentage of NK cells expressing activating receptors NKp30, NKp46, CD16, NKG2D, NKp44, or DNAM-1 (p≥0.47) with TIGIT KO PM21-NK cells maintaining high expression of CD16, NKp30, NKG2D, and DNAM-1 (≥82% of NK cells) (n=3 donors, each average of duplicate). There was also no change in the percentage of NK cells expressing inhibitory receptors TIM3, NKG2A, LAG3, programmed cell death protein 1 (PD-1), CD96, and PVRIG (p≥0.09), or in death ligands FasL and Trail expression (p≥0.77) (figure 2B). Next, mass cytometry was employed to evaluate TIGIT KO FC-NK cell phenotype compared with WT FC-NK cells (n=2 donors; conditions including NK cells alone or NK cells following co-culture with SJGBM2, a tumor target known to highly express TIGIT ligands; online supplemental table 3 and figure 3). The CyTOF panel used in this study evaluated expression of 33 functional NK cell markers, including surface receptors as well as intracellular molecules, to characterize NK cell phenotype. The median metal intensity for each marker was determined (histograms provided in online supplemental figure 4). Across all markers tested, only TIGIT was significantly downregulated in TIGIT KO cells compared with WT controls (p<0.05) (figure 2C). Markers in the panel that had unexpected low detection (CD137, CD16, NKG2C, CD107a) were not included in the analysis. In summary, exNK-cell phenotype of activating and inhibitory receptors for WT and TIGIT KO was unchanged based on donor-matched analysis.
TIGIT KO does not change exNK-cell phenotype based on donor-matched comparisons. (A) Representative flow cytometry histograms are shown overlaying isotype control (black line), WT PM21-NK cells (black outline, gray fill), and TIGIT KO PM21-NK cells (red outline, red fill). (B) TIGIT KO (red triangles) did not significantly change the expression of any NK-cell surface receptors tested compared with WT PM21-NK cells (black circles) (n=3 donors, in duplicate). (C) CyTOF median metal intensities for NK cell markers comparing WT (circles) and TIGIT KO (triangles) FC-NK cells, with (open symbol) and without (closed symbol) exposure to the SJGBM-2 tumor cell line (n=2 donors, with four experimental conditions for each, WT and TIGIT KO NK cells). TIGIT expression was significantly decreased in TIGIT KO NK cells across all conditions (p<0.05), while no significant change in intensity was observed across other markers tested. ExNK, ex vivo expanded NK; KO, knockout; NK, natural killer; TIGIT, T-cell Immunoreceptor with Ig and ITIM domains; WT, wild-type. *P<0.05.
TIGIT knockout increased cytotoxicity of exNK cells against cancer cells
To examine the effect of TIGIT KO on NK-cell cytotoxicity, exNK cells were co-cultured with a panel of pediatric and lung cancer cell lines across multiple E:T ratios. Pediatric cancers were chosen as a population conventionally more resistant to immunotherapeutic approaches and lung cancer cell lines were chosen given the recent failed clinical trials of anti-TIGIT and high-expression of TIGIT ligands on both tumor types24 (online supplemental table 3 and figure 3). Calcein-release cytotoxicity assays were performed against pediatric cancer cell lines CHLA90, CHLA10, RH30, and SJBM2 and live cell-imaging cytotoxicity assays performed against lung cancer cell lines A549, H1299, H358, H1650, and H1975 (online supplemental figure 5) and dose-dependent cytotoxicity curves were determined. All cell lines were killed better by TIGIT KO exNK cells compared with WT exNK cells except CHLA90 (figure 3A–I, online supplemental figures 6 and 7), a cell line that has been indicated to have lower levels of DNAM-1 ligands.50 For pediatric cell lines, across multiple donors (n=6 donors, in duplicate) at a 1.25:1 E:T ratio, a significant increase in cytotoxicity upon TIGIT KO was observed for CHLA10 (p=0.02) and SJBM2 (p=0.017), while RH30 trended toward increased cytotoxicity (p=0.075) and no difference was seen for CHLA90 cells (figure 3J). For lung cancer cell lines, significantly increased cytotoxicity was observed for all cell lines, A549 (p=0.03), H1299 (p=0.03), H358 (p=0.04), and H1650 (p=0.035) except H1975, a cell line highly susceptible to NK cell killing, leaving little room for improvement on TIGIT KO for most donors (n=4 donors, in duplicate) (figure 3K).
TIGIT KO enhances exNK-cell cytotoxicity against pediatric and lung cancer cell lines. The cytotoxicity of WT exNK cells and TIGIT KO exNK cells were determined against a panel of pediatric and lung cancer cell lines. Multiple E:T ratios were used to determine dose-dependent cytotoxicity curves. While no improvement in the killing of CHLA90 (A) was observed, TIGIT KO (red triangles) enhanced FC-NK-cell cytotoxicity against CHLA10 (B), RH30 (C), SJBM2 (D), and PM21-NK-cell cytotoxicity against A549 (E), H1299 (F), H358 (G), H1650 (H), and H1975 (I) compared with their respective WT exNK cells (black circles) (A–I are representative plots from a single donor). (J) Across multiple donors at a 1.25:1 E:T ratio, a significant increase in cytotoxicity was observed for CHLA10 and SJBM2, while RH30 trended toward increased cytotoxicity and no difference was seen for CHLA90 cells (n=6, each average of duplicates). (K) Cytotoxicity was also significantly increased for A549, H1299, H358, and H1650 (n=4 donors, average of triplicates, 1:1 E:T, t=48 h). ExNK, ex vivo expanded NK; KO, knockout; NK, natural killer; TIGIT, T-cell Immunoreceptor with Ig and ITIM domains; WT, wild-type. E:T, effector/target; KO, knockout; NK, natural killer; TIGIT, T-cell Immunoreceptor with Ig and ITIM domains; WT, wild-type. *P<0.05.
The effect of TIGIT KO on the cytotoxicity of exNK cells was further examined by live cell-imaging-based assays using a spheroid model for A549, H1299, H358, H1650, H1975 lung cancer and SK-N-AS pediatric neuroblastoma cells (figure 4A). This approach allowed both kinetic and long-term (7 days) analysis of NK-cell cytotoxicity against cancer spheroids to better mimic the cancer microenvironment. Cytotoxicity over time of TIGIT KO exNK cells was increased compared with WT exNK cells in donor-matched pairs across all cancer spheroids tested (online supplemental figures 8 and 9), with significance reached for A549 (p=0.01) and H1650 (p=0.03) lung cancer spheroids and SK-N-AS neuroblastoma spheroids (p=0.047) analyzed at 72 h post-NK-cell addition (n=3–4 donors, in quadruplicate, with the number of NK cells added optimized for each spheroid type) (figure 4B). There was donor-to-donor variability and each cancer cell line had individual susceptibility to exNK cells; the average relative percent increase in TIGIT KO versus WT exNK-cell cytotoxicity against lung cancer spheroids was 47% for A549, 16% for H1299, 24% for H358, 22% for H1650, and 17% for H1975, and a 24% increase against neuroblastoma SK-N-AS spheroids. Furthermore, half-killing time significantly decreased for all cancer spheroids tested, A549 (p=0.001), H1299 (p=0.01), H358 (p=0.039), H1975 (p<0.001), and SK-N-AS (p=0.004), except H1650 spheroids (p=0.07) (n=3–4 donors, in quadruplicate with NK-cell numbers optimized for each spheroid type) (figure 4C). Dose-dependent cytotoxicity curves were used to determine EC50 values (online supplemental figures 8 and 9). TIGIT KO exNK cells also had significantly decreased EC50 compared with WT exNK cells for all cancer spheroids tested, A549 (p=0.004), H1299 (p=0.039), H358 (p=0.013), and SK-N-AS (p=0.009), except the highly susceptible H1650 and H1975 spheroids against which WT exNK cells had low EC50, leaving little room for further improvement (n=3–4 donors, in quadruplicate with the time point of EC50 determination optimized for each spheroid type) (figure 4D). In summary, TIGIT KO NK cells killed tumors faster requiring fewer cells and resulting in overall greater killing.
TIGIT KO enhances exNK cell cytotoxicity against tumor spheroids. TIGIT KO exNK-cell (PM21-NK cells) cytotoxicity against cancer spheroids was compared with WT exNK cells using kinetic live-cell imaging cytotoxicity assays. (A) Example images of untreated spheroids, with WT exNK cells, and with TIGIT KO exNK cells are shown for each cell line at 72 h post-NK-cell addition. (B) TIGIT KO exNK cells had increased cytotoxicity against all cell lines (n=3–4 donors, average of quadruplicates), with statistical significance reached for A549, H1650, and SK-N-AS. (C) TIGIT KO exNK cells had decreased t1/2 compared with WT exNK cells against all cell lines with statistical significance reached for all except H1650 (n=3–4 donors, average of quadruplicates). (D) EC50 was also decreased for TIGIT KO exNK cells compared with WT exNK cells against all cell lines, except those highly susceptible to NK cell killing, H1650 and H1975 (n=3–4 donors, average of quadruplicates). Each cancer spheroid had individual susceptibility to exNK cells, therefore the number of NK cells added, and time of analyses were used as follows: for cytotoxicity, 3333 exNK cells for A549, and H1650 while 10 000 exNK cells for H1975 and SK-N-AS, and 30 000 NK cells for H1299 and H358 spheroids. Dose-dependent cytotoxicity curves used to determine EC50 were determined at 72 h for H1299, H358, and H1975 and at 48 h for A549, H1650, and SK-N-AS. Data are presented as scatter dot plots with donor-pair lines. Statistical significance was determined by multiple paired t-tests. ExNK, ex vivo expanded NK; KO, knockout; NK, natural killer; TIGIT, T-cell Immunoreceptor with Ig and ITIM domains; WT, wild-type. *P<0.05, **p<0.01.
TIGIT knockout increased basal glycolysis of exNK cells
Metabolic processes, including increased levels of glycolysis and mTORC1 signaling, are critical components in NK-cell activation.51 Given the enhanced cytotoxicity of TIGIT KO exNK cells, the glycolytic rate of TIGIT KO exNK cells was compared with WT cells without restimulation and with stimulation by cytokines (IL-12/IL-15/IL-18) or PVR-expressing K562 cells for 24 h (online supplemental figure 10). TIGIT KO exNK cells had increased basal glycolytic rate compared with WT when they were restimulated with K562-PVR+ cells (p=0.02, n=6 donors, in triplicate) while there were no significant increases in unexposed or IL-12/IL-15/IL-18 stimulated TIGIT KO exNK cells compared with WT exNK cells (figure 5A), suggesting improved metabolic fitness of TIGIT KO PM21-NK cells during tumor exposure. Analysis of RNA-seq data from three donors (online supplemental figure 11) showed upregulation of mTORC1 signaling and glycolysis in TIGIT KO exNK cells compared with WT (figure 5B–C), further supporting TIGIT KO exNK cells are more metabolically fit.
TIGIT KO exNK cells have increased basal glycolysis rate compared with WT exNK cells when exposed to K562-PVR+ cells. (A) WT exNK cells (PM21-NK cells) and TIGIT KO exNK cells were stimulated with either cytokines (IL-12/IL-15/IL-18) or PVR expressing K562 cells (K562-PVR+ cells) for 24 h and compared with unstimulated exNK cells for basal glycolysis levels (n=6 donors, in duplicate). K562-PVR+ cell stimulation significantly increased basal glycolytic rate in TIGIT KO exNK cells (red triangles) compared with WT exNK cells (black circle). Data are presented as scatter plots with donor-pair lines. Statistical significance was determined by multiple paired t-tests. RNA-sequencing analysis with Gene Set Enrichment Analysis confirmed the upregulation of mTORC1 signaling and glycolysis hallmark gene sets in TIGIT KO exNK cells compared with WT exNK cells. Enrichment plots (B) and heatmaps with dysregulated genes (false discovery rate (FDR)<0.1) (C) are shown for these gene sets. ExNK, ex vivo expanded NK; KO, knockout; NK, natural killer; TIGIT, T-cell Immunoreceptor with Ig and ITIM domains; WT, wild-type. *P<0.05.
TIGIT KO exNK cells are resistant to Fc-active anti-TIGIT antibody-driven fratricide
Most therapeutic anti-TIGIT antibodies are Fc-active and can trigger killing of bound cells through CD16 engagement/ADCC or antibody-dependent cellular phagocytosis (ADCP). TIGIT is highly expressed on activated NK cells and therefore could mean depletion of these cells bound by Fc-active anti-TIGIT through fratricide. To evaluate this hypothesis, expanded WT and TIGIT KO exNK cells were incubated with Fc-inactive mouse IgG2b anti-TIGIT antibody or isotype control, or Fc-active anti-TIGIT (tiragolumab, SelleckChem) for 24 h and fratricide was analyzed by determining the percentage of viable NK cells remaining normalized to control samples of unexposed NK cells. Fc-active anti-TIGIT induced fratricide of WT exNK cells resulting in significantly decreased viability (61±16%) compared with isotype control (p<0.002), Fc-inactive exposed WT exNK cells (p=0.016), or TIGIT KO exNK cells exposed to Fc-active anti-TIGIT (p<0.0001) each of which had no significant fratricide relative to exNK cells alone (n=5, in triplicate) (figure 6A). Fc-active anti-TIGIT-exposed TIGIT KO exNK cells viability averaged 104±16%. These data indicated that activated NK cells are vulnerable to Fc-active anti-TIGIT-driven fratricide and TIGIT KO can ameliorate this effect.
TIGIT KO prevents Fc-active anti-TIGIT driven exNK-cell fratricide and prevents the decrease in exNK-cell cytotoxicity when combined with Fc-active anti-TIGIT. A total of 10 000 WT exNK cells (PM21-NK cells) and TIGIT KO exNK cells were co-cultured with Fc-inactive anti-TIGIT or isotype control or Fc-active anti-TIGIT antibodies (10 μg/mL) for 24 h with unexposed exNK cells as a control. (A) Relative viable NK cells in antibody-exposed exNK-cell cultures were determined relative to unexposed exNK cells for each of the WT (black) and TIGIT KO (red) groups for 5 NK-cell donors, in triplicate (average for each donor represented as circles, squares, diamond, hexagon, or triangles in each group). Fc-active anti-TIGIT induced significant fratricide against exNK cells but not against TIGIT KO exNK cells. WT exNK cells and TIGIT KO exNK cells were co-cultured with A549 spheroids in the presence of Fc-inactive or Fc-active anti-TIGIT and cytotoxicity was determined. (B) WT exNK cells showed significantly lower cytotoxicity against A549 spheroids in the presence of Fc-active (red) compared with Fc-inactive anti-TIGIT (blue), whereas TIGIT KO exNK cells alone had increased cytotoxicity compared with WT exNK cells (black), and there was no significant change in cytotoxicity in presence of Fc-inactive or Fc-active anti-TIGIT (n=3 donors, each in triplicate and averages represented as circles, squares, or triangles in each group). Data are presented as scatter dot plots with error bars representing SD. Statistical significance was determined by nested one-way analysis of variance with multiple comparisons and Fisher’s Least Significant Difference (LSD) test post-hoc analysis. *P<0.05, **p<0.01, ****p<0.0001.
To determine if Fc-active anti-TIGIT-driven fratricide of activated NK cells can affect tumor killing, WT or TIGIT KO exNK cells were co-cultured with A549 lung cancer spheroids with either Fc-active or Fc-inactive anti-TIGIT and killing was tracked over time by live-cell imaging (online supplemental figure 12). In WT exNK cells, Fc-inactive anti-TIGIT increased cytotoxicity against A549 spheroids from an average 18±2% to 33±5% (n=3 donors, in triplicate at t=121 h and 10 000 NK cells added p=0.008) but Fc-active anti-TIGIT failed to increase cytotoxicity, averaging at 22±6% (figure 6B). However, TIGIT KO exNK cells were more cytotoxic than WT exNK cells, 38±4% vs 18±2%, and there was no decrease in cytotoxicity in the presence of Fc-active anti-TIGIT at 40±6%. No increase in cytotoxicity was seen, or expected, in the presence of Fc-inactive anti-TIGIT, 45±10%, as TIGIT KO exNK cells have minimal TIGIT expression. Similarly, cytotoxicity against H358 spheroids in the presence of Fc-active anti-TIGIT relative to Fc-inactive was significantly decreased for WT but not TIGIT KO NK cells (p=0.004, n=5 donors, in triplicate) (online supplemental figure 13). Altogether, these observations suggest that Fc-active anti-TIGIT antibodies have the capability of depleting TIGIT+ NK cells and compromise NK-cell antitumor response.
Discussion
TIGIT has garnered great interest as a novel checkpoint with the potential to further improve the response rates and durability of therapies targeting the PD-1/PD-L1 axis. Over 20 anti-TIGIT antibody therapies are in clinical trials for dozens of indications and even more anti-TIGIT modalities are in preclinical development.1 8 9 While all of these prevent TIGIT from interacting with its receptors, they vary in the type of Fc region, using either an Fc-inactive format that does not interact with FcγRs or Fc-active that binds FcγRs on effector cells (ie, NK cells and macrophages) resulting in antibody-dependent cell killing. To date, there is considerable evidence from murine studies to support that an active Fc domain plays a pivotal role in the antitumor efficacy of anti-TIGIT antibodies.17 18 52–54 The proposed mechanism of action may vary depending on antibody type and models used but include ADCC/ADCP-driven reduction of intratumoral Treg populations,17–19 53 55 FcγR-mediated myeloid cell activation,17 52 co-engagement of specific FcγRs on APCs to enhance antigen-specific T-cell responses,54 or potentially through the removal of TIGIT from the surface of T cells via trogocytosis.17 While reduction of immunosuppressive cell populations would be beneficial, any cell highly expressing TIGIT has the potential for Fc-driven depletion by Fc-active anti-TIGIT antibodies, such as has been documented for TIGIThigh CD8+ T cells.56–58 This was considered beneficial to remove exhausted T cells, however other immune cells whose function is critical for an anti-tumor response, such as activated NK cells, have the same or higher levels of TIGIT expression as Tregs or CD8+ T cells.21–24
This study demonstrated that highly activated and cytotoxic NK cells express high levels of TIGIT and can commit fratricide in the presence of Fc-active TIGIT. This resulted in a decrease in the number of viable cells and lower killing in long-term spheroid assays. NK-cell depletion by Fc-active anti-TIGIT antibodies and its consequences have not been considered in studies of TIGIT blockade function and efficacy. Yet, NK cells are the first responders of the innate immune system, and their function is critical for tumor control. They directly target and kill cancerous cells and prime the antitumor response by recruiting and activating the adaptive immune system by secretion of cytokines and chemokines.29–31 ,32 NK cells have been shown to be critical for the efficacy of immunotherapeutic treatment, and low NK-cell numbers also correlate with a lack of response to treatment.20 Taken together, this means NK cells are a vital population of immune cells, that if negatively impacted by treatment regimes, could have detrimental effects on the efficacy of treatments. This holds true in the context of TIGIT blockade therapies with multiple studies reporting that NK cells are a critical immune population required for anti-TIGIT-based antitumor activities.5 17 18 Therefore, depletion of this population can compromise the antitumor effects of the treatment. Negative effects of NK-cell depletion via fratricide have been reported with Fc-active anti-CD38 therapies in the context of multiple myeloma.38 59 Some indications that Fc-active anti-TIGIT can deplete NK cells have been reported. Preillon et al concluded that Fc-active anti-TIGIT preferentially depletes intratumoral Tregs through direct ADCC/ADCP-driven killing due to higher TIGIT+ cell frequency and number of TIGIT receptors per cell than other immune populations.19 However, in the same report, they show NK cells from healthy donors and PBMCs of patient with cancer had similar levels of TIGIT expression; only tumor infiltrated, and likely exhausted, NK cells had somewhat lower levels of TIGIT expression than intratumoral Tregs. This would indicate a potential for Fc-active anti-TIGIT antibodies to deplete these activated peripheral NK cells, although this aspect was not discussed. Additionally, in a study that administered a construct consisting of the IgV domain of PVR fused to Fc in a mouse model of psoriasis, significant NK-cell depletion was noted within the skin 3 days after receiving treatment.60 The study presented here shows exNK cells are susceptible to depletion via fratricide in the presence of Fc-active anti-TIGIT. Furthermore, cytotoxic activity decreased with Fc-active but not Fc-inactive anti-TIGIT. There is also the potential for even further depletion in vivo through other Fc-driven mechanisms such as ADCP. Collectively, this means strategies that can preserve the crucial role of Fc-engagement without depletion of activated TIGIThigh NK cells would greatly benefit anti-TIGIT antibody-based treatments.
Since an Fc-active domain appears critical to the efficacy of the TIGIT therapies, co-administration of adoptive NK cells could provide a viable solution to potentially increase their effectiveness. Over the past decade, development of technologies enabling the use of donor-derived NK cells as an ‘off-the-shelf’ cell therapy has seen rapid growth. NK cells can be ex vivo expanded, viably cryopreserved,61–65 and based on numerous clinical studies are not associated with graft versus host disease,66–68 supporting the safe use of donor-derived material.34 69 ,70 Myriad clinical trials assessing the efficacy of NK cells in various cancer settings are underway71 72 and early results in the leukemia setting are promising.61 73 In this current study, clinically used NK-cell expansion technology with membrane-bound IL-21 stimulation was applied in conjunction with genome editing to develop a cellular product for potential use in combination with Fc-active anti-TIGIT antibodies. A high TIGIT KO efficiency (~90%) in exNK cells was achieved, consistent with other reports for KO of TIGIT in NK-cells lines74 75 and NK cells obtained from human peripheral blood.76 77 However, TIGIT KO exNK cells were resistant to fratricide and maintained cytotoxicity in the presence of Fc-active anti-TIGIT, and they had the added benefit of enhanced metabolic fitness and further increased cytotoxicity compared with WT exNK cells, which are themselves already highly functional and cytotoxic against multiple tumor types.34 49 50 62 78 Previous studies that had functional analyses of NK cells using NK-92 and CD19/CD276-CAR-NK-92 cells did not observe enhancement in cytotoxicity in TIGIT KO constructs, however, these were short-term (2 h) assays and the B-ALL target cell lines tested had below 30% expression for PVRL2 (CD112) and nearly no PVR (CD155) expression, the ligands for TIGIT. Additionally, these were CAR-NK cells derived from an NK leukemia cell line, which may not have the same TIGIT response elements.74 Similarly, in a study that used Cas9-RNP-loaded retroviral particles to simultaneously knockout TIGIT and express EGFR-CAR, no improvement in tumor control was observed in an MDA-MB-231 intraperitoneal xenograft mouse model with TIGIT KO. In both cases, CAR-driven signaling may be less sensitive to TIGIT-mediated suppression.77 In this study assessing cytotoxicity against highly resistant pediatric and lung tumor cell lines in long-term spheroid-based assays that better mimic the tumor microenvironment and native NK-cell activation pathways, improvement with TIGIT KO was more evident.
In summary, Fc-active anti-TIGIT therapeutics can result in the depletion of NK cells potentially contributing to their suboptimal clinical performance. Co-administration of fratricide-resistant TIGT KO exNK cells with current anti-TIGIT antibodies should improve direct tumor killing, enhance recruitment of adaptive immune cell populations, and clear immunosuppressive Tregs for improved overall efficacy. TIGIT KO NK cells also represent a better NK-cell product with improved tumor killing, greater metabolic activity, and lower susceptibility to PVR-driven exhaustion.
Supplemental material
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information. Raw RNA sequencing data are deposited in the NCBI SRA database under accession number PRJNA1049053
Ethics statements
Patient consent for publication
Ethics approval
This study was conducted in accordance with the University of Central Florida Institutional Biosafety Committee and all biological materials used were approved under BARA #19-27 (approved July 17, 2019–February 28, 2023) and Safety Protocol ID: SPROTO202200000044 (approved March 1, 2023). No animal or human studies were included in the manuscript. Furthermore, the use of discarded, anonymized healthy donor human buffy coats (used for isolation of NK cells in experimentation for this study), has been determined by the Nationwide Children’s Hospital and University of Central Florida Institutional Review Boards to not constitute human subjects research.
Acknowledgments
We thank the FL DOH James and Ester King Program (Grant No. 9JK04), the FL DOH Live Like Bella Foundation (Grant No. 22L04), Alex’s Lemonade Stand Foundation (Grant # 22-27258), the St. Baldrick’s Foundation Fellowship Award, the University of Central Florida Preeminent Postdoctoral Program and Kiadis Pharma, a Sanofi Company for the funds that supported this research. Graphical figures were created with BioRender.com.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Twitter @alicjajo
Contributors MFH and ARC: conceptualization, experimentation, data analysis, writing, reviewing, and editing of the manuscript. TJC-P: experimentation, data analysis, writing, reviewing, and editing of the manuscript. JLO, TAD, LDR-C, JEE, SK, BWA, CAC: experimentation, data analysis, review, and editing of the manuscript. MNK and BPT: conceptualization, data analysis, review, and editing of the manuscript. DL and AJC: conceptualization, funding acquisition, project administration, resources, supervision, writing, reviewing, and editing of the manuscript, guarantors.
Funding FL DOH James and Ester King Program (Grant No. 9JK04), the FL DOH Live Like Bella Foundation (Grant No. 22L04), the University of Central Florida, Pre-eminent Postdoctoral Program, Alex’s Lemonade Stand Foundation (Grant # 22-27258), the St. Baldrick’s Foundation Fellowship Award, and Kiadis Pharma, a Sanofi Company.
Competing interests AJC: licensed IP to, consultancy and research support from Kiadis Pharma, a Sanofi company; JLO: licensed IP to, consultancy with Kiadis Pharma, a Sanofi company. MFH, TJC-P: licensed IP to Kiadis Pharma, a Sanofi company; DL: scientific advisory board of and consults for Avidicure; consultancy, licensing, and royalty fees from Kiadis Pharma, a Sanofi Company; IP interests related to NK-cell therapy.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.